1. Suppose you would like to know whether increasing the number of keywords in your ad group can decrease the click through rate of your ads from that ad group to help you make decision on whether you should increase the number of keywords in your ad group. What method should you use for this analysis?
   1. Causal Analysis
   2. Predictive Analysis

2. Which of the following is not a statistics that can help you assess the goodness of fit of your model?
   1. Adjusted R²
   2. Root Mean Square Error
   3. Estimated coefficients
   4. R²

3. Which of the following is an example of Type I error?
   1. Display ads increased your sales, but your model suggests that display ads did not have an impact on your sales
   2. Email promotions did not increase your sales, but your model suggests that email promotions did increase your sales.
   3. Display ads did not increase your sales, and your model suggests that display ads did not have an impact on your sales
   4. Email promotions did not increase your sales, and your model suggests that email promotions did not increase your sales.
4. Suppose you would like to include age groups in your regression. Your data has 5 age groups: 0-10, 10-20, 20-30, 30-40, and 40+. How many age group dummy variables do you need to create and include in your regression?
   1. 1
   2. 2
   3. 3
   4. 4
   5. 5

II. Short Answer Questions

Suppose you observe in data across 500 firms that the firms having a Facebook page have higher sales than the firms not having a Facebook page. You run a regression of sales over Facebook page dummy (which equals 1 if the firm has a Facebook page and 0 otherwise). The estimated coefficient of Facebook page dummy has a p-value of less than 5%. Does this mean that for those firms without a Facebook page, starting a Facebook page can increase sales? Why?