My First Survey

Surveys prove very helpful in collecting quantitative and even qualitative data in a short period of time. The proliferation of the internet has decreased survey costs, but not without a sacrifice. In my experience, thinking of every possible way a survey participant could give you poor data or misinterpret your questions leads to constant survey revisions and second-guessing.

Creating my first survey was more challenging and required more time and patience than I expected. I anticipated this level of dedication would be reserved for the next phase of my research, focus groups. I am a bit terrified.

After several informal interviews to guide our question formation, my team and I started with general insights about our main subject, make-up, then moved on to more detailed needs and brand insights. Finally about where we anticipated survey fatigue (approximately question 17), we concluded with the demographic data.

Determining the best answer choices for the questions was surprisingly time-consuming. Some questions lent themselves to particular answer structures such as, demographic data, brand awareness or product experience and quality. However, there were many other questions where the wording of the answers and/or determining the selection structure (i.e. Likert, open-ended, select one, select all that apply and rank) were quite challenging and the subject of a reasonable amount of discussion. We often used ranking questions where there was a strong likelihood that the respondent would select many of the answers rendering the questions less useful, or if we felt that one particular answer might be significantly stronger than the others. We allowed for only one answer when we felt that the primary answer led to key factors or decisions. Finally we used “select all that apply” questions when knowing every selection was the question goal versus the individual importance of on response over the others.

One concern is limiting the number of questions than insure the respondents could successfully answer them without survey fatigue. Every question was evaluated for its value and effectiveness and whether the same information was provided by another question. If the same or similar information could be gathered by another question, then the question was eliminated. All other questions were ranked, and questions were cut at the fatigue limit, which we estimated at about 20 questions, including demographic questions.

There are other survey shortcomings I hope we can avoid, such as respondents answering untruthfully or inaccurately, or even being offended by questions related to sensitive issues. We intend to validate our survey data with focus to catch and resolve any such problems.