[Python] Pandas crosstab results in wrong Odds Ratio

I have a pandas dataframe with two columns: ['exposure', 'outcome']. both are bool columns. Sample data:

exposure outcome
1 False True
5 True False
7 False False
11 False False
21 True False
13 False False
1 False True
24 False False
15 False False
16 False False
31 False False
5 True True
1 True False
13 True True
40 False False
1 True False


Now when I create a Crosstable table using pandas.crosstab i get the following:

contingency_table = pd.crosstab(data['exposure'], data['outcome'])

outcome False True
exposure
False 12 4
True 20 12


then when doing:

from scipy.stats import fisher_exact
oddsratio, pvalue = fisher_exact(contingency_table.values)


I get the wrong odds ratio (1.8 here) because the order of the crosstab table is wrong. The OR should be 0.55 when I calculate that manually.

What am I missing here?

Continue reading...