tidystats logo

About

I developed tidystats to solve two different problems in statistics reporting—incorrect statistics and incomplete statistics reporting.

Incorrect statistics reporting

Many scientific papers contain statistical inconsistencies. These inconsistencies are detected via software like statcheck. This software detects statistics and re-calculates certain statistics in a paper to see whether the reported statistics match the re-calculated statistics. This technique has revealed that often the statistics don't line up, meaning that the reported statistics are incorrect. This shouldn't be surprising because the most common method for inserting statistics is by simply manually typing them into a text editor or by copy-pasting results from the statistics software. In either case it is easy to make a mistake, causing these statistical inconsistencies.

Incomplete statistics reporting

Not all statistics are reported. Researchers often only report statistics in their papers, which means they need to carefully balance comprehensive reporting with writing a legible text. As a result, statistics are sometimes summarized or omitted in favor of producing a clearer paper. While this may seem like a reasonable compromise, the statistics are still necessary to check conclusions or for use in subsequent work (e.g., meta-analyses).

Solutions

There are several possible solutions to these two problems. To give everyone access to the statistics, one can share the data and scripts of their study so that others can run the analyses themselves and obtain the required statistics. However, this is not (yet) very common, it is often time-consuming to do, and sometimes it is not possible to share the data.

A solution to reducing statistical reporting errors is to write scientific papers using reproducible software solutions like R Markdown or Quarto. However, there are some downsides to using this kind of software. It is often more difficult to collaborate with others because not everyone in the team knows how to work with this kind of software and the software is not as feature rich as more well-known text editors such as Microsoft Word or Google Docs.

tidystats

tidystats is my attempt at producing a user-friendly solution to the problems of incorrect and incomplete statistics reporting.

tidystats works in two steps:

  1. Save the output of statistical tests performed in R
  2. Report statistics in Microsoft Word (and soon in Google Docs)

The tidystats R package is used to create a machine-readable .JSON file that contains all the statistics from your analyses. This file can be read by other software, such as the tidystats Word add-in. This addresses the problem of incomplete statistics reporting because the file contains all the statistics. You can report the statistics easily using this file, or share the file itself so all statistics are shared.

The tidystats Word add-in can read the file created by the R package so that you can insert the statistics in your document using a human-friendly interface. Simply click on the statistic(s) you want to insert and the statistic(s) will be inserted into your document. This addresses the problem of incorrect statistics because it does not require manually typing in the statistics.

If you want to know more, check out the page about the R package, Word add-in, or the Google Docs add-in that's currently in development.

Finally, tidystats is developed by me.