We are looking at frequency of practice (x) and performance (y) across trials. We have two ways of computing practice x and performance y. One way is the sum of the number of times each person practices (sumx) and one is the average x (avgx).

Similarly, perf is calculated by sum of performance scores (sumy) and average (avgx).

the range of sumx and sumy is higher than avgx and avgy. could that be a reason why the correlation between sumx and sumy is way higher than the correlation between avgx and avgy . (see sample data below)

participant x1 x2 x3 y1 y2 y3 sumx avgx sumy avgy
1 3 5 7 80 90 80 15 5 250 83.3
2 1 2 5 70 80 70 8 2.6 220 73.3

Hard to follow ... when you average do you remove outliers (SD +/- 3) ...?

the question is not clear.

bbkaran


We are looking at frequency of practice (x) and performance (y) across trials. We have two ways of computing practice x and performance y. One way is the sum of the number of times each person practices (sumx) and one is the average x (avgx).

Similarly, perf is calculated by sum of performance scores (sumy) and average (avgx).

the range of sumx and sumy is higher than avgx and avgy. could that be a reason why the correlation between sumx and sumy is way higher than the correlation between avgx and avgy . (see sample data below)

participant x1 x2 x3 y1 y2 y3 sumx avgx sumy avgy
1 3 5 7 80 90 80 15 5 250 83.3
2 1 2 5 70 80 70 8 2.6 220 73.3