Bees did not unionize. But the reason you can read two well-sourced stories in the same week, one describing a labor resurgence and the other describing continued decline, is not that one of them is lying. They are usually measuring different things.
There are three separate quantities in play, produced by three institutions for three purposes, and the coverage almost never says which one it is using.
The first is union membership. In the United States this comes from the Bureau of Labor Statistics, which publishes an annual Union Members news release drawn from the Current Population Survey — a household survey, meaning it asks people about themselves rather than counting union records. It reports the number and share of employed wage and salary workers who are union members, and it has trended downward over the long run. It is a stock measure: where things currently stand.
The official series has a floor: it counts stoppages involving a thousand workers or more. Almost every strike you have heard about locally is below it.
The second is union elections. These come from the National Labor Relations Board, which administers representation elections and publishes petition and outcome data. This is a flow measure: how much organizing activity is happening right now. Filings and elections can rise sharply while overall membership share falls, because a surge of successful elections at small workplaces adds a modest number of members to a very large denominator. Both statements are true simultaneously, and a story built on either one alone will feel like it contradicts the other.
The third is work stoppages, and this is where most of the confusion is manufactured. The BLS work stoppages program has a threshold: it counts stoppages involving 1,000 or more workers. That floor exists for good historical and practical reasons, and it means the official series is silent on almost every strike you have heard about at a local hospital, café chain or school district. A year with a great many small strikes and few large ones can produce a headline about a record strike wave and an official count showing nothing unusual, without anyone being wrong.
That gap is why independent trackers exist. The Cornell ILR School’s Labor Action Tracker was built specifically to record labor actions below the federal threshold, and it consistently documents far more stoppages than the official series, because it is counting a different, larger set of events. Comparing the two directly is the error; using each for what it measures is the fix.
None of this requires distrusting any of the sources. All three are well documented and open about their methods. The failure is a reporting convention: a headline says “union” and a number, and drops the noun that would tell you which quantity it is.
So the check is one question, asked of any labor statistic: membership, elections or stoppages? If the piece does not say, the number is not yet meaningful, and the answer is usually in the first linked source rather than the article.
The bees, for the record, took the deal immediately, which is not how bargaining works either.