User:Djr13/OLMS graphing

Brief history: I added 46 of these throughout December 2014. It got overwhelming and I abandoned it for years, only later adding one more in June 2015 and updating six in May 2015(1, 2, 3, 4, 5, 6) and two in February 2016(1, 2). Not a single one of these were updated by another editor, though two were later removed. I looked eagerly toward the newly developing graph functions but was not prepared at the time to explore this new territory. This renewed effort begins (September 2019) thanks to edits by 124.170.216.250 and general maturation of the graphing and data tabulation features.

Old graph vs new[edit]

The old graphs, using Template:Line chart, had extremely redundant and cumbersome syntax, and required manual formatting changes to fit the data, and required data changes to fit the graph's limitations. They did carry a beneficial effect of this, tight control over formatting, to the degree where you could format the graph near perfectly to fit the data, and to align between graphs in a manner that facilitates spotting correlations between finances and membership. Someone even thought the old one was clearer.

Examples of the old template
Typical combination and placement of the graphs. (Source)
Breakdown of membership by category. (Source)
Combination of multiple national official membership records. (Source)

Thanks to 124.170.216.250 many of these old graphs have been recently replaced with a new format. This new format has new limitations, but overall is a huge improvement for maintainability. In the replacement version, numbers can be updated merely by adding a comma followed by the data (without worrying about numbers being too big).

Currently experimenting in User:Djr13/Sandbox and commons:Data:Sandbox/djr13/AFL-CIO.tab. Trying to improve formatting. Hope to use the new tabular data features if possible. Would be wonderful to automate things somehow, but there are always issues, whether with getting the reports, with the data on them, or with matters of factual context surrounding the data.

May be best to templatize these graphs for their particular informational, formatting and presentation needs. A dedicated membership-focused graph could potentially be made, a la Template:Graph:Population history. Being able to label major events (strikes/mergers/splits/legal changes) with visible (and reliably attributable) effects could be a great benefit.

Data notes[edit]

Around 2005, reporting rules changed so that agency fee payers are not counted in a union's total membership. Many unions with lots of agency fee payers thus saw a substantial drop in apparent membership in the report that followed.

It may be possible, and even preferable, to add agency fee payers, and breakdowns of notable membership categories (eg "non-voting", "retirees"). This requires actually looking at each report, identifying data more in-depth than that conveniently provided in OLMS's CSV exports.

I'm not absolutely certain what precise time or time range is used to count members. The most sensible (arguably) would be same as finances, the status as of the end of the reporting period. Note this period is not the same as submission date (is it the same as what OLMS calls "fiscal year"?), and it can even occur twice within the same year if the union adjusted its reporting period. Another sensible measure would be to count all people who had been members at some point during the reporting period (like Union (set theory)!), especially in industries with heavy turnover such as due to seasonal labor (see for example United Farm Workers), but this might not be accepted by OLMS. A union without historical membership records might also just use whatever number is current as of the date of filing. I've also seen obvious examples of estimation. In any case, it's possible the precise meaning of the figure may vary union to union, year to year; known issues should be explained when uploading and using the data.

Revisions happen too, revised data might need to be somehow overridden or discarded.

Overall, the data is expected to be accurate mostly due to threat of perjury.

Old graph vs new[edit]

Data notes[edit]

Index of graphed unions and relevant info[edit]