What this post is about There’s quite the fuzz about ‘objective’ measures of technology use. When we study the effects of tech use on well-being, we often just ask people to report how much they used tech. Turns out, people aren’t really good at estimating their tech use. When we compare their subjective estimates to logged tech use (hence ‘objective’), the correlation between the two is rather low as this meta-analysis reports.
What this post is about Creating data Creating a violin plot (with wrong errors bars) Creating another plot (with error bars that’re still wrong) The last plot (this time correct) What this post is about Personally, I find violin plots with error bars a great way to present repeated-measures data from experiments, as they show the data distribution as well as the uncertainty surrounding the mean.
However, there is some confusion (at least for me) about how to correctly calculate error bars for within-subjects designs.