Correlational and Causal Relationships

 

Correlational and causal research both follow similar basic scientific research design, where a research question is posed, then followed with a hypothesis and a null-hypothesis, where quantitative data either supports the research or null hypothesis (Gonzalez, 2018). However, they differ greatly when it comes to the purpose and outcome of the research. Correlational research attempts to demonstrate a relationship between two or more variables, usually through surveys, but it doesn’t demonstrate causation among variables (SEP, 2016). On the contrary, causal research aims at demonstrating a relationship causal relationship among variables, as in variable A causes variable B, and does so by accounting for extraneous variables by following the experimental method (Srinagesh, 2006). In order to further demonstrate the difference between the two research methods, the following example has been chosen.

Example:

The relationship between age and driving speed

Does the example demonstrate a causal or correlational relationship?

The above example attempts to demonstrate a correlational relationship due to several reasons. The first is that it is vaguely phrased  (Huff, 1973) in a way that indicates that there might be some kind of relationship between the two variables, age and driving speed. If it were to demonstrate a causal relationship, it should be more specific with a clear direction (Dancey & Reidy, 2017).  A more appropriate phrasing if were to attempt to demonstrate a causal relationship would be, as male driver’s age increase, this causes their testosterone level decreases, which causes their average driving speed to decline.  

What’s needed to demonstrate a causal relationship?

The main requirement for demonstrating a causal relationship is to account for extraneous or confounding variables. Such variables are third party variables that aren’t accounted for and might influence the relationship between the variables being studied (age and driving speed) (Srinagesh, 2006). Such variables are accounted for by following the experimental research design, which also entails having randomized samples (Andrade, 2018).

Evaluate whether a chi-squared test or a correlation coefficient would be more suitable to analyze the data in the example.

In order to properly analyze the data in the example, it is important identify the variable types, which will aid in identifying the most suitable test type. Chi-squared test, an association test, is used to demonstrate that the phenomena co-occur among two or more nominal variables (Dancey & Reidy, 2017). Therefore, if the example was phrased as such: how is gender (nominal) associated with driving speed (nominal), then Chi-squared would be most suitable.

Moreover, how the data is collected can sway the researcher to choose one of two types of correlation coefficients in order to evaluate the data. If an ordinal scale was used to collect the data, a non-parametric, is in does not rely of a normal distribution, test called Spearman’s rank order correlation would be the likely option (Schmid & Schmidt, 2007). However, if the data was collected using interval or ratio scale, an alternative parametric correlation coefficient called Pearson’s product-moment correlation is more suitable for analyzing the data (Bollen & Barb, 1981). Since data regarding age and driving speed is most likely to be collected on an interval scale, it would be best to use Pearson’s correlation coefficient.

References

 

Andrade, C. (2018). Internal, external, and ecological validity in research design, conduct, and evaluation. Indian Journal of Psychological Medicine , 498-499.

Bollen, K. A., & Barb, K. H. (1981). Pearson's r and coarsely categorized measures. American Sociological Review, 232-239.

Dancey, C., & Reidy, J. (2017). Hypothesis testing and statistical significance . In Statistics Without Maths for Psychology (7th ed.) (pp. 134-173). Harlow, UK: Pearson.

Dancey, C., & Reidy, J. (2017). Non-parametric statistics . In Statistics Without Math for Psychology (7th ed) (pp. 516-550). Harlow, UK: Pearson.

Gonzalez, K. (2018). What is a null hypothesis? - definition & examples. Retrieved 7 19, 2018, from Study: https://study.com/academy/lesson/what-is-a-null-hypothesis-definition-examples.html

Huff, D. (1973). How to lie with statistics. London: Penguin.

Schmid, F., & Schmidt, R. (2007). Multivariate extensions of Spearman's rho and related statistics. Statistics & Probability Letters, 407-416.

SEP. (2016). Correlational research. Salem Encyclopedia Press, Retrieved from https://liverpool.idm.oclc.org/login?url=https://search.ebscohost.com/login.aspx?direct=true&db=ers&AN=119214045&site=eds-live&scope=site.

Srinagesh, K. (2006). Planning the experiments in statistical terms. In The Principles of Experimental Research (pp. 333-372). Amsterdam : Butterworth Heinmann.