A few days back, Mickey Kaus asked a good question about two conflicting results relevant to the debate about the Employee Free Choice Act (EFCA):
Rasmussen reports that only 9% of non-union workers would like to belong to a union. Labor backers often cite a 2005 Peter Hart survey showing that 53% of "non-managerial workers ... definitely or probably would vote in favor of union representation in their workplace." Hard to reconcile those two numbers, no? The Hart poll was paid for by the AFL-CIO, but that doesn't make it wrong. It would be useful to see the exact Hart questions (and the questions on the same poll that preceded them)
I contacted both Scott Rasmussen and Guy Molyneux of Peter Hart Associates for more details. According to Molyneaux, their most recent results from December 2006 were essentially the same: 53% of "non-managerial/non union workers" would vote to form a union. Except as noted, the Hart data that follow are from the 2006 survey.
As you will see, the apparently wide gap between these surveys is less than meets the eye. Once we calculate the key percentages using a comparable base, factor in the offsetting inconsistency in self-identified union members and consider some important differences in the question text and format, most of the discrepancy disappears. Other more debatable issues include timing and the context set by earlier questions.
1) Different denominators. Much of the gap stems from the different bases used to compute the percentages. Rasmussen tells us that 17% of his sample of adults self-identified as union members, so 83% of adults answered the would-you-like-to-join question. According to Molyneaux, 62% of their adult sample self-identified as "employed for pay" and they classified slightly more than half of those -- or 35% of all adults -- as non-managerial workers.
If we compute both of these statistics among adults, the apparent discrepancy between the two surveys narrows considerably. Seven percent (7%) of adults on the Rasmussen survey say they want to belong to a union (9% x 83%), compared to 19% of adults on the Hart survey who say they would vote to form a union (53% x 35%).
How did the Hart survey define "non-managerial workers?" They used three different questions to identify those who said they are (1) employed for pay and are not either (2) self employed or (3) or a supervisor whose "daily job responsibilities are different" rather than "basically the same as the people you supervise" (see Hart's #F5b after the jump). The reason for that last requirement, according to Molyneux, is the presumption that "a 'supervisor' who does the same job as those under him/her -- like a supervising nurse on hospital floor -- is probably eligible for a bargaining unit."
2) Different estimates of union members. My inquires revealed that the two surveys also produced different estimates of the size of the unionized work force, but in this case the Rasmussen percentages are bigger. Rasmussen reports that 17% of his adult sample were union members. On the Hart survey, 10% answer "yes" when asked, "are you a union member?"
So when we add up union members and those ready to unionize, the gap narrows further: Rasmussen finds 24% that are either union members or would like to join a union. Hart finds 29% that are either union members or would vote in favor of union representation.
We should keep in mind, of course, that Rasmussen's survey offered a bigger of its adult sample the opportunity to say it would like to belong to a union. Had the Hart poll asked its question of 83% of adults (as Rasmussen did) rather than 35%, they would have obtained a bigger percentage. How much bigger? Unless Rasmussen can produce cross-tab results by a similarly defined subgroup of managerial workers, we can only guess.
An aside: Hart also asked union members to answer their would-you-vote-to-unionize question. Among the 41% of all non-managerial workers in the 2006 survey (including union members), 58% said they would vote to unionize their workplace (or 24% of adults). Molyneaux reports that roughly 90% of union members typically say they will vote to unionize. This wrinkle sheds no light on the apparent discrepancy between the two surveys, since the percentages noted above were filtered for non-union, non-managerial workers only.
3) Different questions. Rasmussen asked, "would you like to belong to a labor union where you work?" Hart asked, "If an election were held tomorrow to decide whether your workplace would have a union or not, do you think you would definitely vote for forming a union, probably vote for forming a union, probably vote against forming a union, or definitely vote against forming a union?"
Two big differences here: The first and most obvious is that Hart asks about voting to form a union, Rasmussen asks about belonging to a union. Here is Molyneux's take:
Workers don't "join" a union individually, like joining the Sierra Club or LWV. So I don't know how they interpret Rasmussen's question: could be they think it just means paying dues with no obvious benefit (since you can't get representation and a contract as an individual). Who knows?
Second, and probably more important, the Hart question offers four categories with the two middle choices softened by the the adverb "probably." The full results in December 2006 were 25% of non-union/non-managerial workers definitely for, 28% probably for, 18% probably against, 24% definitely against and 5% not sure. The gentle push into "probably" most likely made their unsure smaller and their total "yes" and "no" percentages bigger.
4) Different timing? Again, the most recent Hart results are from December 2006. Rasmussen's survey was conducted earlier this month. He points out that their update this week shows an eight percent drop since August (from 55% to 47%) in the favorable rating for labor unions. I cannot find any more recent favorable rating measures of unions to confirm the Rasmussen Result.
5) Context? Kaus wondered, appropriately, about the questions that preceded the would-you-unionize items, so I asked. The complete list of questions appears after the jump (Hart provided full text, Rasmussen provided descriptions). Neither lead-in looks like anything I would have expected to influence or bias the results, but the context set in the lead in questions is different and possibly an influence.
The Hart vote-for-unionization question appears after probes of the the family's economic situation and four essentially neutral questions about unions (though some might quibble about whether it's neutral to offer "large business corporations" as the foil to "labor unions" in the favorable ratings. The answer categories on their economic questions also suggest potential answers about Americans "falling behind," "working hard to get ahead" or "having just enough to get by" and so forth.
Rasmussen asked the two union questions as part of a nightly economic tracking survey following five standard items about the condition of the economy and the respondents personal finances. The question that comes just before the two union items asks whether the respondent owns "at least $5,000 in stocks bonds or mutual funds."
The Rasmussen question may suggest, subliminally, that stronger unions would negatively affect the value of stocks and mutual funds -- one of the arguments made by anti-EFCA interests. One could also argue that the context of the Hart questions inadvertently framed unions as means to reduce economic insecurity. Keep in mind that the message that Hart and Molyneux tested, the one that moved Americans to support EFCA, kicks off with a reminder that "supporters say that the system is broken and working people are struggling to make ends meet today, and the middle class is being squeezed" and the explicit argument that unions provide "a way to help average people get their fair share."
I posed this possibility to Molyneux, who responded as follows:
Agreed the context is a bit different. But consider also that we've asked the question SEVEN times over a ten year period, and every time the yes vote was over 40%. The prior questions were different every time (but never included any pro-union message questions or anything like that). So that's at most a minor factor.
I do agree that the context or framing of the lead-in questions is likely a minor piece of this puzzle, although without doing a split sample test it is hard to know for sure. The bottom line is that most of the apparently huge discrepancy between these questions disappears once we do an apples-to-apples comparison of the percentages among all adults, add-in the self-identified union members and consider the differences in the question wording.
Full text of the questions preceding the unions questions follows after the jump.
Questions from December 2006 Hart Research survey sponsored by the AFL-CIO:
1. Are you currently employed for pay?
2. (IF NOT EMPLOYD) Are you a student, do you stay at home, are you retired, or are you unemployed and looking for work?
3. Are you a member of a labor union? (IF "NO," ASK:) Is anyone else in this household a member of a labor union?
4a. Which of the following would you say best describes your family's financial situation? (READ LIST AND ROTATE) Getting ahead, Have just enough to get by OR Falling behind
4b. (ASKED OF RANDOM HALF SAMPLE) Do you think the standard of living for the next generation will be better than today, worse than today, or about the same? Better than today, worse than today OR same as today
4c. I am going to read you two statements. For you personally, which one is the more important economic goal for government today? (ROTATE ITEMS) Statement A: Improving economic security, so people won't lose what they have worked for, OR Statement B: Improving economic opportunity, so people who work hard can get ahead.
5. I'm going to read you the names of several institutions and groups. I'd like you to rate your feelings toward each one as either very positive, somewhat positive, neutral, somewhat negative, or very negative. If you don't know the name, please just say so.
Labor unions (ALWAYS ASKED FIRST)
Large business corporations
The Democrats in Congress
The Republicans in Congress
6a. How much would you say you know about labor unions--a great deal, a fair amount, just a little, or not too much?
6b. Where would you say you have gotten most of your information about labor unions--from personal experience, from people you know who are union members, from media such as newspapers and TV, from the Internet, from your employers, or from somewhere else?
7. In general, do you approve or disapprove of labor unions?
8. If an election were held tomorrow to decide whether your workplace would have a union or not, do you think you would definitely vote for forming a union, probably vote for forming a union, probably vote against forming a union, or definitely vote against forming a union?
[Also cited above]
F5b. (ASK ONLY OF RESPONDENTS WHO SAY THEY ARE EMPLOYED IN Q.1 AND DO NOT SAY THEY ARE SELF EMPLOYED/OWN BUSINESS IN Q.F5a.) Do you personally supervise two or more employees at your job? (IF "YES," ASK:) Are your daily job responsibilities basically the same as those of the people you supervise, or is your job different from theirs?
Questions from the Rasmussen Reports survey from March 2009:
How do you rate the U. S. economy these days? Excellent... Good... Fair... Poor
Are economic conditions in the country getting better or worse these days? Better... Worse... Same... Not Sure
How would you rate your personal finances these days? Excellent... Good... Fair... Poor
Are your personal finances getting better or worse these days? Better... Worse... Same... Not Sure
Is the U. S. economy currently in a recession? Yes... No... Not Sure
Do you own at least $5,000 in stocks bonds, mutual funds? Yes... No... Not Sure
Do most working Americans want to belong to a Labor Union? Yes... No... Not Sure
[Answered Only By Those who are Non-Union Members] Would you like to belong to a labor union where you work? Yes... No... Not Sure