The demands of a market research study are complex. You may need a large number of respondents, but all those respondents need to be the “right” respondents. You don’t want survey responses from outside the target demographic skewing your data, forcing you to start over.
For example, a survey questionnaire we did at my company was recently returned with highly skewed data. We were surveying a school where 78% of the student body was composed of “in-state” students. When we got the results back, only 3% of respondents were in-state students – definitely not representative of the population.
Where we erred: We did not specify what percentage of different groups of people the survey should be sent to, and did not set up quotas to get a certain mix of people. Instead, we just let it fall out naturally, which should have returned results representative of the population.
It is important to specify the type of respondents you need and decide who the email survey will be sent to, and if necessary, set up quotas to ensure a respondent set that matches the population.
It would be almost impossible to cover all the survey pitfalls you can encounter while sampling. My best advice would be to use common sense. When starting the study I’d advise doing a “soft launch” (sending out a small amount of sample) and then check the returns. If you find that the completes from the soft launch are not representative, work to iron out any issues before doing a full launch.
For example, a survey questionnaire we did at my company was recently returned with highly skewed data. We were surveying a school where 78% of the student body was composed of “in-state” students. When we got the results back, only 3% of respondents were in-state students – definitely not representative of the population.
Where we erred: We did not specify what percentage of different groups of people the survey should be sent to, and did not set up quotas to get a certain mix of people. Instead, we just let it fall out naturally, which should have returned results representative of the population.
It is important to specify the type of respondents you need and decide who the email survey will be sent to, and if necessary, set up quotas to ensure a respondent set that matches the population.
It would be almost impossible to cover all the survey pitfalls you can encounter while sampling. My best advice would be to use common sense. When starting the study I’d advise doing a “soft launch” (sending out a small amount of sample) and then check the returns. If you find that the completes from the soft launch are not representative, work to iron out any issues before doing a full launch.
Powered by Compendium Blogware
Comments for Getting the Right People to Respond