The process where resources accrue to those with more resources is called the Matthew Effect. It explains, amongst other things, why the degree distribution of social networks follows a power law.
There's a nice experimental test of this where showing the number of previous downloads a song has makes it more likely to be downloaded (but not to the extent that it entirely overrides the quality of the song.
<https://www.princeton.edu/~mjs3/salganik_dodds_watts06_full....>
> Examples include converting boxplots into violins or vice versa, turning a line plot into a heatmap, plotting a density estimate instead of a histogram, performing a computation on ranked data values instead of raw data values, and so on.
Most of this is not about Python, it’s about matplotlib. If you want the admittedly very thoughtful design of ggplot in Python, use plotnine
> I would consider the R code to be slightly easier to read (notice how many quotes and brackets the Python code needs)
This isn’t about Python, it’s about the tidyverse. The reason you can use this simpler syntax in R is because it’s non-standard-evaluation allows packages to extend the syntax in a way Python does not expose: http://adv-r.had.co.nz/Computing-on-the-language.html
This specific analysis isn’t p-hacking because although they conduct multiple tests, they report all of them rather than just the statistically significant ones.
They should however account for multiple testing. The Bonferroni correction (which is conservative) would set the alpha level to 0.05/5=0.01, for which the 1 day after result is still (just) statistically significant.
The 95% CIs on both timeseries pretty much always cover the baseline number, which is not consistent with the result being statistically significant.