Julie Leones, Extension Economist
AZ1056
September 1998
Arizona Cooperative Extension, College of Agriculture
The University of Arizona
Tucson, Arizona 85721
Table of Contents

1. Introduction
2. General Suggestions on Visitor Surveys
3. Steps in Visitor Survey Design
   • Step 1. Deciding on the Feasibility of a Survey
   • Step 2. Setting Objectives
   • Step 3. Deciding on a Survey Methodology
   • Step 4. Writing Questions
   • Step 5. Formatting the Questionnaire
   • Step 6. Writing Instructions for Completing the Questionnaire
   • Step 7. Pretesting the Questionnaire
   • Step 8. Estimating Your Visitor Population and Selecting Your Sample
   • Step 9. Administering the Survey
   • Step 10. Training Interviewers
   • Step 11. Analyzing Visitor Data
   • Step 12. Reporting Visitor Information
4. Conclusions
5. References

Appendices

1. General Questionnaire (PDF file)
   1. Globe-Miami Questionnaire (PDF file)
2. Determining Adequate Sample Size (PDF file)
3. Conducting Interviews (PDF file)

Acknowledgments

This publication was written after reviewing a publication on community survey efforts written by Dr. Mike Woods of Oklahoma State University and Mr. Gerald Hall of the Oklahoma Department of Commerce. Some of the same material covered in their guide is also covered here and I am indebted to them for ideas about organizing and presenting this information and for some of the topical material. However, there are some special aspects of visitor surveys that require additional discussion and that are not covered in their publication. For this reason, I decided to write this publication on surveying visitors.

This bulletin has been reviewed by Professor Mike Woods, Agricultural and Resource Economist, Oklahoma State University, Mr. Douglas Dunn, Cochise County Cooperative Extension Director, University of Arizona Cooperative Extension, Professor Lay Gibson, Geographer and Regional Scientist, The University of Arizona, Dr. Ed Carpenter, Rural Sociologist and Multimedia and Video Coordinator, The University of Arizona. I thank them for their helpful suggestions and comments. However, any remaining errors in this publication are strictly the fault of the author.

I. Introduction

Although most leaders and officials concerned about tourism in Arizona recognize the value of information from visitor surveys, the cost of such survey efforts generally restricts their use to larger metropolitan areas and major rural tourist attractions such as the Grand Canyon. Using powerful statistical methods, survey data can provide very accurate and useful information. However, if the survey is not carefully designed and conducted following certain guidelines, it can provide information that is inaccurate and even misleading.

The purpose of this bulletin is to provide basic information about conducting visitor surveys. It can be used by community leaders who are interested in tourism either to design and conduct their own surveys or as a basis for putting together a request for proposals for others to do such a survey. It can also be used by private consultants, economic and development professionals and others who conduct visitors' surveys.

Through visitor surveys and careful counts of total visitors, communities can better estimate the total expenditures by visitors in their community and can learn much about visitor characteristics and preferences. This information can be used to better promote tourism in the community and to make strategic decisions about community improvements and additions likely to appeal to visitors.

II. General Suggestions on Visitor Surveys

• Make sure that you really need to conduct a survey: they are expensive and time-consuming to do.
• Keep the survey as short and interesting as possible.
• Pretest the survey before starting the actual survey.
• Look at the survey questionnaires that have been used by others to get ideas about questions to ask and ways to phrase them.
• Get advice and comments on your survey instrument and survey design from several different sources. The advice of a statistician or someone with extensive survey research experience can be especially helpful.
• Make sure that you collect data from a random sample of visitors.
• Make sure that your sample size is large enough to give you the accuracy you desire.
• Design your survey so that the data is easy to enter into a computer using programs such as QPL (Questionnaire Programming Language), a public domain software. Working with the data on the computer will save significant amounts of time when it comes to organizing and analyzing the data.
• Consider offering a small incentive or conducting a raffle using the names of respondents to encourage response.
• To monitor for nonresponse bias, collect several basic pieces of information from the visitor when distributing the survey, including the city and state they are from, whether they are in the area on business or for leisure, how many days (and nights) they plan to stay in the area, their level of education.
• Keep a log with information concerning how the survey was conducted such as who helped with the survey; where, how and when it was distributed; important decisions made along the way, etc.
  
  

III. Steps in Visitor Survey Design
 
Step 1. Deciding on Feasibility of a Survey

Because surveys are so expensive to conduct in terms of time and money, you need to be sure that a survey is necessary before you begin. You also need to be sure that you have adequate resources to complete the survey process.

Some questions you need to answer are:

• What is it that we hope to learn from this survey? Is there some other way of answering these questions?
• Are there existing data on visitors and their expenditures in nearby communities or for the county that could be used in place of survey data for our community? Is there a county level or state level study that can provide the information we need?
• If our resources are limited, would we be better off collecting data to improve estimates of total visitors or conducting a visitor survey?
• Over what period of time will the survey run? A year, several months, a few days?
• Who will be responsible for supervising the survey process? How will they be compensated? Who will help them make visitor contacts, conduct surveys or distribute questionnaires?

Although the costs of doing a survey will vary according to many factors, you can expect the cost of each returned questionnaire to run anywhere from $7.00 to $50.00. For a sample of 1,000 visitors, you can expect a cost for collecting the data of least $7,000. Much of this cost is for labor, although depending on the type of survey, postage, paper and telephone costs may also be a factor.

Step 2. Setting Objectives

1. What is the main purpose of the survey? Examples of goals might include:

• to know more about visitor characteristics for marketing purposes.
• to identify the importance or size of tourism activity relative to other economic activity in the region.
• to learn how visitors view our area and what improvements they feel are important. This will help guide our tourism development efforts.
• to find out how important different types of visitors are to our community (for example, business versus leisure visitors, overnight versus day trip visitors, foreign versus out of state versus in state visitors). This can help us choose where to advertise.
• to find out how visitors spend their time and money while in our area.
• to find out what percent of visitors' total trip expenditures are made in our area. This can help us identify potential business opportunities to increase expenditures in our area.
• to identify where else visitors go outside of our area to help us identify who we should collaborate with regionally for marketing, packaging and other purposes.

2. Break big or complex questions down into smaller questions.

• What are the characteristics of visitors?
  • How old are they?
  • What is their educational attainment?
  • What is their income level?
  • What are their hobbies and interests?
  • What have they come to do in the community?

3. Outline how you will use the information you collect.

Step 3. Choosing a Survey Methodology

1. Decide whether it is best to collect the survey through:

1. mail, with or without short contact with visitor at the site
2. self administered survey conducted while in the community
3. interviews in the community
4. interviews over the telephone

Each of these techniques has advantages and disadvantages:

Mail Surveys

The most popular approach to mail surveys is that used by Dillman. It involves sending out one copy of the survey. If no response is received, then a reminder postcard is sent to the respondent. If still no response is received, the respondent is contacted by phone. Finally, a second survey is mailed to the respondent. Alternatively, if the phone number of the respondent is not known, a third copy of the survey may be mailed to the respondent in place of a phone call follow-up. It works best when a comprehensive list of visitors' names and addresses are available so that a random sample can be drawn from this list. Mail surveys also work well when targeting a specific geographic or interest group where lists of total populations exist.

Advantages:

• Least expensive option.
• Can be mailed out to respondents in wide spread locations.
• Doesn't require interviewers.
• Requires fewer workers to administer.
• Gives the respondent time to think about the questions before responding.
• Can potentially allow for input from more than one family member.

Disadvantages:

• Can result in low rates of response.
• Some questions may be misunderstood.
• Requires that a mailing list of all visitors to the community exists, or that a representative list can be compiled.

Self Administered Surveys

This technique requires contacting a visitor and giving them a copy of the survey questionnaire to answer immediately or to complete and mail in later. If the visitor is completing it on site, it has the advantage over a mail survey of allowing them to ask the surveyor about questions that they may not understand. If the survey is to be returned by mail, some of the same techniques used to increase response rate in mail surveys can be used (i.e. follow up postcards and a second copy of the survey). In either case, since a surveyor has contact with the visitor, they can collect key information that will allow assessing the amount of non-response bias from the survey. Such key information includes purpose of trip, time spent in the area, and the origin of the visitor. This technique works best when local front desk employees are helping to distribute the survey or when dealing with large groups of people at specific locations and times (ex., at festivals or events).

Advantages:

• Response rates may be higher than in mail surveys.
• Less expensive than personal interviews or telephone surveys.
• Do not require a list of visitor names and addresses before conducting the survey.

Disadvantages:

• More expensive than mail surveys.
• Possible bias is introduced when visitor is contacted.

Personal Interviews

This technique is feasible in many visitor surveys, especially if they are designed to be answered in a few minutes time. Visitors can be interviewed as they are leaving an attraction or accommodations and it is possible for an interviewer to enter responses directly into a laptop computer. This eliminates the need for data entry at a later time and can reduce the cost of using personal interviews. This method works best when dealing with large volumes of people in a concentrated area. For example, at festivals and events, at hotel checkout lines, airports, border crossings, car rental return lines. Unless people are not in a hurry or are waiting anyway, response rates can be low because visitors at the end of their trip are anxious to leave, to catch flights, or begin driving home.

Advantages:

• Tend to have higher response rates.
• Can ask more complex questions since the interviewer can explain them further.
• Skillful interviewers can pick up information that might be missed through mail or self administered surveys.
• Can enter response directly into a computer saving data input time and reducing potential inputting errors.

Disadvantages:

• The cost will tend to be higher.
• Bias may be introduced by the interviewer.
• It may be difficult to identify locations and times to interview if the volume of visitors to an area is relatively low.
• Visitors may not be at the end of their visit and may have to give expenditure estimates and information on what they plan to do rather than completed expenditures and what they actually did.

Telephone Interviews

This is a popular approach used by market research firms. Since most people do have telephones it is a viable option. It requires a well defined system for randomly selecting numbers to call, and for handling numbers where there is no response or the line is busy. It requires an accurate current listing for the entire population from which visitors come. Clearly, it is not well suited for surveying international visitors. It is least expensive when used to survey in-state visitors. It is a very viable method if you are interested in a specific tourist market only (for example, visitors to your community from the Metro Phoenix area). It is becoming increasingly difficult to reach people by phone because of widespread telemarketing and telephone surveys. Many people use answering machines and caller ID to screen their calls and refuse to pick up the phone if someone they do not know is on the line. Because of this use of answering machines and other devices to screen calls, it is quite difficult to estimate actual non-response rates for this type of survey.

Advantages:

• Relatively easy to supervise staff.
• Information can be collected rapidly.
• Can enter response directly into a computer saving data input time and reducing potential inputting errors.

Disadvantages:

• Requires a reliable database of residential phone numbers.
• Requires a phone bank and substantial expenditure on phone service.
• People who have no phones are excluded.
• People with unlisted numbers may be excluded.
• Some household members are more likely to answer the phone than others and this may bias the results.
• Respondents may be less honest.
• People may use their answering machine or caller ID to screen your call and not respond.
• Actual response rate is difficult to estimate.
• Bias may be introduced by the interviewer.

Step 4. Writing Questions

1. Characteristics of well written questions:

• neutrally worded to avoid bias
• short and concise
• ask only one thing at a time
• easily understood
• require simple responses
• relate to the study objectives
• open ended (when possible)- the respondent writes in the response and does not choose a response from an existing list.
• Use specific, precise wording to avoid double meanings.

2. Organizing the questionnaire:

• Group questions together according to the following:
  • the questions use the same response category (ex. excellent, good, fair, poor).
  • the questions are related in subject matter (ex. group all demographic questions together, all expenditure questions together, all questions relating to characteristics of this trip together, etc.).

3. Tips for writing questions:

• Describe what units of measure are involved (hours, days, number of people, dollars, cents, percentages).
• Indicate the time period referred to, if appropriate (in the last week, month, year, five years).
• Describe the geographic area to be covered (ex. in this town, in this county, in this state).
• Use standard census content and format for demographic questions (i.e. questions about age, income, education, race and ethnicity). This will allow you to compare responses from the survey to census data (for example, to examine the income level of visitors as compared to that of local residents).
• Include 'undecided', 'don't know', 'other' or 'none' as an option in questions as appropriate. This is important so that you can tell these responses from cases where a question was not answered at all.
• If a question provides categories for responses, try to think of all possible responses and then group them. Make sure that the categories needed to meet the objectives of your study are included. For example, if you are trying to identify people who are coming to your community to buy antiques, you may want to ask specifically if they are antiquing.
• Always include 'other (please specify)' as an option when providing categories
• Leave space at the end of the questionnaire for comments. You can often gain valuable insights from these additional comments concerning the questions you have asked and others you have not asked.
• For written questions that ask for dollar amounts or percentages, use dollar signs or percentage signs in the space for the response. If you ask the respondent to provide percentage responses that total to 100%, put 100% below the spaces for all the responses.
• Keep in mind that although open ended questions produce the richest data, these questions are the most costly to enter, code and later analyze.
• Borrow well constructed questions from other surveys when appropriate. Example survey questions are included in Appendix A.

Step 5. Formatting the Questionnaire

1. Suggestions for the questionnaire format are:

• Start with fairly simple and interesting questions. For example, visitors often enjoy talking about where they have been and where else they are going on this trip. Make sure that the most important questions come soon after a few warm up questions so that they are less likely to be affected by respondent fatigue.
• End with easy questions such as demographic questions about age, education and income.
• Keep the survey as short as possible. If it is a mail survey or self administered, keep in mind that people's willingness to respond will be affected by how many pages they see in the survey. Interviews, whether by phone or in person are best kept to under 10 minutes to avoid respondent fatigue.
• Eliminate questions that do not meet your objectives.
• Make sure that there are no redundant questions (i.e. a question that asks basically the same thing as another question).
• Make sure to include a place in the questionnaire for a survey number. You may also want to mark where the survey was distributed and the date and time.
• Set up a coding system to use when entering responses in the computer before you start using the questionnaire. In a survey on paper (not directly entered in the computer), these codes might be included on the questionnaire itself next to each response. A simpler way to handle coding is to use a program like QPL that allows entry of the responses as they appear on the page, and in the case of multiple choice and checklist questions, is coded by the program. This reduces the likelihood of mistakes in entering data of this type.

Step 6. Writing Instructions for Completing the Questionnaire

1. There are five important pieces of information that need to be included in the instructions:

1. the purpose of the survey.
2. who is sponsoring and who is administering the survey.
3. how confidentiality is being protected. Setting up a system to safe guard confidentiality is an important way to increase the likelihood of honest responses.
4. whether or not there will be a follow-up questionnaire.
5. who the respondent can call or write if they have questions, concerns or want a copy of the survey results.

In mail surveys it is important also to:

• indicate when the survey should be returned (i.e. in two weeks or by a certain date).
• personalize the cover letter or instruction sheet.

2. Specific Question Instructions:

• Make sure it is clear which set of instructions accompany which questions by using appropriate numbering, indentation, etc.
• Explain how the question should be answered (i.e. should the respondent write in an amount, check a response, circle a number).
• Indicate whether you want the individual to respond for him or herself or for the entire visitor party.

Step 7. Pretesting the Questionnaire

• You may want to have several knowledgeable people critique your questionnaire before you pretest it.
• To pretest the survey choose a small sample of people who are representative of the type of people you will be surveying. For example, you might want to include people of different ages, different marital status, with or without children, different income and education levels, different race or ethnicity, depending on the factors that you think will affect an individual's ability and willingness to respond to the survey.
• Encourage pretest respondents to make comments on each of the questions, on the order and format of the questions, or on the nature of the questionnaire. Sometimes this is best done with a phone call to the individual after they have filled out the questionnaire, but before they have returned it. Having several people complete the survey at once and then discussing it in a small group setting can also provide useful insights.
• Take notice of any questions that are frequently unanswered or are frequently misunderstood and answered incorrectly. These questions may be poorly worded, in the wrong place in the questionnaire, too difficult to answer, or too sensitive to answer.
• If possible, time how long it takes for respondents to answer the questionnaire. One indication that a mail survey is too long is if many of the respondents don't complete the entire survey.

Step 8. Estimating Your Visitor Population and Selecting Your Sample

1. Sources of errors in surveys

Sample selection is one of the most important determinants of the quality of information you collect through your survey. Failing to select a random sample can result in high levels of coverage error and biased results. Coverage error results from not randomly sampling from the entire population. In visitor surveys, one common type of coverage error is caused by surveying visitors during only a portion of the year and then estimating annual totals from these surveys. Coverage error can also be caused by not sampling from certain sub-populations, such as visitors who stay in small hotels or motels, visitors who do not visit area attractions, or day trip visitors.

Every survey (unless it is actually a census- a survey of the total population) will also have sampling error. Sampling error can generally be reduced by increasing sample size. There are two other important sources of error in survey research. One critical source is measurement error, when the answers on a survey are incomplete, imprecise or inaccurate. This type of error may be caused by poorly worded questionnaires, poorly trained interviewers (who don't understand the questions) or by error or deception on the part of respondents. Some surveys build in consistency checks as one way of dealing with measurement error once surveys have been collected. For example, they may compare expenditures on fees and admissions to the specific attractions that visitors claim to have visited. They may also decide to omit outliers, or observations that are much different than the average.

A final serious source of error is from nonresponse. This is especially important if certain sub-sets of the population are likely not to respond (for example, less educated or lower income visitors or alternatively, very wealthy visitors). The best way to reduce nonresponse error is to use methods to increase the response rate as much as possible (such as the Dillman approach to mail surveys: see description under step 3). If the names, addresses and phone numbers of non-respondents are available, it may be possible to follow up with at least a sample of non-respondents to check to see if key characteristics such as income, education and expenditure levels are similar to those of respondents.

2. Selecting a sample

For most statistical analysis, it is necessary to select a random sample. A sample is just a collection of people from the general population that you wish to learn something about. A sample for a visitor survey generally needs to be a randomly selected group from all of the visitors to a site, community, county or other geographic area. By random, we mean that selection is designed so that every member of the population has an equal chance of being selected.

Sample selection becomes even more important when collecting expenditure data from visitors. There is wide variation in the total amount and the per person per day amount that visitors spend. Because of this variability, larger sample sizes are required. If sample selection is biased towards visitors who spent a lot, then total estimates of visitor expenditures will overestimate the total amount that visitors spent in an area.

Sample selection is difficult in visitor studies. If you sample only from people visiting major attractions, you miss people who do not visit attractions. If you sample only from commercial accommodations, you miss visitors to private homes and day trip visitors. If you sample from visitors who left their address at the visitors' center, you are sampling from a distinct subset of visitors in the community. Often sampling from a mix of locations (attractions, accommodations, gift shops, malls, restaurants, airports) may help reduce the risk of biasing your sample if you are interested in a sample of all visitors to your community. Alternatively, you may decide that you want information only on visitors staying at hotels and motels in town. This simplifies your sampling strategy, but still presents challenges. Most communities offer a range of lodging alternatives and you will want to be surveying visitors from a variety of different types of accommodations (i.e., resorts, bed and breakfasts, budget hotels, airport hotels, RV parks and campgrounds, etc.).

To avoid increasing the probability that people who stay longer are more likely to receive a survey, be consistent about when you contact visitors. For example, give surveys to visitors either as they check in or as they check out. Do the same when sampling at campgrounds and RV parks.

3. Estimating total visitor population

Identifying your visitor population is a major challenge in visitor surveys. Visitor populations are easiest to identify in the case of attractions that have a limited number of entrances where the visitor either pays an entrance fee or has to go through a turnstile or drive over a traffic counter to get into the attraction. Visitor populations are extremely difficult to identify and estimate in areas where there are many entrances into the area, no entrance fees to pay, no turnstiles or traffic counters.

Traffic counters are used extensively in studies of visitors but are limited in their reliability because:

• They detect the number of vehicles not number of people, thus an estimate of the average number of people per vehicle must be calculated from survey information.
• They can not tell between a visitor and a local. In areas where local traffic is also heavy, it can be hard to estimate what percent of the total traffic is due to visitors. License plate counts can be misleading because even if you are interested in only out of state visitors, you will miss many of the out of state visitors who are driving unmarked rental cars with in-state license plates.
• They can be run over several times by the same visitor during a stay.

Door-counters are also used effectively to keep track of people visiting attractions or retail establishments, but also suffer from the following limitations:

• More than one person may come through a door at a time.
• If the same door is used to enter and exit, then the door count number must be cut in half.
• A door counter will not indicate whether a visitor or a local has come through the door. This information will have to be collected through other means. Similar to the periodic license plate counts, you may want someone to keep track on random days how many people come through the door each time it is opened and how many of the people coming in are local and non-local.

When available, the following types of data can be very useful in either estimating total numbers of visitors or for benchmarking estimates from the survey combined with visitor population counts:

• Average lodging occupancy rates and total number of commercial rooms. These figures tend to be more useful in communities where the percentage of overnight visitors is large compared to day trip visitors.
• State and local revenue data for specific establishments such as commercial lodging and restaurants and bars. In Tucson, Arizona, there is a city tax of one dollar per room night that is extremely useful in estimating total overnight visitor numbers. In addition, there is the standard percentage of total room rate tax. Some states and communities give tax breaks to visitors who spend a longer time (such as more than 30 days) in commercial accommodations. If your community does waive bed taxes for long term visitors and you have a lot of long term visitors in commercial accommodations, then you need to make sure that you don't underestimate visitor numbers and expenditure figures based on bed and room night taxes.
• Traffic count data from a typical and an event day and police crowd estimates can be useful if the survey is of visitors in the community for an event.
• Visitor counts from paid attractions and/or attractions with turnstile counters are especially valuable. However, if survey respondents go to more than one attraction while in the area, this must be adjusted for in estimating total visitors to the area. This is why asking about local attractions visited is a key question on any visitor survey.

Step 9. Administering the Survey

1. Administering the survey involves:

• Making key decisions at the outset concerning when to survey, where to survey and how long to survey.
• Managing interviewers and any businesses or other organizations that are cooperating in the survey.
• Managing databases of respondent information.
• Arranging follow up calls and mailings.
• Selecting a data inputting system and managing survey data input.

2. Deciding when to survey:

• Because there are usually pronounced seasonal differences in visitors to an area or community, it is best to try to contact visitors over the course of a year.
• Tourism seasons are not necessarily the four seasons. Some communities have only a high and a low season. Others have high, low and shoulder seasons, yet others have high, shoulder, low, and second shoulder seasons.
• These seasons may vary significantly in length and in terms of the number of visitors who are in the community. It may actually take more labor to reach low season visitors than high season visitors because there are fewer of them to contact.
• Shoulder seasons exist where numbers of visitors taper off or increase slowly rather than dropping off or rising suddenly from one month to the next.

3. Deciding who should distribute or administer the survey:

• Because of the high level of interviewer fatigue from running a survey daily for a year, you may want to consider randomly selecting a three to six week period (depending on the length of the season and the ease of contacting visitors) in each season to interview intensively. This works best if you are using staff who are employed doing other things the rest of the time. It could prove costly in terms of training time if you have to hire different surveyors for each of these survey periods. If there were unusual weather patterns or other events during the survey periods, it could bias your data.
• Doing visitor surveys frequently requires the cooperation of local businesses and attractions. In some communities, local businesses and attractions have been enlisted to give surveys to their customers or visitors. However, without compensation, the people involved often tire quickly of giving surveys to customers and of keeping track of the number of acceptances and refusals. You could use a similar strategy as mentioned above and only survey visitors over short periods during each season.
• Local businesses and attractions may allow a paid interviewer to contact their customers or guests on site. This requires good coordination between the staff at the business or attraction and the interviewers in arranging dates and times to contact guests.
• Using volunteers to survey works best if done for short periods of time during the year, rather than continuously.

4. Respondent information to track:

• Lists of respondents (with their addresses and phone numbers) and when they returned their survey.
• Total distributed, returned and completed surveys, possibly by where and when the visitor was contacted to complete the survey. This information is useful to monitor whether you have collected enough surveys by season and by type of location.
• Respondent acceptance and refusal of the survey. There are several types of refusal that should be noted: is the refusal due to language (the visitor is not fluent in English if this is the only language in which the survey is available), is the refusal due to the respondent being a local not a visitor or is the refusal simply because they do not want to complete the survey. If you know that there are a significant number of foreign visitors to your area, it would help to have versions of your survey available in other languages, particularly German, Spanish, French and Japanese. However, translating the surveys well can be expensive as can translating the response received to the survey.

5. Arranging for follow-up calls and mailings:

• If possible, survey visitors as they leave the community or very soon after leaving the area, so that visitors will be better able to recall details of their visit. Likewise, if mail back surveys are not returned within a few weeks, it is important to send out a reminder postcard or a second survey right away.
• Set up a routine system of sending out reminder postcards and additional copies of the survey if a mail back questionnaire is used.
• Keep records of when postcards, additional questionnaires or phone calls were made.

6. Selecting a data input system and managing data input:

• Select a system before beginning the survey. A variety of software programs are available for this purpose. A free U.S. General Accounting Office (GAO) program called Questionnaire Programming Language (QPL) is available on the Web at http://www.gao.gov/qpl/qpl.htm.
• Test the selected system with pretest data. Both enter and do some analysis with the pretest data.
• Enter data as it is collected when possible. This will make any serious problems with the questionnaire apparent early.
• If this is a year long study, consider doing an interim report so that you start the task of editing and analyzing the data early and can catch problems with the data early. For example, you may discover that one key question was left off of the questionnaire. If the survey is still going on, it may be possible to revise the questionnaire.
• If possible, have the data entered twice and checked. Or have one person enter and another person check the entered data against the original questionnaire. This is not possible for data that are entered directly into a computer from a face to face or telephone interview.

Step 10. Training Interviewers

1. In hiring interviewers look for candidates who:

• Have good communication skills: enunciate well, use language appropriate for interviewing visitors.
• Have good interpersonal relations skills.
• Are socially mature.
• Are friendly and outgoing.
• Are good at keeping conversations on track.
• Are good independent workers with a strong work ethic.
• Are able and willing to work irregular hours (such as evenings or weekends).
• Are comfortable using computer programs for data entry and record keeping if this will be part of their work.

2. In training interviewers make sure that you:

• Explain the objectives of the study and what the main questions are that you wish to answer.
• Go through the survey instrument thoroughly.
• Have the interviewer practice interviewing you and other interviewers before interviewing a visitor.
• Train interviewers in the use of the data entry program you are using and have written instructions on how to use the program.
• Show interviewers how to save data files and help them understand how and why to make back up copies of data files.
• Teach interviewers about the different components of an interview.
• Explain to interviewers that how they ask questions will affect response.
• Train them in good interviewing technique (see appendix C for an introduction to interviewing).
• Show the interviewers what records they must keep and why these are important

3. In supervising interviewers make sure that you:

• Meet with interviewers regularly.
• Collect and check their data files regularly.
• Monitor them at work periodically.
• Encourage them to find ways to do the survey more efficiently or effectively.
• Praise and reward them for good work.
• Warn them and then help them if they are having trouble doing the work involved.
• Give them some flexibility with regard to the days and hours that they work.

Step 11. Inputting and Analyzing Visitor Data

At this stage, you find the answers to the questions that made you decide to conduct a survey. However, this is not always easy in a visitor survey because of the diversity of visitors that may be in your area. The process involved in analyzing data includes:

• Having data entered as it is collected and then compiling it and checking it for any obvious errors will save time later during data analysis.
• Editing the data to make sure that missing responses are properly coded, expenditure data makes sense on a per person per day basis, there are no easily detectable errors in data entry.
• Deciding how to handle outliers such as visitors who stay unusually long periods of time or make single very large expenditures (for a major piece of art, a car, an RV, repairs on a vehicle, etc.). Generally, these observations should be handled separately, or if they are a very small percent of the total, removed from the data set. For example, in a study of visitors to Tucson, Arizona, one visitor purchased an $ 80,000 recreational vehicle. Including this visitor's purchase increased average expenditures by over 100%. Since RV dealers were reluctant to provide information concerning how many RVs they sell to out of town visitors, it was decided that this observation should not be included when estimating average expenditures.
• Deciding on weighting and summarizing scheme. For example, if business visitors are under represented in the sample, should they be weighted? Should the data be summarized by season? Should it be summarized separately for distinct types of visitors such as day trip and over night visitors, leisure and business visitors? By separating data into subgroups, the measure of error on each group increases because you have fewer observations.
• Summarizing and exploring data using a statistical package or statistical options in a spreadsheet using such statistics as:
  • means and standard deviations
    • The mean or average is the sum of all values given for a variable (such as age) across all surveys divided by the number of observations or surveys that provided a value for the variable. The standard deviation is a measure of how much variation there is around the variable mean. It is equal to the square root of the squared difference between the variable given in each survey from the survey mean summed across all surveys and divided by one less than the total number of surveys providing data on this variable.
  • cross tabulations and matrices
    • Cross tabulations and matrices are ways of organizing data in tables with rows and columns. For example, you might want a table that shows the percent of all visitors, of business visitors and of leisure visitors by origin.
  • t-tests, F-tests, Chi squares, etc.
    • These are all statistical tests that are used to determine if there are significant differences between groups of data. T-tests and F-tests are used for continuous data and chi square is typically used for categorical data (ex. yes or no responses). For example, we may want to know if there is a statistically significant difference between the income of day trip and overnight visitors. A paired t-test can help us determine this. Alternatively, we might want to know if there is a statistically significant difference between the percent of business and leisure visitors who play golf during their stay in the community. A chi-square test can help determine this.
  • regression analysis
    • Is another statistical analysis used to determine if variables are correlated or may affect the value of other variables. For example, regressions on expenditures are often used to determine if visitor characteristics such as age, income, distance from home, interests or hobbies, purpose of trip affect how much they spend while on a trip.
• Checking the summarized data to see that it makes sense. Do percentages total to 100 percent? Are some of the standard deviations unusually large, and if so, did you miss an outlier for that variable?
• Examining the summarized data to look for patterns and answers to key questions and to begin writing up the results of the study.

If you do not have the in-house capability to handle the analysis of the survey data, ask for help early on. Most individuals who analyze data prefer to be involved right from the survey design and sampling stage so that they know how the data were collected. A variety of organizations may be able to offer help including Universities, Community Colleges, state, federal and local agencies, EDA development centers, economic development staff affiliated with your local utilities, and non-profit organizations.

Step 12. Reporting Visitor Information

• Choose a variety of ways of reporting the results:
  • formal report
  • executive summary of report
  • press releases
  • talks and slide presentations
  • workshops and seminars
  • documents on your Website
• Consider including the following in your report:
  • an executive summary with the highlights of the study
  • an introduction
  • the results presented in text, tables, and graphs
  • conclusions and recommendations
  • an appendix with sampling methodology and analytical methods
  • an appendix with a copy of the survey instrument
• Some tips for how to present information in the report are:
  • Use graphics if they make the data easier to interpret.
  • Use tables to organize data.
  • Compare survey data to other data. For example, you might compare visitor income to local income levels, or the age distribution of visitors compared to the U.S. age distribution.
  • Compare the information between different types of visitors in your study. For example, between day trip and overnight visitors or between business and leisure visitors.
  • Use a recommendation section to interpret the results of your survey and share what you think are the implications for the community and for local businesses. For example, if visitors consistently rated customer service in the community as low, there may be a need for an intensive hospitality training effort in the community. If visitors are not in town during the times that shops are open, perhaps shops need to reconsider their hours of operation if visitors are important customers.
• Make enough copies of this report to share or sell to not only people and businesses currently interested in tourism in the community, but also for individuals who may be inquiring about opportunities for tourism related businesses in the area and for key contacts outside the community (the state office of tourism, university recreation and tourism departments, non-profit tourism associations).
• Enlist the help of your local media people to get the word out to the general community. The individuals and businesses affected by tourism are often widespread in a community and the mass media may be a good way of reaching them, particularly if the survey covers a large area or is conducted in an area with a large population. Also, consider breaking out information from the report into small chunks for press releases. This way you can inform people without overwhelming them with too much detail at once.
• Consider preparing a set of slides with information from the survey that you or others can use in presentations related to tourism in your area. Interspersing survey results in graphic and bullet form with pictures of the area and major attractions can enhance the appearance of your presentation and remind people in your audience of why visitors come to your area.

IV. Conclusions

In this bulletin, twelve steps to conducting a visitor survey have been described in some detail. The ideas for this bulletin come from various published sources listed in the reference section and from the author's experience. The purpose of the bulletin is not to be a bible for visitor surveys, but rather to give the reader a quick overview of the survey process and to provide tips for conducting an effective and useful survey. The author recommends that if you are planning a visitor survey, that you seek out people with experience and training in survey design and implementation to advise you. If you are planning to hire someone else to do a survey, this bulletin will give you an idea of what steps they will need to complete to be successful.

Several appendices were mentioned earlier in the bulletin. These include Appendix A with two example visitor surveys, Appendix B with information on calculating the necessary sample size to get accurate results from your survey, and Appendix C with interviewing tips.

References:

Dillman, Don A. Mail and Phone Surveys: The Total Design Method. John Wiley & Sons, New York, NY. 1978.

Mendenhall, William and James E. Reinmuth. Statistics for Management and Economics. Fourth Edition. Duxbury Press, Boston, MA. Chapter 15. 1971.

Ritchie, J.R. Brent and Charles R. Goeldner. Travel, Tourism, and Hospitality Research. A Handbook for Managers and Researchers. Second Edition. John Wiley & Sons, Inc., New York. Chapter 12. 1994.

Salant, Priscilla and Don A. Dillman. How to Conduct Your Own Survey. John Wiley and Sons, New York, NY. 1994.

Snedecor, George W. and William G. Cochran. Statistical Methods. Seventh Edition. The Iowa State University Press, Ames, IA. 1982.

Woods, Mike D. and Gerald Hall. A Guide for Local Community Survey Efforts. Circular E-895. Cooperative Extension Service, Division of Agriculture. Oklahoma State University, Stillwater, OK. 1990.

---

Issued in furtherance of Cooperative Extension work, acts of May 8 and June 30, 1914, in cooperation with the U.S. Department of Agriculture, James A. Christenson, Director, Cooperative Extension, College of Agriculture, The University of Arizona.

The University of Arizona College of Agriculture is an equal opportunity employer authorized to provide research, educational information and other services only to individuals and institutions that function without regard to sex, race, religion, color, national origin, age, Vietnam Era Veteran's status, or disability

This document located at http://ag.arizona.edu/arec/pubs/guidevisitorsurveys/az1056.html