different results when 'rows fixed' or 'columns fixed' in Bayesian contingency tables
When conducting a Bayesian contingency tables, different BF10/01 values are prouced if the 2 variables are either placed in rows then columns, or then reversed.
Any advice on how to interpret these differences? What would the difference mean?
Thank you
Comments
Suppose you want to compare two proportions.
Proportion 1: 0 successes out of 5 trials
Proportion 2: 1 success out of 6 trials
So:
Successes Failures
Condition 1 0 5
Condition 2 1 5
If we switch the rows and columns, we have:
Proportion 1: 0 successes out of 1 trial
Proportion 2: 5 successes out of 5 trials
These are different comparisons.
E.J.
Our analysis maps onto the first of these scenarios (success/failure is the columns variable). Does this correspond to 'rows fixed' or 'columns fixed' in JASP?
Thanks EJ !
You need "rows fixed" then (as the rows denote the conditions)
EJ