top of page

CASE 3

Introduction

This case is based on a Facebook advertising dataset.  It contains data from a large-scale field experiment of advertisements for various clothing brands on the Facebook platform.

image.png
Objective

The firm is interested in identifying the factors they should use to get the largest number of clicks per dollar.

Solution

Interaction effect analysis:

To figure out this objective, it would be better to run the regression between clickPerDollar (dependent variable) and independent variable factor to understand what effect the factor would have on the clickPerDollar, so we can know which factor could get the largest number of clicks per dollar.

It’s worth noting that starting from a realistic point of view, people of different ages react differently to advertisements from different categories, so it would be rigorous to add an interaction effect in the regression.

​

lm(formula    =    clickPerDollar    ~    factor(adType)    +    factor(placement)    +

factor(keywords)    +    factor(body)    +    factor(category)    *    ageMean,

data    =   hw2Data)

Result Interpretation

When the targeted age increases by 1, the eCom category receives 0.36 fewer clicks per dollar than the cosmetics category. Older customers prefer cosmetics to eCom.

bottom of page