0
$\begingroup$

I am unsure how to interpret the second of two scatterplots I created.

The first scatterplot shows a slight negative correlation between the variables (n=14, population not sample). This interpretation is reinforced by the Pearson r.I also calculated a Spearman rho. The first scatterplot is below. The dashed line is a linear trend and both the Pearson r and Spearman's rho are listed on the chart.

enter image description here

On a whim, I also created a scatterplot of the rank of the two variables for each observation. This second scatterplot shows two clusters with each cluster having a moderate positive correlation. This scatterplot is below. enter image description here

Is this second scatterplot meaningful? If so, how should I interpret it?

Note: I am already aware of issues related to the small number of observations. I am only interested in whether the scatter plot of variable rank has any meaning.

$\endgroup$

1 Answer 1

0
$\begingroup$

It means what you probably think it means: It's a scatter plot of the ranks.

Both plots (but especially the second) show that the data seem to be in two groups, with positive relationships in each group and a negative relationship between the groups. This is a definite warning sign about interpreting either correlation coefficient because both of them are about linear relationships and the relationship here doesn't look linear.

I don't know what "uncompensated care" and "FAP richness" are, but I think you should think about the data set and why it might be in two groups.

$\endgroup$
1
  • $\begingroup$ Thank you! As an aside, FAP richness is a measure of a hospitals financial assistance program (i.e. charity care program). The higher the score the more "generous" the program. Uncompensated care is the sum of charity care and bad debt as a percentage of gross revenue. Our state is small and the number of hospitals (points) examined is the total number of hospitals in the state. $\endgroup$
    – Eric
    Dec 2, 2019 at 13:50

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.