This article forms part two of The Crimson’s four-part series analyzing exit polling data collected during Harvard’s historic 2018 unionization election. Read part one here, part three here, and part four here.
Arts and Humanities and Social Sciences students were much more likely to vote to unionize in Harvard’s election last week than were Sciences and Engineering and Applied Sciences students, according to exit polling data collected by The Crimson.
Harvard held its second-ever unionization election over the course of two days on April 18 and 19. The final results tipped in favor of unionization, with 1,931 ballots cast for the union and 1,523 against. This outcome means over 5,000 eligible graduate and undergraduate students can now begin to collectively bargain with the University as members of Harvard Graduate Students Union-United Auto Workers.
The Crimson conducted an exit poll throughout the day on both election days, stationing reporters outside voting sites for 78 percent of all hours the sites remained open. Of the 3,454 students who cast a ballot in the unionization election, The Crimson collected exit surveys from 1,295 voters—representing more than a third of the voting population.
The 23-question surveys asked students how they voted, various demographic questions, and their views on various campus issues. The surveys were later processed electronically with open-source software.
Exit poll data showed that Arts and Humanities and Social Sciences students overwhelmingly voted in favor of unionization. Of survey respondents, 91.1 percent of Arts and Humanities students and 88.5 percent of Social Sciences students voted pro-union, respectively.
By comparison—of survey respondents—49.2 percent of Sciences students and 28.2 percent of SEAS students voted in favor of unionization.
The Crimson did not adjust these results for response bias (see our methodology below).
Broadly, exit polling data also suggested students in their mid to late thirties were most likely to vote in favor of unionization, while students in their late teens and early twenties were least likely to vote in favor of unionization.
A significant majority of respondents between the ages of 32 and 39—89.1 percent—voted to unionize. Slightly less than half, or 47.5 percent, of respondents between the ages of 18 and 22 voted in favor of unionization.
Respondents between the ages or 23 and 28 and between the ages of 29 and 33 voted pro-union at 66.6 percent and 82.6 percent, respectively. Of respondents over 40, 73.7 percent voted to unionize.
Respondents from the Kennedy School and the Graduate School of Education favored unionization by the greatest percentage—with 91 and 90.4 percent yes votes, respectively—while respondents from the the Medical School were the least likely to vote yes, at 38.5 percent.
In addition to the Kennedy School and Ed School, four other schools saw a majority of affiliated respondents vote in support of unionization. School of Public Health respondents voted yes at a rate of 88.7 percent, Law School respondents at a rate of 88.5 percent, Graduate School of Design respondents at 80.7 percent, and Graduate School of Arts and Sciences respondents at 66.4 percent.
By contrast, a majority of respondents from the College and from the Faculty of Arts and Sciences—which includes the School of Engineering and Applied Sciences—voted against unionization. Of respondents falling into these categories, less than half—or 48.5 percent—voted to unionize.
The number of survey respondents affiliated with the Dental School, the Business School, the Divinity School, and the Extension School was too small to generate reliable data on the students’ voting patterns.
Women were slightly more likely to favor unionization than men, according to the exit polling data. Of survey respondents who identified as female, 73.9 percent voted in favor of unionization, compared to 64.8 percent of respondents who identified as men.
Exit polling data also suggests new voters were more likely to support unionization than those who voted in Harvard’s first unionization election, held Nov. 2016. Of respondents who reported voting in the last election, 64.9 percent said they voted pro-union this time, compared to 76.2 percent of those who reported they had not voted in the last election.
The final results of the 2016 election saw more votes against unionization than in favor. But lawyers for HGSU-UAW later challenged that outcome, leading to more than a year of legal battles between Harvard and pro-union advocates. The National Labor Relations Board ultimately sided with the union, mandating in Jan. 2018 that the University must hold a second election.
Of respondents who reported voting in the last election, 11.7 percent reported changing their vote between the 2018 election and the 2016 iteration. Of those who reported changing their mind, 78.8 percent indicated they voted in favor of unionization this time.
These shifts in support for unionization likely influenced the final tally—at least in part.
After the last ballot was tallied Friday, the 1,523 ballots recorded as cast against unionization remained roughly constant compared to the 1,526 “No” votes students cast in the Nov. 2016 election.
By contrast, the number of ballots cast in favor of unionization increased by roughly 500: rising from 1,396 votes in the Nov. 2016 election to 1,931 votes in April 2018.
The data presented in this story is entirely sourced from the exit poll of eligible voters The Crimson conducted during the two days of the April 2018 unionization election.
Crimson analysis of the raw exit poll data indicated voters in favor of unionization were more likely to fill out the survey than were voters who voted against unionization. Specifically, The Crimson calculated an oversampling factor of 1.32 for respondents who reported voting “yes.” The Crimson’s survey specifically asked students whether they voted in the Nov. 2016 election and, if so, whether they had changed their mind on unionization since the first vote.
Using the responses to these questions, The Crimson calculated the 32 percent oversampling by taking the percentage of respondents who reported they voted yes in both the April 2018 and Nov. 2016 elections (out of all respondents who voted in the Nov. 2016 election and indicated they did not change their minds in the interim), then dividing it by the actual percentage of voters who voted in favor of unionization in Nov. 2016. The Crimson utilized this response bias factor, in addition to a geographic correction that reweighted the data by polling site, to infer that approximately 50.6 percent of eligible voters who cast ballots April 18 and 19 voted in favor of unionization.
The official vote count Friday revealed that roughly 56 percent of those who cast ballots voted in favor in unionization, leading to an overall oversampling factor of 1.23. Unless otherwise indicated, the data reported in this story is the raw data from the survey and has not been corrected for this apparent response bias.
Crimson editors Brian P. Yu and Phelan Yu conducted data analysis for this story. Questions regarding the survey or methodology can be directed to firstname.lastname@example.org and email@example.com.