Fighting the Agenda in Enemy Territory

(Let me preface this post by admitting that I realize that it isn't very polished, does little more than gripe about statistics, and is so narrow of focus that it won't be of much interest to very many readers. However, I suspect that the group in question intends to take a central role as the gay marriage battle moves into Rhode Island, so it may very well prove useful to my side to have some preliminary analysis of the group's data and statistical techniques.)

I had only intended to mention a Providence Journal article about a new advocacy group for homosexuals called Equity Action to make a few minor points. I stress, at the outset, that my intention, here, is to address organizations, not people.

The first minor point is to note that, in contrast to this free advertisement for a fund-seeking organization, I still have yet to see within the paper a single mention of the rallies Massachusetts against gay marriage. The second is to suggest that the data being presented is inevitably skewed, not only by the standard survey disclaimers, such as that people will tend to give the "right" answers for personal questions, but also by the fact that this survey is obviously intended for advocacy purposes, something stated in the instructions of the survey. And the third is to challenge, out of hand, the contention that homosexuals in Rhode Island have any legitimate major complaints, as a group.

But when I began to look at the specifics, mostly out of curiosity, the distortion that I discovered in the study's summary and the Projo article proved so egregious that anything either organization says on this matter ever must be considered suspect in the extreme. The degree to which the authors sought to exaggerate the supposed difficulty of being gay in Rhode Island and to minimize any factors that might suggest that gays are doing just fine is ridiculous.

The summary document (PDF) lumps together all household income between $25,000 and $100,000 in order compare the 70% total for Rhode Islanders in the U.S. Census with 65.8% of the homosexual respondents. Of course, the writers can't do otherwise than admit that homosexuals have higher incomes, but the numbers as presented aren't dramatic given that more than three-quarters of the survey respondents were over 30.

However, whoever put together this particular table, although he or she managed to calculate the raw numbers of people correctly, included the $15–25K data from the Census chart in the $25–100K percentage. Therefore, the <$25K comparison should be 30.1% (Rhode Islanders) versus 12.9% (RI homosexuals); the $25–100K should be 58.5% versus 65.8%; and the >$100K is correct at 11.4% versus 21.4%. Looking at the raw data (PDF) for the homosexual survey, it isn't surprising that the middle income group has significantly more representation toward the higher end than does the public at large.

To be fair, the lumping of that huge income range was likely done, in part, because the categories differ between the two sources by increments of $5,000. That's a fortuitous oversight on the surveyors' part. Similarly, the calculations for the report's table were likely a mistake (although I wonder why the person with the calculator wouldn't notice that he'd added up three rows to get the raw number, but only two to get the percentage), but it's a fortuitous one.

Another area of fortuitous skew is the monogamy number. According to the survey, 53% of respondents say they are currently in monogamous relationships. Of course, once again, the age ranges contribute to this result. It's also worth noting that 3.5% are in straight marriages. And it should also be remembered that the respondents knew the advocacy purpose of the survey. On top of these points, the survey's methodology makes it likely that a significant number of the people in committed relationships who answer this question are, in fact, committed to each other.

But that's all "fun with numbers" analysis. The truly egregious distortion comes into play with the dark side of being gay in Rhode Island. Projo writer Karen Lee Ziner declares:

And yet, the report states, they endure frequent harassment and discrimination -- at work, at school, on the street, in doctors' offices, in public places and in their places of worship. Some have experienced physical violence, or fear it, because of their sexual orientation.

Referring to the press release (PDF) from which she likely got this impression, one notices that the numbers are extremely small for all but the monthly and yearly instances categories, and even here the numbers are all under 23% for monthly and 33% for yearly. One who has already looked at the raw data might also notice that the never column has been dropped from the official table. Here's the never data, with "not applicable" in parentheses:

At work: 47.2% (18.9%)
At school: 16.8% (65.6%)
At home: 64.3% (19.7%)
On the street: 27.9% (6.3%)
At a public establishment: 31.6% (6.2%)

Moreover, the "at school" row of the table is footnoted as "only those who indicated being a student," even though those respondents who answered the question all had some reason for doing so. (Otherwise, they would have selected "not applicable.") The effect? The first number is the press release/summary table, and the second number is the raw data:

More than 1x/day: 6.5% versus 2.0%
Daily: 9.1% versus 2.3%
Weekly: 9.1% versus 2.3%
Monthly: 19.5% versus 5.8%
Yearly: 18.2% versus 5.2%

Not only has the data been selectively presented, but one very important fact was left unsaid: 44.6% of the respondents said that none of the homophobia/harassment was directed at them. 34.7% said only a little was; some, 15.2%; a lot, 3.5%; and all, 2.0%. So a great many of these daily, weekly, monthly, and yearly instances of homophobia and harassment could be off-color jokes, misunderstood comments, or imagined bias.

The press release goes on to claim that "LGBTQ individuals are afraid to reveal their sexual orientations or gender identities even to family members," but one need only look at the accompanying table (noting that the question is being "out," not the reason) to see how ridiculous this claim is. The raw data for "support and acceptance" reveals this even more.

More interesting, however, is the summary's handling of healthcare. It includes a paragraph about the dangers of providers' discrimination, which "results in delayed, infrequently, or avoided use of these services" and leads to "late diagnoses, misdiagnoses and missed opportunities for healing and prevention." Looking at this question in the raw data, however, it isn't difficult to understand why the summary doesn't offer any of the data relevant to this question. Of the 80.9% of respondents who have told some or all healthcare providers about their being homosexual, 12.7% say that their care improved, with only 2.0% saying that it worsened. In contrast, of the 19.1% who haven't mentioned it to a healthcare provider, only 13.5% (2.6% of the total) cited reasons having to do with discrimination or other people finding out.

Now, if some group wants to get together and study Rhode Island's gay community for the purpose of finding out what they actually require, then I wouldn't be inclined to object. This (non-profit) group, however, has the stated purpose of lobbying for civil rights legislation. For a wealthy, largely comfortable segment of the population that has a powerful advocate already in the state's only major newspaper. Using statistics that are either next to useless or incompatible with the "interpretation."

If Rhode Island's busybodies wish to investigate an oppressed minority that feels intimidated, undersupported, and even uncomfortable telling people about the thing that makes them different, they should perform a similar survey among conservatives. We certainly could use a little more representation in this state.