- Franco Arda

# Are we promoting more men than women?

We've promoted 117 women and 203 men. Are we favoring men in promotions? Is the result statistically significant? To answer this question, we need to do a hypothesis test. The classical test would be to calculate the p-value while the modern approach is to simulate the confidence interval by using a computational method called bootstrap.

I'm using the R bootstrap package __infer__. I'm a huge fan of the package as it empowers us to visualize the confidence intervals nicely in ggplot2.
First, I load the dataset and calculate the difference in promotions manually. I get a promotion difference of around 3% favoring men.

Next, we simulate with replacement 1,000 promotions.

The result is that we can expect a difference in promotions of -1.2% and 7.3% within a 95% confidence interval.

The final step is to visualize the bootstrap method:

The visualization in R shows us that a difference of around 3% is very well within the "range" (confidence interval) of a simulate promotion.

To push this analysis in production, I have chosen __Tableau__. Tableau allows the users (most likely HR) to drill deeper into the analysis.

The Tableau Dashboard is inspired by Dee Naidoo (__Udemy Tableau course__). I also used her dataset but added an additional column with promotions.

Franco