Exercise

Intuition behind statistical significance

In this exercise you will work to gain an intuitive understanding of statistical significance. You will do this by utilizing the get_pvalue() function on a variety of parameter sets that could reasonably arise or be chosen during the course of an A/B test. While doing this you should observing how statistical significance results vary as you change the parameters. This will help build your intuition surrounding this concept, and reveal some of the subtle pitfalls of p-values. As a reminder, this is the get_pvalue() function signature:

def get_pvalue(con_conv, test_conv, con_size, test_size):  
    lift =  - abs(test_conv - con_conv)

    scale_one = con_conv * (1 - con_conv) * (1 / con_size)
    scale_two = test_conv * (1 - test_conv) * (1 / test_size)
    scale_val = (scale_one + scale_two)**0.5

    p_value = 2 * stats.norm.cdf(lift, loc = 0, scale = scale_val )

    return p_value

Instructions 1/3

undefined XP
  • 1

    Find the p-value with initial conversion rate of 0.1, test conversion rate of 0.17, and 1000 observations in each group.

  • 2

    Find the p-value with control conversion of 0.1, test conversion of 0.15, and 100 observations in each group.

  • 3

    Now find the p-value with control conversion of 0.48, test conversion of 0.50, and 1000 observations in each group.