Sentiment of the Week: Anger in the Cinema Predicts Oscar Wins

By: Dheepan Ramanan

February 20, 2015

Tags:
CEM Programs
CEM Solutions
Customer Engagement
Sentiment Analysis

The Oscar awards are almost always full of upsets and surprises, especially the contest for Best Picture. The process is shrouded with the equivalent secrecy of a papal election, replete with votes cast by a shadowy voting block and opaque voting criteria never made transparent to the public. Entertainment analysts often use vague criteria to massage out at best a hindsight rationale to the selection. For example, it could be that sci-fi movies never win best picture, or the winning film deals with a serious issue facing society.

Today, I am doing something different, using purely data to determine this year’s Oscar’s best picture winner.

Methodology

To conduct this analysis, I collected data from the movie review websites Rotten Tomatoes and IMDB. From Rotten Tomatoes, I looked at Critics Score (aggregate critics review rating) and Audience Score (aggregate audience review rating). From IMDB, I gathered 95,000 user reviews to conduct detailed text analysis.  To walk you through my prediction choice, I will begin by using structured, numerical data analysis and then proceed to use text analytics— a technique that should take the surprise out of this year’s best picture winner.

Looking at the Rotten Tomatoes review and audience scores there are not too many patterns that pop out of the past winners. On average, the movies have a high Critic and Audience score (around 89 on average for both), a sentiment that ranges from -.46 (very negative) to .55 (very positive) and averages around .22, the difference between audience scores and critics scores ranges from 13 to -18.

This data demonstrates how poor a choice “Crash” was to win Best Picture. The 2005 film is the lowest scoring movie on all three metrics and it was the 7th lowest scored movie by critics from 2000 to 2015. From IMDB reviews, Crash was the lowest scored movie by sentiment in the entire data set.

What we know:

  • Best picture will have an audience score and critics score of around 89
  • Sentiment will likely be positive (11/15 are positive) with an average of .2 or higher

Let’s turn to the historical dataset of all Oscar Nominated Movies from 2000 to find some more tendencies.

 

Historical Data                 

A 90 critic rating seems to be the magic number for best picture winners:  10 out of 15 have a critic review rating of 90 or above, and seven movies had an audience and critic review of 90 or above. The seven movies that scored an 89 did not win best picture.

Despite the criticism of the Oscar movie committee, the highest rated movie has won the award in the majority of years past. Furthermore, of the movies that have won an Oscar, the deviation between the audience and the critics is smaller than the biggest snubs by over a point.  By this data, it looks like 2003’s “Pianist,” which had a review and critic scores high in the 90s, was the biggest snub for Best Picture in the past 15 years.

Unfortunately, despite the strong correlation between movies that score 90s on both review rating and audience rating, there are exceptions to this strong trend, and years where multiple movies fall into that rarified group.

2014 was one example:

Three movies had 90s in both categories in 2014, with 2 more scoring a 90 with the audience. This is where parsing the IDMB reviews reveals further insights that will allow us to make a prediction on this year’s batch of movies.

 

Text Analytics & Reviewer Sentiment

When analyzing the reviews for all movies that were nominated for a best picture, there are clear differences between Oscar winners and nominees.

In terms of how people are talking about these movies, “Oscar worthiness” was mentioned 55% more often in Best Picture winners compared to just nominees. The other substantial difference in these groups is mentions of the director or direction in the reviews.  Whatever the secret sauce is in a movie considered being a Best Picture, IMDB reviewers seem to all have some sort of agreement around the idea.

IMDB reviews for Best Picture winners reveal some truth to commonly held beliefs around Oscar-worthy movies. Winning films induce much less humor, and lesser amounts of love and sadness. Instead, they are more emotionally explosive with anger and violence. This finding points to why “12 Years A Slave” was so popular with critics and audiences alike—the negative emotions with slavery made the film resonate emotionally. This also explains why a movie like “Gladiator,” replete with action-packed violence, produced a high entertainment value and fit the mold of an Oscar winner.

 

Portrait of a Best Picture Winner

In summary, a Best Picture winner has the following profile on Rotten Tomatoes and IMDB:

  • At least 90 critic score from Rotten Tomatoes, ideally at a 90 from both critics and audiences
  • Movie likely creates feelings of anger, or commentary around violence towards events or characters and has elevated mentions of entertainment and is less likely to be funny or depressing
  • Reviews frequently mention the Oscar Best picture merit of the film and its strong direction, this characteristic is the most strong among Best Picture winners

This year’s lineup is stocked with critics’ favorites. Five of the nominees have a critic score of 90 or above. The only nominee with both a critic and an audience review in the 90’s is “Whiplash.” Recognizing that a 90 or above is necessary from critics, we can eliminate “The Imitation Game” and “The Theory of Everything” from contention. “Selma” has the lowest sentiment of reviews, while “The Theory of Everything” has the highest.

Although “Whiplash” and “Selma” both induced high levels of anger toward events and characters in the film, Selma also was loved the least by its audience. Reviewers thought “Birdman” was confusing, but still the most entertaining of the nominees, and the second funniest. “The Grand Budapest Hotel” was the most humorous movie by its critics, mentioning humor 314% more often than the rest of the movies.

Given that none of the movies fit the average profile of other best pictures, I will only eliminate “The Grand Budapest Hotel,” since it would be the most humorous Oscar winner since 2000—and humorous movies are less likely to be winners based on historical data.

The Oscar and direction comments are much more telling.

A Best Picture nominee with high mentions of Oscar worthiness is the most important indicator of actually winner an Oscar. Of this year’s crop of movies that score a 90 or above from critics, only “Selma” and “Birdman” have higher than average mentions of this trait. In this regard, “Birdman” has the highest average of these mentions, and compared to “Selma” people speak more positively around the deservedness of best picture. Interestingly, despite “Boyhood” generating considerable Best Picture buzz by the media, the average reviewer thought it was less deserving for the award.

And the winner is. . .

2015 will be a tight Oscar race for best picture, however looking at the data there can only be one winner—“Birdman.”

“Birdman” has the most mentions of Oscar worthiness, and compared to “Selma” it not only has more mentions but more positivity in speaking about its chances to win best picture. “Birdman” also has a smaller gap between the score of its audience and the critics compared to “Selma,” another trait most Oscar winners share. In regards to audience emotion, reviewers found it the most entertaining movie of the best picture nominees, but felt less happiness watching the film, these are both traits further indicate a best picture winner.

Learn how to best deploy text and sentiment analytics in your business by watching our webinar: 10 Best Practices for Deploying Text Analytics.

 


Dheepan Ramanan is a data scientist at Clarabridge. Follow him on Twitter @DheepanRamanan.
2016-1486_clarabridge_demo_request_cta