Tuesday, April 05, 2005

A question of 'fairness'

If it wasn't for Michelle Malkin, I would have never known about the new Zogby International poll showing overwhelming support for not removing Terri Schiavo's feeding tube. I couldn't find mention of this poll on any of the Alphabet networks, and when I looked for mentions of it on google news, I could only find the poll in Christian/pro-life publications. Even Zogby himself didn't have the poll on his site; he merely reproduced the article written by LifeNews in its entirety (you can read it here) so it's impossible to know what Zogby's motives and methods for this poll are and whether in truth, the questions are designed to test reaction to the Terri Schiavo case.

Malkin used the LifeNews' reporter's own words in characterizing this as a 'fairer' poll, and this is where I take exception. It's ridiculous to read these questions and not see them as poorly worded, using inflammatory language, and having an unavoidable bias that can only prompt one answer. It reminds me of the time I received a survey from [insert political party of choice here] and one question asked was, "Do you support education initiatives?" Well, yes. Mind you, the question didn't ask specifics, but rather a general overview question; no doubt, [political party of choice] was overwhelmed with the response to that question -- how much do you want to bet nearly everyone said yes?

The problem with questions like that is they lead respondents to a conclusion desired by the researcher herself and are not truly indicative of what the respondent actually feels. Zogby's questions are a perfect example of this. To wit:

If a disabled person is not terminally ill, not in a coma, and not being kept alive on life support, and they have no written directive, should or should they not be denied food and water.

The word 'disabled' is the first red flag in this question, the first answer as to the question of bias. 'Disabled' means many, many things to many different people, covering the range from chronic back pain to persistent vegatative states. It is impossible to answer this question 'no' without considering the wide gauntlet of definition; most sane and caring people are going to disagree with this statement (though, I would argue that the issue in this case was not so much a question of disability but rather who should make decisions for a patients and whether Congress overstepped its bounds; ask those questions and you'll get a very different answer).

Second, while the question isn't necessarily double-barreled, there is no reason to make the reader run through a maze of 'should or should they not'; in my opinion, that's bad question writing, especially when you can start a question with the word "Should..." and continue from there. The respondent is also forced to consider several factors all in the same sentence: not terminally ill, not in a coma, not being kept alive on life support, and have no written directive.

This poses a research problem as we don't know what factor is more important to the respondent than the other, what tipped their viewpoint into the 'no' column, not to mention the confusion the respondent faces in dealing with such a complex question. And indeed, without direct mention of Terri Schiavo within the question, I find it hard-pressed to apply the results of this study to her situation.

Going on to the next question: When there is conflicting evidence on whether or not a patient would want to be on a feeding tube, should elected officials order that a feeding tube be removed or should they order that it remain in place?

I don't know what it is with Zogby and his penchant for asking two questions in one. A good pollster would not do that. In addition, a definition of who 'elected officials' are is missing. Judges, such as the brutally vilified Judge Greer, can be elected. Tom DeLay, God bless his ethically misplaced soul, is an elected officials, as is Bill Frist. Once again, the question relies on the respondent to draw her own conclusions and definitions. 'Conflicting evidence' also assumes parties are conflicted and an independent source has not made an assessment. This question, however, is not quite as preposterous as the previous one, but again, with direct mention of Terri Schiavo, it's not easy to apply the results of this question to her case.

I'm not saying polls are without bias; polls do have a certain amount of bias and in any report, that bias is noted at the end and a margin of error is given to account for that bias. However, without a margin of error figure, what/who the sample population was, or even the original questions or context of the poll, I'm left with only the questions LifeNews has reiterated and from those, the word 'fair' is not the proper adjective to use in describing this particular survey.

Incidentally, Zogby had a new poll out on my favorite congressperson, Tom DeLay, yesterday; you can read it here, but note Zogby gives the sample population figures and demographics. Also note this line: But nearly 69 percent of people in the poll, including substantial majorities of Democrats and Republicans, said they opposed the government's intervention in the long-standing family battle. I'm just waiting for LifeNews' take on that.

No comments: