Politics: WSJ’s “The Numbers Guy” on statistics & Thai polling”

By Pooky, Thai Intel’s economics journalist

Thailand is a totally corrupt country-from top to bottom-and everything in between-with only pockets of transparency and good governance.

As a business and economics journalist, I have hoped that at least, the polling and surveys done in Thailand-are professional.

But get this, the common and accepted belief in Thailand-is that the elite ruler of Thailand has its poll and also that the opposition and the Red Shirt has its poll. This poll is on this side and that poll is on that side-is what the Thais say.

Here in Thailand, when a poll comes out-it is not the result of the poll that the Thais ask first-but it is “Who did the poll.”

There is a serious problem here in Thailand concerning the “Chinese Wall.” For example, with brokers, the research department and the IPO related services are supposed to have a “Chinese Wall” seperating the two-to ensure objectivity. The same can be said about Thai media group-where there is suppose to be a “Chinese Wall” between the editorial department and the owners and advertisers.

But in fact, in Thailand, the “Net-Work System” breaks all the “Chinese Wall” down to bits.

And thus in the polling and surveying business-it is the same thing. Pollsters here in Thailand, can structure a poll and survey-for practically any result they want. And that result they want-is just simple-it is what their “Net-Work” wants.

Takes ABAC University poll, for example. The pollster-is actually an adviser who meets regularly with the Thai prime minister-and gives advice to the Thai prime minister. And ABAC poll on something, could go this way or that like anything.

Thus it is so often in Thailand-that 2 or 3 pollsters-could be polling the Thai people on very much the same thing-but the 2-3 polls-could come out with a 180 degree differences in result.

The following is from the Wall Street Journal’s “Numbers Guy” and it can give Thai Intel readers-a glimpse into the complexities of polling.

From the WSJ Number’s Guy:

My print column this week examines the fascinating numbers and tricky statistical issues that accompany sex surveys, such as a high-profile one published by Indiana University researchers this week.

This study in many ways is a descendant of one conducted by the University of Chicago’s National Opinion Research Center in 1992. Such studies are so infrequent because sexual research that looks beyond fertility and sexual health issues can be hard to fund, researchers say.

The latest study emerges in a different research climate than the one that produced the 1992 study. The Indiana study makes use of online polling, which wasn’t an option last time around. Knowledge Networks, which conducted the survey, uses mailing addresses to recruit its panel and provides Internet access to households that need it. This helps overcome some potential methodological drawbacks to online research, which also has an important potential advantage for research asking about sensitive topics: It allows researchers to ask about sex without ever personally asking anyone about sex. On the phone or in person, there is “more risk that respondents and research subjects are going to be less honest in their responses,” because of their desire to please interviewers, or not be embarrassed with them, said J. Michael Dennis, executive vice president at Knowledge Networks.

Another important advantage for online polling is cost. “For face-to-face interviews, to get really large numbers, it is considerably expensive,” said Debby Herbenick, research scientist at Indiana University and co-author of the recent study.

Then there are all the inherent challenges with getting people to talk about sex, for which pollsters have devised a wide range of strategies. For the 1992 survey, interviewers avoided asking directly about sensitive topics, either by giving respondents some questions in writing or by using a clever way to cloak their responses. When asking about which kinds of sexual behavior was involved in respondents’ most recent sexual encounter, interviewers would hand over a card where letters corresponded to various options, such as oral sex, and respondents would tick off letters. Interviewers didn’t have the key to the card.

In addition to the interview mode, other characteristics differ between the two surveys — notably questions asked. That lessens the ability to track trends between the surveys.

Funding also differed. The older study was planned with the promise of Congressional money, but when North Carolina Republican Jesse Helms led an effort to block in the Senate, researchers had to delay and turn to other sources, including foundations. The newer study bypassed governmental sources, receiving funding for data collection from Church & Dwight Co., the makers of Trojan condoms. Condoms are a major topic in the research. Herbenick said the company didn’t dictate study design, though it did offer advice.

These surveys nonetheless are broader and in some ways more reliable than the federal government’s efforts to track sexual behavior, which generally are part of broader studies. The Centers for Disease Control and Prevention’s Youth Risk Behavior Surveillance System, administered in schools, asks just seven questions about sex. It also doesn’t define sex, out of fear that more schools or parents would opt their students out of responding. “We want to ask appropriate questions, but at the same time, we don’t want to create any unnecessary red flags,” said Laura Kann, who helps direct the survey.

Meanwhile, the National Survey of Family Growth, run by the CDC’s National Center for Health Statistics, focuses on reproduction, so it covers only ages 18 to 44, and only in 2002 did it begin to survey men in addition to women.

One finding many of these surveys share is that the stories men and women tell about their sex lives don’t always match up as we’d expect. Typically, men report more sex partners than women do. Also, in the latest survey and in recent studies of college students, more men report their partners achieve orgasm in their latest sexual encounters than the proportion of women who report achieving orgasm, a finding which has received extensive press coverage.

Researchers who have found these results attribute them generally to poor communication between men and women in bed. “We all need to do a better job of communicating,” Herbenick said.

But such findings may be problematic when certain groups are excluded. Robert T. Michael, an economist at the University of Chicago and a principal investigator on the study, pointed out that prostitutes are unlikely to tell strangers about their sex habits, and those with many clients may have a major effect on estimates of average number of sex partners.

His collaborator on the 1992 study, Edward Laumann, said that because of such discrepancies, “people could be describing the same thing,” yet their answers still could add up to different numbers for each gender.

William Mosher, a statistician at the National Center for Health Statistics, added that these surveys generally are fixed on a certain age range, and don’t ask people outside the country about their sex habits — both factors that could lead to mismatches between male and female responses.

This doesn’t necessarily mean that the orgasm reporting gap isn’t real, or even bigger than found — just that there are potential confounding factors. Stanford University sociologist Paula England, who has collaborated on the college studies, said these findings hold up among various subcategories of respondents — which suggests that at least in this population, if not necessarily in the overall population, sex isn’t always perceived the same way by both partners. And Pepper Schwartz, a sociologist at the University of Washington, said a study she worked on in the 1980s found a similar effect, though not as large, among couples — who presumably were reporting on the same sexual encounter. “The direction or percentage might have some wiggle room, but the message is pretty clear,” Schwartz said.