amc a list

Date Night Data: Analyzing Movie Trends and Stats

/ rkelleyc / Leave a comment

Over the past year, Brittany and I have embarked on a cinematic journey, watching 46 movies in theaters over 365 days. From Halloween thrillers to box office flops, this has sparked our curiosity to learn more about the trends and insights into the movies we love. Join us as we dive into the data behind movie budgets, ratings, and profits, along with a brief refresher from your high school statistics class.

We prefer to go on Thursday’s when new movies are released. We saw 24 of 46 movies on opening weekend

Using the AMC A List (movie subscription pass), we paid ~$5.60 per ticket; 59% less than a standard ticket ($13.36) and 73% less than an Imax/Dolby ticket ($20.41)

Most of those movies we saw were dramas, action movies, or comedies (note some movies had more than one genre)

We saw six movies from both Paramount Pictures and Sony Pictures

Ready to dive deeper into the stats?

You could watch the shortest movie (Bottoms) 2.2x before the longest movie is over (Killers of the Flower Moon).

Of all the movies, these were the biggest financial hits (at least for opening weekend), generating more box office revenues during opening weekend (red) than the film’s budget (blue)

Generally, the production company gets to keep 50-60% of the box office revenues from opening weekend and the theater owner get to keep the remaining 40-50%. However, this split is contentious and often varies based on the strength of the film and the bargaining power of both the production company and the theater. After opening weekend, that split often shifts over subsequent weeks such that the production company’s cut decreases and the theater’s share increases.

There are plenty of exceptions but opening weekend box office revenue carries an outsized importance because it often sets the tone for the film’s entire theatrical run. By Sunday of opening weekend, studio executives can predict with great accuracy what a movie will earn by the time it leaves the cinema. Ticket sales are the largest driver of income for movies, although not necessarily the most profitable because of taxes and the cut owed to the theater. Other sources of revenue for the production company include DVD sales, on-demand rentals, licensing income from streaming services, and merchandising.

This reliance on opening weekend box office revenues is changing though. Film critic Bilge Ebiri explained, “Hollywood is in the midst of a transition… [they’re creating] Fewer gigantic productions that need massive opening weekends to justify their humongous costs; [and] more solid films that can turn a profit over a few weeks and months thanks to good word of mouth. The smash-and-grab opening-weekend strategy was never going to be sustainable, and the industry had become alarmingly reliant on an increasingly small handful of titles saving their bottom lines.”

This transition may include a larger reliance on sequels and related works. Of the five movies above, four aren’t entirely original stories. Barbie has the doll, Five Nights at Freddy’s is based on a video game, and Spider man and Maxxxine are both sequels. In total, 14 of the 46 movies we watched were direct sequels or part of a movie series, and several more were based on books or other works. TheFilmAutopsy explained sequels “just make more money,” and the limited data we collected tends to agree. On average, a sequel generated box office revenues equal to ~50% of its total budget during opening weekend, whereas novel movies only returned ~41%.

In general, longer movies had moderately better reviews (r=0.52, p=.002)

The correlation coefficient (aka “r” value) measures the strength and direction of the linear relationship between two variables. The value can range from -1 to 1. A value of 0 would indicate that two variables aren’t related at all (like your shoe size and reading ability), whereas a value of 1 indicates a perfect positive relationship between two variables (like degrees Fahrenheit and degrees Celsius). The r value of of 0.52 here indicates there’s a moderately positive relationship between a movie’s run time and it’d IMDb rating.

The p-value helps determine the significance of the results. In this scenario, the p-value is the probability of observing the results below if run time and IMDb rating were not correlated at all (meaning an r-value of 0). This dataset has a p-value of 0.0002 (0.02%), giving us confidence that these results are not random and that there is positive relationship between run time and IMDb rating. To be considered statistically significant, most fields look for a p-value of less than .05.

The line of best fit (aka regression line) shown in red is a straight line that best represents the data on the scatter plot below. The formula for the straight line can be line described by the formula y=mx+b, where y is the IMDb rating, m is the slope of the line, x is the movies run time, and b is a constant representing where the line crosses the y axis. Altogether, we can use this formula to estimate a movie’s IMDb rating as “IMDb Rating=0.016×Run Time+4.817.” The constant of 4.817 isn’t meaningful in itself, but suggest a movie with a run time of 0 minutes (a hypothetical scenario) would have an IMDb rating of 4.817. For every incremental minute, the slope of the line suggest that the IMDb rating would increase by 0.016 point.

This data makes intuitive sense given that longer movies have more time to develop characters and storylines, which could potentially lead to a better overall movie and higher ratings.

But strangely, getting better reviews doesn’t necessarily mean that more people are going to go see that particular movie on opening weekend (r=0.243, p=0.104)

The trend line in red does indicate a weak positive correlation between a movie’s IMDb rating and opening weekend box office revenue. However, the p value of 0.104 suggests that there’s a 10.4% of obtaining this result (or something even more extreme) even if the variables weren’t related at all. Based on that, we can’t confidently claim that higher IMDb ratings correlate with higher opening weekend box office revenues.

Why aren’t these variables more correlated? I had a few theories:

  • Opening weekend box office revenue is probably more related to how much the production company spent on marketing and advertising for the movie beforehand. You’re probably not going to go see a movie that you’ve never heard of, right? I’ll investigate this further below. Note I would have liked to use total box office revenue instead of just opening weekend box office revenue, however it wouldn’t have been fair since several of these movies are still in theaters.
  • More people want go to the movies around certain holidays or peak seasons (like Halloween or Valentine’s Day), even if the movies aren’t that great. For example, Brittany and I went to see the horror movie “Five Nights at Freddy’s” around Halloween and the romantic comedy “Anyone But You” around Valentine’s Day, even though we weren’t particularly excited about either film.
  • Reviews are a lagging indicator. I’ve noticed that a movie’s IMDb rating tends to start out very high and gradually decrease over time. For example, if you go see a movie on opening night, you probably already know that you were going to like the movie before it even started. But as more people go see a movie, the rating usually decreases closer to it’s truer and long-term average. Since these movies all came out at different times, this probably isn’t a fair dataset. Or maybe IMDb is just rigged.

Just like politics, it’s all about the money. Higher movie budgets don’t lead to better reviews (r=.089, r=0.55), but they do lead to higher opening weekend box office revenue (r=0.467, p=.001).

The chart on the left is all over the place, driving home the point that higher budgets don’t correlate with higher IMDb ratings. The r value indicates a weak correlation, however even if that were true, the p-value of 0.55 indicates that there’s a 55% chance that the observed data could occur even if budget and IMDb were totally unrelated. Therefore, we conclude that there’s no significant evidence that higher budgets lead to higher IMDb ratings.

On the other hand, the chart on the right indicates that higher budgets do lead to higher opening weekend box office. This isn’t a surprise. With more money, studios will bring in more famous actors, hire better producers, and spend more money on advertising. The r-value of .467 suggests a moderately positive correlation between a movie’s budget and opening weekend box office revenue, and the p value of .001 gives us high confidence that the correlation is statistically significant. The slope of the trend line (m=.202) suggests that for every additional million dollars in a movie’s budget, the opening weekend box office revenue is expected to increase by approximately $0.202 million (or $202,000).

So what does all this mean?

Our sample size is too small to reach any major conclusions other than we’ve had a great time on our 46 date nights at the movies. We’re looking forward to more movies coming up (Joker 2, Wicked, Bettlejuice, etc.), which may turn into more blog insights, but until then, check out my last movie blog where I’m “Reviewing Movie Reviewers.”

Reviewing Movie Reviewers

/ rkelleyc / 1 Comment

It all started with the “Barbenheimer.”

And now, 60 days later, I’m exhausted. Brittany and I have seen 11 movies (in theater) over the past two months. After learning that movie tickets can now cost up to $20.41 (no joke), we signed up for the AMC “A list,” a subscription service where you can see up to three movies per week for a monthly fee of $21.50. Our mission, which we chose to accept, was to see as many movies as possible to get our monies’ worth out of the subscription service. With the period coming to an end, I wanted to answer three key questions about movie reviewers:

1. What’s the difference between the different rating sources?

User Based Reviews:

IMDb scores are based on a weighted-average rating of all registered users (meaning everyday people). This is supposed to give you a good idea of what normal consumers think of the movie. However not all votes carry the same weight, which was designed to prevent individuals (or groups) from rigging the rating. IMDb says they don’t disclose that calculation “to ensure [their] rating mechanism remains effective.” Like many other user-based review sites, the biggest pitfall is that most people only submit a review when they have very strong positive or negative feelings about a movie, which skews the ratings in favor of either enthusiastic supporters or strong critics.

Audience Score, by Rotten Tomatoes, is similar to IMDb in that it represents the percentage of everyday users who rated a movie or TV show positively. There isn’t much information available on how the final score is tallied or if there are any weightings. Regardless, similar to IMDb, this score is susceptible to review bombing or inflated ratings by franchise cults.

Brittany’s Ratings. Brittany is my most trusted movie companion to see all these movies. Not only do we share similar tastes, but we get to experience these movies together, which whether we admit or not, does matter. For example, the theater was freezing cold during Haunted Mansion which literally created a chilling atmosphere. Or during The Equalizer 3, the projector was out of focus for the first 45 minutes of the film, leaving us both annoyed.

Critic-based Reviews:

The Tomatometer, by Rotten Tomatoes, is a score based on the opinions of hundreds of film and television critics. It gives a quick and reliable idea of whether a movie is worth watching. However, the biggest issue with the Tomatometer is that it breaks down complex opinions into a “Yes” or “No” score, and takes the simple average. So if every critic scored a movie 2.5 of out of 4 stars, the Tomatometer would consider all of those positive reviews and give the film a 100% rating, whereas a simple average would give the movie 62.5/100.

Metacritic collects reviews from a broad range of critics and aggregates them into one “metascore.” The individual scores are averaged but somehow weighted according to a critic’s popularity, stature, and volume of reviews through a secret process. Several people still consider this the most balanced aggregate score.

Source

2. Why are the ratings so different?

Using the 11 movies Brittany and I saw over the past 60 days, we can pull out the following takeaways:

A. User-based review sites seem more likely to be impacted by manipulation.

  • Compared to a professional movie critic, individual user ratings from IMDb and “Audience Score” seem more likely (or easier) to be influenced by hype, controversy, or organized efforts to flood a score with either overly positive or negative reviews to manipulate the score.
  • For example, the “Audience Score” seems particularly unreliable. It gave “Haunted Mansion” an 8.4/10, a surprisingly high rating compared to Metacritic (4.7), the Tomatometer (3.8), and my own rating (5). The film was notoriously a box office flop, only grossing $24M at the box office during opening weekend. Could Disney have paid or influenced users to leave positive reviews on the “Audience Score” to artificially inflate the movie’s score?

B. Weighted averages tend to lead to lower average scores.

  • IMDb and Metacritc both openly state that their scores are subject to some sort of behind-the-scenes weighting formula, whereas the Tomatometer is based on a simple average. The “Audience Score” doesn’t say whether it’s weighted or not, so I’ll assume it’s a simple average.
  • The average rating was 7.28 for IMDb and 6.66 for Metacritic; both lower than the simple averages taken from Tomatometer (7.68) and “Audience Score” (8.45).  This could be because the former sites exclude (or dilute) outliers and suspicious reviews like we saw with “Haunted Mansion” in section A. This also may help explain why user-based scores for IMDB and “Audience Score” are so different; because of weighted averages.  

C. Critics often rate movies lower than everyday movie goers.

  • Critics and audience members often have different criteria for evaluating films. For example, critics often consider cinematography, artistic value, and other technical aspects. Audience members, on the other hand, may be more influenced by sampling basis (i.e. only going to movies they’re likely to enjoy and rate highly), herd mentality (i.e. if Brittany likes a movie I’m inclined to agree with her), or the entertainment factor (i.e. the number of explosions).  
  • This discrepancy was most apparent for “Gran Turismo.” Metacritic’s score of 4.8 was significantly lower than the user based reviews from the “Audience Score” (9.8) and IMDb (7.4). Brittany and I also rated the movie high at 9.3 and 9.5 respectively. On the other end of the spectrum, Metacritic rated the artistic film “Asteroid City” at 7.4, higher than IMDb (6.7) and the “Audience Score” (6.2).

3. Which rating service most closely aligns with my own ratings?

My own ratings most closely align with the Tomatometer, however, I don’t think the Tomatometer tells the entire story by itself. I’d look to IMDb first given that A. My personal rating is more likely to align with other audience members (rather than critics) and B. IMDb appears to do a good job of sorting out outliers and manipulation.

To wrap it all up:

Ratings don’t always make sense. They can be good guides, but the magic of the cinema is largely rooted in your own personal taste and connection to the film. My favorite part of the movie-going experience has been the excitement of being in a sold out theater on opening night, always having something to talk about around the office water cooler, and having a weekly date with Brittany.

Oh and Nichole Kidman is annoying.