Statistics questions

Movie Reviews Submitted by Glenn Henshaw for MAT 119/120 (revised 5/23/2018)

1 Introduction

We all want to know how good a movie is going to be before we commit our two hours and $15 to watch it in theaters. To help us decide if a movie is going to be good or not there are several websites that gather reviews from critics and average the scores together to produce a rating. These websites are called review aggregators. You may have heard of Rotten Tomatoes, Fandango, IMDB, and Meta Critic. Browse the following websites to get a sense of how their movie ratings systems work.

• rottentomatoes.com

• imdb.com

• metacritic.com

• fandango.com For this project you will be investigating questions like: What do the rating

distributions look like in terms of center, spread, and shape? Which rating system seems to be the most trustworthy? What influence does the business model of a website have on the distribution of its ratings?

Class discussion:

1. Pick your favorite movie and check its rating on the different sites. Which site rated your movie the highest and which one was the lowest?

2. How could you compare scores coming from different sites since they have different scales?

2 Getting the data

1. Download the dataset, “movie reviews.csv”. It is located at github.com/ghenshaw/datasets/

(a) click on the file “movie reviews.csv”

(b) right click on the “raw” button

(c) click save-as to save it to a convenient place on your computer.

2. Open “movie reviews.csv” using Microsoft Excel, SPSS, or R.

1

The data has 147 rows. The first 11 rows should look like this:

Each row is a single movie and each column value is its rating for the different sites.

3 Data normalization

When you visited the websites you might have noticed that different sites have different scales for their reviews. Rotten Tomatoes, uses a 100 point scale in its rating system, while Fandango uses a 5 point scale. This makes it hard to compare the scores. To make the scores comparable we normalized the scores so they all are based on a 5-point scale. To do this we used the following formula:

normalized score = 5 · score maximum score

For example, Rotten Tomatoes rates the 2016 movie “Doctor Strange” as a 90 on its 0 − 100 scale. Using the above formula we can calculate its score on a 0 − 5 point scale.

5 · 90 100

= 4.5

4 Questions

Question 1: What would the normalized score (0-5) be for the 1999 movie “Eyes Wide Shut” which was given a score of 74 on Rotten Tomatoes? Question 2: If you were asked to normalize Rotten Tomato scores to a (0-10) point scale how would the formula above have to change?

2

Question 3: Use Excel, SPSS, or R to make a boxplots for each of the columns of the dataset on the same axis. Question 4: Compare the distributions of ratings (boxplots) for each column in terms of center, spread, and shape. Which website (Rotten Tomatoes, IMDB, fandango or meta critic) seems to consistently give movies higher ratings? Question 5: Do some research online to find out how each of the movie review aggregators makes money. How could their business models effect the ratings they give to each movie?

5 Essay

Write a essay that summarizes the results you obtained above. The essay should have an introduction that describes the ratings systems of the four ratings aggre- grators (Rotten Tomatoes, IMDB, fandango or meta critic) and clearly describes the main question addressed in this project. Next, include a picture of the box plots you generated followed by a comparison of the center, spread an shape of the distributions. Conclude with a discussion of the results: Which website (Rotten Tomatoes, IMDB, fandango or meta critic) seems to consistently give movies higher ratings? How does the business model for this website differ from the business model of the other web sites?

6 Further reading

For a more in-depth discussion search for the article on fivethrityeight.com en- titled, ”Be Suspicious Of Online Movie Ratings, Especially Fandango’s.”

3