by Phil Wilson

What’s happening with voter turnout in NLRB elections?

This week Bloomberg News reported that union win rates in NLRB elections have remained steady over the course of this year (pre- and post-Covid), meaning that mail-ballot elections haven’t helped unions as much as employers may have first feared. Our data from LRI RightNow agrees – unions won 67.7% of elections certified after the NLRB re-opened elections on April 4. Unions won 68.5% of elections certified this year prior to the shutdown.[1]

That wasn’t what popped out to me in the story, however. What caught my attention was the voter turnout numbers. The article stated that voter turnout in mail ballot elections this year was just over 42%. That’s shockingly low. The article also noted that average turnout over the prior five years was 47%. The article stated:

Voter turnout was 42.1% for both mail-ballot and in-person elections, the analysis of 2020 election results showed. That represents a dip from the 47% turnout for union elections from 2015 to 2019.


“Voter turnout has been unusually low so far this year, but that’s understandable, given the logistical challenges brought on by the pandemic,” Bloomberg Law labor and employment analyst Robert Combs said. “What’s surprising is that these challenges haven’t affected one method of voting more than another, in terms of either participation by voters or the outcome of the elections.”

I scratched my head for a while on this one. After all, we handle nearly 100 elections each year and I’ve been doing this work for longer than I want to admit. I can’t remember an election where turnout was anywhere near as low as 40%. I know low turnout does occasionally happen. But the political science major in me has always loved how NLRB elections represent what elections in the US should be like. Every vote matters, and the turnout is usually quite high especially when voters fully understand how the election works.

Even with the mail ballot elections we’ve worked I’ve been surprised at voter turnout. I finally concluded something must be wrong with the data.

In our database we show voter turnout numbers significantly higher than 42% and 47% mentioned in the article. We show a total of 254 elections certified prior to the re-opening on April 6. In those elections there were 12,085 eligible voters and a total of 9,470 votes cast, which is turnout of 78.3%.

After elections re-opened (and virtually all of these were mail ballot elections) there’ve been 475 elections certified, with 31,368 voters eligible and 21,159 votes cast, which is turnout of 67.4%.

Finally, from 2015-2019 we show 7,509 elections with 465,529 eligible voters, with 361,139 total votes cast, or a turnout of 77.6%. That’s basically the same as the pre-Covid elections certified this year.

Which numbers are right? I’m still investigating, but I’m pretty sure there was a mistake in the Bloomberg article. It looks like they are relying on reports from the NLRB, so it could be a mistake made in those reports. Or I also considered if they only counted union votes instead of all votes cast. I’ll let you know what I find out.

There is a bigger lesson here. The political science major in me also knows better than to make any political predictions for the upcoming elections. However, I think it is safe to say that there will be a LOT of statistics thrown around about union elections over the next few years. These will be used to justify or dispute the need for major labor law reform.

Just like in the world at large, there will be a fair amount of “fake news” about what happens in NLRB elections. That’s why it is more important than ever for practitioners to be careful about taking what they read or hear at face value. And that includes this article – maybe I’ve made a mistake. And rest assured that if I have, I’ll let you know.


[1] Our numbers differ slightly from the election win percentages in the Bloomberg article. This could be for any number of reasons. We look at the date an election is certified, but they may be looking at the date the election is held. We included all elections, but they may only be looking at RC elections. And with so many elections in the database there could be a mistake in the records. But our numbers are usually very close.