The Most Common Survey Errors

When carrying DIY market research careful planning is required to avoid potential errors that can bias your data and lead to incorrect decision making.


A common error made by DIYers is surveying the wrong respondents, that is interviewing those who do not accurately represent your target market (population). By understanding how some of these errors occur you will more likely be able to avoid them in the future.  Here I would like to discuss four of the most common errors

 Four of the Most Common Survey Errors

  1. Sampling Error – This situation occurs when the sample of the target population you select to participate in your survey does not accurately represent the target market for your business or your product.

For example – Let’s say your target market consisted mostly of those aged 25 – 50 years.   You want to carry out a survey to find out whether your business should include another product in its range.   So you used twitter to find a sample of respondents.  However when you get your survey results in you find all your responses are from those aged 22-35 years.

Your sample will not be representative of your target market and may over or under estimate the demand for the product.

  1. Selection Error – This error can occur when you are not randomly selecting respondents from your target market.     For example when you are interviewing people in a mall, you may target those who don’t seem too busy or are sitting and having a cup of coffee.  Or you are only interviewing in the middle of the day.  This means your sample may only contain respondents not working which can cause a problem if your target market consists of those in paid employment as well.
  2. Sampling Frame Error – A sampling frame is a source of contacts for a population from which a random sample is drawn for surveying.  For example your customer data base could be used as a sampling frame if you wanted to draw a random selection of customers to be interviewed.   The perfect sampling frame contains all members of the target population.  Therefore if you randomly draw sample of respondent from the sample frame then your sample should be representative.

However, it is difficult to find a list of an entire population.  It used to be common to use the phone book as a sampling frame if your target market was the entire population of a city.  However, this method is flawed as many have unlisted numbers or only have mobile phones nowadays therefore this is not the way to get a random sample.

  1. Non Response Error – This can occur when those who participate in the interview are different from those who don’t.   There are two ways this could happen.  For example when some members of the sampling frame are not available for interviewing, such as families with children who may be away over the summer holiday break.

Alternatively non response error can occur when those who participate in the study have different opinions than those who do not.  This is more likely to happen when the subject material is sensitive i.e.  politics, race or religion.  Those who participate in surveys are more likely to have stronger views (either negative or positive) than those who don’t potentially polarizing results.


There is always the potential for sampling error in your study; however there are some steps you can take to minimise it.   You can

  1. Sample as many people as your budget will permit, the higher the sample size the lower the response bias.
  2. Ensure that your sample of respondents closely matches your target population in terms of demographic make-up (i.e. age, gender, location etc).

If you are launching a new business and your target market (population) is accountants in the CBD. Then you should have an idea of the demographic makeup of the population of accountants in the Sydney CBD.  Perhaps 25% are sole traders, 55% work in medium-sized firms and 20% work for large firms.   Therefore if you wanted to interview a random sample of 100 accountants in the CBD.  25 should be sole traders, 55 should work in medium-sized businesses and 20 for large corporations.

Having a truly representative sample will help ensure more accuracy in your data.

No comments yet.

Leave a Reply