Jason Taylor4 said:You edited the post. There wasn't a link there when I replied.
My bad.
I thought it might be useful to include a link that provides a brief overview of the methodology.
The methodology itself isn't actually that complicated. It’s the statistical analysis techniques that have become increasingly complex. You rarely see research relying on straightforward tests like t-tests or ANOVA anymore; nowadays, everything revolves around multivariate techniques. But that’s beside the point. What really matters is that the researcher knows exactly which technique to apply, why they're using it, and where its limitations lie. As for the actual math? The computer handles that.
There’s no mystery to it.
Suppose we want to determine the correlation between personality traits and intelligence. What would our next step be?
The simplest route—and the one most people take—is to just pair an intelligence test with a personality questionnaire.
Suppose we utilized Raven's intelligence test to measure general intelligence alongside Eysenck's EPQ to assess two specific personality dimensions: Neuroticism and extraversion. These traits are relatively well-supported by neurophysiological studies on temperament, effectively corresponding to negative and positive affectivity.
Alright, we have a research problem on our hands: determining how the E and I dimensions correlate with general intelligence. One working hypothesis suggests that Neuroticism will show a negative correlation with intelligence.
So, what’s our next move? We need to select our sample group. To ensure we get a sufficiently homogeneous set, I’m proposing we target healthy individuals between the ages of 25 and 30. At this stage, people are typically at their physical peak. Intelligence hasn't begun its gradual decline, and personality traits haven't fully solidified yet.
Sample size? Honestly, 300 men and 300 women should be plenty.
The margin of error is sitting at about 4%, which is perfectly acceptable.
It’s actually quite straightforward. We gather a group and administer one test followed by another. By rotating which test comes first for different participants, we effectively neutralize any carryover effects from the initial assessment.
Once we have the results, we plug them into the computer and run the analysis using specialized software like SPSS. If you don't have access to those programs, Excel will get the job done.
What was the point of this study? We wanted to tackle a specific question: is there actually a link between intelligence and personality? Using statistical correlation analysis, we found that the correlation between Intelligence and extraversion 0.11 isn't statistically significant. Essentially, there is a 95% probability that no connection exists between these two traits in the general population. However, the relationship between Intelligence and Neuroticism yielded a negative correlation of -0.35, which did prove statistically significant. This means we can be 95% certain that this negative link holds true for the broader population as well. No significant correlation was found between extraversion and Neuroticism. From here, we can pivot to a more granular analysis by separating men and women to see if gender influences these patterns or shifts how strongly these traits intersect.
Now we have the numbers—the actual results that require interpretation. The correlation we've identified suggests that individuals scoring higher in Neuroticism tend to perform poorly on intelligence tests. A potential reason for this could be [bla bla bla]. To be clear, this is a hypothetical study. The data I’m presenting here is purely dummy data, created solely to illustrate the most basic type of psychological research.
Correlational research.These methods establish correlations between phenomena, but they don't actually provide answers regarding cause and effect.
You can try to argue causality here, but you have to tread carefully. If you really want to establish cause and effect, an experiment is the way to go—but let's save that discussion for another time.
Generally speaking, correlational studies remain the workhorse methodology across most branches of psychology. The actual processing methods are far more intricate than my brief summary suggests, often involving multiple variables and their complex interactions, but the fundamental logic remains essentially the same as what I described above.