IF YOU CAN, REPLACE TABLE() WITH TABULATE() IN YOUR #rstats CODE!

data <- sample(1:3, 100, replace = TRUE)
microbenchmark::microbenchmark(table(data), tabulate(data), times = 10000)

I'm almost angry at how much faster this is, and why I was today years old when I learned this 🤬
1/2

So, in exploratory analyses, we explicitly modelled the effect of intrusions on false memories. We found that only in the sleep group, higher memory performance is associated with more false memories - but only when intrusions are high.

7/10

Contrary to Diekelmann et al., we find INCREASED false memories when memory performance is high - especially in the sleep group, and especially when looking at intrusion-adjusted memory performance, i.e., taking into account the # of responses that are neither list words nor false memories.

6/10

Participants' memory was tested either after a period of wake (day) or sleep (night). Consistent with Diekelmann et al., we find no effect of sleep on false memory production or veridical memory performance (whether the number of correctly recalled words was adjusted by intrusions or not).

5/10

In an online study, N = 398 participants completed a DRM paradigm: They learned different word lists; all words within a list were associated with a "critical lure" that was never presented. Participants tend to falsely recall this critical lure in later memory tests.

4/10

I have a lot of these fake signal update notes in my timeline ('Hidden "invite random journalist" button moved to avoid accidental traps'). And yes, I DO get the joke and think it's funny, but I wanted to point out that the ACTUAL signal update notes are funny as well.

github.com/signalapp/Si...

Lastly, screenshot of the code that generated the plots in the second post.

If I keep annoying chatGPT, it WILL give me the correct code, but apparently thinks (like me until recently, admittedly) that height = 0 is the default.

ChatGPT, notoriously bad at ggplot2, will btw make the same mistake.

In the first plot shown here, I set a jitter width. In the second one, I set the height explicitly to 0. Note how the y-positions differ! (I set a seed for comparability)

The default behaviour of geom_jitter() and position_jitter() is to jitter the width/height if no argument is provided!

Enter patchwork!
The plot areas are now exactly the same size, without me having to say a word!
Code below.
2/4

I am very late to the party, but can we talk about the absolute LEGEND patchwork by @thomasp85.com?!

I have two plots, one with a hidden y axis (it's the same for both plots). But the plot area will greedily take up the space freed up by the axis, so it appears larger in cowplot's plot_grid().
1/4