Editor's note: Gayle Kaplan is senior research associate with National Information Services (NIS), Minneapolis. NIS is a division of National Computer Systems, Inc.

Whether you're collecting data on your own for the first time or you're an old pro and have hired a professional marketing research firm, the validity of your research is highly dependent on the thoroughness of your pre-research work. This article offers 15 tips and techniques to think about before beginning your research to obtain useful and valid information.

1. A survey is not a crystal ball.

A survey gives you better information to help you make business decisions. When you do a survey (or other marketing research), you are checking with your marketplace before you decide to manufacture a new product, choose an advertising strategy or set up a customer satisfaction program. The information you receive from the survey allows you to understand your market’s requirements and opinions. Through survey research, you are better equipped to make the right decisions the first time. However, it is important to remember that survey results do not predict the future; the economy may improve, a competitor may issue a new product, a hurricane may hit, or, a Coca-Cola showed with the new Coke, you might not ask all the right questions.

2. GIGO: Garbage in, garbage out.

Take time to think about what you want. Spend time and effort to plan your survey. Bad information is worse than no information at all. By the time you decide you need information, you usually need it immediately. But the time and effort you take up front to plan your survey ensures that you obtain good information to help you make your business decisions.

3. Set up a survey design team.

Involve employees who will use or be affected by the survey information in planning the survey. There are two reasons to do so. First, employees who will use the survey results are probably quite knowledgeable about either the subject or the customers you plan to survey. They can prove valuable information in determining the scope of the survey and in designing the actual survey instrument. Second, survey information is no good unless people use it. One way to encourage company employees to use the results is to involve them from the beginning and get their buy-in to the project.

4. Choose the survey methodology that meets your information needs and budget.

Use the research methods that best give you the information you want. Don't choose a method because you have always done it that way. If you want to ask a lot of probing "why" questions, think about using in-depth personal interviews or focus groups. If you want to survey all your customers, consider a direct mail survey instead of telephone. If you want to know how your company compares to its competitors, try perceptual mapping. Budget is always a constraint, but you can certainly acquire useful information with a limited budget. Budget constraints make it all the more important to consider different options as to how to obtain the information you want.

5. Sampling.

Surveying a sample instead of your entire population allows you to obtain valid information at a reasonable cost. However, if you want information that is statistically valid and reliable, you must choose a statistically valid sample. The two key issues for sampling are randomness and size. Randomness means that each person in your desired survey population has an equal chance of being chosen for the survey. For size, you need to decide with what level of confidence you are comfortable: 99%, 95%, 90%, etc. 95% is the usual business choice.

6. Decide at the beginning what breakdowns or groups you want to use to analyze your data.

The number and type of breakdowns you want determine the number and sizes of samples you need for statistically valid data. For example, if you want to analyze results by region, you need to stratify your sample so that you will obtain a sufficient number of response for each region. In addition, determining the breakdowns helps you to decide what kind of specific information you want to include in the questionnaire.

7. Keep the words simple.

In designing the questionnaire content, a good rule of thumb is to use words that are one or two syllables in length. Obviously, there are longer words that almost everyone will understand, e.g., "presentation," but, in general, shorter words are better. Keep the use of technical or industry terms to a minimum. When you do use a technical or industry term, give a brief definition of it; do not assume that everyone you are surveying understands the term.

8. Focus each question on one issue.

If you ask respondents to agree with the statement that "customer service representatives are polite, friendly, and helpful," what does the resulting data mean? If your respondents agree, are they agreeing that CSRs are polite or friendly or helpful or some combination of the three? You do not receive precise and actionable data from one question that deals with two or three issues. Focus your question to obtain information you can use to make decisions.

9. Design questions that give you interval or ratio data where possible so you can do further statistical analysis later.

In data analysis, there are four types of data: nominal, ordinal, interval, and ratio. Interval and ratio data allow you to do more sophisticated statistical analysis, such as a regression analysis. Even though you may not want to do such analysis immediately, it may be useful later to have the right type of data.

10. Learn what statistical analysis can do for you.

Many people do not think of doing statistical analysis because they have forgotten all their math, or they never learned what probability meant, or they think statistics does not provide any useful information. Statistics, as with almost anything, can be misused. But, properly used, statistical analysis is a powerful tool to provide you with valuable information you will not obtain from simple tables. You do not have to become a statistician to learn what statistics can do for you. A good researcher can explain the different techniques in terms of the information you will receive.

11. Make the questionnaire look good.

You may have a wonderful questionnaire that asks all the right questions, but if it looks crowded or the type is very small, most of your respondents will simply not answer it. To make a questionnaire look good, consider such elements as type size (not too small), type font (a fancy font can be difficult for some people to read), sufficient white space, color, and artwork. Using three colors (e.g., red and blue with black type) is an excellent way to present a questionnaire that looks interesting to the respondent. Artwork also gets peoples' attention and piques their interest in reading and filling out the questionnaire. In addition, try to use different type fonts, color shading and graphic symbols to emphasize and supplement your written instructions. A visual cue helps respondents to follow your directions.

12. For telephone surveys, interview the respondent when it is convenient for them.

It will come as no surprise that many people see telephone surveys as intrusive. One method for dealing with this issue is to first introduce yourself and the survey and then ask to make an appointment to conduct the interview at their convenience. If you don't ask to do the survey immediately, you don't put pressure on people. Some people will just tell you to go ahead; most of the others will make an appointment. This technique is especially useful in business-to-business research.

13. For a mail survey, use a reminder post card, additional waves or incentives to increase response rate.

Response rate is an especially critical issue in direct mail surveys because you have no control over how many returns you will get. There are different techniques that encourage people to fill out and return their questionnaires. Research done on these techniques has shown that each may increase response rate from five to fifteen percent. One simple and fairly inexpensive method is to send everyone a reminder post card. The post card is sent to each respondent a week after the survey is mailed. It is not personalized. It simply reminds the respondents that you sent them a questionnaire, that their input is very valuable to you and you hope they will answer and return the questionnaire. You may put a telephone number on the card for people to call if they need another copy of the questionnaire. The best, but also most expensive, method for obtaining a high response rate is to just keep mailing out the survey. If you want to mail the survey to non-respondents only, you need to set up a system to track those people who do respond and match them against your survey population list. A third way to increase response rates is to use incentives. Sending a dollar bill out with each questionnaire is expensive and less effective than it used to be. Instead, you can offer a gift to each respondent or you can offer them a chance to enter a sweepstakes with a prize such as a personal computer. If you choose the gift or prize option, you still want to maintain respondents' anonymity. Include a separate return post card for respondents to send back to receive their gift or enter the prize drawing.

14. Pre-test your questionnaire.

This is one of the most important items on the list. Pretesting a questionnaire involves giving it to a small group of people who are similar to your survey respondents and asking for their feedback on the questionnaire. The pre-test is vital as an assessment of how well potential respondents can understand and respond to the questionnaire. It allows you to check for ambiguity, clarity, and bias in the directions as well as the actual questions. The pre-test will allow for fine-tuning of the questionnaire to ensure accurate and complete response as well as question validity. Even after you have designed a good questionnaire, it is still useful to pre-test it as a final check before you do the survey.

15. Maintain anonymity and confidentiality for your respondents.

People today are very concerned about privacy. They want to know that their answers will be kept anonymous and confidential. This issue is especially sensitive in employee and internal customer surveys. In your cover letter and on the questionnaire itself, let respondents know that you will keep their results confidential and anonymous. If you need to track individual respondents for some reason, you can still assure your respondents of confidentiality. Set up a system to ensure confidentiality and make sure that the system is implemented.