Tuesday, October 28, 2008

What was the question?

According to this morning's Tennessean article (front page, above the fold) "English-only measure splits voters in Davidson" they commissioned a survey of Nashvillians and got a result they were looking for. As far as I can tell from their website and the dead tree version, the Tennessean fails to provide the actual question and fails to provide any more information about the breakdown of the measly 200 participants. I can only assume this important clarifying information was purposefully omitted.

From their article they asked likely voters about:

"a measure that would bar Metro government agencies from translating written materials into other languages or offering interpreters to the public."
But the actual wording of the amendment from the last petition sent out by the English First folks doesn't say the city would be barred from translating or providing interpreters. It says:
“English is the official language of the Metropolitan Government of Nashville and Davidson County, Tennessee. Official actions which bind or commit the government shall be taken only in the English language, and all official government communications and publications shall be in English. No person shall have a right to government services in any other language. All meetings of the Metro Council, Boards, and Commissions of the Metropolitan Government shall be conducted in English. The Metro Council may make specific exceptions to protect public health and safety. Nothing in this measure shall be interpreted to conflict with federal or state law."
It doesn't say 'bar' it says you don't have a right to a translation. It also says exceptions may be made for health and safety. That's a looooong way from 'bar'. If they used 'bar' in their questions it was misleading and nullifies the survey completely.

1 comment:

N.S. Allen said...

First of all, it's not uncommon for polls to not report the exact question asked or to not offer as much additional information as one might like. If you poke around just about any national polling topic, you'll find a lot of numbers and few questions. That's not a good thing - obviously, it helps us to understand the poll if we know what was asked and what demographics were hit - but it's not exactly some devious trick that is out of mainstream, polling practices.

Secondly, while a poll with two hundred participants is on the small side of things, I don't see why that size should be immediately suspect. State polls conducted about the national election by major pollsters often only involve a few hundred more, and, presumably, those pollsters have some statistical basis for thinking their numbers are sound.

Thirdly, the proposal says the official, government publications would all be printed in English. That pretty much bans translation of written materials into other languages. Presumably, "official actions" which "bind or commit" the government refers to something more higher up than basic, social services...but it's hardly clear, and it could conceivably be stretched to ban interpreters.

Which is all to say that 1) while polling methodology may be strange, immediately conflating it with bias is, at the least, presuming something without evidence, and 2) the proposal is hardly written in the straight-forward language that supporters suggest. There's a reasonable argument to be made that it extends beyond what they say it would cover.

The fact that the question (whatever it was) might not have used the rosy interpretation you'd have liked it to doesn't mean the question was biased. It just means it wasn't biased in your views' favor.