Poor Students scored lower in STEM and why

Source: https://www.pnas.org/content/116/5/1553

Poor students are statistically doing worse than Black students in STEM. Literature review suggested that the low self confidence on something emphasis rank and status throws the spanner in the works. Study hypothesized the above model to explain and improve the anxiety caused disadvantage.

The way this model presented could be possibly perceived as “poor test performance” enhance the “performance anxiety” which is not (the main point) addressed in this study.

(I expect author who submits to PNAS would have bothered to put a p-value bar in graph but) The difference in scoring between lower and higher income students (column 1 and 3) was significant, and the difference after intervention was also significant (column 1 and 2). The hypothetical anxiety intervention did not cause any changes in the higher income group.

Does that mean rich kids fear not about the exam? or they simply have all the resources to prepare for the exam? What is the anxiety easing factor offered by self-evaluated well preparation? Did the intervention even work to improve the anxiety in higher income group (is there other read-out for anxiety)? What else did the intervention change in lower income student? Which means that anxiety using score as read-out was the only factor being considered in this analyze.

The intervention significantly increase the passing rate for lower income group (column 1 and 2) but have no effect in higher income group (column 3 and 4). Some other fancy numbers showed that statistically, the intervention closed the gap between lower and higher income group but did not totally shut it tight.

The passing rate in the higher income group was already close to 100%. I wonder if an improvement from 95% to any percentage would reflect in any statistic using a pool of 890 students. Anyway, more importantly is that the intervention worked in the lower income group, which is great.

Basically this survey after the exam result asked those kids if they agree with better anxiety management the better they would have scored. Lower income group agrees less than higher income group and the intervention enhanced this feeling in lower income group

I am not sure how the subjects understand about this study before taking it, but it will definitely be confounding if they know the measurement was for easing anxiety or converting the negative anxiety to positive excitement (whatever you name it). This will become a positive feedback that — I took the intervention, I have done better than those did not and because those are for easing the anxiety so now I agree I can be superman if I keep thinking positively. Well, the only point is that, this read-out sucks. Better one would be an assessment of their ability in reappraising their anxiety in similar setting but for other STEM subjects, for example. The assumption would become they find this a good stuff and keep applying this strategy and the premise is their learning ability is not significantly different which I believe had already been controlled.

This kind of circular argument is not uncommon even across different academic field. But this would be too strong for a critique in the professional setting and the professional term for this would be “validation” although I believe that should have been done in the beginning of the study or in the appendix of the article.