This post explores a potential legal concern that a disproportionate number of sampled customers are men. This is particularly troublesome since the target for this promotion is women, since they are the most likely to try Sun Basket offerings.
If a customer is randomly selected from the Facebook marketing audience, the probability that the customer is a woman, P(W), is 944,643/1,574,405, or 0.6. This marginal probability indicates that roughly 60% of all potential customers marketing to on Facebook for this campaign are women.
Let’s say that Sun Basket decides to focus a campaign, at first, on dedicated “Mediterranean Diet Followers” group. Given that the customer is from this specialized group, the probability that the customer is a woman, P (W | M) is 27,491/45,819, or .6. The proportion of women in this specialized group is the same as in the larger population. Luckily, since Facebook provides the option to select gender when targeting ads, Sun Basket need not be concerned about this high probability (40%) of randomly selecting a man to see highly targeted ads aimed at women. Facebook allows ad buyers to select an audience not just based on liked topics, but also based on gender and income level. So it will be easy to rule out men for whom these ads are less likely to work.
Post #4 in this series explores a lottery analysis.
Tech Twire. 