Avoid these 4 Common Errors when Analyzing Survey Data

Analyzing survey data is an important function in advising business decision makers. Logically, errors made in the analysis can lead to incorrect conclusions which in turn may lead to unanticipated and often undesirable outcomes.

Consider these 4 common errors when analyzing survey data.

 

1. The word “Most”
It may seem so basic but misuse of the word most is a common mistake when analyzing survey data. Take this example: 45% of people thought that drinking soft drinks every day is bad for your health. 30% thought that drinking more than one soft drink a day is bad for your health and the rest had not thought about it at all.

In this instance it is safe to say that the most frequent response was that drinking soft drinks every day is bad for your health. However, it would be incorrect to say that most people thought that drinking soft drinks is bad for your health because statistically that amount, in this context, is not greater than 50%.

2. Causation vs correlation
Most survey data is the result of correlational research as opposed to experimental research. Unlike, experimental research, which manipulates a variable to measure its effect on other variables, (hence determining causation), correlational research measures the relationship between different variable to measure its correlation (hence the name).

Understanding the difference between causation and correlation determines the conclusions one can draw from a given data set. For example, if we take some sample data and estimate that 30% of males and 60% of females were using the internet last week we can’t say the difference is because of their sex. We can only say that internet use is associated with their gender. The reason for the difference may be caused by something else entirely.

3. Avoid quoting percentages only
Percentages are a familiar and practical way to look at data results but be careful of not including the underlying base on which the percentages are determined. This is especially important when reporting changes in percentages, particularly if the statistical sample is too small.

4. Focusing on the average
Relying solely on averages can be misleading. Regression toward the mean may indicate the sample period was too short to be meaningful. Or, the reason behind an average is often more important than the number itself. An average result, without context, is often meaningless in supporting an evaluation of the underlying process you are analyzing.

Leave a Reply

Your email address will not be published. Required fields are marked *

*

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>