Is my association significant?

#1
I am analyzing a program where people participate with the hopes of having event occur(event=1). The program has phases that occur in sequence but the sequence is different across the participants. I am analyzing one specific phase and trying to determine if its sequence position is significantly impacting the results.

For example, if the phase happened first are we more likely to see the desired outcome versus if the phase happened 5th?

I created a frequency table of the sequence position by the event outcome and calculated the Peason Chi-Square and Mantel-Haenszel Chi-Square test statistics. My understanding is the Mantel_Haenszel statistic is a stronger measure for ordinal values. The results are below.

The Cramer's V is .075 and the Spearman Correlation is -0.075. I could use some help properly interpreting these results. Both the chi-square test statistics suggest there is a significant association (at a .05 significant level). However both the Cramer's V and Spearman Correlation are very close to 0. Does this mean that there is strong evidence to suggest there is no correlation between the program outcome and the position of the phase being studied?

1584132192544.png

1584132311174.png

1584132192544.png
 

Karabiner

TS Contributor
#2
I cannot see much sense in performing inferential statistical tests with a sample size n=41572.
The standard error of any statistic is near zero, so why bother.

You can interpret your table right away. Position one has a markedly larger rate of result "1"
than the 4 other positions. Whether this difference is practically important I don't know.

With kind regard

Karabiner