Designing a Focused Survey

It is impossible to evaluate a statistic without understanding the methods used to acquire the data. For example, imagine that you hear the statistic that 3 out of every 4 husbands make dinner for their wives. What would you want to know about how those numbers were arrived at?

An important consideration in the design of this statistical study would be: Who was sampled to acquire this data? For example, were the respondents all located in a suburb of Chicago? If yes, that might be fine if the results of the study are only generalized to that suburb or suburbs with similar characteristics. Hence, we hit upon a defining characteristic of samples—they must be representative of the population to which the data will be generalized.

Different types of data require different forms of analysis. In addition to the different types of data that are encountered in statistical studies, one has to look at the methods for measuring data and measuring error in data.

Define your data collection strategy, which includes developing survey questions. To allow you to focus your time and effort on the actual survey tool and analysis, this course uses a software program to provide the data. Using technology to generate survey data causes hypothetical restrictions in your survey design. As a result, the number and type of questions you write for your survey must follow the Data Collection Template.

Directions

Develop six questions for your survey. Keep in mind that you will not be conducting the survey by mailing or e-mailing questionnaires. Your survey responses will be determined by an electronic survey generator, which accepts only certain types of questions; therefore, the types of questions you can ask for this assessment are very specific and must adhere to the guidelines listed below. In the future, as you design surveys to solve problems or answer key questions in your professional life, you will likely draw from a broader range of question types.

Guidelines for the Questions

• All six survey questions should be stated in question form. They should be written like they are being asked directly to a survey participant. For example, you would ask "How many cats do you own?" rather than stating "The number of cats owned."
• The first four questions (1–4) must be binary—that is, they have only two possible responses. Examples of this are yes/no, true/false, and male/female (to name a few).
• The last two questions (5 and 6) must be quantitative. These questions have to elicit a single numeric response. Survey participants can respond with only one number.
• There are certain types of questions that are not within the parameters of the project. The following types of questions cannot be asked:
  • Do not ask questions where survey participants have to explain something. This includes short answer and fill-in-the-blank questions.
  • Do not use a Likert scale. That is where a participant would choose from responses such as strongly agree, agree, neutral, disagree, and strongly disagree.
  • Avoid questions where a participant chooses a number on a scale, such as 1 to 10 or 1 to 5.
  • Do not ask any questions that are contingent on another question. Each question must have a stand-alone response.
  • Do not ask questions that have already been answered by how you have defined your population. For example, if your population includes only males, do not ask "Are you male or female?"

Guidelines for the Minimum and the Maximum of the Questions

The minimum is the lowest number that you think a participant can or will respond with. The maximum value is the largest value that you think a participant can or will respond with. To determine these values, make an educated estimate based on your population and the research you have conducted on the issue. For example, if we are surveying Capella students, we might ask "What is your age?" In this situation, a minimum age might be 16, and a maximum age might be 85. Note that it is possible to be outside these ranges. The minimum and the maximum are approximations, or likely ranges, of what you expect.

• Your binary questions will not have a minimum or a maximum.
• Your will need to set a minimum and a maximum for each of the quantitative questions.
• The minimum and the maximum are each one single number.

Guidelines for the Expected Values for Typical Responses to the Questions

The typical responses will help the program that generates your hypothetical data produce more realistic participant responses based on your knowledge of the issue you are studying. The typical response will fall between the minimum and the maximum. It is the value you think will be the most common response from a survey participant. Base the typical response on your knowledge of what you are surveying and the research you have conducted on the issue.

• You will not need to fill in a typical response for questions 1 through 4.
• The typical responses for the quantitative questions should be what you think the mean (average) of all the responses will be.
• The typical response is one single number.

Defining Your Data Collection Strategy

• Use the Data Collection Template (linked in the Resources under the Required Resources heading) to define your strategy.
• Include the following items in your strategy:
  • Your target population for the survey.
  • Your sampling strategy and how you would attempt to conduct your survey. Be sure to include some rationale for your strategy and any potential issues that might affect your survey results.
  • Your six survey questions following the guidelines presented above. Be sure to document the expected value along with a reasonable minimum value and maximum value in the table, as you will use these pre-survey values in later project components.
• In Table 1 of the Data Collection Template, fill in every box that says "you fill in" and then remove that message once you have done this. Your final product should have a response in every box; there should be no empty boxes in the table.