Johan Brynedal Öckinger @ockinger.bsky.social · 10/12/2024Yes but also that bar charts aren't the best choice if you data is anything but a count. Too much of the underlying data is not represented. 110
Richard Gill @gill1109.bsky.social · 10/12/2024I disagree. If you want to visualise the differences between four relative frequencies, four bars with confidence intervals to indicate precision is not bad. Sure, there are other displays which give more information but which require more technical understanding. 210
Johan Brynedal Öckinger @ockinger.bsky.social · 10/12/2024I don't know exactly what type of data you are using. And maybe what you describe is the norm in your field. But to correctly interpret a confidence interval or standard deviation requires a deeper statistical understanding than to interpret a boxplot, for example. 220
Rob Campbell @raacampbell.com · 10/12/2024Does not solve the problem. Confidence intervals for really skewed data are happening. It requires minimal technical understanding to overlay scattered raw data or do violin plots. 010
