Let's discuss how to present/discuss the results in the mosaic plots! For those of you who might not recall, these figures, while complex, show important contingencies in data. I'd like to have a section on these figures, mostly to push forward some of the salient points they bring up.
In a nutshell, mosaic plots like this are interpreted by starting on one axis, each of which represents a category, and looking at the corresponding rows/columns of categories one is interested in seeing correlations in. There's more information here: https://en.wikipedia.org/wiki/Mosaic_plot
The example plot I show above is quite confusing, because it has a number of things all showing at once. In the paper we may choose to stick to one or two more simple examples, such as the following:
Here we see that women are much more likely to ask a question if the first question is from a woman. We can combine this with another plot that shows that the first question is more likely to be from a woman if the chair is a woman (likely due to the fact that the chair often asks the first question in quieter sessions) to argue that having more women chair sessions will help to improve representation in Q/A sessions.
Thoughts? I can generate any combo of the following variables in making these mosaic plots: chair, speaker, question, and first question.
Let's discuss how to present/discuss the results in the mosaic plots! For those of you who might not recall, these figures, while complex, show important contingencies in data. I'd like to have a section on these figures, mostly to push forward some of the salient points they bring up.
In a nutshell, mosaic plots like this are interpreted by starting on one axis, each of which represents a category, and looking at the corresponding rows/columns of categories one is interested in seeing correlations in. There's more information here: https://en.wikipedia.org/wiki/Mosaic_plot
The example plot I show above is quite confusing, because it has a number of things all showing at once. In the paper we may choose to stick to one or two more simple examples, such as the following:
Here we see that women are much more likely to ask a question if the first question is from a woman. We can combine this with another plot that shows that the first question is more likely to be from a woman if the chair is a woman (likely due to the fact that the chair often asks the first question in quieter sessions) to argue that having more women chair sessions will help to improve representation in Q/A sessions.
Thoughts? I can generate any combo of the following variables in making these mosaic plots: chair, speaker, question, and first question.