Content

• Steps
• Part 1 of 3: Develop Questions
• Part 2 of 3: Conduct a survey
• Part 3 of 3: Adjust your survey
• Tips

If a company, non-profit organization, or politician needs to know what project participants, constituents, or clients think, they create and conduct a survey. The results can lead to image changes, decision making and policy changes if the comments and suggestions are substantiated. Creating a survey may seem like a very simple task, but if not properly designed, the results can be skewed and unreliable.

Steps

Part 1 of 3: Develop Questions

1. 1 Decide what you want to know by taking a survey. Ask yourself what data you get and how you use it. This will help you formulate useful questions and the order in which you will ask them. Ideally, the survey should be short, so decide which goals are significant and which may not be necessary.
2. 2 Plan questions to help you get the information you need. Start with a large number of questions, and then narrow the list down until each one is relevant to your goal in a certain way. Keep your questions and answers simple and use as few words as possible if possible. You can ask both questions requiring a detailed answer, and questions that allow a monosyllabic answer.
3. 3 Use non-detailed questions to get specific answers. These questions have a set of alternatives that respondents can choose from. These can be questions that can be answered “yes” or “no”, questions for true or false statements, or questions in which respondents must agree or disagree. Monosyllabic questions may look the same as long-range questions, but they will have only a few answer choices for respondents. Such questions might look like this:
   • "Have you shopped here before?"
   • "If so, how often do you buy something here?" (with such questions, the respondents will have several given answers, from which they will choose - from "once a week" to "once a month", for example).
   • "How satisfied are you with our services today?" (just as in the previous version, there will be a limited number of answers to this question - from "I liked it very much" to "I did not like it very much").
   • "Would you recommend this store to a friend?"
4. 4 Use open-ended questions to get comments and suggestions. Open-ended questions offer answers that you cannot foresee, and you do not have a list of answers to choose from. Open-ended questions are a chance for respondents to communicate their special requirements or expectations. Here are examples of such questions:
   • "How will you use your purchase?"
   • "Where else do you shop?"
   • "Who recommended this store to you?"
   • Such questions may well clarify the previous answer - "Why do you feel this way?"
5. 5 Ask questions in an understandable manner and so that it is impossible to give an evasive answer. Do not push for an answer, as this means that the questioner is waiting for a definite answer, and this, in turn, will limit the options that the respondents could use. Either ask the possible answers, or change the structure of the question so that it does not turn out that you are leading the respondent to a specific answer.
   • You can ask the same question in different ways, which can reduce the overall bias of the respondents and give you a chance to know what the person really thinks about the topic.
   • Questions should be structured so that everything is perfectly clear. Confused respondents will give you the wrong data, so the questions should be as clear as possible. Avoid double negatives, unnecessary phrases, or obscure subject and object agreements.

Part 2 of 3: Conduct a survey

1. 1 Think about how to conduct your survey. There are many options. You can use the online service to design a survey. Then you can send out the links to the survey by email. You can interview respondents by email or phone. Or conduct the survey in person with professionals or volunteers.
2. 2 Design your survey based on how you conduct it. Each method has pros and cons, as well as limitations on what you can do. Ask yourself which method of conducting the survey best suits the subject of the survey and how you would best collect the data. For example:
   • Surveys conducted by computer, telephone, mail target a wide range of people, while in-person surveys are time intensive and limit the number of people who take part (which can be useful).
   • Surveys that are conducted by computer, by mail and in person may involve the use of pictures, while telephone surveys do not.
   • Respondents may hesitate to answer some of the questions over the phone and in person. Decide if you want to clarify the questions if the respondent does not understand something; only interviews that are conducted by a real person can include clarifications.
   • A computer survey assumes that respondents have access to a computer. If the survey is personal, it is best to conduct it using a computer.
3. 3 Consider the order of the questions. The form of your survey is as important as the content. You should organize the order of the questions so that it is logical and that there are sequential transitions from section to section. Other types of questions can affect how the respondent completes the questionnaire.
   • You should organize the questions so that if the person answers yes or no to a question, they skip further questions that are not relevant to them. This will help the respondent focus and the survey will take less time.
   • Determinants are questions that lead respondents away from completing other questions. Place them at the beginning of the survey.
   • If demographics are the most important thing in your survey, put questions related to that at the top.
   • Leave personal or difficult questions at the end of the survey. Respondents will not be overwhelmed by them and are more likely to be more open and honest.
4. 4 Decide if you intend to use incentives when completing the survey. It is always easier to attract respondents if you offer them something in exchange for their time. Online, mail, or telephone surveys may assume a coupon is available after the survey is completed. The survey in person may suggest some goods in exchange for participation. Polls are also a good way to draw attention to newsletters or membership offers that might otherwise go unnoticed by respondents.
5. 5 Test your survey before starting your research. Friends, employees, and family members are great subjects. You can ask them to test the survey while it is still in development, or they can try out the final version.
   • Ask those taking the survey for suggestions and comments. They can point you to the sections that confused them. The respondent's impressions of the survey are as important as the survey itself.
   • After testing, work with spreadsheets to make sure you are collecting the data you need. If you are having trouble getting the information you want, redesign the survey. You may need to rephrase some of the questions, add an introduction, reorganize, add or remove questions to make the survey fit your purpose.

Part 3 of 3: Adjust your survey

1. 1 Review the data to understand what your survey really was about. Remember that polls are usually part of a larger campaign. They can be modified and used many times to get different demographics, ask different questions, or better define goals. After analyzing the results, you may find that while your questions are meaningful, they are not quite right for your purpose.
   • For example, you may find that a question such as "How often do you shop here?" restricts your respondents to those who shop at point of sale. If you want to see how people buy a particular product, you can expand the questions to include online shopping.
   • Your implementation method can also be limited to data. For example, surveys that are conducted online usually include respondents with above-average computer skills.
2. 2 Analyze the questions further. Some of your questions will work during testing, but will not work for the survey itself. Your questions should be clear to the specific social group you are targeting. Check if your respondents understand the questions well enough, or if your survey is so standardized that respondents don't even have to think.
   • For example, a question such as "Why are you shopping here?" may be an overly broad question that may confuse respondents. If you want to know if the decor in a store has an impact on the number of purchases that are made at it, you can ask respondents to describe how much they like the decor and decor in the store.
3. 3 Check out the long-range questions. Consider if these questions are working right for you. They can be too open, in which case the respondents will respond incoherently. Or they may not be sufficiently open, in which case the data obtained will not be very valuable. Ask yourself what role these questions play in your survey and redesign them appropriately.
   • As stated above, advanced questions such as "How do you feel shopping here?" may not give respondents the right direction. Instead, you can ask, for example, another question: "Would you recommend this store to your friends? Why? Why not?"
4. 4 Decide how you will respond to the missing information. Not all respondents will answer all questions that may or may not be problematic for you. Ask yourself which questions were missed or which questions were incomplete, if any. This could be due to the order of the questions, the wording of the questions, or their subject matter. If the missing information is important, try reformulating the missing questions to make them more or less specific.
5. 5 Analyze what suggestions and comments you receive. See if there are any unusual directions in your data and determine if this is real or due to flaws in the survey. For example, your closed-ended questions will restrict respondents from the information they can give you. Your answers may be so limited that a strong opinion looks the same as a weak one, or you may have an incomplete list of necessary answers.
   • For example, if you are asking respondents to rate an event, you must provide them with the options "very good" and "very bad", as well as other options in between.

Tips

• You can also add a "Don't know" answer to questions that respondents may not have an honest opinion about. This will help avoid collecting incorrect data.
• Select respondents strategically. No matter how well designed your survey is, the results will not be as useful if the test group is somehow bifurcated. For example, conducting a survey about computer use over the Internet can give a completely different result from a survey over the phone, as the test group will be more knowledgeable about computers.
• If possible, offer something for completing the survey. Or let the respondents know how their answers will be used. Such incentives will motivate respondents.