Provoke Insights

A Brand Consultancy & Market Research Firm

Blog

Research survey, marketing
4 Watch Outs When Conducting a Survey
Sep 10, 2015

“There are errors everywhere,” my market research professor claimed on the first day of class.    He continued and said, “as a market researcher, your job is to reduce the errors as much as possible.”  As someone who has worked in the industry for fifteen years, I believe that this lesson is still extremely important.

If research errors are not found, the results will lead to inaccuracies.    Ultimately, the insights garnered from the study may lead a brand to the wrong conclusions.

If you are developing a survey, here are a few errors that you need to watch out for. 

1.     Inaccurate Objectives

When constructing your research objectives, it is always important to think how the end results will be used.

2.     Improper Research Methodology

Sample size, language issues, and goals of research are aspects you need to consider when determining the methodology. For example, a survey with only 25 respondents only provides directional information and does not represent a portion of the population.   You may want to consider using more open-ended research approaches such as focus groups.  

3.    Bad Survey Writing

Survey length is too long, leading with sensitive questions (i.e. household income), too many open ends, inconsistent or wrong survey scales are just a few of the problems that occur when writing a survey.   The best way to solve survey errors is asking others to take your survey. The more people who take the survey, the more errors you will be able to find.  Also, double check to ensure the survey logic is accurate.      

4.    Sampling Bias

How you acquire your sample can impact survey results.    Your friends or colleagues may provide biased results if they took your survey.  Also, only researching a certain segment of your population may also skew results. For example, if you were targeting a national population, but only reached New Yorkers the results won’t be representative.