Can Phi Coefficient be used as an effect size for Chi-Square Goodness of Fit tests?

#1
Flip a coin 100 times -> 48 H, 52 T
Null - coin follows a specified distribution
Run a chi-square goodness of fit test using 48, 52 and 50, 50

I get a high P-Value which tells me to accept the null (coin is fair).

Why then is the Phi-Coefficient low (2%)...meaning that the association between expected and observed is weak?

Thanks a lot Guys.
Love ya!
 

trinker

ggplot2orBust
#2
Re: Can Phi Coefficient be used as an effect size for Chi-Square Goodness of Fit test

P value and effect size tell us two different things. One is an overall test of significance (yes no) the phi is a strength of effect that answers how much.