An editorial in today’s New York Times referenced a Pew poll in which a number of people expressed anti-Muslim sentiments. I’m not going to get into that here, because I think we can agree that, duh, bigotry is bad. Nothing new to say there.
No, the offense I’m about to get on my soapbox about is an offense to mathematics and common sense. You see, one of the numbers quoted was that half the respondents agreed that “at least some” American Muslims are anti-American.
At least some.
At least some.
That may be the most godawful meaningless stupid poll option I have ever seen.
At least some. As in, more than one. Well gosh golly, yes, I’m sure there are at least two Muslim citizens who hate America. Also two or more unpatriotic Christians, Jews, Buddhists, atheists, apatheists, vegans, pole dancers, science fiction fans, redheads, and grammarians.
“At least some” housepets have rabies. “At least some” wives cheat on their husbands. “At least some” Homo sapiens have set other H. sapiens on fire.
Good going, Pew Research Center. I’m sure that at least some of your people know what they’re doing.
It could be a useful gauge though of the opposite. It means that there are people that are willing to say that ‘NO American Muslim is anti-American.’ I think that that’s a position that people might take and I would wager quite a few do take. It gives you an idea of how truthful people are being or whether they are using the poll simply as a virtue signal or way to advance their agenda. It’s not uncommon for polls to have questions like that that gauge whether the poll is being taken seriously or not.
“Some”,“About half”, and “Almost all/most” were aggregated under the label “At least some”, yielding the 49% figure. I get the motivation for this kind of aggregation, but I think they shouldn’t have bothered.
I can’t really see how an individual respondent can push an agenda. Sure, the poll can push an agenda by how it is worded, but, as a respondent, voicing my “agenda” is exactly the same thing as picking my preferred option in the poll.
I mean, if I support Black Lives Matter, and a poll asks a question about whether I think black people are mistreated by police, and I answer yes, I guess you can say I’m “pushing my agenda.” But I’m also doing exactly what I’m supposed to do: answering the poll.
Now, if the pollster actually asked the question in the OP, then I might suspect the pollster had an agenda. It’s so obviously true that it implies that they are pushing for a specific answer by their choice of wording.
I agree it’s confusing, but I’m also struggling to come up with a better wording. With the last two, “half or more” or “at least half” would work, but I can’t come up with anything including all 3 options. “More than a few” still has the same problem as “at least some,” to my ear.
Glad to know it wasn’t Pew that messed up, but very bad wording by the NYT. I’m not sure it would require even two American Muslims to be anti-American. One might be enough. In formal logic, some is generally taken to mean “at least one”. I’m not sure how to interpret at least some. Does it mean “at least at least one”? If so, is that any different from “at least one”?
“Significant number”? Admittedly, you’d have to define “some” (which is marginally better than “at least some”, but still too ambiguous). “Half or more”?
I agree. A few years ago someone linked to a “What are your politics?” webpage that had questions just as ridiculous.
Interpretations of polls can get silly too. Some decades ago I read in a foreign newspaper of a poll showing that a whopping 57% of Indians opposed mass violence against Muslims. The headline was “Indians favor religious tolerance.” :smack:
OK. Maybe Pew’s questions aren’t so bad, but some polls’ are.
I have another Pew peeve. Their raw data is available, once it’s aged a year or two, so I downloaded some. (I was thinking of dusting off my Karhunen-Loeve code and finding the American eigenvoters! ) You have to run the Pew data through R, then run it through Excel just to get a .csv file (which didn’t even have commas :smack: … except within quotes). After that, when I got around to looking at the data, I found it to be in a peculiar anal-retentive format.
For example in a raw survey of 5009 interviewees, 2371 answered “No, does not describe Republican Party” to one or more of the questions. Thanks, Pew! Did it occur to any of you that since the responses are in a fixed format, we already know what the question was and anything beyond “No” was redundant? (Or was this a for-free bonus from the -->R–>Excel–> mumbo-jumbo? BTW, 1856 interviewees gave that response to at least two questions; 394 to at least three; and 74 interviewees answered four questions this way.)
Yes, the data was there, but I found everything about their approach to “raw data” so unappetizing that I lost interest in finding the eigenvoters.
