Correlational Does Not Imply Correlation: Correlational Research Design vs. Correlation Analysis

Posted June 18, 2018

In quantitative research, there are three major types of research design: descriptive, experimental, and correlational. Of course, this is simplifying things a bit and schools vary in their classification of designs, but we will discuss these three for now. Within descriptive research, researchers are primarily interested in describing an existing situation rather than making statistical inferences. Within experimental research, researchers are primarily interested in determining probable causal relationships in a controlled setting. Within correlational research, researchers are primarily interested in determining non-causal relationships amongst variables. More specifically, the correlational research design is a type of non-experimental study in which relationships are assessed without manipulating independent variables or randomly assigning participants to different conditions. Now, we have all heard the phrase “correlation does not imply causation,” but I’d like to introduce a new phrase: “correlational does not imply correlation.”

What I mean is, do not get caught up in the word “correlational” when thinking of your research design. Just because you have a correlational research design does not mean that you are limited to using a correlation analysis for your study. In other words, a correlational design is not a correlational analysis! More broadly, while your research design informs what analyses you will perform (and vice versa), the analysis is not the design.

To illustrate, say you were doing a non-experimental comparative design, in which you propose to use an ANOVA to determine differences in job satisfaction amongst types of job positions at a zoo. You would not describe your study as having a quantitative methodology with an ANOVA design. You would say: “I propose to perform this study using a quantitative methodology with a non-experimental comparative design, and I propose to analyze my data using an ANOVA.” However, many seem to make this mistake when discussing correlational research designs. A correlational design looks at relationships, so you could appropriately consider, yes, correlations, but also regressions, path analyses, various nonparametric analyses that are based on similarity of ranks, correspondence analysis, or canonical analysis, to name a few.

So, do not think that if you have a correlational research design that you are limited to a bivariate correlation analysis. This highlights the importance of being very specific and consistent in your dissertation proposal; in research, terms often refer to very specific things (think design vs. analysis), even though they may seem similar. However, I have often seen reviewers get confused about this correlational design vs. correlation analysis issue even after explanation. Sometimes you may be better off describing your study as a non-experimental design, rather than the more specific term correlational design.

Pin It on Pinterest

Share This