TV Movie Ratings Case Study (g-3)

Statistical Analysis with p-values

Analysis Scope: CBC Network movies (30 observations)
Dependent Variable: Rating (— avg)
Available Variables: Fact, Stars, Previous Rating, Competition
Status: Uses sample variances for t-tests; df-based critical values; real OLS regression with p-values (jStat)

The Case and the Questions

The Network has collected data on the ratings of their TV movies. They have kept track of whether the movies had stars, if the plots were based on real-life situations, what the ratings were for the lead-in show to the movie and what the ratings were for the competition at the same time.

You can generate a new data set, then download it to do your own analysis. You can also use the two analysis tabs to use the metrics calculated by the tool. The executives have the following questions:

Question 1: Should the network choose to hire stars to improve their ratings or not?

Choose yes if the one-tailed t-test shows a significant positive effect of stars.

Q1 options

Question 2: Should the network choose plot-lines based on real life situations or not?

Choose yes if the one-tailed t-test shows a significant positive effect of reality-based plots.

Q2 options

Question 3: Which strategy increases ratings more — hiring stars or using reality-based plot-lines?

Select the factor with the larger estimated effect size in a two-variable regression.

Q3 options

Question 4 [Advanced]: Which single factor should the network prioritize as their primary programming strategy?

Suggested Strategy:

  1. Run the regression with all four variables using the “Impact Analysis of Various Factors” tab (Tab 3).
  2. Identify the independent variable with the largest coefficient and compare its t-statistic to the two-tailed 95% t-critical value.
  3. If that variable is significant (|t| ≥ t-critical) and has the highest coefficient, declare it the best choice and confirm it matches your answer.
  4. If the highest-coefficient variable is not significant, remove it, rerun the regression with the remaining variables, and repeat the comparison.
  5. Continue until a variable is both significant and has the highest coefficient among those left. If all independent variables are not significant, the answer should be “none of the variables are significant so programming strategy must target something else.”
Q4 options

A/B Comparison using One-Tailed T-Tests

Test whether one group gets higher ratings than another using one-tailed t-tests (equal variances assumed, 95% confidence). Uses sample variances and df-based critical values.

Select Comparison:

Multiple Linear Regression Analysis

Select which independent variables to include in your regression model:

Selected: Fact, Stars, Previous Rating, Competition (4 variables)