Language matters

Well, at least I got the result of the poll right at #16. Thanks, nutrax, for setting the facts out.

Shilgia - you may have missed the part where I said that I wasn't saying that all polls are inaccurate. Obviously, there have been some that were accurate. We are a stubborn race but if something continually doesn't work, we do tend to drop it eventually.

I'm just saying that, no matter how well a poll is executed there is a potential for the results to be skewed. It may be that the amount that the poll is off is negligible, i.e. 4 people out of 100 change their minds after answering the question, wouldn't make much of a difference. I was also saying that not all polls are well executed.

you may have missed the part where I said that I wasn't saying that all polls are inaccurate.

I did. Where did you say it?

I was responding to

There's alway a large margin of error on surveys.

and

No matter how many people you survey, you'll never get enough to make for true statistics.

I said it in number 9. I guess I was getting a bit carried away before that. I get occasional bouts of over-skepticism. Sorry.

Ah, and I guess I skipped #9. Sorry.

No problem. I don't blame you. Like I said, I did go rather overboard before that point. I probably would have also missed number 9 if I were you.

My statistic professors in grad school taught us that we should never look at a survey or poll as a perfect result, since it's (almost) impossible to ask everyone in your target group your series of questions. But statistics are how you measure how far off your results may be.

So instead of thinking "the margin of error is high, therefore the results are wrong" we should be thinking "the margin of error is high, which lets us know the degree to which we can accept these numbers as-is." Acknowledging a high margin of error is honest and professional. It's like a doctor telling you about possible side effects of medicine you're taking. But the existence of side effects doesn't mean that there's something wrong with the drug. Being told about it up front is a good thing. And anyone with ethics who is interpreting data will take the margin into account.

Those who are against same sex relationships in Britain tend to use the word 'homosexual'; those who are relaxed about them or don't care tend to say 'gay' or 'lesbian'.

I suspect that the use of the terms will influence the response from some. I am suggesting that asking 'neutrals' about 'homosexuals' will more likely result in a negative response whereas asking them about 'gays' or 'lesbians' will more likely result in a positive response.

It wouldn't surprise me if similar results were obtained by, say, using the terms 'niger' or 'black person' in another survey. Think about it: 'Do you think nigers should have...?' or 'Do you think black people should have...?' It's obvious what the questioner 'wants' and many will respond accordingly.

Many simply respond to (what they perceive to be) the expectation of the questioner and give the 'right' answer.

One of the problems with a referendum is the wording of the question.

Exactly Diana and tony.

I suspect that the use of the terms will influence the response from some.

You don't have to suspect it, the evidence is in the OP. That was the point of the OP, in fact. (Although both phrasings elicited a positive response.)

And I don't know about the UK, but I think if you tried to take a survey in the US that used the word n!gger in the question, you'd get very few people staying on the line to answer.