Examining how director experience predicts IMDb ratings and how this relationship differs between drama and action movies.
This study’s aim is to analyse whether a directors experience leads to higher IMDb ratings and whether this link depends on the type of movie (drama vs action). Directional experience will be measured as the cumulative number of minutes of movies directed in the past, rather than the number of movies. This choice reflects the fact that movies vary greatly in length. This measure accounts for both the quantity and scale of the director’s past work.
Besides analysing the main effect of director’s experience on movie ratings, this study also investigates this relationship with the moderating effect of action- and drama movies.
Academically Relevant
This study contributes to research on the movie industry by examining how directors’ cumulative experience predicts movie evaluations. The IMDb ratings capture consumer evaluation patterns. Prior research has shown that the director’s experience also influences the movie outcome. Directors with more experience are more likely to produce movies that achieve higher ratings (John et al., 2014). Furthermore, genre could play a crucial role, Francemone et al. (2023) examined 9,068 movies over twenty years and found out that action movies tend to be more popular with a higher audience causing higher box office numbers and more IMDb reviews than drama movies. However, drama movies are strongly linked with critical acclaim and get higher ratings (Francemone et al., 2023). This evidence shows the importance of considering genre when analysing how directional experience relates to movie ratings.
Managerially Relevant
For studios and investors making decisions, insights into how directors´ experience influences audience evaluations are a clear practical value. If director experience correlates with IMDb ratings, this makes hiring decisions easier. Experience may become a signal of quality and predictability of outcomes. Moreover, it is essential to consider differences in genres. Dramas typically rely more on director-driven elements (Francemone et al., 2023). In this context, director experience is more likely to enhance quality and reduce both artistic and commercial risk. Action movies, by contrast, often depend more heavily of technical production capacities. Here, managers may mitigate risk more effectively by investing in technical teams, rather than relying solely on director experience. For decision-makers, this implies differentiated allocation of talent and budget.
To what extent does a director’s prior experience (measured as the cumulative minutes of movies directed) predict a movie’s IMDb rating, and does this relationship differ between action and drama movies?
We used the following datasets:
name.basics.tsv
title.basics.tsv
title.crew.tsv
title.ratings.tsv
The data was obtained from the IMDb database.And after the data-engineering we ended with two final datasets:
- imdb_movies_direct with 114685 observations
- imdb_movies_mod with 60361 observations
in this table you will find the columns and a small description of each.
| Variable | Description |
|---|---|
| directors | Key ID of the director |
| total_runtime | Total experience of the director in minutes |
| avg_rating | Average rating of the movies directed |
| n_films | Number of movies the director has directed |
| most_common_genre | Predominant genre associated with the director |
Multiple linear regression analysis
To address our research question, multiple linear regression analysis is appropriate because it allows us to examine the relationship between a metric independent variable (directional experience) and a metric dependent variable (IMDb rating), while incorporating a non-metric moderator (movie genre: action vs. drama). This approach enables estimation of both direct and interaction effects within a single model and allows inclusion of control variables, improving interpretability and reducing potential confounding. Moreover, multiple regression provides a flexible framework to test whether the effect of director experience differs across genres. By measuring effect sizes and accounting for multiple predictors at the same time, the method offers meaningful insights into how experience and genre jointly influence movie ratings. In the preview you can find the visualization of this output.
The output of the project is visualized in these two scatterplots. With these two visualizations we can conclude the following:
- The slope of the red trend-line is slightly negative, which indicates that there is a negative correlation between the total director experience and average rating.
- The result is also influenced by the moderating effect of genre (action vs drama). In action movies the slope is slightly positive, while in the drama genre the slope is slightly negative. This shows that the main effect is depended on the moderating effect of genre.
Overall we can see that the effect of director experience and ratings differ across the two genres.
├── data/
│ ├── *- imdb_movies_dataset.csv*
│ └── *- raw_data/*
│ └── *- urls*
├── plots/
│ ├── - conceptualmodel.jpeg
│ ├── - Runtime_on_average_rating.jpg
│ ├── - Moderating_effect.jpg
│ └── - missing_values_imdb_movies.jpeg
├── reporting/
│ ├── - rmarkdown_group10.Rmd
│ ├── - report.Rmd
├── src/
│ ├── data-preparation/
│ │ ├── - makefile
│ │ ├── - Clean_and_merge.R
│ │ ├── - dataprep_directeffect.R
│ │ ├── - dataprep_modeffect.R
│ │ └── - download.R
│ └── analysis/
│ ├── - Analysis.R
│ └── - makefile
├── - makefile
├── *- gen/*
│ ├── *- temp/*
│ │ ├── *- ratings.csv*
│ │ ├── *- crew.csv*
│ │ ├── *- name_basics.csv*
│ │ ├── *- title_basics.csv*
│ │ ├── *- imdb_movies.csv*
│ │ ├── *- imdb_movies_direct.csv*
│ │ └── *- imdb_movies_mod.csv*
│ └── *- output/*
│ ├── *- visual_directeffect.png*
│ └── *- visual_moderatingeffect.png*
├── - .gitignore
├── - .Rhistory
├── - makefile
├── - README.md
├── -installpackages.R
├── *-Rplots.pdf*
└── - team-project-team-10.Rproj
*are created when running the workflow*
library(dplyr)
library(GGally)
library(ggplot2)
library(tidyr)
library(tidyverse)
To run this workflow, follow these steps:
- Fork this repository on GitHub.
- Open your command-line interface (e.g., Git Bash, Terminal, or Git GUI).
- Clone the repository to your local machine by running:
git clone <repository_URL>
- Run the install.packages.R to make sure all packages are installed correctly
- Navigate into the project directory and execute:
make
This command will run the entire workflow automatically.
- To clean the repository run ht efollowing code:
make clean
Now all generated files are deleted and you can run make again.
- Francemone, C. J., Kryston, K., & Grizzard, M. (2023). Examining genre success, co-occurrence, release, and production of 9,068 films over twenty years. Poetics, 99, 101794. https://doi.org/10.1016/j.poetic.2023.101794
- John, K., Ravid, S. A., & Sunder, J. (2014). Managerial ability and success: Evidence from the career paths of film directors. Journal Of Corporate Finance. http://dx.doi.org/10.1016/j.jcorpfin.2014.11.001
This project is set up as part of the Master's course Data Preparation & Workflow Management at the Department of Marketing, Tilburg University, the Netherlands.
The project is implemented by team 10
Members:
Daan Bankers snr:2100281
Luc Vesters snr:2145531
Seppe Bogers snr:2080522
Melvin Hong snr:2146253
Wouter van der Linden snr:2116790
Marvin Herwig snr:2138521