Fig 1.
Data set by audience and author type.
Each block represents one article.
Fig 2.
Data set by year of publication and audience.
Fig 3.
Correspondence analysis biplot of 353 articles discussing reproducibility, analyzed for 29 themes.
Articles that are close together have similar narrative profiles. The closer an article appears to the center of the plot, the more closely it resembles the mean profile for all articles. The further away a theme is from the origin, the more variation there is in how authors discuss that theme. The color of an article’s plotted point (a circle) indicates the main term used in the article, and the size of a theme’s plotted point (a square) represents the contribution of that theme to constructing the dimensions. The eight most contributing themes are labeled. Supplementary variables (not used to construct the dimensions) are labeled in red.
Fig 4.
Correspondence analysis biplot depicting dimensions 1 and 3 of the analysis.
Articles that are close together have similar narrative profiles. The closer an article appears to the center of the plot, the more closely it resembles the mean profile for all articles. The further away a theme is from the origin, the more variation there is in how authors discuss that theme. The color of an article’s plotted point (a circle) indicates their label in a hierarchical clustering, based on Euclidean distance in the latent dimensions. The size of a theme’s plotted point (a square) represents the contribution of that theme to constructing the dimensions. The eight most contributing themes are labeled.
Fig 5.
Multiple factor analysis of bootstrap samples drawn a set of 353 articles on reproducibility.
Shaded areas indicate the peeled convex hulls of points based on 1000 bootstrap replicates plus the original sample, showing an approximate 95% confidence region for each theme. Twenty-nine codes were included in the analysis, but only select codes are displayed for legibility. Colored squares indicate the position of each theme based on the original sample, and black squares indicate the position of each theme based on the overall bootstrap analysis.
Fig 6.
Multiple factor analysis of 353 articles discussing reproducibility, analyzed for 29 themes, with articles grouped by author type.
The size of the square indicates the degree of within-theme inertia. The eight themes with the greatest within-theme inertia are labeled, and partial points for those themes are displayed. Red points indicate the location of those themes in the group of articles authored by scientists, and blue points indicate the theme location for the journalist group. Grey points indicate the location of the articles on the first factor plane.
Fig 7.
Multiple factor analysis of 353 articles discussing reproducibility, analyzed for 29 themes, with articles grouped by intended audience.
The size of the square indicates the degree of within-theme inertia. The eight themes with the greatest within-theme inertia are labeled, and partial points for those themes are displayed. Red points indicate the location of those themes in the group of articles aimed at a scientific audience, and blue points indicate the theme location for the popular audience group. Grey points indicate the location of the articles on the first factor plane.