statistics

Some folks like to look at current polls and poll averages and make a mental adjustment to the results and see how they’d look if the polls missed the way they did in the recent past. But I think even for people who follow this stuff pretty closely, it can be hard to remember all those details. My goal here is to help with that kind of thing while also showing that the narrative about polling errors is probably an overgeneralization.

Some folks like to look at current polls and poll averages and make a mental adjustment to the results and see how they’d look if the polls missed the way they did in the recent past. But I think even for people who follow this stuff pretty closely, it can be hard to remember all those details. My goal here is to help with that kind of thing while also showing that the narrative about polling errors is probably an overgeneralization.

When you want to find out who the NFL’s best kickers are, the tools available to you tend to be pretty crude, with not much more than a record of makes and misses. We all know from watching the games that some kicks are harder than others, but how do we synthesize that into a few easy to understand statistics?

Here I discuss statistical and logistical details involved in my attempts at modeling kicker talent. This includes model selection as well as a brief history of research. I use some quantitative assessments to compare my approach to previous attempts and it does quite well.

Observers note that in the 2020 Democratic primary, candidate support in polls is closely correlated with the amount of attention candidates get in the news. This begs the question of whether and how these two things might be causally related, particularly whether news coverage is helpful to candidates. Here I show some evidence that news coverage does indeed help.

It has been just a little more than a day since I announced my new R package, panelr, to the wider world. There’s at least one comparable package for R, called plm, which is very good and should be particularly appealing for economists. This leads to the understandable question as to how panelr differs from plm.

It has been a long time coming, but my R package panelr is now on CRAN. Since I started work on it well over a year ago, it has become essential to my own workflow and I hope it can be useful for others. panel_data object class One key contribution, that I hope can help other developers, is the creation of a panel_data object class. It is a modified tibble, which is itself a modified data.