The Bechdel Test measures the following three criteria: (1) A movie features two women, (2) Two women talk to each other, (3) They talk to each other about something other than a man. A movie can get anywhere from a rating of 0 to 3 depending on how many of the criteria they meet. A score of 3 is passing and a score below 3 is failing. Pick a seat from the theater below and learn more about how movies performed during that year and decade.
Project Repository: https://github.com/washuvis/bvis
The Bechdel test checks to see if a woman in a movie's character is entirely about the men in the movie. It specifically asks three questions of movies: 1. Are there two or more female characters in the story? 2. Do they talk to each other, and 3. Do they at any point have a conversation unrelated to a man? This is an important baseline test to diagnose representation in movies though the criteria for passing or failing the test might seem simple to most, the amount of popular movies and works of fiction that fail this test are numerous. Hollywood movies are extremely influential to US culture, and in many cases, emphasize stereotypes and cultural norms. We wanted to design a webpage that makes use of the Bechdel test to highlight these disparities in movies through our data visualizations and try to map trends across time.
We were inspired to do this project as 50% of our two person group identifies as female, and 100% of people who watch movies are affected by the influence that movies have on society and cultures worldwide. As of milestone 1, we don't have any specific studies that we saw that inspired us besides viewing this as an interesting topic and one that would be suitable to study with data visualization. However, if you've seen Avengers: Endgame, there's a certain scene in the third act of the movie where all the female Avengers band together which raises some questions as to what fair representation for women in movies looks like, especially in that movie.
Our preliminary research questions are as follows:
As of milestone one, we've used the following datasets:
Here are some sources we used to help write the code.
Libraries used:
The Bechdel Dataset is a website that has data on around 8000 movies. Anyone is able to submit movies along with data on their Bechdel rating (between 0-3), and each entry also includes information on a movie's imdb value and rating.
The Movies Dataset contains data on over 45,000 movies and has lots of metadata for each movie. We matched entries in the Bechdel set with entries in the Movies set by imdb id and kept in information about revenue, budget, genre, and imdb user votes.
In our initial design we thought to use a bar chart to determine if there were differences between average Bechdel rating within decades/years to compare to one another. It's difficult to glean anything of value at that high level given the nature of the Bechdel dataset being operated by a community, and some years having significantly more movies available than others. We then decided to switch up our approach to have our visualizations to allow users to get deeper into the data and put each decade/year more on an even field with the others.
We started with the idea of having a bar chart be the main controller for the other visualizations. The idea was to start at a high level, and then go deeper into the data and time periods, down to the movie, as the user interacts with the charts. Here are two images of that concept:
.png)
.jpeg)
We realized that we could be more creative with how we hook a user onto our webpage, and that there are better ways to do that. We decided instead to create a visualization based on nodes and thematically tie it into movies. Each node would represent a year, sized by proportion of movies that pass/fail the test and colored by decade. Clicking on one of these takes the user further down the page to see more information about that particular year and decade through the other visualizations Here are some sketches that indicate our ideas for the orientation
As of milestone one, here is our current primary visualization: it utilizes the concept of the spotlight sketch but flipped over.
This visualization only includes the nodes from the 90s for Milestone 1, but will be reshaped into a different movie-related object or fancier design once we reintroduce the other 60 nodes. We also intend to adjust the positioning of the svg and images to be responsive and centered.
For milestone 2, we will also include sketches and implementation of the search and compare visualization. It will include a form that allows users to try and search for movies within our database (from any year available), and then display all the different attributes we have such as pass/fail, budget/revenue, genres, rating, etc.
As of Milestone 2, here is our updated initial visualization, designed to emulate reserving seats from a theater online.
We also added in a search function that we stated our intentions for in the milestone 1 push. It displays four categories of data in four different quadrants (general, financial, Bechdel, and popularity/rating). Here is a sketch of the initial thought process, as well as a screenshot of the execution.
We've covered the changes and the implementation strategy behind the primary visualization. We also created a line chart that draws two lines over the span of a decade: the points on the lines represent the cumulative revenue of movies that pass or failed the Bechdel test in a given year. This revenue line chart correlates revenue to the movies that passed and movies that failed the test in a given year, and allows users to see trends in whether or not movies that pass or fail the test make more money or not. We calculated the sum of movies that pass/failed the test in a year and put them into a csv. Here you'll find our initial sketches for the revenue chart as well as the current prototype
.jpeg)
In our update for milestone 2, we intend to restyle the axes on the chart, make the points more distinct, and add an interactive tooltip component that will show a user the exact value of a given point on hover.
We implemented a bubble chart that groups the movies that pass or fail the Bechdel test for a clickable year. In Milestone 1, we have the data for each year in the movies from the 90s. Here you'll find our initial sketches as well as the current prototype.
.jpeg)
For milestone 2, for interactivity, we will implement a tooltip that shows the user (on hovering over a node) what movie they are hovering on. We will also fix the chart to have an encompassing green and red circle over the other nodes and adjust opacity, as well as increase the overall size, positioning, and responsiveness (to adjusting screen dimensions) of the chart.
For milestone 2, we added an interactivity element and converted to a pack chart to prevent nodes from overlapping.
We wanted to also determine if certain genres had more movies that passed or failed the Bechdel test. In our proposal we envisioned this as being a pie chart where we had genres occupy a proportion of the overall circle. Here you'll find our preliminary sketches for the idea:
.jpeg)
We realized that it would be more effective to convey this information in a bar chart, with genres on the x-axis and count of movies that pass/fail on the y-axis. Each genre would have two bars, one for passing and one for failing movies. Here is a sketch of what the revised version would look like
Our final version will implement two bars along with a tooltip that shows how many movies pass/fail a category on hover.
The milestone 1 and 2 versions can be seen here, with the milestone 2 version having both bars per genre.
We believe our visualizations can provide answers to all of the questions of interest we outlined when starting our project. The visualizations all provide different kinds of insights into the data surrounding movies, and a user can easily access our website and interact with the visualizations, movie data for different years/decades, and even specific movies as well. Our tool provides a lot of opportunities for user engagement whether they're curious about multiple years' worth of data or individual movies themselves, and our nav bar which we added after user feedback gives them a quick way to move from visualization to visualization.
From our visualizations, we were able to see different trends in years/decades. If you click on one of the earlier decades and compare them to a later year (such is starting with a year in the 60s vs a year in the 2010s), you can see the visual difference in how many movies passed or failed the Bechdel test through our bubble chart. The bar chart is also a place to draw interesting conclusions from as some genres like action have a high proportion of movies that fail the Bechdel test across multiple years. Of course, our tool would be better with more accurate and complete data though from a conceptual standpoint it does what it sets out to do.
In user testing we received feedback on clarifying what the Bechdel test was, and also feedback surrounding styling, clarity, and usability. We implemented the navbar, updated some styles, and added more information wherever we could such as legends and tooltips to help make the site more interactive and clear. If we were to further improve, we would overhaul the styling and make the graphs and colors look a bit cleaner as well as make some small adjustments to how some numbers are displayed.