By Lisa Keen
A survey released this month suggests that most gay, lesbian, and bisexual voters are supporting Hillary Clinton's bid for the Democratic nomination. But the web-based survey has its limitations and its critics.
The survey was conducted by a professional pollster and a long-time gay history professor at Hunter College using a pre-selected panel of 768 people who self-identified as lesbian, gay, or bisexual to a web-based consumer research firm.
Of that panel, 579 (or 75 percent) said they would "definitely" or "probably" vote in their state's presidential primary or caucus. As with previous polls concerning party affiliation, this one found a significantly majority -87 percent– expected to participate in a Democratic primary or caucus, versus 13 percent in a Republican one.
When the smaller subset of 501 "Democrats" were asked for whom they would vote "If the 2008 presidential primary or caucus in your state were being held today," 63 percent said Hillary Clinton, 22 percent said Barack Obama, with the remaining 15 percent being spread out over five other Democratic candidates. Less than one percent of the Democratic subset declined to answer the question. (The margin of error is said to be plus or minus four points.)
The LGB support for Clinton in this survey, taken November 15-26, is much stronger than in polls of Democratic voters generally nationwide taken during the same time period. For instance, while 63 percent of the LGB Democratic voters said they would support Clinton, only between 38 and 44 percent of Democratic voters in polls of all voters said they would.
Two-horse race
Support among LGB Democratic voters for Obama is roughly similar to that of Democratic voters generally. While recent polls in the general population indicate that between 21 and 27 percent of Democratic voters support Obama, 23 percent among LGB voters do. That suggests, then, that at least one reason Clinton's numbers are higher among LGB Democratic voters is because LGB support for candidates other than Clinton and Obama is relatively weak. For instance, while between 12 and 13 percent of Democratic voters generally support John Edwards, only 6.5 of the 501 Democratic LGB voters planned to support him.
With only 78 people indicating they expected to support a Republican candidate, the reliability is much reduced, however, 50 percent indicated they would support Rudy Giuliani, 23 percent John McCain, 11 percent Mitt Romney, 10 percent Fred Thompson, and the remaining six percent was spread among the remaining candidates. Again, less than one percent declined to answer the question.
The survey also suggests the presidential race in the LGB community is over once the Democratic nominee is selected. Asked generically, who the LGB Democratic voters would "probably vote for" if the general election were held today, 91 percent said the Democratic nominee, compared to 8 percent for the Republican nominee. Polls of voters nationwide generally indicate it will be a much more contested race. A Fox poll during the same time period as the "Hunter College Poll" found 43 percent of voters would choose the Democratic nominee and 39 percent the Republican.
Hunter College also asked LGB voters how they would vote if the general election were a match-up between Democrat Clinton and Republican Giuliani. An overwhelming majority (88 percent) chose Clinton over Giuliani (11 percent). Several national polls indicate a Clinton-Giuliani match-up in the general election show Clinton with only a five-point lead over Giuliani, at best.
The survey also showed that LGB voters identified "LGB rights" as the "issue mattering most" to them in choosing a 2008 presidential candidate (21 percent), followed by the "economy and jobs" (19 percent) and health care (16 percent). Iraq came in third with 14 percent.
The nitty and the gritty
The survey was created by professional pollster Murray Edelman of Rutgers University, political science professor Ken Sherrill of Hunter College, and assistant professor of politics Patrick Egan at New York University. The researchers conducted the study with a $50,000 grant from the Human Rights Campaign Foundation, said Edelman, but without any input from the national gay political organization.
As soon as the results were released, November 29, criticism quickly emerged that HRC's involvement with the survey may have skewed the results in Clinton's favor, even though HRC has never endorsed any candidate in the presidential race. Adding fuel to that fire was the fact that the survey researchers used the same polling firm -Knowledge Network– that HRC had used for a survey just prior to the U.S. House vote on the Employment Non-Discrimination Act (ENDA). From that survey, HRC touted that "70 percent of the GLBT community supports passage of "ENDA without including protections based on gender identity. And fanning the flames was the fact that, shortly after the poll was released, the Clinton campaign had a press release out, drawing attention to the results favorable to the senator from New York.
Edelman dismissed the criticism, saying that the investigators on the Hunter College Poll worked independently from HRC and that the organization was not involved at all in the actual survey, only in funding it.
Edelman, who is also Director of Statistics in the CBS News Election and Survey Unit, said being able to work independently was important to him and the other investigators professionally and that they made that clear to HRC from the start.
"From the beginning, we made it very clear that this had to be an independent study," said Edelman. "I needed it myself because I work with the media and can't be involved with an advocacy group. …This was not an HRC poll."
Edelman pointed out that the press release from Hunter College about the survey results even contradicts the HRC-ENDA poll -saying that 60 percent of LGBs said it was wrong to "remove protections for transgendered people" to pass ENDA.
"As anyone knows, HRC could not have been very happy with that," said Edelman.
But there were other criticisms, too –that the polling sample was unusual -49 percent identified as "bisexual," compared to 51 percent as "gay" or "lesbian" or "homosexual. Fifty-one percent were women, versus 49 percent men.
Gary Gates, a respected UCLA researcher in LGBT demographics, said Knowledge Networks is "a completely reputable survey firm."
"While Internet-based surveys can be biased because Internet users are different (e.g., more educated, higher incomes) from the general population," explained Gates, "Knowledge Networks is one of the few companies that tries to correct this problem by providing computers and Internet access to respondents who are not online. Overall, I think this is among the better Internet-based surveys out there."
Gates also noted that the most recent government survey to ask about sexual orientation to a random sample -the National Survey of Family Growth– "found that about half of GLB people identified as bisexual" and there was "about a 50/50 split between men and women."
A spokesman for Networks, David Stanton, said the company has about 50,000 people who are part of the company's "Knowledge Panel," which, according to the company web site, is "the only online panel that is representative of the entire U.S. population." He emphasized that the overall national panel is not reliant on people volunteering or opting-in, but is created through random phone calls from a national database of phone numbers. A representative sample of people who answer the phone are invited to be part of a regularly surveyed "Knowledge Panel."
Stanton would not say when or how the sexual orientation of Knowledge Panel members is determined and referred all other questions to the Hunter College researchers. But Edelman said the sexual orientation question is asked during Knowledge Networks' initial screening of potential panelists.
Asked why the Hunter College Poll did not include transgender people in the survey, Edelman said it was because the survey sample was seeking panelists based on their sexual orientation, not gender identity, which would have included heterosexuals.
