3 ms·

Whether or not it's a survey that's extrapolated is not the point. But it's an extrapolation of people's self-observation and, even more unreliable, their predi

by morsch 13y ago

Whether or not it's a survey that's extrapolated is not the point. But it's an extrapolation of people's self-observation and, even more unreliable, their predictions regarding their own behaviour.

belorn 13y ago
11% said that they had already abstained from using phone, cell phone or e-mail in certain occasions That's not extrapolated about predictions of the future. That is about the past situation at the time of the poll. 52% said they probably would not use telecommunication for contacts like drug counselors, psychotherapists or marriage counselors because of data retention That's is how they would react at the present time of the poll, not predictions about the future. 6% believe to receive less communication since the beginning of the data retention That is a prediction of the future. So we got data from past, present and future. One is prediction, one is what they would do today, and one is about the past. Its a common standard for survey. If it has been about voting, the questions would be: What did you vote last time? if it was election today, which party would you vote on? In the next election, who do you think will win? Past, present and future. If someone want to ignore it, one is always free to wait until election day and tally the votes. We do not actually use the survey data to decide who wins elections. The survey will always be more unreliable than the actually tallied votes. The question one might then ask is why government pay for poll survey, and why they tend to be correct even if they are made from unreliable self-observation of a "few" questioned people. The answer would probably be somewhere in the profession of statisticians, and as Im not one, I don't have it :).
- morsch 13y ago
  No, one is a prediction, one is what they think they would do today and one is about how they think they behaved in the past. I was simply making the point that people aren't necessarily reliable observers of their own behaviour. It's a common mistake in evaluating these kinds of surveys. (Also, even when they observe correctly, they might lie, but lets not get into that.) An interesting statistic lacking this particular source of error would be a survey of e.g. suicide hotlines about the number of calls they get (I'm sure they keep such statistics).

by morsch 13y ago

Whether or not it's a survey that's extrapolated is not the point. But it's an extrapolation of people's self-observation and, even more unreliable, their predictions regarding their own behaviour.

belorn 13y ago

11% said that they had already abstained from using phone, cell phone or e-mail in certain occasions That's not extrapolated about predictions of the future. That is about the past situation at the time of the poll. 52% said they probably would not use telecommunication for contacts like drug counselors, psychotherapists or marriage counselors because of data retention That's is how they would react at the present time of the poll, not predictions about the future. 6% believe to receive less communication since the beginning of the data retention That is a prediction of the future. So we got data from past, present and future. One is prediction, one is what they would do today, and one is about the past. Its a common standard for survey. If it has been about voting, the questions would be: What did you vote last time? if it was election today, which party would you vote on? In the next election, who do you think will win? Past, present and future. If someone want to ignore it, one is always free to wait until election day and tally the votes. We do not actually use the survey data to decide who wins elections. The survey will always be more unreliable than the actually tallied votes. The question one might then ask is why government pay for poll survey, and why they tend to be correct even if they are made from unreliable self-observation of a "few" questioned people. The answer would probably be somewhere in the profession of statisticians, and as Im not one, I don't have it :).

morsch 13y ago
No, one is a prediction, one is what they think they would do today and one is about how they think they behaved in the past. I was simply making the point that people aren't necessarily reliable observers of their own behaviour. It's a common mistake in evaluating these kinds of surveys. (Also, even when they observe correctly, they might lie, but lets not get into that.) An interesting statistic lacking this particular source of error would be a survey of e.g. suicide hotlines about the number of calls they get (I'm sure they keep such statistics).