I AM YUEYIN JI
CASE1
Objective
A retail company wants to release an advertisement for their products. They want me to evaluate the effectiveness of this advertisement.
Experiment Design
I used diff-in-diff analysis, which is to evaluate the effectiveness of advertising by looking at changes in sales data before and after the advertisement was canceled. I created two groups: one that does not receive ads (the treatment group) and one that does (the control group). Both groups had similar sales patterns before advertising. By comparing the sales data of the two groups after the advertisement was canceled, we can evaluate the effectiveness of the advertisement.
Data Preprocessing
I set the dummy variable for the treatment group and the treatment period. 1 for the treatment group and 0 for the control group. 1 for the treatment period (the advertisement was canceled) and 0 for the non-treatment period.
In-depth Data Analysis
-
Randomization check
I ran the regression between the treatment group and the control group’s revenue when the non-treatment period.
Result interpretation:
The analysis shows that there is no statistically significant difference between the two groups(p-value=0.57), meaning that it is plausible that there is no difference between the two. So, Any observed difference must be due to the treatment effect.
-
Effectiveness analysis
I ran the regression between the treatment group and the control group’s revenue when the treatment period.
Result interpretation:
The coefficient estimates imply that the presence of advertising increases the revenue by roughly 0.7%. Since we cannot reject the null hypothesis, it is plausible that advertising has no effect. However, because of the size of the standard error, advertising could still plausibly increase revenue by as much as 3%.
Conclusion
Therefore, while there is evidence showing a positive correlation between advertising and revenue growth, we need to interpret this result more carefully and consider conducting more extensive studies to verify the robustness of this finding. Future studies may require larger sample sizes, longer time series data, or other statistical techniques to reduce estimation errors and obtain more accurate assessments of AD effectiveness.