Analysis of Movies and their Trailer Release Dates
This is the analysis about the determinants of the period between movie trailer release dates and actual movie release dates.
Let’s define the Promotion Period as the period between the official movie trailer release date and the movie release date. The picture above visually shows what the promotion period used in this analysis means. The next picture displays the distribution of this promotion period for the movies released in the US between 2017 and 2019.
As seen in the graph above, there is a large variation in the promotion period. Although about 30 to 60 days have the most movies, many movies still have more than 100 days.
What makes this variation? Movie distributors use various channels to promote their movies. This article starts with the question of whether the variation in the promotion period is one of the strategies for promoting the movies. The goal of this post is to examine the factors related to the decision about the length of the promotion period.
The Data Sets
I scraped the two movie websites: 1) boxofficemojo.com and 2) traileraddict.com. The general method of scraping these information is described in this article.
The movies used for this analysis played in 2017~2019 in theaters. The total numbers of movies scraped from the boxofficemojo and the traileraddict are 2,711 and 25,047, respectively. The reason why the number of the movies scraped from the latter is much larger than the former is that the latter is the list of trailers for movies and each movie has several versions of trailers. The data sets scraped from the movie sites are as follows:
The promotion period was defined as the period between the official trailer release date (trailer_date in trailderaddict)and the movie release date (release_date in boxofficemojo). The promotion period is calculated in terms of days. In order to remove outliers, only the movies with the promotion periods less than 365 days are used here. Then, our data set has 947 rows.
There are several issues regarding combining the data sets. It is well explained in detail here.
Which Factors Are Related to the Decision about the Promotion Period?
Distributors
The first factor to be checked is the size of distributors. To see if there are significant differences in the promotion period across distributors, I picked 4 major distributors and 4 minor distributors. ‘Walt Disney Studio’, ‘Warner Bros.’, ‘Sony Pictures’, and ‘Universal Pictures’ are the movie distributors with market share greater than 10% respectively. The market share of ‘Well Go USA Entertainment’, ‘The Orchard’, and ‘Strand Releasing’ is below 0.1% respectively. ‘IFC Films’ is the company which distributes mainly independent films and its market share is around 0.2%.
As seen in the graphs above, the promotion period for the major distributors is generally greater than those for the minor distributors. This implies that the size of the distributors is related to the length of the promotion period.
Budget
Regardless of the actual quality of the movies, a budget can be a good proxy for the expectations that distributors have in regards to the future success of the movie. We have 241 entries with non-missing values in budget. Let’s see the scatter plot between the budget and the promotion period.
The graph above slightly reveals the positive relationship between budget and the promotion period. The size of markers for each distributor represents its rank in the market share. The correlation coefficient between the promotion period and budget is 0.2935.
The correlation coefficient is weakly positive at the 1% significant level. In the previous section, distributors appear to be related to the variance in the promotion period.
Here, we can has a question about if the size of distributors can affect the relationship between the promotion period and budget. To see this, let’s draw the same scatter plots and calculate the correlation coefficients for each distributor.
The order of the distributors in the table is by the market share. Except for ‘Walt Disney’, as the distributor’s market share increases, the correlation between the promotion period and budget gets stronger and more significant.
MPAA
The Motion Pictures Association of America (MPAA) film rating system is used in the US to rate movie suitability for certain audiences based on the movie content. The number of each MPAA rating in our data set is as follows:
Since the rating ‘G’ and ‘Not Rated’ have very small numbers, they are excluded in the analysis. Let’s see how the distributions of the promotion period for each rating group are different.
The graphs above show that the distributions between ‘PG’ and ‘PG-13’ movies seem to not have any significant difference. However, ‘R’ rated movies seem to be slightly lower than others. To see if the difference is statistically significant, let’s do the t-test and Kolmogorov-Smirnov(KS) test.
The t-statistic between PG and PG-13 movies to test the difference of means between two samples is not statistically significant even at 10% level. This means that we cannot reject the null hypothesis that the means of the two distributions are equal. KS-test is to measure the maximum distance between the two cumulative distributions to see if the two distributions are different. As KS-statistic is bigger, the difference between the two distributions gets larger. In the case above, the p-value for KS-statistics is 0.4548, which means that the difference is not significant at the 10% level.
In regards to comparing ‘PG-13’ and ‘R’ rated movies, the mean of the promotion period for ‘R’ rated movies appears to be significantly lower than the mean of ‘PG-13’ movies.
To sum up, the distributions of the promotion period are likely to be correlated to MPAA between ‘PG-13’ (or ‘PG’) and ‘R’.
Genres
The genres for each movie are expressed as the combination of several base genres.
For example, the genres for ‘Avengers: Endgame’ are recorded as ‘Action, Adventure, Drama, and Sci-Fi’. To check the relationship between the promotion period and genres, I created dummy features for each base genre. If the genres of a movie contain ‘Action’, the value of Action feature for the movie becomes 1 and otherwise, it becomes zero.
The total frequency for each base genre in the data set is as follows:
This is how to examine the relationship between the promotion period and genres. For each genre, I divide the data into the two groups: One group with the value of the genre = 1 and the other group with the value of the genre = 0. Then, I test the t-statistics and KS-statistics between the two groups. The results for t-statistic and KS-statistic for each genre are as follows:
The t-statistics and KS-statistics for some genres are significant. Especially, the distributions of the promotion period for the movies with Action, Adventure, Sci-Fi, Fantasy, Family, or Animation are relatively long compared to other genres.
Conclusion
The promotion period defined in this analysis is not a topic often covered by others. Using the data scrapped from the boxofficemojo.com and traileraddict.com, I explored how this promotion period is correlated to other factors of the movies: distributors, budget, MPAA, and genres. The results of the graphs or the tests imply that the promotion periods are not randomly determined, but are related to the size of distributors and budgets, MPAA, or the genres of the movies.
The data and the code for this analysis can be found here.
