The goal of this project is to examine the relationship between movie running time and audience ratings, and how this relationship may differ between a selection of movie genres.
The movie industry continuously tries to understand what factors influence audience ratings, as they play an important role in determining commercial success and popularity. One characteristic that might impact viewers’ perceptions is running time. Longer films can allow for more extensive storytelling and character development; however, they may also demand higher mental effort and time commitment from the audience. As a consequence, the relationship between running time and audience ratings remains unclear and worth investigating.
Previous research in media and entertainment studies suggests that structural movie characteristics can influence audience evaluations and satisfaction (Cini & Abela, 2025).
Furthermore, audience expectations may differ between movie genres. For instance, viewers might expect longer runtimes in genres such as drama, whereas faster-paced runtimes may be expected in low cognitive effort genres such as comedies. Genre-based expectation theories propose that consumers develop category-specific expectations that influence their evaluations (Jeffres, Atkin, & Neuendorf, 2023).
In addition, theories related to cognitive load and information processing suggest that increased exposure time may influence judgments due to differences in mental effort.
Based on these notions, the main research question that will be examined is described as:
- What is the relationship between running time (IV) and audience ratings (DV) in movies, and how does this differ between movie genres?
To examine the research question, two publicly available datasets from the IMDB website were utilized and later merged.
The data sets used:
After merging, the final dataset contains 63.619 observations after filtering for movies, their running time (<=360 minutes) and number of votes (>100), where missing values for genre's and audience ratings were filtered out.
- Include a table of variable description/operationalisation.
The main variables that will be examined:
• Movie running time (IV) - primary runtime of the title, in minutes
• Audience rating (DV) - weighted average of all the individual user ratings
• Genre (Categorical Moderator) - includes up to three genres associated with the title
For data preparation, data wrangling principles were administered. Firstly the dataset title.basics.tsv.gz was filtered to only include movie titles, thereby excluding other included film-types in the dataset such as tv-shows and short-films. Next, the dataset title.ratings.tsv.gz was merged. This merged dataset was structured in ascending order of release year.
Moreover, next to filtering for missing values for genre and audience scores, the observations were filtered to have a running time of up to 360 minutes and a minimum number of 100 recorded audience votes.
As for the method of analysis, Multiple Linear Regression (MLR) with interaction terms will be used. MLR is an appropriate analysis method for a continuous IV and DV, allowing examination of interaction terms for movie genre as a moderator.
genre-runtime-effect/
│
├── data/ # Data-related scripts and raw data handling
│ ├── placeholder.txt # Placeholder until data is downloaded
│ └── raw/ # Raw downloaded datasets
│
├── reporting/ # Final report and application files
│ ├── report.Rmd # Main analysis report
│ ├── group project Dprep week 3.Rmd # Initial research proposal
│ ├── start_app.R # Option to run the report using a Shiny app
│ └── makefile # Makefile for rendering the report
│
├── src/ # Source code
│ ├── analysis/ # Statistical modeling and analysis scripts
│ │ ├── analysis.R # Main analysis script
│ │ └── makefile # Makefile for running the analysis
│ │
│ └── data-preparation/ # Data cleaning and preparation scripts
│ ├── download-data.R # Script to download raw data
│ ├── data-cleaning.R # Data cleaning script
│ └── makefile # Makefile for data preparation
│
├── README.md # Project documentation
├── makefile # Main workflow automation
└── .gitignore # Files excluded from version control
The analysis reveals that movie runtime has a small but significant positive effect on audience ratings overall. Longer movies tend to get a slightly higher rating. However, the relationship differs based on genres.For example, Thriller films benefit the most from longer runtimes, compared to Comedy or Adventure films that tend to receive lower ratings as runtime increases. Drama and Horror films show no significant interaction with runtime.
The findings are deployed as a reproducible PDF report generated with RMarkdown, which includes regression tables, visualizations of interaction effects, and clear interpretations of the results.
These findings are relevant for filmmakers, editors, and streaming platforms, as they suggest that the runtime for a movie is not universal but depends heavily on genre. Understanding these can help professionals take informed decisions regarding their productions.
To recreate and run this repository's workflow, a recent version of R Studio is necessary.
The main packages needed to run this workflow in R Studio:
- R Package: readr - used to read selected datasets
- R Package: tidyverse - used for data wrangling and analysis
- R Package: ggplot2 - used for creating visualizations
- R Package: broom - used for tidying model outputs
- R Package: knitr - used for rendering tables in the report
- R Package: rmarkdown - used for rendering the RMarkdown report
To load these packages in R:
# Load libraries
library(readr)
library(tidyverse)
library(ggplot2)
library(broom)
library(knitr)
library(rmarkdown)
In addition, make is required to run the automated workflow. It can be installed via:
- Windows: install from https://gnuwin32.sourceforge.net/packages/make.htm
- Mac: available via Xcode Command Line Tools (
xcode-select --install)
Pandoc is required to render the RMarkdown report to PDF. Pandoc can be downloaded from:
- Windows (recommended version 3.9.0.2): https://github.com/jgm/pandoc/releases/tag/3.9.0.2
A LaTeX distribution is also required for PDF rendering. The easiest option is to install TinyTeX from R:
install.packages("tinytex")
tinytex::install_tinytex()
To run this repository's workflow, you first need to clone the GitHub repository.
Using Command Prompt on Windows/ Terminal on Mac, clone this repository:
git clone https://github.com/course-dprep/genre-runtime-effect.git
cd genre-runtime-effect
Run the full workflow (download data, clean, analyze, and generate the PDF report) with a single command:
make
This will execute all steps in the correct order: data preparation → analysis → report generation.
To remove all generated output files (the gen/ and data/ folders), run:
make clean
After following these steps, any interested people should be able to recreate, or further analyse the IMDB datasets appropriately.
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 4, members: Dalyl Lachkar, Ahmad Meirghadam, Francesca Stoica, Youri Vandenbussche, Wanja Koekman