• Boating Holidays
  • Asia & Pacific
  • Australia & New Zealand
  • Central America
  • Indonesia & Indian Ocean
  • Middle East
  • Scandinavia
  • South America
  • Business Travel
  • Culinary Travel
  • Wellness Travel
  • Golf Travel
  • Travel Style
  • Travel Gear
  • Travel Fashion
  • Real Estate
  • Destination Weddings
  • Fishing Holidays
  • Tricks & Hacks
  • Travel News
  • Read the All Issues
  • Working with DRIFT Travel
  • Submit Your Photos to DRIFT Travel
  • DRIFT Travel Magazine Media Kit

travel survey questions


How to Conduct a Travel Survey: A Guide for Agencies and Tour Operators

travel survey questions

People who work in travel industry come across many issues every day. Essentially, they have heard it all from “The receptionist was not nice” to “The bed was too small”. The feedback they receive is most often aimed at bad things, no matter how small or rare they are. With traveler surveys, agencies and operators can actually get comprehensive experiences from travelers that range way beyond a few unsatisfactory ones. Surveys offer an insight into traveler psychology and show you how best to meet the expectations of people. How and why you and your agency should start using surveys?

Why conduct a travel survey?

Travel surveys have many benefits. Firstly, they indicate certain trends in the travel and hospitality industry and show people’s needs and wants. Thanks to surveys, agencies and operators insight into changes they need to implement and knowledge they are missing. Brans can use the knowledge to plan their future growth.

Surveys also show your clients that you care about their experiences and opinions. With surveys, you can find out what makes a destination special, what the destination is missing, how your services can be improved and how you can reach more customers.

Before writing any questions, it’s important to figure out what information you want to learn from your survey. Do you want to learn more about your target audience? Do you want to learn about a certain destination and how it reflects on the travelers? Or maybe you want to learn how your clients are satisfied with your tour operators , offers, etc? Make sure to have your goal clear and you’ll compile your survey much easier.

Handle demographics

It’s crucial to know who your base audience is, but this part can be quite boring and awkward for your clients. Luckily, if you choose a professional tool for surveys , you can allow your users to run through this part and check boxes or fill gaps quickly and efficiently. Make sure to include all important information like age, location and certain characteristics. It’s crucial to be respectful, especially when asking about gender, race, sexuality or any other sensitive question.

Organize your questions

When creating questions for your survey, keep your clients in mind and make sure to include questions they are bound to understand. All the words should be present in everyday use and questions should be straightforward. Your survey should be organized logically with a nice flow of questions. After all, your clients are doing you a favor by filling out these surveys, so the least you could do is provide them with a quality and logical survey.

Find out about their travel habits

One of the main reasons why agencies choose to give out surveys is to gain the knowledge necessary for the creation of new packages, services and offers. Therefore, make sure to ask about your clients’ travel habits: Are they DIY travelers or do they prefer tours with guides? Do they travel alone or with someone? Do they prefer short or long trips? Do they plan their trips well in advance?

Ask about satisfaction

Satisfaction questions make clients feel appreciated and give you valuable feedback. So make sure to include questions that concern their satisfaction with first impressions, staff, hygiene, communication, service value, expectations, overall satisfaction, etc. If you want to improve your services, this is the most valuable information for you.

How did they hear about you?

Here’s another important question to ask: How did they learn about your agency ? For instance, let’s say you’ve been pouring money into TV commercials, yet people on the survey said they didn’t see your commercial. This is your cue to change marketing strategies.

Ask about challenges

Travelers face a lot of challenges while on the road and it’s your job to try and make their lives easier. Ask whether your users need more help with planning activities and discovering new places. Or maybe they struggle with finding authentic experiences or catching rides while traveling. Identify their challenges and try to find solutions.

And finally, make sure to end your survey with a question such as: “Do you have anything else to share or add?” If your customers have anything they want to get off their chest, this will be a nice way to do so. All in all, you will learn much from your survey, and if you know how to use that information, it can be amazing for your agency.



How to Ensure Cyber Security While Traveling


Navigating the Aftermath: What to Do After a Car Accident While on Vacation


Demystifying Personal Safety: Your Key to Fearless Traveling

Privacy overview.


BLUETTI’s First Ecosystem SwapSolar Keeps Your Gadgets Powered and Drinks Chilled...

travel survey questions


How to build business travel surveys travelers will actually take

Asking business travelers what they think is an effective way to get the insights you need to build a strong corporate travel program and improve policies already in place.

Creating winning business travel survey questions is one of the secrets to optimizing your corporate travel program. Rolling out travel surveys will help you identify your program’s shortcomings so that you can have a stronger, more agile travel program for all your business travelers.

Building surveys with the goal of taking the key learnings and improving the overall experience for everyone involved is crucial since happy travelers mean productive travelers. So, let’s take a look at the  10 key steps you can use to implement winning business travel surveys.

Step 1: Set the goals for your corporate travel survey questions

When it comes time to designing a survey,  make sure you have a goal in mind.  For example: as you consider resuming or expanding your travel program, your goal might be to understand your traveler’s new working approach while traveling for business. Keep the goal in mind when planning your questions.

No matter whether your goal is to  optimize your travel policies , to help your travelers make the most of their business trips, or to simply get a feel for the health of your corporate travel program,  having a goal in mind will make your surveys more effective.  

Step 2: Define stakeholder and survey review process

Involve relevant stakeholders in the survey process, from helping to build the questions, to reviewing the data insights extracted from it. You might Implicate colleagues in technology, data analysis, travel risk management, meetings, or finance who can help you with different aspects of the process. In the case of data analysts, they can help you better understand and utilize the data that is derived from the business travel survey.

Step 3: K.I.S.S. Keep it simple and specific

Keep the travel survey length and questions short and straightforward. No one will want to answer surveys that are too long, so  the fewer the corporate travel survey questions, the better.  Our insider experts suggest creating surveys that take no longer than eight minutes to complete. The questions should offer a range or scale of response options as open-ended questions take more time to answer and are harder to quantify. 

Step 4: Select a user-friendly survey tool 

We’ll leave it up to you to decide what survey platform to go with, since there are tons of suitable options out there. Just keep in mind to check reviews and make sure the tool is intuitive. It should meet your needs for the types of questions you wish to ask and offer built-in analytics capabilities. You should  spend time creating great questions instead of trying to get your platform to work. 

Step 5: Plan survey frequency and timing

Survey your travelers as often as you need to, but don’t overwhelm them.   Your survey frequency should depend on your goals. If the travelers are engaged and highly motivated (e.g., by incentives) they may have more incentive to participate monthly. Monitor responses and adjust survey plans accordingly.

Distribute the business travel survey questions when it makes the most sense

For example you may consider launching a business travel survey prior to travel policy rollouts or changes; following the rollouts to gauge effectiveness; or before major events or holidays where the targeted audience is likely to miss the survey communication.

Decide how long your survey should remain open

The best practice for the GetGoing internal research team is  generally a two-week survey window.  Send reminders three days to one week after the survey opens. If the response is underwhelming, it’s okay to send a gentle reminder email. Don’t forget: a catchy subject line with a call to action is always a good idea.

travel reports

Step 6: Test it

Practice makes perfect , and this rule most definitely applies to business travel surveys. Make sure you pass along the survey “practice test” to your colleagues, to look out for content or formatting mistakes, unclear questions or responses, and overall practicality. This will help work out any kinks before you actually launch the questionnaire. 

Step 7: Report back

Grow confidence in your business travel program, by sharing survey results and demonstrating how they’ll be put into practice across your organization. Organize this report by writing a short summary of your findings, including visuals to support key data and help put the survey results in context.

Step 8: Think long term 

To help get an optimal number of responses each time you send a survey, focus on creating a diverse audience in terms of geolocation, language, gender, age, and tenure. The better your survey participants represent the actual company of business travelers, the more relevant your insights.

Keep in mind that this global representation won’t happen after a single survey. It will improve over time, especially if you can demonstrate how your survey insights directly influence travel program policies.

Step 9: Make it worthwhile

Encourage participants to take the survey by offering some form of incentives ,  such as recognition badges, gifts, discounts, or entering them into a sweepstakes to win a prize. Make sure your invitations and incentives are targeted to the right traveler profiles. Otherwise, you might attract less helpful attention from go-getters searching for freebies. 

Step 10: Review, refine and repeat

Once you have reached a shared understanding of the results, it’s important to review, refine and repeat the process. Surveys are an iterative process, meaning that it involves building, improving and refining. There is no such thing as a perfect survey or marketing campaign, so be sure you learn and refine as you go. After all, practice makes perfect. 

What did you learn about business travel survey questions?

Now that we’ve gone over the 10 steps to create a winning survey with the right corporate travel survey questions, you have all that’s necessary to nail your business travel program. Surveys give you a hand in obtaining valuable information of a large sample of individuals in a limited time and with limited resources required. 

Looking for more ways to optimize your travel program?  Contact GetGoing  and we’ll help you get traveling. 

Go share the news:

Sign up now to receive exciting news & updates.

travel survey questions

Table of Contents

What is a travel questionnaire, the value of collecting information, 14+ travel questionnaire templates in google docs | word | pages | numbers | pdf, 1. pre travel questionnaire template, 2. sample travel questionnaire template, 3. medical travel questionnaire template, 4. pre travel questionnaire template, 5. pre-travel questionnaire in pdf, 6. international travel questionnaire, 7. international travel medical questionnaire, 8. pre-travel questionnaire form, 9. simple travel questionnaire, 10. domestic travel survey questionnaire template, 11. foreign travel questionnaire template, 12. basic travel questionnaire template, 13. residency and travel questionnaire, 14. printable travel questionnaire in pdf, 15. professional travel questionnaire, how to create a travel questionnaire, questionnaire templates.

With over 7 trillion U.S. dollars global economic contribution last 2016, it is undeniable that one of the highest-earning industries all over the world is travel and tourism. Aside from that, the international tourist arrivals were forecasted to exceed 1.8 billion travelers by 2030 from 1.19 billion in 2015. However, these figures wouldn’t be realized without the efforts of business people. And being in this industry entails writing and preparing documents such as a travel questionnaire to keep it running. Learn more about it as we walk you through the basics of the document.

travel survey questions

  • Medical Travel Questionnaire: This type of questionnaire collects information regarding the medical condition or medical history of an individual. Usually, the details provided in the form is essential for insurance companies in determining the coverage and cost of the travel insurance policy.
  • Pre-Travel Questionnaire: A pre-travel questionnaire includes questions regarding the travel plans or itinerary of the travelers. They may also include their medical needs to reduce the risk of health issues while traveling.
  • Post-Travel Questionnaire: After the tour, business owners can distribute business questionnaire to tourists for them to evaluate the services rendered by the company. The feedback about the trip provided by the customer can be the basis for businesses for the improvement of their services.

travel questionnaire template

Step 1: Specify the Purpose

Step 2: start with the personal information, step 3: list down questions, step 4: keep it short and simple, step 5: distribute, more in questionnaire templates, travel & tour bi-fold brochure template, travel brochure template, travel sale tri-fold brochure template, travel agency bi-fold brochure template, travel & tour tri-fold brochure template, beach vacation rental bi-fold brochure template, condo apartment vacation rental tri-fold brochure template, vacation rental advertising tri-fold brochure template, vacation rental advertising bi-fold brochure template, resort vacation rental tri-fold brochure template.

  • 20+ Work Questionnaire Templates in PDF | DOC
  • 11+ Security Assessment Questionnaire Templates in MS Word | MS Excel | PDF
  • 10+ Pre-Employment Health Questionnaire Templates in PDF | MS Word
  • 6+ Email Marketing Questionnaire Templates in PDF | MS Word
  • 19+ Child Questionnaire Templates in PDF | MS Word
  • 10+ Food Survey Questionnaire Templates in PDF | MS Word | Pages | Google Docs
  • 9+ Business Client Questionnaire Templates in PDF | MS Word
  • 8+ Driver Health Questionnaire Templates in PDF | MS Word
  • 15+ Pre-Employment Questionnaire Templates in PDF | MS Word
  • 11+ Project Evaluation Questionnaire Templates in PDF | MS Word
  • 5+ Client Satisfaction Questionnaire Templates in PDF | MS word
  • 7+ Student Experience Questionnaire Templates in PDF | MS Word
  • 10+ Financial Planning Questionnaire Templates in MS Word | PDF
  • 9+ Financial Management Questionnaire Templates in PDF | MS word
  • 5+ Strategic Planning Questionnaire Templates in PDF | MS word

Travel Agency Survey

The job of a travel agency is to reduce the stress of planning a vacation. Find out how well your agency does in satisfying your customer’s expectations. Gather feedback on how they got to know about you and their rating on the service provided by you. Gain insights on whether they are likely to recommend you to other travelers.

Travel agency survey questionnaire template

Why should travel agencies conduct surveys?

These surveys provide a window into the traveler’s experience, allowing agencies to identify areas of excellence and pinpoint potential shortcomings. By understanding what travellers appreciate and where improvements are needed, travel agencies can tailor their services to meet traveler expectations. Ultimately, this feedback-driven approach not only enhances customer satisfaction but also strengthens the agency’s competitiveness in the travel industry.

Travel agency servey

How often do you book your travel with us?

How likely are you to book us again?

Stay ahead of customer expectations

Travel trends and customer expectations are constantly evolving. Surveys allow travel agencies to stay attuned to these changes by capturing feedback on emerging preferences, desired amenities, and innovative experiences. By understanding these evolving expectations, travel agencies can adapt their offerings and stay ahead of the competition.

Engage with customers during the pre- and post-travel phases

Surveys offer an opportunity to engage with customers not only during their trip but also before and after their travel experience. Pre-travel surveys can help agencies gather preferences, tailor itineraries, and ensure a seamless trip planning process. Post-travel surveys capture feedback on the overall experience, allowing agencies to address any issues, offer post-trip assistance, and maintain an ongoing relationship with customers.

Identify and address the customer pain points

Surveys provide an avenue for customers to express their concerns, highlight areas of dissatisfaction, or suggest improvements. By actively seeking feedback travel agencies can identify pain points in their services and take corrective measures. Addressing these issues promptly and effectively helps in improving overall service quality and customer satisfaction.

Build strategies to enhance Customer Retention

Surveys provide a direct avenue for customers to express their concerns, highlight areas of dissatisfaction, or suggest improvements. By actively seeking feedback, travel agencies can identify pain points in their services and take swift corrective measures. Satisfied and loyal customers not only generate repeat business but also become brand advocates, attracting new customers through positive recommendations.

Leveraging customer testimonials for business growth

Surveys serve as an effective platform to gather testimonials and reviews from satisfied customers. Positive feedback collected through surveys can be utilized as testimonials on the website, social media platforms, and marketing materials. These testimonials and reviews not only enhance the agency’s credibility but also build trust with potential customers.

Customer-centric pricing: Maximizing revenue through insights

Surveys provide insights into customers willingness to pay for various travel services and experiences. By incorporating pricing-related questions in surveys, travel agencies can gauge customer perceptions of value and adjust their pricing strategies accordingly. This data-driven approach helps agencies optimize their revenue streams, ensuring that their pricing aligns with customer expectations. By offering competitive and customer-centric pricing, travel agencies can maximize revenue while maintaining customer satisfaction.

Revolutionize your data collection with Zoho survey!

Travel agency survey

Rating questions

Multilingual surveys, offline surveys, sms distributions, discover insights, craft experiences..

New : flexible, templated dashboards for more control.  Meet Dashboards

Learn / Blog / Article

Back to blog

Survey questions 101: 70+ survey question examples, types of surveys, and FAQs

How well do you understand your prospects and customers—who they are, what keeps them awake at night, and what brought them to your business in search of a solution? Asking the right survey questions at the right point in their customer journey is the most effective way to put yourself in your customers’ shoes.

Last updated

Reading time.

travel survey questions

This comprehensive intro to survey questions contains over 70 examples of effective questions, an overview of different types of survey questions, and advice on how to word them for maximum effect. Plus, we’ll toss in our pre-built survey templates, expert survey insights, and tips to make the most of AI for Surveys in Hotjar. ✨

Surveying your users is the simplest way to understand their pain points, needs, and motivations. But first, you need to know how to set up surveys that give you the answers you—and your business—truly need. Impactful surveys start here:

❓ The main types of survey questions : most survey questions are classified as open-ended, closed-ended, nominal, Likert scale, rating scale, and yes/no. The best surveys often use a combination of questions.

💡 70+ good survey question examples : our top 70+ survey questions, categorized across ecommerce, SaaS, and publishing, will help you find answers to your business’s most burning questions

✅ What makes a survey question ‘good’ : a good survey question is anything that helps you get clear insights and business-critical information about your customers 

❌ The dos and don’ts of writing good survey questions : remember to be concise and polite, use the foot-in-door principle, alternate questions, and test your surveys. But don’t ask leading or loaded questions, overwhelm respondents with too many questions, or neglect other tools that can get you the answers you need.

👍 How to run your surveys the right way : use a versatile survey tool like Hotjar Surveys that allows you to create on-site surveys at specific points in the customer journey or send surveys via a link

🛠️ 10 use cases for good survey questions : use your survey insights to create user personas, understand pain points, measure product-market fit, get valuable testimonials, measure customer satisfaction, and more

Use Hotjar to build your survey and get the customer insight you need to grow your business.

6 main types of survey questions

Let’s dive into our list of survey question examples, starting with a breakdown of the six main categories your questions will fall into:

Open-ended questions

Closed-ended questions

Nominal questions

Likert scale questions

Rating scale questions

'Yes' or 'no' questions

1. Open-ended survey questions

Open-ended questions  give your respondents the freedom to  answer in their own words , instead of limiting their response to a set of pre-selected choices (such as multiple-choice answers, yes/no answers, 0–10 ratings, etc.). 

Examples of open-ended questions:

What other products would you like to see us offer?

If you could change just one thing about our product, what would it be?

When to use open-ended questions in a survey

The majority of example questions included in this post are open-ended, and there are some good reasons for that:

Open-ended questions help you learn about customer needs you didn’t know existed , and they shine a light on areas for improvement that you may not have considered before. If you limit your respondents’ answers, you risk cutting yourself off from key insights.

Open-ended questions are very useful when you first begin surveying your customers and collecting their feedback. If you don't yet have a good amount of insight, answers to open-ended questions will go a long way toward educating you about who your customers are and what they're looking for.

There are, however, a few downsides to open-ended questions:

First, people tend to be less likely to respond to open-ended questions in general because they take comparatively more effort to answer than, say, a yes/no one

Second, but connected: if you ask consecutive open-ended questions during your survey, people will get tired of answering them, and their answers might become less helpful the more you ask

Finally, the data you receive from open-ended questions will take longer to analyze compared to easy 1-5 or yes/no answers—but don’t let that stop you. There are plenty of shortcuts that make it easier than it looks (we explain it all in our post about how to analyze open-ended questions , which includes a free analysis template.)

💡 Pro tip: if you’re using Hotjar Surveys, let our AI for Surveys feature analyze your open-ended survey responses for you. Hotjar AI reviews all your survey responses and provides an automated summary report of key findings, including supporting quotes and actionable recommendations for next steps.

2. Closed-ended survey questions

Closed-end questions limit a user’s response options to a set of pre-selected choices. This broad category of questions includes

‘Yes’ or ‘no’ questions

When to use closed-ended questions

Closed-ended questions work brilliantly in two scenarios:

To open a survey, because they require little time and effort and are therefore easy for people to answer. This is called the foot-in-the-door principle: once someone commits to answering the first question, they may be more likely to answer the open-ended questions that follow.

When you need to create graphs and trends based on people’s answers. Responses to closed-ended questions are easy to measure and use as benchmarks. Rating scale questions, in particular (e.g. where people rate customer service or on a scale of 1-10), allow you to gather customer sentiment and compare your progress over time.

3. Nominal questions

A nominal question is a type of survey question that presents people with multiple answer choices; the answers are  non-numerical in nature and don't overlap  (unless you include an ‘all of the above’ option).

Example of nominal question:

What are you using [product name] for?

Personal use

Both business and personal use

When to use nominal questions

Nominal questions work well when there is a limited number of categories for a given question (see the example above). They’re easy to create graphs and trends from, but the downside is that you may not be offering enough categories for people to reply.

For example, if you ask people what type of browser they’re using and only give them three options to choose from, you may inadvertently alienate everybody who uses a fourth type and now can’t tell you about it.

That said, you can add an open-ended component to a nominal question with an expandable ’other’ category, where respondents can write in an answer that isn’t on the list. This way, you essentially ask an open-ended question that doesn’t limit them to the options you’ve picked.

4. Likert scale questions

The Likert scale is typically a 5- or 7-point scale that evaluates a respondent’s level of agreement with a statement or the intensity of their reaction toward something.

The scale develops symmetrically: the median number (e.g. a 3 on a 5-point scale) indicates a point of neutrality, the lowest number (always 1) indicates an extreme view, and the highest number (e.g. a 5 on a 5-point scale) indicates the opposite extreme view.

Example of a Likert scale question:

#The British Museum uses a Likert scale Hotjar survey to gauge visitors’ reactions to their website optimizations

When to use Likert scale questions

Likert-type questions are also known as ordinal questions because the answers are presented in a specific order. Like other multiple-choice questions, Likert scale questions come in handy when you already have some sense of what your customers are thinking. For example, if your open-ended questions uncover a complaint about a recent change to your ordering process, you could use a Likert scale question to determine how the average user felt about the change.

A series of Likert scale questions can also be turned into a matrix question. Since they have identical response options, they are easily combined into a single matrix and break down the pattern of single questions for users.

5. Rating scale questions

Rating scale questions are questions where the answers map onto a numeric scale (such as rating customer support on a scale of 1-5, or likelihood to recommend a product from 0-10).

Examples of rating questions:

How likely are you to recommend us to a friend or colleague on a scale of 0-10?

How would you rate our customer service on a scale of 1-5?

When to use rating questions

Whenever you want to assign a numerical value to your survey or visualize and compare trends , a rating question is the way to go.

A typical rating question is used to determine Net Promoter Score® (NPS®) : the question asks customers to rate their likelihood of recommending products or services to their friends or colleagues, and allows you to look at the results historically and see if you're improving or getting worse. Rating questions are also used for customer satisfaction (CSAT) surveys and product reviews.

When you use a rating question in a survey, be sure to explain what the scale means (e.g. 1 for ‘Poor’, 5 for ‘Amazing’). And consider adding a follow-up open-ended question to understand why the user left that score.

Example of a rating question (NPS):

#Hotjar's Net Promoter Score® (NPS®) survey template lets you add open-ended follow-up questions so you can understand the reasons behind users' ratings

6. ‘Yes’ or ‘no’ questions

These dichotomous questions are super straightforward, requiring a simple ‘yes’ or ‘no’ reply.

Examples of yes/no questions:

Was this article useful? (Yes/No)

Did you find what you were looking for today? (Yes/No)

When to use ‘yes’ or ‘no’ questions

‘Yes’ and ‘no’ questions are a good way to quickly segment your respondents . For example, say you’re trying to understand what obstacles or objections prevent people from trying your product. You can place a survey on your pricing page asking people if something is stopping them, and follow up with the segment who replied ‘yes’ by asking them to elaborate further.

These questions are also effective for getting your foot in the door: a ‘yes’ or ‘no’ question requires very little effort to answer. Once a user commits to answering the first question, they tend to become more willing to answer the questions that follow, or even leave you their contact information.

#Web design agency NerdCow used Hotjar Surveys to add a yes/no survey on The Transport Library’s website, and followed it up with an open-ended question for more insights

70+ more survey question examples

Below is a list of good survey questions, categorized across ecommerce, software as a service (SaaS), and publishing. You don't have to use them word-for-word, but hopefully, this list will spark some extra-good ideas for the surveys you’ll run immediately after reading this article. (Plus, you can create all of them with Hotjar Surveys—stick with us a little longer to find out how. 😉)

📊 9 basic demographic survey questions

Ask these questions when you want context about your respondents and target audience, so you can segment them later. Consider including demographic information questions in your survey when conducting user or market research as well. 

But don’t ask demographic questions just for the sake of it—if you're not going to use some of the data points from these sometimes sensitive questions (e.g. if gender is irrelevant to the result of your survey), move on to the ones that are truly useful for you, business-wise. 

Take a look at the selection of examples below, and keep in mind that you can convert most of them to multiple choice questions:

What is your name?

What is your age?

What is your gender?

What company do you work for?

What vertical/industry best describes your company?

What best describes your role?

In which department do you work?

What is the total number of employees in your company (including all locations where your employer operates)?

What is your company's annual revenue?

🚀 Get started: gather more info about your users with our product-market fit survey template .

👥 20+ effective customer questions

These questions are particularly recommended for ecommerce companies:

Before purchase

What information is missing or would make your decision to buy easier?

What is your biggest fear or concern about purchasing this item?

Were you able to complete the purpose of your visit today?

If you did not make a purchase today, what stopped you?

After purchase

Was there anything about this checkout process we could improve?

What was your biggest fear or concern about purchasing from us?

What persuaded you to complete the purchase of the item(s) in your cart today?

If you could no longer use [product name], what’s the one thing you would miss the most?

What’s the one thing that nearly stopped you from buying from us?

👉 Check out our 7-step guide to setting up an ecommerce post-purchase survey .

Other useful customer questions

Do you have any questions before you complete your purchase?

What other information would you like to see on this page?

What were the three main things that persuaded you to create an account today?

What nearly stopped you from creating an account today?

Which other options did you consider before choosing [product name]?

What would persuade you to use us more often?

What was your biggest challenge, frustration, or problem in finding the right [product type] online?

Please list the top three things that persuaded you to use us rather than a competitor.

Were you able to find the information you were looking for?

How satisfied are you with our support?

How would you rate our service/support on a scale of 0-10? (0 = terrible, 10 = stellar)

How likely are you to recommend us to a friend or colleague? ( NPS question )

Is there anything preventing you from purchasing at this point?

🚀 Get started: learn how satisfied customers are with our expert-built customer satisfaction and NPS survey templates .

Set up a survey in seconds

Use Hotjar's free survey templates to build virtually any type of survey, and start gathering valuable insights in moments.

🛍 30+ product survey questions

These questions are particularly recommended for SaaS companies:

Questions for new or trial users

What nearly stopped you from signing up today?

How likely are you to recommend us to a friend or colleague on a scale of 0-10? (NPS question)

Is our pricing clear? If not, what would you change?

Questions for paying customers

What convinced you to pay for this service?

What’s the one thing we are missing in [product type]?

What's one feature we can add that would make our product indispensable for you?

If you could no longer use [name of product], what’s the one thing you would miss the most?

🚀 Get started: find out what your buyers really think with our pricing plan feedback survey template .

Questions for former/churned customers

What is the main reason you're canceling your account? Please be blunt and direct.

If you could have changed one thing in [product name], what would it have been?

If you had a magic wand and could change anything in [product name], what would it be?

🚀 Get started: find out why customers churn with our free-to-use churn analysis survey template .

Other useful product questions

What were the three main things that persuaded you to sign up today?

Do you have any questions before starting a free trial?

What persuaded you to start a trial?

Was this help section useful?

Was this article useful?

How would you rate our service/support on a scale of 1-10? (0 = terrible, 10 = stellar)

Is there anything preventing you from upgrading at this point?

Is there anything on this page that doesn't work the way you expected it to?

What could we change to make you want to continue using us?

If you did not upgrade today, what stopped you?

What's the next thing you think we should build?

How would you feel if we discontinued this feature?

What's the next feature or functionality we should build?

🚀 Get started: gather feedback on your product with our free-to-use product feedback survey template .

🖋 20+ effective questions for publishers and bloggers

Questions to help improve content.

If you could change just one thing in [publication name], what would it be?

What other content would you like to see us offer?

How would you rate this article on a scale of 1–10?

If you could change anything on this page, what would you have us do?

If you did not subscribe to [publication name] today, what was it that stopped you?

🚀 Get started: find ways to improve your website copy and messaging with our content feedback survey template .

New subscriptions

What convinced you to subscribe to [publication] today?

What almost stopped you from subscribing?

What were the three main things that persuaded you to join our list today?


What is the main reason you're unsubscribing? Please be specific.

Other useful content-related questions

What’s the one thing we are missing in [publication name]?

What would persuade you to visit us more often?

How likely are you to recommend us to someone with similar interests? (NPS question)

What’s missing on this page?

What topics would you like to see us write about next?

How useful was this article?

What could we do to make this page more useful?

Is there anything on this site that doesn't work the way you expected it to?

What's one thing we can add that would make [publication name] indispensable for you?

If you could no longer read [publication name], what’s the one thing you would miss the most?

💡 Pro tip: do you have a general survey goal in mind, but are struggling to pin down the right questions to ask? Give Hotjar’s AI for Surveys a go and watch as it generates a survey for you in seconds with questions tailored to the exact purpose of the survey you want to run.

What makes a good survey question?

We’ve run through more than 70 of our favorite survey questions—but what is it that makes a good survey question, well, good ? An effective question is anything that helps you get clear insights and business-critical information about your customers , including

Who your target market is

How you should price your products

What’s stopping people from buying from you

Why visitors leave your website

With this information, you can tailor your website, products, landing pages, and messaging to improve the user experience and, ultimately, maximize conversions .

How to write good survey questions: the DOs and DON’Ts

To help you understand the basics and avoid some rookie mistakes, we asked a few experts to give us their thoughts on what makes a good and effective survey question.

Survey question DOs

✅ do focus your questions on the customer.

It may be tempting to focus on your company or products, but it’s usually more effective to put the focus back on the customer. Get to know their needs, drivers, pain points, and barriers to purchase by asking about their experience. That’s what you’re after: you want to know what it’s like inside their heads and how they feel when they use your website and products.

Rather than asking, “Why did you buy our product?” ask, “What was happening in your life that led you to search for this solution?” Instead of asking, “What's the one feature you love about [product],” ask, “If our company were to close tomorrow, what would be the one thing you’d miss the most?” These types of surveys have helped me double and triple my clients.

✅ DO be polite and concise (without skimping on micro-copy)

Put time into your micro-copy—those tiny bits of written content that go into surveys. Explain why you’re asking the questions, and when people reach the end of the survey, remember to thank them for their time. After all, they’re giving you free labor!

✅ DO consider the foot-in-the-door principle

One way to increase your response rate is to ask an easy question upfront, such as a ‘yes’ or ‘no’ question, because once people commit to taking a survey—even just the first question—they’re more likely to finish it.

✅ DO consider asking your questions from the first-person perspective

Disclaimer: we don’t do this here at Hotjar. You’ll notice all our sample questions are listed in second-person (i.e. ‘you’ format), but it’s worth testing to determine which approach gives you better answers. Some experts prefer the first-person approach (i.e. ‘I’ format) because they believe it encourages users to talk about themselves—but only you can decide which approach works best for your business.

I strongly recommend that the questions be worded in the first person. This helps create a more visceral reaction from people and encourages them to tell stories from their actual experiences, rather than making up hypothetical scenarios. For example, here’s a similar question, asked two ways: “What do you think is the hardest thing about creating a UX portfolio?” versus “My biggest problem with creating my UX portfolio is…” 

The second version helps get people thinking about their experiences. The best survey responses come from respondents who provide personal accounts of past events that give us specific and real insight into their lives.

✅ DO alternate your questions often

Shake up the questions you ask on a regular basis. Asking a wide variety of questions will help you and your team get a complete view of what your customers are thinking.

✅ DO test your surveys before sending them out

A few years ago, Hotjar created a survey we sent to 2,000 CX professionals via email. Before officially sending it out, we wanted to make sure the questions really worked. 

We decided to test them out on internal staff and external people by sending out three rounds of test surveys to 100 respondents each time. Their feedback helped us perfect the questions and clear up any confusing language.

Survey question DON’Ts

❌ don’t ask closed-ended questions if you’ve never done research before.

If you’ve just begun asking questions, make them open-ended questions since you have no idea what your customers think about you at this stage. When you limit their answers, you just reinforce your own assumptions.

There are two exceptions to this rule:

Using a closed-ended question to get your foot in the door at the beginning of a survey

Using rating scale questions to gather customer sentiment (like an NPS survey)

❌ DON’T ask a lot of questions if you’re just getting started

Having to answer too many questions can overwhelm your users. Stick with the most important points and discard the rest.

Try starting off with a single question to see how your audience responds, then move on to two questions once you feel like you know what you’re doing.

How many questions should you ask? There’s really no perfect answer, but we recommend asking as few as you need to ask to get the information you want. In the beginning, focus on the big things:

Who are your users?

What do potential customers want?

How are they using your product?

What would win their loyalty?

❌ DON’T just ask a question when you can combine it with other tools

Don’t just use surveys to answer questions that other tools (such as analytics) can also answer. If you want to learn about whether people find a new website feature helpful, you can also observe how they’re using it through traditional analytics, session recordings , and other user testing tools for a more complete picture.

Don’t use surveys to ask people questions that other tools are better equipped to answer. I’m thinking of questions like “What do you think of the search feature?” with pre-set answer options like ‘Very easy to use,’ ‘Easy to use,’ etc. That’s not a good question to ask. 

Why should you care about what people ‘think’ about the search feature? You should find out whether it helps people find what they need and whether it helps drive conversions for you. Analytics, user session recordings, and user testing can tell you whether it does that or not.

❌ DON’T ask leading questions

A leading question is one that prompts a specific answer. Avoid asking leading questions because they’ll give you bad data. For example, asking, “What makes our product better than our competitors’ products?” might boost your self-esteem, but it won’t get you good information. Why? You’re effectively planting the idea that your own product is the best on the market.

❌ DON’T ask loaded questions

A loaded question is similar to a leading question, but it does more than just push a bias—it phrases the question such that it’s impossible to answer without confirming an underlying assumption.

A common (and subtle) form of loaded survey question would be, “What do you find useful about this article?” If we haven’t first asked you whether you found the article useful at all, then we’re asking a loaded question.

❌ DON’T ask about more than one topic at once

For example, “Do you believe our product can help you increase sales and improve cross-collaboration?”

This complex question, also known as a ‘double-barreled question’, requires a very complex answer as it begs the respondent to address two separate questions at once:

Do you believe our product can help you increase sales?

Do you believe our product can help you improve cross-collaboration?

Respondents may very well answer 'yes', but actually mean it for the first part of the question, and not the other. The result? Your survey data is inaccurate, and you’ve missed out on actionable insights.

Instead, ask two specific questions to gather customer feedback on each concept.

How to run your surveys

The format you pick for your survey depends on what you want to achieve and also on how much budget or resources you have. You can

Use an on-site survey tool , like Hotjar Surveys , to set up a website survey that pops up whenever people visit a specific page: this is useful when you want to investigate website- and product-specific topics quickly. This format is relatively inexpensive—with Hotjar’s free forever plan, you can even run up to 3 surveys with unlimited questions for free.

travel survey questions

Use Hotjar Surveys to embed a survey as an element directly on a page: this is useful when you want to grab your audience’s attention and connect with customers at relevant moments, without interrupting their browsing. (Scroll to the bottom of this page to see an embedded survey in action!) This format is included on Hotjar’s Business and Scale plans—try it out for 15 days with a free Ask Business trial .

Use a survey builder and create a survey people can access in their own time: this is useful when you want to reach out to your mailing list or a wider audience with an email survey (you just need to share the URL the survey lives at). Sending in-depth questionnaires this way allows for more space for people to elaborate on their answers. This format is also relatively inexpensive, depending on the tool you use.

Place survey kiosks in a physical location where people can give their feedback by pressing a button: this is useful for quick feedback on specific aspects of a customer's experience (there’s usually plenty of these in airports and waiting rooms). This format is relatively expensive to maintain due to the material upkeep.

Run in-person surveys with your existing or prospective customers: in-person questionnaires help you dig deep into your interviewees’ answers. This format is relatively cheap if you do it online with a user interview tool or over the phone, but it’s more expensive and time-consuming if done in a physical location.

💡 Pro tip: looking for an easy, cost-efficient way to connect with your users? Run effortless, automated user interviews with Engage , Hotjar’s user interview tool. Get instant access to a pool of 200,000+ participants (or invite your own), and take notes while Engage records and transcribes your interview.

10 survey use cases: what you can do with good survey questions

Effective survey questions can help improve your business in many different ways. We’ve written in detail about most of these ideas in other blog posts, so we’ve rounded them up for you below.

1. Create user personas

A user persona is a character based on the people who currently use your website or product. A persona combines psychographics and demographics and reflects who they are, what they need, and what may stop them from getting it.

Examples of questions to ask:

Describe yourself in one sentence, e.g. “I am a 30-year-old marketer based in Dublin who enjoys writing articles about user personas.”

What is your main goal for using this website/product?

What, if anything, is preventing you from doing it?

👉 Our post about creating simple and effective user personas in four steps highlights some great survey questions to ask when creating a user persona.

🚀 Get started: use our user persona survey template or AI for Surveys to inform your user persona.

2. Understand why your product is not selling

Few things are more frightening than stagnant sales. When the pressure is mounting, you’ve got to get to the bottom of it, and good survey questions can help you do just that.

What made you buy the product? What challenges are you trying to solve?

What did you like most about the product? What did you dislike the most?

What nearly stopped you from buying?

👉 Here’s a detailed piece about the best survey questions to ask your customers when your product isn’t selling , and why they work so well.

🚀 Get started: our product feedback survey template helps you find out whether your product satisfies your users. Or build your surveys in the blink of an eye with Hotjar AI.

3. Understand why people leave your website

If you want to figure out why people are leaving your website , you’ll have to ask questions.

A good format for that is an exit-intent pop-up survey, which appears when a user clicks to leave the page, giving them the chance to leave website feedback before they go.

Another way is to focus on the people who did convert, but just barely—something Hotjar founder David Darmanin considers essential for taking conversions to the next level. By focusing on customers who bought your product (but almost didn’t), you can learn how to win over another set of users who are similar to them: those who almost bought your products, but backed out in the end.

Example of questions to ask:

Not for you? Tell us why. ( Exit-intent pop-up —ask this when a user leaves without buying.)

What almost stopped you from buying? (Ask this post-conversion .)

👉 Find out how HubSpot Academy increased its conversion rate by adding an exit-intent survey that asked one simple question when users left their website: “Not for you? Tell us why.”

🚀 Get started: place an exit-intent survey on your site. Let Hotjar AI draft the survey questions by telling it what you want to learn.

I spent the better half of my career focusing on the 95% who don’t convert, but it’s better to focus on the 5% who do. Get to know them really well, deliver value to them, and really wow them. That’s how you’re going to take that 5% to 10%.

4. Understand your customers’ fears and concerns

Buying a new product can be scary: nobody wants to make a bad purchase. Your job is to address your prospective customers’ concerns, counter their objections, and calm their fears, which should lead to more conversions.

👉 Take a look at our no-nonsense guide to increasing conversions for a comprehensive write-up about discovering the drivers, barriers, and hooks that lead people to converting on your website.

🚀 Get started: understand why your users are tempted to leave and discover potential barriers with a customer retention survey .

5. Drive your pricing strategy

Are your products overpriced and scaring away potential buyers? Or are you underpricing and leaving money on the table?

Asking the right questions will help you develop a pricing structure that maximizes profit, but you have to be delicate about how you ask. Don’t ask directly about price, or you’ll seem unsure of the value you offer. Instead, ask questions that uncover how your products serve your customers and what would inspire them to buy more.

How do you use our product/service?

What would persuade you to use our product more often?

What’s the one thing our product is missing?

👉 We wrote a series of blog posts about managing the early stage of a SaaS startup, which included a post about developing the right pricing strategy —something businesses in all sectors could benefit from.

🚀 Get started: find the sweet spot in how to price your product or service with a Van Westendorp price sensitivity survey or get feedback on your pricing plan .

6. Measure and understand product-market fit

Product-market fit (PMF) is about understanding demand and creating a product that your customers want, need, and will actually pay money for. A combination of online survey questions and one-on-one interviews can help you figure this out.

What's one thing we can add that would make [product name] indispensable for you?

If you could change just one thing in [product name], what would it be?

👉 In our series of blog posts about managing the early stage of a SaaS startup, we covered a section on product-market fit , which has relevant information for all industries.

🚀 Get started: discover if you’re delivering the best products to your market with our product-market fit survey .

7. Choose effective testimonials

Human beings are social creatures—we’re influenced by people who are similar to us. Testimonials that explain how your product solved a problem for someone are the ultimate form of social proof. The following survey questions can help you get some great testimonials.

What changed for you after you got our product?

How does our product help you get your job done?

How would you feel if you couldn’t use our product anymore?

👉 In our post about positioning and branding your products , we cover the type of questions that help you get effective testimonials.

🚀 Get started: add a question asking respondents whether you can use their answers as testimonials in your surveys, or conduct user interviews to gather quotes from your users.

8. Measure customer satisfaction

It’s important to continually track your overall customer satisfaction so you can address any issues before they start to impact your brand’s reputation. You can do this with rating scale questions.

For example, at Hotjar, we ask for feedback after each customer support interaction (which is one important measure of customer satisfaction). We begin with a simple, foot-in-the-door question to encourage a response, and use the information to improve our customer support, which is strongly tied to overall customer satisfaction.

How would you rate the support you received? (1-5 scale)

If 1-3: How could we improve?

If 4-5: What did you love about the experience?

👉 Our beginner’s guide to website feedback goes into great detail about how to measure customer service, NPS , and other important success metrics.

🚀 Get started: gauge short-term satisfaction level with a CSAT survey .

9. Measure word-of-mouth recommendations

Net Promoter Score is a measure of how likely your customers are to recommend your products or services to their friends or colleagues. NPS is a higher bar than customer satisfaction because customers have to be really impressed with your product to recommend you.

Example of NPS questions (to be asked in the same survey):

How likely are you to recommend this company to a friend or colleague? (0-10 scale)

What’s the main reason for your score?

What should we do to WOW you?

👉 We created an NPS guide with ecommerce companies in mind, but it has plenty of information that will help companies in other industries as well.

🚀 Get started: measure whether your users would refer you to a friend or colleague with an NPS survey . Then, use our free NPS calculator to crunch the numbers.

10. Redefine your messaging

How effective is your messaging? Does it speak to your clients' needs, drives, and fears? Does it speak to your strongest selling points?

Asking the right survey questions can help you figure out what marketing messages work best, so you can double down on them.

What attracted you to [brand or product name]?

Did you have any concerns before buying [product name]?

Since you purchased [product name], what has been the biggest benefit to you?

If you could describe [brand or product name] in one sentence, what would you say?

What is your favorite thing about [brand or product name]?

How likely are you to recommend this product to a friend or colleague? (NPS question)

👉 We talk about positioning and branding your products in a post that’s part of a series written for SaaS startups, but even if you’re not in SaaS (or you’re not a startup), you’ll still find it helpful.

Have a question for your customers? Ask!

Feedback is at the heart of deeper empathy for your customers and a more holistic understanding of their behaviors and motivations. And luckily, people are more than ready to share their thoughts about your business— they're just waiting for you to ask them. Deeper customer insights start right here, with a simple tool like Hotjar Surveys.

Build surveys faster with AI🔥

Use AI in Hotjar Surveys to build your survey, place it on your website or send it via email, and get the customer insight you need to grow your business.

FAQs about survey questions

How many people should i survey/what should my sample size be.

A good rule of thumb is to aim for at least 100 replies that you can work with.

You can use our  sample size calculator  to get a more precise answer, but understand that collecting feedback is research, not experimentation. Unlike experimentation (such as A/B testing ), all is not lost if you can’t get a statistically significant sample size. In fact, as little as ten replies can give you actionable information about what your users want.

How many questions should my survey have?

There’s no perfect answer to this question, but we recommend asking as few as you need to ask in order to get the information you want. Remember, you’re essentially asking someone to work for free, so be respectful of their time.

Why is it important to ask good survey questions?

A good survey question is asked in a precise way at the right stage in the customer journey to give you insight into your customers’ needs and drives. The qualitative data you get from survey responses can supplement the insight you can capture through other traditional analytics tools (think Google Analytics) and behavior analytics tools (think heatmaps and session recordings , which visualize user behavior on specific pages or across an entire website).

The format you choose for your survey—in-person, email, on-page, etc.—is important, but if the questions themselves are poorly worded you could waste hours trying to fix minimal problems while ignoring major ones a different question could have uncovered. 

How do I analyze open-ended survey questions?

A big pile of  qualitative data  can seem intimidating, but there are some shortcuts that make it much easier to analyze. We put together a guide for  analyzing open-ended questions in 5 simple steps , which should answer all your questions.

But the fastest way to analyze open questions is to use the automated summary report with Hotjar AI in Surveys . AI turns the complex survey data into:

Key findings

Actionable insights

Will sending a survey annoy my customers?

Honestly, the real danger is  not  collecting feedback. Without knowing what users think about your page and  why  they do what they do, you’ll never create a user experience that maximizes conversions. The truth is, you’re probably already doing something that bugs them more than any survey or feedback button would.

If you’re worried that adding an on-page survey might hurt your conversion rate, start small and survey just 10% of your visitors. You can stop surveying once you have enough replies.

Related articles

travel survey questions

User research

Data-driven vs. data-informed decision-making: which should your product team use?

Data is a crucial part of building a product your customers love. But just how much should you rely on data, and how do you ensure it works in the best interests of your team—and end-users?

Hotjar team

travel survey questions

5 steps to defining your company’s ideal customer profile

You can only choose a wonderful present if you really know the person you’re buying it for. By the same token, you can only create an outstanding product or service when you really know the person you’re building it for. 

That’s the theory behind the ideal customer profile (ICP) framework. 

travel survey questions

7 reasons your website users are frustrated (and how to fix them)

You've painstakingly built a website to capture users' heads and hearts. But despite your effort, the site's bounce rates are high, conversions could be better, and users abandon their shopping carts faster than a kitten chasing a laser pointer. 

This is the story of many website designers, UX designers, and online businesses—and these all-too-common issues are often symptoms of user frustration. 

travel survey questions

Travel Survey Examples – PDF

travel survey examples

  • Survey Questionnaire Examples
  • Survey Examples

What is a Travel Survey?

architecture buildings business 745243 e1523929083177

Recent or Continuous City-Wide Travel Surveys

Travel survey example.

sample travel survey

Simple Travel Survey Example

simple travel survey

  • Pleasure/Personal
  • Internal company business (visit regional office, plant or attend a company meeting)
  • External company business (sales call, visit customer or vendor)
  • Official government/military business
  • Attend industry meeting/trade show
  • Attend a convention
  • Accompany family member on business/convention
  • Other business (not described above)
  • Incentive travel (company-paid pleasure trip)
  • Visit friends/relatives
  • Visit resort
  • My trip will (did) include plans to sightsee
  • My trip will (did) include plans to ski
  • My trip will (did) include plans to golf/tennis
  • Personal affairs (moving; to/from school)
  • Personal emergency (illness, etc.)
  • Vacation or other personal reason
  • Curbside baggage check-in
  • (Company) ticket counter
  • Express baggage/Seat check-in counter
  • Security checkpoint
  • Red Carpet Room
  • Boarding gate check-in counter
  • Aircraft boarding
  • Ticket counter agent
  • Express baggage agent
  • Security checkpoint personnel
  • Red Carpet Room receptionist
  • Boarding gate agent(s)
  • Flight attendant (during boarding)

city 3323160

More Design

3+ job satisfaction survey examples, 3+ research survey examples, 2+ hotel/restaurant survey examples, examples of writing a brand awareness survey, examples of online questionnaire, examples on how to conduct a business survey, what is the importance of a questionnaire, what is a survey questionnaire, 9+ income questionnaire examples.


Related Articles

  • id; ?>)" rel="noopener" role="button" tabindex="0" aria-label="postclick">35+ Health Questionnaire Examples
  • id; ?>)" rel="noopener" role="button" tabindex="0" aria-label="postclick">34+ Questionnaire Examples in Word Format
  • Mobile Forms
  • See 100+ integrations
  • See more Integrations
  • See more CRM Integrations


  • See more Storage Integrations
  • See more Payment Integrations

Mad Mimi

  • See more Email Integrations
  • Jotform Teams
  • Admin Console
  • Enterprise Mobile
  • Prefill Forms
  • HIPAA Forms
  • Secure Forms
  • Assign Forms
  • Online Payments
  • See more features
  • Multiple Users
  • White Labeling
  • See more Enterprise Features
  • Contact Sales
  • Contact Support
  • Help Center
  • Jotform for Beginners
  • Jotform Academy NEW

Get a dedicated support team with Jotform Enterprise.

Apply to Jotform Enterprise for a dedicated support team.

  • Sign Up for Free
  • Travel Questionnaire Form

A travel questionnaire form is used by travel agencies to get to know their customers’ preferences, what destinations they want to travel to, the best time of the year to visit these destinations, etc. If you manage a travel agency, you can streamline your customer service with our free Travel Questionnaire Form template. Whether you want to find out what type of travel plans customers are interested in or what destinations are popular amongst your clients, use our online travel questionnaire form to get your customers’ feedback easily.

After taking the form, you can share the responses with your customers or send them to your email account or to any other platform you use. Simply customize the form by adding your logo, selecting the colors, updating the fields, or adding partner logos—or even adding a social media page to make it easier for customers to get in touch with you. If you’d like to pull in data from other accounts to get more information about your customers, 100+ integrations are at your disposal!

More templates like this

Form Templates: Flight Reservation Form

Flight Reservation Form

A flight reservation form is used by an airline or travel agency to gather information needed to book air travel for clients. Ready to see your booking rates soar? Streamline the flight booking process for passengers and take your business to new heights by collecting the information you need in a single, secure online Flight Reservation Form. Form responses are stored in your Jotform account, creating an easy-to-access database of passenger information — and if you’d like to boost functionality and send form responses to your other accounts, just link your form to a few of our 130+ app integrations.  Give your Flight Reservation Form a unique look with Jotform Form Builder! Add your company logo, change fonts and colors to match your branding, or include any terms and conditions your customers may need to know. If you charge a fee for your service, accept payments through your online form via trusted payment gateways like Square or PayPal. Get ready for your bookings to take off — by bringing your reservation process online with our free Flight Reservation Form, you’ll save time, reduce unnecessary back-and-forth calls and emails, and smooth out the process for your customers.

Form Templates: Vacation Itinerary Form

Vacation Itinerary Form

This Vacation Itinerary form is normally utilized by travel agencies or person who just wants to travel and would like to have a detailed itinerary form. This vacation itinerary form will be a great help for those who are planning for a trip or travel agencies looking for a standard template they can use when organizing a trip for their clients. The form will need information such as destination, duration of stay, arrival time, mode of transport and cost.

Form Templates: Guest Registration Form

Guest Registration Form

A guest registration form is used to collect information from hotel guests. Whether you run a small, private bed and breakfast or an internationally renowned five-star luxury resort, get new customers and boost repeat business with our free Guest Registration Form template! All you need to do is embed the form on your website, share it with a link, or have your guests fill it out on your tablets or computers — then watch as guests submit their information directly to your booking system!If you’d like to accept card payments online, integrate with a trusted payment processor like Square, PayPal, or Stripe — we have 100+ other apps you can choose from, too. Integrate your hotel’s otherwise scattered data with a free Guest Registration Form — track guests, add notes, and stay organized with Jotform!

  • Form Templates /
  • Booking Forms /
  • Travel Booking Forms /

Travel Booking Forms

Travel booking forms, also known as travel reservation forms, are used by travel agencies to help book travel for clients. Whether your agency specializes in planning tours, cruises, family vacations, or business travel, our free Travel Booking Forms will speed up the reservation process by collecting important travel information online. Clients can provide details about the number of attendees, arrival and departure dates, and important travel demands all at once, eliminating the need for back-and-forth phone calls and emails. Each submission will be safely stored in a single, secure Jotform account, easily accessible by staff on any device.

Your travel agency makes sure to tailor to clients’ needs, so why not tailor your travel booking forms to your needs? Our easy-to-use Form Builder only needs you to drag and drop to make modifications to any Travel Booking Form below. You can change the form fields, template layout, background image, and even upload your logo for a professional look. Need to collect reservation fees? Just integrate your form with one of our payment gateways to process fees instantly — Jotform offers more than 30 processing apps, including Square and PayPal. You can even connect the form to other online accounts, such as your Google Drive or Slack workspace, to speed up your workflow. Start booking more clients online and improve your agency’s reservation process with our free Travel Booking Forms!

Gather passenger information with an online Flight Reservation Form. Free to customize and embed in your website. Great for airlines or travel agencies.

A guest registration form is used to collect information from hotel guests.

Form Templates: Travel Information Form

Travel Information Form

Collect travel requests online with a free Travel Information Form. Great for travel agencies. Easy to customize, embed, and share with your clients.

Form Templates: Xstream Paycation Travel Booking Request Form

Xstream Paycation Travel Booking Request Form

For gathering clients travel booking request information.

Form Templates: Travel Agency Booking Form Template

Travel Agency Booking Form Template

A travel agency booking form is a service reservation form used by travel agencies to book hotels, flights, or cruise packages. It is a useful tool to improve your hotel or airline booking services. Take your services to the next level!

Form Templates: Travel Registration Form

Travel Registration Form

Are you travel experts or do you provide travel assistance? Use this travel registration form to plan and organize your clients trip. You can provide this travel tour registration form to your clients so they can easily register their trip ahead and fill out the required information about their travel preferences. This trip registration form template asks personal information and travel information such as cruise and hotel name, cabin types, hotel bedding and emergency contact number. Send this trip registration form to your clients now to start accepting travel registrations!

Form Templates: Service Fee Agreement Template

Service Fee Agreement Template

This is for travel agents who charge fees (service fees or consultation fees) or for those who are exploring charging a fee. What it Does: The contract is not only legally binding, but it helps you be very clear about your fee structure so your clients don't get sticker shock when they book with you! It outlines what you charge for fees and when they need to pay it. Final Thoughts: This contract is thorough, and like other Jotform forms, you can modify the conditional logic to tailor it to the needs of your agency. See the whole article at HostAgencyReviews.com/blog/free-travel-agent-forms-clients



This template was created by Queen Raw Toni Ward. This is to qualify clients.

Form Templates: Trip Details Confirmation Form

Trip Details Confirmation Form

A trip details confirmation form is a travel confirmation form used by clients to receive a travel itinerary. Send a trip details confirmation form to clients with a single click and make the most of their trip with Jotform!

Form Templates: Travel Quote Request Form

Travel Quote Request Form

A travel quote request form is a document that allows customers to request a quote for a travel-related product or service, such as a cruise or a flight.

Form Templates: Travel Request Booking Form

Travel Request Booking Form

Form Templates: Travel Plan Form

Travel Plan Form

Travel Plan Form composed of areas for your clients to provide their personal/contact information with attendee details, departure, and arrival information and flight details which can be used as your official travel plan format.

Form Templates: Client Passenger Information

Client Passenger Information

Here is a travelers' information form where you can gather passengers' personal and contact information, name, gender, birth date, contact number, email address, home address, and their additional comments and questions.

Form Templates: Travel Interest Survey

Travel Interest Survey

An online travel interest survey can be used by travel agents or travel agencies to gather information about the interests, experience, and past travel of a potential client. Stay ahead of the competition with Jotform!

Form Templates: Travel Request Form

Travel Request Form

A travel request form is a form used by businesses to collect info about a request a client is making.

Form Templates: Disney World Planning And Booking Cheatsheet

Disney World Planning And Booking Cheatsheet

Disney World Planning and Booking Cheatsheet is an agent-facing form to help agents keep track of their Disney World client details. Save time, and be more efficient with Jotform!

Form Templates: LB Credit Card Authorization Form

LB Credit Card Authorization Form

Authorization Form

Form Templates: Travel Inquiry Form

Travel Inquiry Form

A travel inquiry form is a questionnaire given to customers to enquire about travel details such as destination, dates, and prices.

Form Templates: Tour Booking Form

Tour Booking Form

A tour booking form enables customers to book tours and activities through a travel website. Whether you guide kayaking, hiking, or cycling tours through your website, streamline the process with the Tour Booking Form template.

Form Templates: Open House Guest Registration Form

Open House Guest Registration Form

Real estate agents can seamlessly collect open house requests with this free Open House Guest Registration Form. Embed on your website to collect leads fast.

Form Templates: Vacation Review Form

Vacation Review Form

A vacation review form is used by travel agents to give to clients post-travel to help process the highs and lows of their trip and share them with you.

Form Templates: Client Profile

Client Profile

Do you want to know about your customers what are they thinking about travel? This client profile template is really helpful for you! Travel agent client profile form has six different question group which are air travel, hotel, car rental, booking information, cruising information and travelers information. Also, this client profile form template asking for your personal information.

Form Templates: Customer Information Form

Customer Information Form

A travel customer information form is used to collect contact information, travel plans and other details from travel customers for future reference.

Your account is currently limited to {formLimit} forms.

Go to My Forms and delete an existing form or upgrade your account to increase your form limit.

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • View all journals
  • My Account Login
  • Explore content
  • About the journal
  • Publish with us
  • Sign up for alerts
  • Data Descriptor
  • Open access
  • Published: 29 January 2024

Multidimensional well-being of US households at a fine spatial scale using fused household surveys

  • Kevin Ummel 1 ,
  • Miguel Poblete-Cazenave   ORCID: orcid.org/0000-0001-8280-910X 2 , 3 ,
  • Karthik Akkiraju 4 ,
  • Nick Graetz 5 ,
  • Hero Ashman 1 ,
  • Cora Kingdon 1 ,
  • Steven Herrera Tenorio 1 ,
  • Aaryaman Sunny Singhal 1 ,
  • Daniel Aldana Cohen 1 &
  • Narasimha D. Rao 3 , 4  

Scientific Data volume  11 , Article number:  142 ( 2024 ) Cite this article

Metrics details

  • Interdisciplinary studies
  • Social sciences

Social science often relies on surveys of households and individuals. Dozens of such surveys are regularly administered by the U.S. government. However, they field independent, unconnected samples with specialized questions, limiting research questions to those that can be answered by a single survey. The presented data comprise the fusion onto the American Community Survey (ACS) microdata of select donor variables from the Residential Energy Consumption Survey (RECS) of 2015, the National Household Travel Survey (NHTS) of 2017, the American Housing Survey (AHS) of 2019, and the Consumer Expenditure Survey - Interview (CEI) for the years 2015–2019. This results in an integrated microdataset of household attributes and well-being dimensions that can be analyzed to address research questions in ways that are not currently possible. The underlying statistical techniques, designed under the fusionACS project, are included in an open-source R package, fusionModel, that provides generic tools for the creation, analysis, and validation of fused microdata.

Background & Summary

Ideally, social scientists would have access to a “comprehensive survey” that employs a large sample size, asks many questions on various topics, is representative of the general population, and enjoys perfect recall and accuracy. Such a survey would allow researchers to examine spatial patterns at higher resolution, analyze differences across detailed population subgroups, explore relationships among a wide range of phenomena, and build detailed micro-simulation models to anticipate policy impacts across households and communities. Unfortunately, a truly comprehensive survey is impossible. Budgets and sample sizes are limited; respondent participation suffers if too many questions are asked; and the scope of social phenomena is too large for a single survey instrument. In practice, a diverse collection of surveys exists at any one time, varying in size, subject matter, structure, and provenance.

Practitioners regularly impute or otherwise predict a variable or two from one dataset on to another. Piecemeal, ad hoc data fusion is a common necessity of quantitative research. Proper data fusion, on the other hand, seeks to systematically integrate two different samples into one microdata set. The desire to “fuse” or otherwise integrate independent datasets has a long history, dating to at least the early 1970’s 1 , 2 . The most prominent examples of data fusion have involved administrative record linkage 3 , 4 , 5 , 6 , 7 . This consists of exact matching or probabilistic linking of independent datasets, using observable information like social security numbers, names, or birth dates of individuals. Record linkage, the gold standard, can yield important insights and high levels of statistical confidence. However, it is rarely feasible for the kinds of publicly available microdata that most researchers use day-to-day (nevermind the difficulty of accessing administrative data). The challenge and promise recognized 50 years ago by Nancy and Richard Ruggles remains true today:

Unfortunately, no single microdata set contains all of the different kinds of information required for the problems which the economist wishes to analyze. Different microdata sets contain different kinds of information… great deal of information is collected on a sample basis. Where two samples are involved the probability of the same individual appearing in both may be very small, so that exact matching is impossible. Other methods of combining the types of information contained in the two different samples into one microdata set will be required 1 .

Here we present the first results of the fusionACS project, a group of synthetic microdatasets 8 that fuses selected data on energy consumption, appliances, and insecurity from the Residential Energy Consumption Survey (RECS) of 2015; on transportation costs from the National Household Travel Survey (NHTS) of 2017; on dwelling characteristics and vulnerabilities from the American Housing Survey (AHS) of 2019; and expenditures on various household goods and services from the Consumer Expenditure Survey - Interview (CEI) of 2015–2019, to the corresponding years of the American Community Survey (ACS) (Fig.  1a ). These synthetic data offer a new spatially granular characterization of American households’ multidimensional well-being and their living conditions. Specifically, the large sample size of the ACS allows for better spatial resolution than any other survey, resulting in a de facto estimation of the imputed variables at the level of individual Public Use Microdata Areas (2,300 nationwide), a significant spatial gain compared, e.g., to the US Census Division level information of the RECS of 2015 (Fig.  1b ). These data have the potential to advance research on multidimensional poverty and improve justice-oriented policy design.

figure 1

fusionACS: Output Data Schematic: ( a ) Input Surveys and Actual vs Simulated Outputs; ( b ) Representation of the Simulated Responses of fusionACS in New York City at the PUMA level.

Overall, the principal aim of the fusionACS project is to maximize the amount of information that can be extracted from the existing array of U.S. social surveys. This is accomplished through statistical “fusion” of disparate surveys in an attempt to simulate a more comprehensive survey. The technique uses the American Community Survey (ACS) – the largest U.S. household survey – as the “data backbone” of this process. Variables in “donor” surveys are fused onto ACS Public Use Microdata Sample (PUMS) microdata to produce simulated values for variables unique to the donor. This results in probabilistic estimates of how ACS respondents might have answered the donor survey questionnaire. Respondent characteristics that are common to both the donor and the ACS (e.g. income) – as well as spatial information that can be merged to both (e.g. characteristics of the local built environment) – are used to model donor variable outcomes using machine learning techniques.

In the context of fusionACS, we are interested in the following problem:

We have microdata from two independent surveys, A and B, that sample the same underlying population and time period (e.g. occupied U.S. households nationwide in 2018). We specify that A is the “recipient” dataset and B is the “donor”. Survey A is the American Community Survey and invariably has a larger sample size than B (N a  >  N b ). The goal is to generate a new dataset, C, that has the original survey responses of A plus a realistic representation of how each respondent in A might have answered the questionnaire of B. To do this, we identify a set of “harmonized” variables, X, that are common to both surveys; in practice, these are often things like household size, income, respondent age, race, etc. We then fuse a set of variables unique to B – call them Z, the “fusion variables” – onto the original microdata of A, conditional on X .

This has generally been posed as a “statistical matching” problem 9 whereby records from the donor microdata ( B ) are matched to a statistically-similar record in the recipient ( A ). Variables common to both datasets ( X ) are used to calculate similarity between records. For each record in A , a set of similar records are identified in B ; e.g. using a k -nearest neighbor algorithm. A single record in B is selected from this set and the variables unique to the donor ( Z ) are added (fused) to the matched record in A . A “mixed method” variant of this approach (see, e.g., Section 3.1.3 in Lewaa et al . 10 ) fits statistical models to B to estimate the conditional expectation of Z | X . The models are used to predict Z | X for both A and B ( Z × a and Z × b , respectively), possibly adding a random residual. The similarity of donor and recipient records is then calculated using Z × a and Z × a (rather than X ) and the ultimate fusion of Z proceeds as in the statistical matching case. The mixed method is effectively an implementation of predictive mean matching (PMM) first developed by Rubin 11 in the context of statistical matching and then extended to missing data imputation by Little 12 . Mixed, PMM-based techniques offer a number of advantages, including some protection against model misspecification (in the stochastic case) and a more defensible (and fast) calculation of record similarity, since it avoids calculating similarity across X variables of possibly mixed types and varying levels of relevance in explaining Z .

Statistical matching techniques – mixed or otherwise – generally fuse complete records from the donor. This is a practical advantage, since it ensures that multivariate relationships among the fused variables are not obviously erroneous. But complete matching also introduces the possibility that donor observations will be repeated – possibly many times – in the fused dataset, increasing the risk that real-world variance is under-represented in the fused dataset. Intuitively, matching of complete records is most sensible when the donor’s sample size is at least as large as the recipient’s ( N b ≥ N a ) and the number of variables to be fused is small. Neither condition holds for fusionACS use cases. A useful variant comes from the imputation literature 13 , where the insertion of complete records is impossible due to the typical sparsity of missing data. Imputation techniques usually proceed sequentially, filling in missing values one variable at a time or, alternatively, by sequential “blocks” of variables that are imputed jointly (e.g. see the popular mice imputation package 14 ). A related literature in the area of data synthesis for statistical disclosure control 15 also relies on sequential (“chained”) generation of synthetic variables. For example, Reiter 16 introduced the use of machine learning decision trees 17 to create wholly synthetic versions of survey microdata that do not rely on record matching 18 , 19 . However, the goal in these cases is the synthesis of a single dataset for purposes of disclosure control, not the fusion of separate datasets.

The fusion strategy implemented in the fusionModel package borrows and expands upon ideas from the statistical matching 9 , imputation 13 , and data synthesis 15 literature to create a flexible data fusion tool. It employs variable- k , conditional expectation matching that leverages high-performance gradient boosting algorithms. The methodology and code is tailored for intended fusionACS applications, allowing fusion of many variables, individually or in blocks, and efficient computation when the recipient (the ACS in the case of fusionACS) is large relative to the donor. Specifically, the goal was to create a data fusion tool that meets the following requirements: [noitemsep]

Accommodate donor and recipient datasets with divergent sample sizes

Handle continuous, categorical, and semi-continuous (zero-inflated) variable types

Ensure realistic values for fused variables

Scale efficiently for larger datasets

Fuse variables “one-by-one” or in “blocks”

Employ a data modeling approach that: [noitemsep]

Makes no distributional assumptions (i.e. non-parametric)

Automatically detects non-linear and interaction effects

Automatically selects predictor variables from a potentially large set

Ability to prevent overfitting (e.g. cross-validation)

There are practical limits to this process, generally reflected in declining confidence in results as more is asked of the underlying data. For this reason, uncertainty estimation (via multiple implicates and associated analytical tools) is an important part of fusionACS’s development. Ideally, researchers are able to ask any question of fusion output and then decide if the answer’s associated uncertainty is suitable for the intended analysis.

General strategy

Consider the simple case where we fuse a single, categorical variable Z consisting of v classes. Using the notation from above, we fit a model to the donor data, \(G=f(Z| {X}_{b})\) . G is used to predict conditional expectations for each recipient observation, \({D}_{a}=G({X}_{a})\) . In this case, D a is a N a × v matrix of conditional probabilities from which N a simulated class outcomes ( Z a ) are probabilistically drawn. The statistical model, G , consists of a LightGBM 20 gradient boosting model that minimizes the cross-validated log-loss. The categorical case is comparatively straightforward and easily implemented.

Now consider fusing a single, positive continuous variable Z . In this case, we use multiple models to estimate the conditional distribution of Z | X . Let \({G}_{u}=f({Z}_{u}| {X}_{b})\) estimate the conditional mean and \({G}_{q}=f({Z}_{q}| {X}_{b})\) estimate conditional quantiles ( q ) associated with p equally-spaced percentiles. This yields p  + 1 cross-validated LightGBM models. G u minimizes the cross-validated squared error (L2) loss; G q minimizes the cross-validated quantile (pinball) loss. Training models for large p is expensive; by default, we use p  = 3 with percentiles {0.166,0.5,0.833}. The conditional expectations of the recipient observations, D a , consists of a N a × ( p  + 1) matrix of conditional mean and quantiles.

Unlike in the categorical case, there is no obvious way to simulate Z a from D a . Common parametric assumptions are not ideal, since the conditional expectations imply unknown and (quite often) decidedly non-normal distributions. One option is to extend PMM to the current context, resulting in generalized “conditional expectation matching”. In this case, we derive D b by predicting G u and G q back onto the original training data, then find the k nearest neighbors ( k NN) in D b associated with each observation in D a . This is analogous to conventional PMM, except that we use Euclidean distance based on p  + 1 conditional expectations to find the nearest neighbors. Each Z a is then sampled randomly from the k nearest neighbors in the donor.

There are drawbacks to this approach. First, it fundamentally differs from that used for a single categorical variable. In the categorical case, D a provides a complete description of the conditional distribution. Ideally, we’d have something analogous in the continuous case; i.e. non-parametric, conditional distributions consistent with the conditional expectations from which to draw simulated values. Second, as with any PMM approach, the appropriate value of k is not clear. The literature on preferred k (see Van Buuren 21 , Section 3.4.3) is based on simulation studies and general recommendations. Third, the expense of the k NN operation increases with N a , N b , p , and k . The fusionACS context assumes (at a minimum) large N a , leading to concerns about computation time (in practice, the fusion operation works with MN a rows of recipient data, where M is the number of implicates. So, the effective row size passed to the k NN operation is >50 million for a single year of ACS households given typical M  = 40).

To address these issues, we modify the approach outlined above. First, we find the K nearest neighbors in D b associated with each observation of D b (not D a ). Since the fusionACS context implies N b ≪ N a , the k NN step using D b is not usually a problem (later we introduce an option for handling even large N b ). This yields a N b × K matrix (call it S ) of observed Z values, where each row contains values sourced from donor observations with the most-similar conditional expectations.

Note that the conditional expectations can exhibit widely-varying magnitudes. To ensure that the k NN step gives approximately equal weight to each expectation, we scale the columns of the input matrices. If x is column j of input matrix D , the transformed values are:

where med ( x ) and mad ( x ) are the median and median absolute deviation, respectively, and ε  = 0.001. This results in robust scaled values such that med ( D j ) = 0.5 with range approximately {0,1}.

Next, for each row in S , we find the unique integer value k* ( k* ≤ K ) that yields the best empirical approximation of the conditional distribution of Z|X b . That is, for each row in S , we find k* such that the first k* values result in mean and quantile values most similar to those in D b . This is done by minimizing an objective function for each row in S .

Let x contain the first k values from row i of matrix S . We calculate measures of divergence between x and the conditional mean and quantiles ( u and Q 1: p ) from row i of D b . The divergence from the conditional mean is:

We then calculate a measure of divergence for each of Q 1: p conditional quantiles:

where τ  =  P p when P p  > 0.5 and τ  = 1 − P p otherwise.

The overall divergence:

The deltas are calculated for each value of k in 1: K , and the optimal \({k}_{i}^{* }\) is that which minimizes Δ. The use of ϕ (normal PDF) and Φ (normal CDF) do not imply any parametric assumptions about the shape of the conditional distribution itself. The derivation of σ from the conditional quantiles assumes a normal distribution 22 , but this is done only to plausibly scale the mean divergence to {0,1}. Note that both of the deltas are bounded {0,1} and equal zero when there is perfect agreement between x and the conditional expectations, allowing them to be summed. Critically, \({k}_{i}^{* }\) can be determined using maximally-efficient matrix operations, even when S is large.

This operation produces a list ( L ) of N b variable-length (i.e. variable- k* ) vectors of observed Z values that give an empirical approximation of each donor observation’s conditional distribution for Z | X b . For each row of D a , we find the row index \(i\in \{1,N+b\}\) of the single nearest neighbor in D b . A simulated value is then randomly drawn from the observed Z values in L i . Finding the single nearest neighbor is fast.

To recapitulate: For each donor observation, we construct an empirical approximation of the conditional distribution, Z | X b , using observed Z values. Conditional expectations are modeled for each recipient observation. Each recipient is matched to the donor observation with the most similar conditional expectations. Finally, simulated Z values are drawn from the empirical conditional distribution of the matched donor observation. Figure  3 shows schematic diagrams of the process for categorical and continuous variables.

This “variable- k ” approach has desirable properties: it does not require a fixed k ; it explicitly uses the conditional expectations to approximate a non-parametric conditional distribution; computation time is not unduly influenced by N a ; and the simulated values are drawn from observed Z , ensuring valid outcomes.

In principle, it is preferable to use D a in the initial K nearest neighbors step, resulting in S being a N a × K matrix containing observed Z values. However, we find that k* is typically much larger than the k  = 5 or k  = 10 used in conventional PMM. With k* regularly on the order of 100 to 300, K needs to be large enough to ensure we capture a good approximation of the conditional distribution ( K  = 500 by default in fusionModel). If both N a and K are comparatively large, the required k NN operation may be unduly slow when using D a directly (as is the case for fusionACS applications).

fusionModel includes an additional option to speed up calculations in the event that N b is large. In this case, we can first perform k -means clustering on D b to reduce it to some smaller number ( r ) of cluster centers. With D b reduced to an r × ( p  + 1) matrix, the calculations proceed as above but significantly faster when r   ≪   Nb .

Semi-continuous Z that is inflated at zero is common in social surveys, especially variables related to dollar amounts. We use a two-stage modeling approach in this special case. A categorical (binary) model is first used to simulate zero vs. non-zero outcomes. Then p  + 1 mean and quantiles models and the variable- k approach described above are used to simulate outcomes, conditional on Z ≠ 0.

If there are multiple fusion variables, Z 1: n , they are fused sequentially such that \({G}_{i}=f\left({Z}_{i}| X,{Z}_{1:i-1}\right)\) . Fusion variables earlier in the fusion sequence become available as predictors. This allows within-observation dependence among the fusion variables to be modeled explicitly (at least for Z that occur later in the sequence), as well as being mediated through X .

Sometimes it is useful to fuse variables in “blocks”. This is most relevant when there are fusion variables that are structurally linked. For example, if a set of continuous variables need to sum to one at the household level, they must be fused in a block to ensure this identity is preserved in the output. Variable blocks can contain any variable type (categorical, continuous, semi-continuous). For computational convenience, fusion of blocks employs the fixed- k conditional expectation matching approach first described. That is, k is fixed to some user-specified integer ( k  = 10 by default). In this case, D b and D a include the conditional expectations of all variables in the block. If all Z are in a single block, then the fusion process equates to sampling complete records of Z from the donor using fixed k .

Modeling details

Successful fusion hinges on the amount of information that can be extracted from X . The Data Preparation section describes how we maximize the amount of potentially useful data in X . Our ability to then extract useful information depends critically on the modeling strategy used to estimate f ( Z | X ).

The fusionACS project uses LightGBM gradient boosting models (GBM) 20 , because they are flexible and efficient – functionally, computationally, and in terms of predictive ability. By changing the loss function, we can use a single modeling framework for prediction of conditional probabilities, means, and quantiles. GBM’s do not require a specified functional form and make no parametric assumptions. They can handle many predictor variables and automatically detect important predictors, interaction effects, and non-linear relationships. Tuning and cross-validation during training results in models that exhibit state-of-the-art predictive ability. And since LightGBM was designed for large-scale machine learning applications, even comparatively large fusionACS exercises compute efficiently.

Gradient boosting is (largely) a “black box” machine learning strategy ideal for contexts that demand high predictive ability but care little about inference. That is not generally the case in academic settings, but it is a good description of the fusionACS context. Since the platform seeks to accommodate and convincingly model any variable from any donor survey, GBM’s ability to perform well under what we might call “hands off, kitchen sink” conditions is an advantage.

The primary danger here is that a model could “overfit” to the training data, learning spurious patterns that are a result of random noise instead of legitimate signal. This issue receives little attention in the larger synthetic data literature, because it is largely focused on creating synthetic versions of the donor survey itself; i.e. replication of noise in the donor is not necessarily a problem. In the fusionACS case, the overarching goal is to estimate how ACS respondents might have answered the donor survey questionnaire. This implies learning generalizable patterns in the donor data (i.e. avoidance of overfitting). Or, to put it differently, overfit models will underestimate the amount of variance that we would reasonably expect ACS respondents to exhibit if they actually completed the questionnaire.

To protect against overfitting, we train each LightGBM model using 5-fold cross-validation to find the number of iterations (i.e. number of tree learners) that minimizes the out-of-sample loss metric. The final model is fit to the complete data set using this optimal number of iterations. In addition, we test three different tree sizes (number of leaves: 16, 32, 64), subsample 80% of predictors in each iteration, and set the minimum number of node observations to 0.1% of N b (minimum 20). All of these settings are designed to reduce the risk of overfitting during training. In addition, we employ a “prescreen” step that selects a unique subset of the predictor variables in X to use with each fusion variable in Z . This helps reduce both the risk of overfitting and computation time. While there is no penalty to making X as data-rich as possible (in general), we don’t ask the GBM modeling process itself to handle potentially hundreds of predictors. Doing so would unnecessarily increase the chance of a model learning a spurious pattern.

The prescreen step fits LASSO models 23 using the complete X and choosing the model that explains 95% of deviance, relative to a “full” model that includes all potential predictors. Since the LASSO shrinks coefficients towards zero, the selected model utilizes only a subset of X , and it is a useful screening strategy in the presence of highly-correlated predictors – as is often the case for fusionACS applications given the large number of correlated spatial attributes present in X .

Implementation details

The fusionACS pipeline produces ACS PUMS microdata with donor survey variables fused (simulated) for each respondent household (~1.3 million per year). This output can be used to perform any kind of analysis typically applied to microdata, with the added benefit that analyses can use variables from both the ACS and donor survey questionnaires. The output microdata can be used to produce estimates for specific locales at the level of individual Public Use Microdata Areas (PUMA’s, ~2,300 nationwide), a higher level of spatial resolution than available in most donor surveys. By passing the microdata through an additional spatial downscaling step 24 , estimates can be produced for areas as small as individual block groups.

The fusionACS “platform” consists of two packages written in the R programming language. The fusionModel package provides an open-source interface for general data fusion (i.e. modeling and analytical tools). A separate, data processing package (fusionData) is used to generate the data inputs needed to fuse variables from a range of U.S. social surveys onto ACS microdata. For a given candidate donor survey, the data processing and analytical “pipeline” consists of the following steps (see Fig.  2 ):

Ingest raw survey data to produce standardized microdata and documentation.

Harmonize variables in the donor survey with conceptually-similar variables in the ACS.

Prepare clean, structured, and consistent donor and ACS microdata.

Train machine learning models on the donor microdata.

Fuse the donor’s unique variables to ACS microdata.

Validate the fused microdata to gauge the quality of the fusion process.

Analyze the fused microdata to calculate estimates and margins of error.

figure 2

fusionACS: Flow of data.

figure 3

fusionACS: Schematic of the fusion process: ( a ) Fusion of a single categorical variable; ( b ) Fusion of a single continuous variable; ( c ) Chained fusion; ( d ) Generation of multiple implicates.

Steps 1–3 are part of the fusionData package. Steps 4–7 are carried out using the fusionModel package.

The methodology described above is implemented in the open-source fusionModel R package as two primary functions: train() and fuse() . The former encompasses a LightGBM model fitting to a donor dataset and the variable- k calculations, while the latter makes conditional expectation predictions for a recipient dataset and then draws simulated outcomes.

The train() function was written to enable maximum speed and memory efficiency via forking on Unix-like systems (e.g. Linux servers). On Windows machines, OpenMP-enabled multithreading is used within the LightGBM model training step only (forking is not possible on Windows). We have found forking to be faster for typical donor microdata, and this is what we use for production runs on Linux servers.

The fuse() function takes advantage of LightGBM’s native multithreading regardless of platform, since the expensive step is prediction of the numerous GBM’s for the ACS recipient microdata. To accommodate intended fusionACS applications, fuse() intelligently “chunks” operations depending on available system memory and writes output to disk “on the fly”. This makes large-scale fusion tasks possible (even if they cannot fit in physical RAM) and allows the multithreading to operate near peak efficiency.

Both train() and fuse() include an approximate nearest neighbor search, for which they use the ANN library ( http://www.cs.umd.edu/ mount/ANN/) implemented via the RANN package ( https://github.com/jefferislab/RANN ). LASSO models are fit using the glmnet package 23 , 25 . More generally, fusionModel relies on the data.table ( https://Rdatatable.gitlab.io/data.table ; https://github.com/Rdatatable/data.table ) and matrixStats ( https://github.com/HenrikBengtsson/matrixStats ) packages for the key data manipulation steps. All of these packages – as well as LightGBM – are maximized for efficiency and written in low-level C code. So even though fusionModel itself is written in R , the vast bulk of the computation is optimized for speed and memory usage.

Data preparation

A significant amount of effort is required to prepare raw survey microdata so they can be used within the fusionACS pipeline. The production of standardized microdata, harmonized variables, spatial datasets, and associated documentation across multiple surveys is a major contribution of the fusionACS project. Here, the fusionACS outputs generated combine openly available data from the Residential Energy Consumption Survey (RECS) of 2015 ( https://www.eia.gov/consumption/residential/data/2015/index.php ), the National Household Travel Survey (NHTS) of 2017 ( https://nhts.ornl.gov/downloads ), the American Housing Survey (AHS) of 2019 ( https://www.census.gov/programs-surveys/ahs/data/2019/ahs-2019-public-use-file-puf-.html ), and the Consumer Expenditure Survey - Interview (CEI) ( https://www.bls.gov/cex/pumd_data.htm ), with data from the American Community Survey for the years 2015, 2017, and 2019 ( https://www.census.gov/programs-surveys/acs/news/data-releases.html ). None of these datasets, nor the fused variable outputs, contain any sensitive or identifying human-derived data. The code used to prepare the raw data inputs is housed in the fusionData github repository ( https://github.com/ummel/fusionData ).

In principle, any survey of U.S. households or individuals circa 2005 or later is a candidate for fusion. Ideal donor surveys are those with larger sample sizes, respondent characteristics that overlap with ACS variables, and more detailed information on respondent location. Absence of these factors does not preclude usage but will affect the associated uncertainty.

“Ingestion” of a survey requires transforming raw survey data into standardized microdata that meet certain requirements. This entails writing custom code for every donor survey, vintage, and respondent type (household and/or person), often involving tedious “pre-processing” tasks like (among others things): replacement of integer codes with descriptive variable levels; replacement of “valid blanks” and “skips” with plausible values; imputation of missing observations; appropriate “classing” of variables (e.g. defining ordered factors); and documentation of variables. This process sometimes identifies errors or irregularities in the raw survey data, which suggests that we are thoroughly interrogating the data during the ingestion step.

Ingestion results in processed microdata observations that meet the following conditions:


Once a donor survey has been successfully ingested, it can then be “harmonized” with the ACS in preparation for fusion. The harmonization step identifies variables common to both the donor survey and the ACS and is the statistical linchpin of the fusion process. It consists of matching conceptually similar variables across surveys and determining how they can be modified to measure similar concepts. The harmonization process should be as exhaustive as possible, since the predictive power of subsequent LightGBM models depends on the amount of information in the shared/harmonized predictor variables.

In general, harmonization is time-consuming and error-prone. To address these concerns, we built a custom “Survey Harmonization Tool” – a web app within the fusionData package – to make the harmonization process faster and more robust. The harmonization tool can define complex harmonies, including “one-to-one”, “many-to-one”, and “many-to-many” linkages, as well as “binning” (discretization) of continuous variables in one survey to create alignment with a categorical variable in another. Importantly, the harmonization process makes use of both household- and person-level variables, when available. This is true even if fusion occurs only at the household level. For example, it is common for donor surveys that solicit person-level information to ask for the age of each household member. This variable can be harmonized with an analogous variable in the ACS person-level microdata. Even if the eventual fusion step models and simulates household-level variables for each ACS respondent household (as is typical), the person-level harmonies are still utilized. In this case, the underlying code automatically constructs a household-level variable reporting the age of the householder/reference person (constructed from the person-level microdata and associated harmonies). In this way, we leverage maximum information that is common to both the donor and ACS.

Spatial predictors

Part of the data processing involves adding spatial variables to the donor and ACS microdata to expand the number of predictor variables available for the modeling step. Spatial variables help to characterize a household’s location/environment, as opposed to the respondent-specific characteristics used in the harmonization process. For example, knowing something about the population density of a household’s general location can help explain patterns in the variables being fused that might not be “picked up” by a model using only respondent characteristics.

In principle, there is no limit to the amount, nature, or resolution of third-party spatial information that can be utilized by the fusionACS platform. The only requirement is that a spatial dataset must have national coverage. To date, we have focused on readily-available datasets likely to be useful in explaining the kind of socioeconomic phenomena measured by the donor surveys ingested so far (see Table  1 ).

Spatial variables are merged to the donor and ACS microdata at the level of individual PUMA’s. This is because the ACS PUMS only discloses respondent location for PUMA’s, so this is as precise as we can be with the spatial variables. For example, the EPA-SLD dataset provides variables describing features of the built environment for individual block groups. These variables are summarized at the PUMA-level prior to merging to the donor and ACS microdata. They are then available as LightGBM predictor variables in both the training (on donor microdata) and prediction (on ACS microdata) steps.

Due to confidentiality constraints, the donor surveys utilized here do not disclose the PUMA of respondents. Consequently, we impute each donor respondent’s PUMA given observable information. We make use of disclosed location information (e.g. respondent’s state of residence) as well as the suite of harmonized respondent-level variables. The latter are used to perform a probabilistic imputation, using Gower’s distance 26 as a similarity measure between donor and ACS respondents. That is, we assign a PUMA to each donor respondent by matching to an ACS respondent (using its observed PUMA) within the same disclosed location (e.g. state), where the probability of selection is proportional to the similarity of the donor and respondent on observable (harmonized) characteristics. This allows us to leverage all available information in the PUMA imputation process.

Uncertainty estimation

The fusion process attempts to produce a realistic representation of how each ACS respondent household (or individual) might have answered the questionnaire of the donor survey. The fused values are inherently probabilistic, reflecting uncertainty in the underlying statistical models.

In order to fully capture this uncertainty, fusionACS output consists of M multiple “implicates”. A single implicate contains a simulated response for each fused variable and ACS-PUMS respondent. Each implicate provides a unique, plausible set of simulated outcomes. Multiple implicates are needed to calculate unbiased point estimates and associated uncertainty (margin of error) for any particular analysis of the data, making it the standard approach in the literature 27 .

The use of multiple implicates is conceptually akin to that of replicate weights in conventional survey analysis. Replicate weights quantify uncertainty (variance) by keeping the response values fixed but varying the weight (frequency) associated with each respondent. Conversely, when imputing (or fusing) data, the primary sample weights are typically fixed while the simulated values vary across implicates.

Since proper analysis of multiple implicates can be rather cumbersome – both from a coding and mathematical standpoint – the fusionModel package provides a convenient analyze() function to perform common analyses on fused data and report point estimates and associated uncertainty. Potential analyses currently include variable means, proportions, sums, counts, and medians, (optionally) calculated for population subgroups.

Point estimates for any particular analysis are simply the mean of the M individual estimates across the implicates. In general, higher M is preferable but requires more computation and larger output file size. For fusionACS production runs, we currently use M  = 40 as a reasonable compromise.

Uncertainty for a given estimate reflects standard errors “pooled” across the implicates. A number of pooling rules for implicates have been introduced in the imputation and synthesis literatures, beginning with that of Rubin 28 for multiple imputation contexts. The closest analog to the fusionACS context is that considered in Reiter 29 . The pooling formulae in Reiter 29 assume a two-stage simulation strategy with parametric models that is not straightforward to apply to fusionACS output. However, that paper shows that the original Rubin 28 pooling formulae result in somewhat positively biased variance compared to the “correct” formulae. Consequently, the fusionModel analyze() function uses the Rubin 28 method to conservatively estimate uncertainty and associated margin of error (MOE). The MOE returned by analyze() reflects a 90% confidence level, consistent with how the Census Bureau reports MOE for native ACS-based estimates.

The unpooled standard errors (SE’s) that are used within the pooling formulae are calculated using the variance within each implicate. For means (and sums), the ratio variance approximation of Cochran 30 is used, as this is known to be a good approximation of bootstrapped SE’s for weighted means 31 . For proportions, a generalization of the unweighted SE formula is used. For medians, a large- N approximation is used when appropriate 32 and bootstrapped SE’s computed otherwise.

The analyze() function can also (optionally) include uncertainty due to variance in the ACS PUMS replicate weights. This is generally preferable, since it captures uncertainty in both the fused (simulated) values and the sampling weight of ACS households within the population. We introduce replicate weight uncertainty by assigning a different set of replicate weights to each of the M implicates (there are 80 PUMS replicate weights, so we use half of the replicate weights when M  = 40). We then estimate the additional across-implicate variance when using replicate weights (compared to the primary weights), and add this to Rubin’s pooled variance.

Data Records

The fusionACS dataset, available in figshare 8 include:

Fusion of 12 select donor variables from RECS 2015 to ACS 2015.

Fusion of 6 select donor variables from AHS 2019 to ACS 2019.

Fusion of 5 select donor variables from NHTS 2017 to ACS 2017.

Fusion of 5 select donor variables from NHTS 2017 to ACS 2015.

Fusion of 44 household consumption-expenditure and tax variables from CEI 2015–2019 (pooled) to ACS 2019.

The fused RECS 2015, AHS 2019, and NHTS 2017 microdata consists of both a single .fst and a single gzipped.csv file per survey, each containing 40 implicates. The RECS energy expenditure variables were fused in a second step using only the consumption and location variables as predictors to attain local consistency in energy prices.

The fused CEI 2015–2019 microdata consists of a .fst file containing 30 implicates. The data include a single “tax” variable derived from the CEI’s native before- and after-tax income variables.

The fused variables were selected to provide examples of socially-relevant variables of differing data types. The list of fused variables and their description can be found in Table  2 . Graphical examples of some of the capabilities of fusion outputs can be seen in Fig.  4 . Figure  4a presents an example of the higher spatial granularity that is possible, allowing estimates to be generated for each of the 2,351 PUMAs available in the ACS as compared to just 10 large US Census divisions in the original RECS. Figure  4b illustrates how fusionACS outputs can be used for multidimensional analyses of variables from multiple surveys.

figure 4

fusionACS: examples of enhanced capabilities of the fused microdata: ( a ) Annual Electricity Consumption (MWh) in RECS vs RECS-ACS fusion; ( b ) Households experiencing both energy (RECS 2015-ACS 2015) and travel insecurity (NHTS 2017-ACS 2015).

Technical Validation

Prospective users of fusionACS require information about the quality of fused data outputs. This necessitates validation exercises to demonstrate the expected “utility” of fusion-based analyses. In the data synthesis literature, a distinction is drawn between “general” (global) and “specific” (narrow) utility of synthetic datasets 33 . The former provides an overall statistical measure of the similarity of synthetic and observed data, while the latter refers to specific comparisons of synthetic and observed data for realistic analyses. General utility is a rather low bar to clear and (more importantly) it does not provide the kind of intuitive and familiar “proof” of data quality that can inspire confidence in prospective users. Consequently, we focus on validation exercises using measures of specific utility.

Internal validation

Internal validation consists of analyzing synthetic data produced by fusing variables “back” onto the original donor microdata. It is analogous to assessing model skill by comparing predictions to the observed training data.

The fusionModel package includes a validate() function to perform specific (non-general) internal validation tests on synthetic variables. For fusionACS production runs, the fusion models are simulated back onto the donor data and the result passed to validate() . All of the fusion datasets described in the Data Records section undergo the same internal validation process. The results are used to confirm that the fusion output behaves as expected and to characterize how data utility/quality changes across prospective analyses.

Utility in this case is based on comparison of analytical results derived using the multiple-implicate fusion output with those derived using the original donor microdata. By performing analyses on population subsets of varying size, validate() estimates how the synthetic variables perform for analyses of varying difficulty/complexity. It computes fusion variable means and proportions for subsets of the full sample – separately for both the observed and fused data – and then compares the results.

The plots in Fig.  5 show validate() output for a sample of five household expenditure variables fused from the CEI 2015–2019 donor survey. The population subsets are constructed using the six predictor variables that provide the closest analogs for income; race/ethnicity; education; household size; housing tenure; and respondent location. The internal validation trends presented in Fig.  5 are illustrative of results obtained for the other donor surveys as well.

figure 5

Example internal validation plots for five household expenditure variables from CEI 2015–2019. The results illustrate how fused variables compare to the original donor variables across population subsets of varying size. The variables are: “airshp” (Air and ship travel), “cloftw” (Clothing and footwear), “eathome” (Food eaten at home), “elec” (Electricity), and “gas” (Gasoline). ( a ) Median absolute percent error of point estimates. Variables “eathome”, “elec” and “gas” exhibit relatively low error, implying that they are well-captured by the underlying models. Variables ‘‘cloftw’’ and “airshp” exhibit higher error, suggesting caution may be warranted when analyzing in detailed analyses; ( b ) Median “value-added” based on comparison of fusion point estimates to those of a naive (null) model. Value-added is generally strong (>0.8) across all five variables, though it is noisier for “airshp” and ‘‘cloftw’’; ( c ) Median ratio of fusion point estimate uncertainty to that of original donor. The fusion point estimates typically result in about 20% higher uncertainty, reflecting the additional uncertainty associated with the modeling process.

Smaller population subsets are more susceptible to outliers in the observed data, causing the discrepancy between observed and simulated estimates to generally increase as subset size declines. For validation purposes we want to know what the general trend looks like, ignoring noise/outliers in the observed data. The validate() function plots results using a robust median smoother, in order to convey the expected typical (median) performance at a given subset size. It reports results for three different performance metrics, explained below.

Figure  5a shows how the observed and simulated point estimates compare, using median absolute percent error as the performance metric. We consider this the easiest-to-interpret error metric for practical purposes. Note that the x-axis is not linear; it is scaled to reveal more detail for small population subsets. The y-axis gives the (smoothed) median absolute percent error at each subset size.

The discrepancy (error) between the observed and simulated point estimates exhibits the typical pattern, increasing as subset size declines, but there is considerable variation. The variables “eathome”, “elec”, and “gas” exhibit quite low percent error, even for small subsets, implying that the explanatory patterns driving these variables are strongly identified by the underlying, cross-validated LightGBM models. The variables “cloftw” (clothing and footwear) and “airshp” (air and ship travel) exhibit higher error, especially for smaller subsets. These results suggest caution might be warranted if using “airshp” (and possibly “cloftw”) in high-resolution or complex analyses.

The validate() output plot in Fig.  5b presents an alternative way to gauge fusion quality, using a “value-added” metric that compares fusion output to that of a naive (null) model. Given simulated point estimate y s and observed estimate y o , we define the value-added ( V ) as:

where E ( y 0 ) is the full-sample mean. That is, V measures the extent to which the simulated estimates out-perform the naive estimate of a null model. This is conceptually similar to the approach used to define the canonical coefficient of determination ( R 2 ) (note that V uses absolute instead of squared error, hence, V is lower than R 2 , all else equal). V  = 1 when y s  =  y o and V  = 0 when y s is worse than the naive estimate. V is calculated for each individual analysis and then the median smoother applied.

We observe high value-added (>0.8) throughout most of the subset size range, though it is noisier for “airshp” and “cloftw”. Value-added helps to isolate the performance of the underlying fusion process, controlling for the degree of variance across population subsets. Some types of analyses may exhibit both high percent error (Fig.  5a ) and high value-added. This is indicative of a fusion variable whose underlying modeling process is performing close to optimal; i.e. this is probably “as good as it gets”, given the available survey data.

Finally, validate() outputs a comparison of simulated and observed relative uncertainty (MOE divided by the point estimate). This is useful for confirming that the simulated margin of errors exhibit plausible magnitudes. Figure  5c indicates that the fused data typically result in MOE about 20% higher than we observe in the training data, reflecting the additional uncertainty associated with the modeling process. The “airshp” MOE’s inflate for smaller subsets, reflecting the relative difficulty (also observed in the other plots) in modeling air and ship travel expenditures given the available predictor variables.

External validation

External validation consists of comparing specific results produced using fused data with analogous results from an entirely independent data source. We further validated the RECS fusion output with household natural gas and electricity consumption data at the county level available both publicly and from utility companies for the states of New York (NYSERDA, https://maps.nrel.gov/slope/data-viewer?layer = energy-consumption.net-electricity-and-natural-gas-consumption res = state year = 2017 filters = %5B%5D), California (California Energy Commission, http://www.ecdms.energy.ca.gov/elecbycounty.aspx ), and Massachusetts (masssavedata, https://www.masssavedata.com/public/home ) (Fig.  6a,b ). In general, we observe a good correlation between the simulated metrics and metrics obtained externally for the total electricity and natural gas consumption metrics and that the ranked metrics for the 3 states are largely preserved. Also, the averaged metrics for natural gas and electricity consumption showed larger deviation (Fig.  6c,d ), especially for the counties in New York, due to the lower sampling of RECS in these areas.

figure 6

External validation of fusionACS using ACS-RECS fused dataset: Comparison of total ( a ) simulated natural gas and ( b ) simulated electricity consumption metrics versus the actual total consumption metrics obtained from external sources; Comparison of average ( c ) simulated natural gas and ( d ) simulated electricity consumption metric versus the actual consumption metrics obtained from external sources for counties in New York, California, and Massachusetts; ( e ) Comparison of the rank of energy insecurity variable versus the rank of the number of 311 calls made at the PUMA-level for New York City, Philadelphia, Austin, and Boston.

We also compared publicly available 311 distress call data made during 2014–2016 reporting heat and/or hot water complaints in New York City (Heat/Hot Water) obtained from NYC Open Data (311 Service Requests from 2010 to Present, https://data.cityofnewyork.us/Social-Services/311-Service-Requests-from-2010-to-Present/erm2-nwe9 ), Austin (Building A/C & Heating Issues), City of Austin (PARD 311 Data, https://data.austintexas.gov/dataset/PARD-311-Data/t2t7-xusa ), Philadelphia(Heat) obtained from OpenDataPhilly (311 Service and Information Requests, https://opendataphilly.org/datasets/311-service-and-information-requests/ ), and Boston (Heat - Excessive Insufficient or Heat/Fuel Assistance) obtained from Analyze Boston (311 Service Requests, https://data.boston.gov/dataset/311-service-requests ) to the energy insecurity burden metric insec obtained from RECS-ACS fusion for 2015 at the PUMA-level (described in Table  2 ) (Fig.  6e ). Here we find reasonable agreement (corr.coeff = 0.36–1.00) between the rank of insecurity and the number of 311 distress calls which further validates the strength of the fused energy insecurity indicator.

Usage Notes

All generated fusion outputs are available as.fst files containing 40 implicates (30 implicates for CEI 2015-2019 fusion) along with the household id from the corresponding ACS, which can be accessed using the fst R package ( http://www.fstpackage.org ). Instructions on how to use the various functions of the fusionModel package are available in the corresponding github repository ( https://github.com/ummel/fusionModel ).

The validation exercises confirm overall high data quality/utility while also revealing variance across variables and analyses. This is due to unavoidable variation in the predictive ability of the underlying models, given the nature of the phenomena being fused and the available (harmonized) predictor variables. The Uncertainty Estimation section describes how fusionModel propagates this variance (uncertainty) into multiple fusion implicates.

It is recommended that users of fusionACS data outputs make use of the multiple implicates to calculate uncertainty (margin of error) alongside point estimates of interest. The margin of error should be taken into consideration when deciding if the results of any particular analysis are “good enough” for the intended application. For example, point estimate uncertainty may be acceptable when analyzing a particular fusion variable by income, but unacceptably high when analyzing by both income and race. There is no “shortcut” to this determination other than calculating uncertainty for the desired analysis. This is best practice for any analysis of survey microdata, fused or otherwise.

The analyze() function in the fusionModel package was built for this purpose and enables users to correctly calculate means, proportions, sums, counts, and medians, (optionally) for population subsets (e.g. by race) along with associated uncertainty. There are always limits to how much one can ask of the available data, and uncertainty estimation is the principal tool for detecting when an analysis has “gone too far” – the definition of which can only be specified by the analyst. If only point estimates are required, it is sufficient to simply derive the mean of the M estimates calculated for each of the independent implicates.

Code availability

The data preparation codes and the specific codes to generate the fused datasets presented in this study are on the fusionData github repository ( https://github.com/ummel/fusionData ). The generalized codes for the fusion, analysis, and validation of the datasets are available on the fusionModel github repository ( https://github.com/ummel/fusionModel ).

Ruggles, N. & Ruggles, R. A strategy for merging and matching microdata sets. In Annals of Economic and Social Measurement , vol. 3(2), 353–371 (NBER, 1974).

Alter, H. E. Creation of a synthetic data set by linking records of the canadian survey of consumer finances with the family expenditure survey. In Annals of Economic and Social Measurement , vol. 3(2), 373–397 (NBER, 1974).

Chetty, R., Hendren, N., Kline, P. & Saez, E. Where is the land of opportunity? the geography of intergenerational mobility in the united states. The Quarterly Journal of Economics 129 , 1553–1623 (2014).

Article   Google Scholar  

Chetty, R., Hendren, N. & Katz, L. F. The effects of exposure to better neighborhoods on children: New evidence from the moving to opportunity experiment. American Economic Review 106 , 855–902 (2016).

Article   PubMed   Google Scholar  

Chetty, R., Friedman, J. N., Saez, E., Turner, N. & Yagan, D. Income segregation and intergenerational mobility across colleges in the united states. The Quarterly Journal of Economics 135 , 1567–1633 (2020).

Meyer, B. D. & Mittag, N. Using linked survey and administrative data to better measure income: Implications for poverty, program effectiveness, and holes in the safety net. American Economic Journal: Applied Economics 11 , 176–204 (2019).

Google Scholar  

Medalia, C., Meyer, B. D., O’Hara, A. B. & Wu, D. Linking survey and administrative data to measure income, inequality, and mobility. International journal of population data science 4 (2019).

Ummel, K. et al . Multidimensional well-being of US households at a fine spatial scale using data fusion with national household surveys: fusionACS, figshare , https://doi.org/10.6084/m9.figshare.c.6998172.v1 (2024).

D’Orazio, M., Di Zio, M. & Scanu, M. Statistical matching: Theory and practice (John Wiley & Sons, 2006).

Lewaa, I., Hafez, M. S. & Ismail, M. A. Data integration using statistical matching techniques: A review. Statistical Journal of the IAOS 37 , 1391–1410 (2021).

Rubin, D. B. Statistical matching using file concatenation with adjusted weights and multiple imputations. Journal of Business & Economic Statistics 4 , 87–94 (1986).

MathSciNet   Google Scholar  

Little, R. J. Missing-data adjustments in large surveys. Journal of Business & Economic Statistics 6 , 287–296 (1988).

Little, R. J. & Rubin, D. B. Statistical analysis with missing data , vol. 793 (John Wiley & Sons, 2019).

Van Buuren, S. & Groothuis-Oudshoorn, K. mice: Multivariate imputation by chained equations in r. Journal of statistical software 45 , 1–67 (2011).

Drechsler, J. Synthetic datasets for statistical disclosure control: theory and implementation , vol. 201 (Springer Science & Business Media, 2011).

Reiter, J. P. Using cart to generate partially synthetic public use microdata. Journal of official statistics 21 , 441 (2005).

Breiman, L., Friedman, J., Olshen, R. & Stone, C. Classification and regression trees. Group 37 , 237–251 (1984).

Nowok, B., Raab, G. M. & Dibben, C. synthpop: Bespoke creation of synthetic data in r. Journal of statistical software 74 , 1–26 (2016).

Bowen, C. M. et al . A synthetic supplemental public use file of low-income information return data: methodology, utility, and privacy implications. In Privacy in Statistical Databases: UNESCO Chair in Data Privacy, International Conference, PSD 2020, Tarragona, Spain, September 23–25, 2020, Proceedings , 257–270 (Springer, 2020).

Ke, G. et al . Lightgbm: A highly efficient gradient boosting decision tree. Advances in neural information processing systems 30 (2017).

Van Buuren, S. Flexible imputation of missing data (CRC press, 2018).

Cook, J. Determining distribution parameters from quantiles. UT MD Anderson Cancer Center Department of Biostatistics Working Paper Series. Working Paper 55 (2010).

Friedman, J., Hastie, T. & Tibshirani, R. Regularization paths for generalized linear models via coordinate descent. Journal of statistical software 33 , 1 (2010).

Article   PubMed   PubMed Central   Google Scholar  

Graetz, N., Ummel, K. & Aldana Cohen, D. Small-area analyses using public american community survey data: A tree-based spatial microsimulation technique. Sociological Methodology 52 , 53–74 (2022).

Simon, N., Friedman, J., Hastie, T. & Tibshirani, R. Regularization paths for cox’s proportional hazards model via coordinate descent. Journal of Statistical Software 39 , 1–13 (2011).

Gower, J. C. A general coefficient of similarity and some of its properties. Biometrics 857–871 (1971).

Rubin, D. B. Multiple imputation after 18+ years. Journal of the American Statistical Association 91 , 473–489 (1996).

Rubin, D. B. Multiple imputation for nonresponse in surveys (John Wiley & Sons, 1987).

Reiter, J. P. Multiple imputation when records used for imputation are not used or disseminated for analysis. Biometrika 95 , 933–946 (2008).

Article   MathSciNet   Google Scholar  

Cochran, W. G. Sampling techniques (John Wiley & Sons, 1977).

Gatz, D. F. & Smith, L. The standard error of a weighted mean concentration–i. bootstrapping vs other methods. Atmospheric Environment 29 , 1185–1193 (1995).

Article   ADS   CAS   Google Scholar  

Doob, J. L. The limiting distributions of certain statistics. The Annals of Mathematical Statistics 6 , 160–169 (1935).

Snoke, J., Raab, G. M., Nowok, B., Dibben, C. & Slavkovic, A. General and specific utility measures for synthetic data. Journal of the Royal Statistical Society. Series A (Statistics in Society) 181 , 663–688 (2018).

Download references


This research was supported by funding from the Environmental Protection Agency through RTI International with Grant GR114933 RTI/EPA and the National Science Foundation Growing Convergence Research Grant GR117886. The authors are also grateful to the following groups for funding earlier rounds of research, which made this particular contribution possible: the Population Studies Center, the Office of the Vice Provost for Research, and the Kleinman Center for Energy Policy, all at the University of Pennsylvania; the Eunice Kennedy Shriver National Institute of Child Health and Human Development; the New York State Energy Research and Development Authority; and the Canadian Institute for Advanced Research (CIFAR).  

Author information

Authors and affiliations.

University of California, Berkeley, Socio-Spatial Climate Collaborative, Berkeley, 94720, USA

Kevin Ummel, Hero Ashman, Cora Kingdon, Steven Herrera Tenorio, Aaryaman Sunny Singhal & Daniel Aldana Cohen

VU Amsterdam, Institute for Environmental Studies (IVM), Amsterdam, 1081, Netherlands

Miguel Poblete-Cazenave

International Institute for Applied Systems Analysis (IIASA), Energy, Climate, and Environment Program, Laxenburg, A-2361, Austria

Miguel Poblete-Cazenave & Narasimha D. Rao

Yale University, School of the Environment, New Haven, 06511, USA

Karthik Akkiraju & Narasimha D. Rao

Princeton University, Princeton, 08544, USA

Nick Graetz

You can also search for this author in PubMed   Google Scholar


K.U., N.G., D.A.C. and N.D.R. conceived the initial framework. K.U., M.P.-C., K.A., N.G., H.A., C.K. and S.H.T. collected and prepared data. K.U. implemented the model, with M.P.-C. and N.G. providing methodological feedback. K.U., M.P.-C. and N.D.R. wrote the manuscript, with figures provided by K.U. and K.A. All authors reviewed the manuscript.

Corresponding author

Correspondence to Miguel Poblete-Cazenave .

Ethics declarations

Competing interests.

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/ .

Reprints and permissions

About this article

Cite this article.

Ummel, K., Poblete-Cazenave, M., Akkiraju, K. et al. Multidimensional well-being of US households at a fine spatial scale using fused household surveys. Sci Data 11 , 142 (2024). https://doi.org/10.1038/s41597-023-02788-7

Download citation

Received : 24 April 2023

Accepted : 24 November 2023

Published : 29 January 2024

DOI : https://doi.org/10.1038/s41597-023-02788-7

Share this article

Anyone you share the following link with will be able to read this content:

Sorry, a shareable link is not currently available for this article.

Provided by the Springer Nature SharedIt content-sharing initiative

Quick links

  • Explore articles by subject
  • Guide to authors
  • Editorial policies

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

travel survey questions

Politics latest: Supermarket boss accused of 'sour grapes' after switching to Labour

Rishi Sunak has rejected criticism from Liz Truss and Boris Johnson of the government's ban on tobacco and disposable vapes. Elsewhere, Lord Cameron is primed for a return to the Middle East for talks on the crisis there - and the government's Rwanda plan is being scrutinised in the Lords.

Monday 29 January 2024 23:18, UK

Please use Chrome browser for a more accessible video player

  • PM rejects criticism of tobacco and disposable vape bans
  • Starmer 'delighted' as Iceland boss endorses Labour
  • But minister accuses supermarket chief of 'sour grapes'
  • Cameron returning to Middle East as talks ramp up
  • 'Horrifying' allegations of sex assaults in mental health hospitals
  • Jon Craig: 'Will of people' vs 'will of God' in Rwanda Lords debate
  • Poll tracker:  Reform UK overtakes Lib Dems in polls
  • Politics At Jack And Sam's : The Week... Sunak takes on The Plot
  • Live reporting by Faith Ridler and (earlier)  Ben Bloch

Thanks for following along on an extremely busy day in Westminster.

Before you go, here are the highlights:

  • The second reading of the Safety of Rwanda (Asylum and Immigration) Bill has passed the House of Lords and will move on to the next stage;
  • A Liberal Democrat-backed motion which served as the law's first major hurdle in the Lords was also comfortably defeated;
  •  The Archbishop of Canterbury has made a powerful intervention in the debate over the government's Rwanda plan. Justin Welby says the bill to declare the country safe for asylum seekers to be sent leads the UK "down a damaging path";
  • The protesters outside were to be expected but no one anticipated details of the DUP's highly confidential executive meeting to be live tweeted from inside, our Ireland correspondent David Blevins reported;
  • An independent inquiry published its report into claims of corruption linked to the controversial Teesworks development.

We'll be back from 6am with all the latest.

One year ago, Rishi Sunak made five pledges for voters to judge him on.

The prime minister met his pledge to halve inflation by the end of 2023 - even though there was a surprise increase in the 12 months to December - leaving four pledges outstanding.

When it comes to the NHS, while both waiting lists and waiting times are shrinking slightly, the waiting list is still longer than in January last year, and the junior doctor pay dispute has not been resolved.

And as the Rwanda plan divides the Tory party, small boat crossings continue.

So, with the general election approaching, how is the prime minister doing on delivering his other promises?

You can see the progress for yourself below.

Last week, the government's plan to grant new oil and gas licences every year passed its latest hurdle in parliament.

The bill's supporters say it will improve energy security, as the UK still relies heavily on fossil fuels. 

Critics, however, argue it will not cut energy bills and instead will break the UK's promise to phase out fossil fuels.

One of those critics is Chris Skidmore ,  who recently resigned as a Conservative MP over the government's net zero strategy and its decision to boost new oil and gas production.

Mr Skidmore, who led the independent government review into net zero, talked about the decision and Rishi Sunak's climate agenda with Sky's science and technology editor Tom Clarke on the latest episode of ClimateCast.

The second reading of the Safety of Rwanda (Asylum and Immigration) Bill has passed the House of Lords and will move on to the next stage.

This came after six hours of debate by peers, and a defeated amendment vote.

An amendment to the Rwanda bill proposed by Liberal Democrat peer Lord German has been defeated in the Lords tonight.

The motion had been designed to block the legislation.

There were 206 not content votes to 84 content, marking a significant government majority of 122.

After a six-hour debate, peers are now voting on a proposed Liberal Democrat amendment to the Safety of Rwanda (Asylum and Immigration) Bill.

This was proposed by Lord German.

By David Blevins, Ireland correspondent

The protesters outside were to be expected but no one anticipated details of the DUP's highly confidential executive meeting to be live tweeted from inside.

The running commentary, shared on X by loyalist activist Jamie Bryson, made a mockery of the party's attempt to debate a potential deal in private.

But his account must be read in the context of him being the arch-opponent of any compromise to restore power-sharing government in Northern Ireland.

"DUP meeting descends into mayhem," he writes. "Jeffrey Donaldson saying texts being sent to Jamie Bryson who is giving a blow-by-blow account of the meeting.

“DUP meeting now halted.  Fury and mayhem.  Jeffrey Donaldson says Police Service of Northern Ireland brought in to try and block phone signals."

The Democratic Unionists have boycotted power-sharing government for two years in protest over the power-Brexit trade border in the Irish Sea, between Great Britain and Northern Ireland.

But their negotiations with the UK government are now understood to have concluded and Party Officers have been weighing up a possible return to Stormont.

Having kept the venue for tonight's meeting secret until the last minute to limit opposition outside, it ironic that the disruption has come from inside.

The surge of Reform UK continues as the party overtakes the Liberal Democrats in the polls for the first time.

The Sky News live poll tracker - collated and updated by our Data and Forensics team - aggregates various surveys to indicate how voters feel about different political parties.

Labour is still sitting comfortably on a roughly 20-point lead, averaging at 44.3% in the polls, and the Tories on 24.6%.

In third is Reform UK on 10.3%, followed by the Lib Dems on 9.8%, the Greens on 6.2%, and the SNP on 3.1%.

See the latest update below - and you can read more about the methodology behind the tracker  here .

By Faye Brown , political reporter

Rishi Sunak kicked off 2024's political season with a hint at when the next general election will be - saying earlier this month that it's his "working assumption" it will happen in the second half of the year.

Speculation has been rife for months about when the prime minister will choose to go to the polls, with some pundits  believing he would call one in May to coincide with the local elections.

UK general elections have to be held no more than five years apart, so the next one must take place by 28 January 2025 at the latest.

This would be five years from the day since the current parliament first met (17 December 2019), plus the time required to run an election campaign.

The phrase  "working assumption"   does give Mr Sunak wriggle room should circumstances change, and he has not ruled out a spring election.

Sky News spoke to pollsters about the factors the prime minister will be weighing up in making his decision - and when they think the election should be.

Kenneth Clarke's mobility is severely limited these days, and he perches on a walking frame when he attends the House of Lords.

But there's nothing slow moving about his intellect, as he demonstrated with a powerful speech opposing the government's Rwanda bill.

His reliance on his walking frame means he doesn't sit with his colleagues from John Major's cabinet on the front bench below the gangway on the government side of the chamber.

Instead, he and his frame sit directly in front of the cross-bench peers, with Lord Clarke facing the wool sack and the throne, rather than facing the opposition benches like the rest of the Tory peers.

But while this physical limitation means there's no opportunity for any theatrics, dramatic arm-waving gestures or the sort of rhetorical flourishes for which he was famed in his pomp in the 1980s and '90s, at 83 he's still a class act.

A grandee and national treasure these days, Lord Clarke said the bill was "a step too far" and it risked moving the UK towards an "elective dictatorship".

And on Rwanda's suitability for deportations, he claimed the country had been ruled by a "one-man dictatorship" and had "a dodgy record on human rights".

Vintage Clarke: the kind of plain speaking for which he has been renowned for decades.

Criticising the government's response to the Supreme Court ruling on the Rwanda policy, he said: "They have decided to bring an act of parliament to overturn a finding of fact made by the supreme court of this country."

And vowing to back amendments to the bill, he said the government had "dug a hole for itself" and it was guilty of  "follies in crashing on" with the policy.

Of course, we're used to listening to Lord Clarke speaking with gusto and flair. 

But he wasn't the only member of John Major's cabinet to attack the bill. And it was someone we might not have expected to speak out so passionately.

Douglas Hogg, now Lord Hailsham, made a fiery speech of the kind we didn't hear from him when he was a cabinet minister in the 1990s (see 20.10 post).

"This bill trashes our reputation for domestic and international probity," he declared angrily.

And on ditching human rights protections, he quoted Martin Niemoller, a German pastor who spoke of his failure to speak out against the Nazis.

He admitted the circumstances were very different from those of the 1930s, but warned "best not to step on to a slippery slope because it can end in some very murky places."

But another member of the Major cabinet and a former home secretary like Lord Clarke, Kenneth Baker, backed the bill, as did Boris Johnson's former chief of staff, Eddie Lister.

Well, it was Eddie's old boss who launched the Rwanda plan back in 2022.

Also backing the bill was Lord Frost, who has been threatened with losing the Tory whip and being dumped from the candidates' list for would-be MPs over claims he's linked to the plotters against Rishi Sunak.

Well, up to a point, he backed it. 

"I think it would certainly have been better if it had been amended to strengthen the exclusion of international law as proposed in the Commons, in my view we will one day have to go there in this area," he said.

"But it's done now, the Commons debated it fully and has now spoken, I support the government in bringing it into force swiftly.”

He appeared to be saying he backed the rebel amendments proposed by Rober Jenrick and Sir Bill Cash in the Commons. On that, he was a lone voice in the House of Lords.

When Tory rebels were plotting against John Major, Ken Clarke famously declared at a Tory conference: "Any enemy of John Major is an enemy of mine."

That was more than 30 years ago. These days, despite his mobility being impaired, Lord Clarke's ability to make a decent speech certainly isn't.

Be the first to get Breaking News

Install the Sky News app for free

travel survey questions


  • Çözümler Sanayi Oyun Otomotiv Spor ve etkinlikler Eğitim Devlet Seyahat ve Konaklama Finansal hizmetler Sağlık hizmeti Esrar Teknoloji Kullanım durumu NPS+ Topluluklar Kitle InsightsHub Anında Yanıtlar Canlı Anketler Üye Deneyimi GDPR Pozitif İnsan Bilimi 360 Geri Bildirim Anketleri
  • Kaynaklar Blog e-Kitaplar Anket Şablonları Durum çalışmaları Eğitim Yardım Merkezi
  • fiyatlandırma

Free Travel Surveys: Questions & Templates

Pick A Template

Select any of our free templates or start building surveys from scratch.

Sign Up For Free

Sign Up for free account and access it right away. No copy/ pasting!

Plug & Play

Customize, Distribute, Collect, and Analyze using your Survey Templates.

List Of Travel Survey Templates

  • Müşteri Memnuniyeti Anketleri
  • Müşterinin Sesi Anketleri
  • Ürün Anketleri
  • Müşteri Değerlendirme Anketleri

Çalışan Deneyimi

  • Çalışan Deneyimi Anketleri
  • İş Memnuniyeti Anketleri
  • Eğitim Değerlendirme Anketleri
  • Yeni Ürün / Konsept Test Anketleri
  • Konferans Geri Bildirim Anketleri
  • Odak Grup İşe Alım Anketleri
  • Donanım Yazılım Anketleri
  • Web Sitesi Anketleri
  • Otel Restoran Anketleri
  • B2B Anketleri
  • Perakende Anketleri
  • Seyahat Anketleri
  • Sağlık Araştırmaları
  • Sigorta Anketleri
  • Sosyal Anketler
  • Psikografik Anketler
  • Kişisel Anketler
  • Eğitim Anketleri
  • Okul Anketleri
  • Üniversite Anketleri

Kar amacı gütmeyen

  • Kar Amacı Gütmeyen Etkinlik Anketleri
  • Kâr Amacı Gütmeyen Kuruluş Anketleri
  • Kar amacı gütmeyen gönüllü Anketler
  • COVID 19 Sağlık Değerlendirme Anketleri
  • COVID 19 Çalışan Değerlendirme Anketleri

Travel Surveys

QuestionPro is the leader in travel surveys. These travel survey template questionnaires are created with extensive direction by travel industry experts, making them highly optimized and geared towards delivering best quality survey responses. The templates include airline service evaluation, travel tour, flight survey and a general travel survey template. You can also use these templates as survey examples or samples. Alternatively, just pick any template of your choice, make your edits if needed and send it directly out to your audience. Get started!

Airline Service Evaluation Survey Template

Airline Service Evaluation Survey Template offers questions to evaluate all aspects of air travel service. In order to improve the flying experience, most airlines prefer taking inputs for questionnaires. This sample contains a list of comprehensive questions that can help in gathering feedback about scope of improvement in terms of time management, food quality or customer service.

Travel Survey Template

Travel Survey Questions template analyzes the purpose of the trip, service problem identification, and overall satisfaction of a travel industry customer. This sample questionnaire has a thorough list of questions that can provide travel organization insights on how to improve the respondent's travel experience. A couple of examples from this questionnaire are "What is the purpose of your travel?", "Describe the nature of your trip," etc. According to the recent study by the Social Toaster, by 2022, it's estimated that total travel gross bookings will hit $440 billion.

Travel Tour Evaluation Survey Template

Travel Tour Evaluation Survey Template offers questions the evaluate the entire tour experience and willingness to recommend. Survey creators can edit the questionnaire sample according to the exact purpose of the survey. For example, asking the customers to provide feedback about the food arrangements during the tour can help the travel company to serve their future customers better.

Airline Flight Survey Template

Airline flight survey template offers questions for aircraft evaluation, flight selection, amenities, and in-flight service. This sample questionnaire can be used to let passengers share their flight experience so that the airlines can gather actionable intelligence. For example, asking the customers to provide feedback on the in-flight arrangements can help the airline serve their future customers better. Use this free airline flight survey template as it is or customize it to suit your needs.

  • Sample questions
  • Sample reports
  • Survey logic
  • Integrations
  • Professional services
  • Survey Software
  • Customer Experience
  • Communities
  • Polls Explore the QuestionPro Poll Software - The World's leading Online Poll Maker & Creator. Create online polls, distribute them using email and multiple other options and start analyzing poll results.
  • Research Edition
  • InsightsHub
  • Survey Templates
  • Case Studies
  • AI in Market Research
  • Quiz Templates
  • Qualtrics Alternative Explore the list of features that QuestionPro has compared to Qualtrics and learn how you can get more, for less.
  • SurveyMonkey Alternative
  • VisionCritical Alternative
  • Medallia Alternative
  • Likert Scale Complete Likert Scale Questions, Examples and Surveys for 5, 7 and 9 point scales. Learn everything about Likert Scale with corresponding example for each question and survey demonstrations.
  • Conjoint Analysis
  • Net Promoter Score (NPS) Learn everything about Net Promoter Score (NPS) and the Net Promoter Question. Get a clear view on the universal Net Promoter Score Formula, how to undertake Net Promoter Score Calculation followed by a simple Net Promoter Score Example.
  • Offline Surveys
  • Customer Satisfaction Surveys
  • Employee Survey Software Employee survey software & tool to create, send and analyze employee surveys. Get real-time analysis for employee satisfaction, engagement, work culture and map your employee experience from onboarding to exit!
  • Market Research Survey Software Real-time, automated and advanced market research survey software & tool to create surveys, collect data and analyze results for actionable market insights.
  • GDPR & EU Compliance
  • Employee Experience
  • Customer Journey
  • Executive Team
  • In the news
  • Testimonials
  • Advisory Board

QuestionPro in your language

  • Encuestas Online
  • Pesquisa Online
  • Umfrage Software
  • برامج للمسح

Awards & certificates

The experience journal.

Find innovative ideas about Experience Management from the experts

  • © 2021 QuestionPro Survey Software | +1 (800) 531 0228
  • Privacy Statement
  • Terms of Use
  • Cookie Settings


  1. FREE 8+ Sample Travel Survey Templates in PDF

    travel survey questions

  2. FREE 7+ Travel Questionnaire Forms in PDF

    travel survey questions

  3. Online Business Travel Survey Template

    travel survey questions

  4. Más de 30 plantillas de cuestionarios y diseños en Microsoft Word (2023)

    travel survey questions

  5. Travel Survey Examples

    travel survey questions

  6. 14+ Travel Questionnaire Templates in Google Docs

    travel survey questions


  1. Travel Survey Questions and Templates

    Is your locale known for its hospitality? Find out what makes a visit to your fair city or town memorable, or what it might take to make yours a preferred travel destination. Packages and promotions. What do your customers want? Flights? Flights and hotels? Flights, hotels, rental cars? What about access to Uber or Lyft drivers?

  2. 42 Informative Travel Survey Questions

    1. What primarily encouraged you to book this trip? 2. Did you travel for leisure, business, or family matters? 3. How important was finding a relaxing destination to you? 4. Did you seek an adventurous or thrill-seeking experience? 5. Were local cultural experiences a significant part of your travel decision? 6.

  3. 9 Tips for Travel Survey Questions

    1. Organize questions by topic and keep it simple When you craft a survey, you should always keep the people taking the survey — your valued customers — in mind. Never use terms or abbreviations they won't completely understand. Keep the wording of your questions simple and straightforward.

  4. Travel survey questions

    Travel Survey Questions template analyzes the purpose of the trip, service problem identification, and overall satisfaction of a travel industry customer. This sample questionnaire has a thorough list of questions that can provide travel organization insights on how to improve the respondent's travel experience.

  5. Vacation survey template

    See Curiosity at Work Find out about people's vacation and travel habits. Use our expert-certified vacation survey template today for FREE and find out where people are going.

  6. Travel Survey Examples

    Learn what a travel survey is, why it is important, and how to conduct one. Download two sample travel surveys in PDF format. See examples of travel survey questions for different purposes and scenarios, such as business, pleasure, or personal travel.

  7. Free Travel Surveys: Questions & Templates

    Travel Surveys

  8. How to Conduct a Travel Survey: A Guide for ...

    When creating questions for your survey, keep your clients in mind and make sure to include questions they are bound to understand. All the words should be present in everyday use and questions should be straightforward. Your survey should be organized logically with a nice flow of questions. After all, your clients are doing you a favor by ...

  9. How to build business travel surveys

    April 29, 2022 By: Liudmila Buga Asking business travelers what they think is an effective way to get the insights you need to build a strong corporate travel program and improve policies already in place. Creating winning business travel survey questions is one of the secrets to optimizing your corporate travel program.

  10. Tourist Survey: What It Is + Tips with Examples

    LEARN ABOUT: Free Travel Surveys: Questions & Templates. Tourist Survey Tips with Examples. If you are going to create a survey for tourists, we recommend: Remember that only some people who will respond to the survey speak the same language as you. With QuestionPro, you can create multilingual surveys for travelers to understand and answer ...

  11. Free Online Travel Survey Questionnaire Template

    A travel survey questionnaire can help travel companies understand the preferences and needs of travelers and identify ways to improve the travel experience. With forms.app, you can create a customized travel survey questionnaire in just a few minutes! ... People will answer a set of questions to see which traveling option is best for them ...

  12. Travel Satisfaction Survey Form Template

    An online travel satisfaction survey is a questionnaire used to evaluate the experiences of travelers. Whether you're a cruise operator, travel agency, airline, or hotel manager, use this free Travel Satisfaction Survey to collect information from your customers, both in the app and on your website! Just customize the survey questions to ...

  13. 14+ Travel Questionnaire Templates in Google Docs

    What Is a Travel Questionnaire? Travel questionnaires vary in content and purpose. While some travel companies make use of these to gather customer feedback, other organizations utilize travel survey questionnaires to collect information about an individual's travel behavior.

  14. Travel Agency Survey Questionnaire & Template

    Travel Agency Survey. The job of a travel agency is to reduce the stress of planning a vacation. Find out how well your agency does in satisfying your customer's expectations. ... By incorporating pricing-related questions in surveys, travel agencies can gauge customer perceptions of value and adjust their pricing strategies accordingly. This ...

  15. Survey Questions: 70+ Survey Question Examples & Survey Types

    15 Jan 2024 Reading time Share This comprehensive intro to survey questions contains over 70 examples of effective questions, an overview of different types of survey questions, and advice on how to word them for maximum effect.

  16. Business travel survey template

    Visit Help Center Check out our expert-certified Business Travel survey template. From sample questions to powerful analytics, we make it easy to get feedback.

  17. A database of travel-related behaviors and attitudes before, during

    The survey questions cover a wide range of topics including commuting, daily travel, air travel, working from home, online learning, shopping, and risk perception, along with attitudinal ...

  18. National Household Travel Survey

    National Household Travel Survey. Conducted by the Federal Highway Administration, the NHTS is the authoritative source on the travel behavior of the American public. It is the only source of national data that allows one to analyze trends in personal and household travel. It includes daily non-commercial travel by all modes, including ...

  19. Travel tour evaluation survey questions + Sample questionnaire template

    Get real-time analysis for employee satisfaction, engagement, work culture and map your employee experience from onboarding to exit! Customer Survey Software Robust, automated and easy to use customer survey software & tool to create surveys, real-time data collection and robust analytics for valuable customer insights.

  20. Travel Survey

    Travel survey is a study of things people prefer during travelling. It includes collecting details of the places that one liked, hotels, travelling methods and expenditure. The most important detail that one provides in a travel survey is the diary of their journey i.e. mode & purpose of travel, start & end location, start & end time, accompaniment etc. Travel surveys are not conducted ...

  21. Travel Questionnaire Form Template

    A travel questionnaire form is used by travel agencies to get to know their customers' preferences, what destinations they want to travel to, the best time of the year to visit these destinations, etc. If you manage a travel agency, you can streamline your customer service with our free Travel Questionnaire Form template.

  22. Travel Planning Questionnaire

    Travel Questionnaire Thank you for taking the time to fill out this online questionnaire to get started planning your travels. The more detailed and thoughtful you can be while filling out the questions, the better I can personalize an itinerary for you. Before filling out the form, please review my planning fees page.

  23. Multidimensional well-being of US households at a fine spatial scale

    Ideally, social scientists would have access to a "comprehensive survey" that employs a large sample size, asks many questions on various topics, is representative of the general population ...

  24. Politics latest: Iceland boss accused of 'sour grapes' over Labour

    Back to the Lords, where a busy afternoon of debate on the Rwanda bill has just entered its fifth hour. This evening, we expect a vote on a proposed amendment, and a shout on the second reading of ...

  25. Free Travel Surveys: Questions & Templates

    Travel Survey Questions template analyzes the purpose of the trip, service problem identification, and overall satisfaction of a travel industry customer. This sample questionnaire has a thorough list of questions that can provide travel organization insights on how to improve the respondent's travel experience. A couple of examples from this ...