Charlie Cook is ignorant about the power of the internet

[donsu]

Charlie Cook of the Cook Political Report was on c-span's Wash. Journal this morning talking about the various people running for prez. from both sides.

he was giving his opinions of each one's chance of winning.

the host asked him about a poll from one of the news shows and Cook said opinion polls on cable and the internet don't mean anything. he was very negative about internet polls.

I take individual polls with a grain of salt, depending on who is running the poll; but, polls taken together do tell us something. and internet polls sure had a lot of power in re: the last election.

wonder why he was dissing and disregarding internet polls.

he is supposed to be a neutral reporter but he seemed to lean right to me.

anyone else see him this morning?

[The Magistrate]

1. Any Poll Where Respondents Select Themselves, Ma'am

Will be dismissed by competent analysts. There is never any reason to believe such polls reflect the views of the public at large. If they measure anything, they measure the intensity of claques, and the efficiency of their internal communications. This could have some small bearing on events, but is not the same thing as the feelings of the whole voting public on a matter.

[donsu]

4. what do you mean 'polls where respondents select themselves' sir

what does that mean?

guess you could call backers of the various people running for prez 'claques'. what the 'efficiency of their internal communications' has to do with the subject at hand eludes me?

aren't the 'feelings of the whole voting public' shown in various and numerous polls and announcements from 'claques'?

[boobooday]

8. It simply means that the scientific method is not used for sampling respondents

In an internet poll -- that is, all the respondents have chosen to participate rather than being selected at random.

This is true -- internet polls are not scientific, and whatever the results are, they can only be assumed to represent the people who elected to respond, and not the public at large. But they are interesting sometimes.

[badgerpup]

9. Think of it this way...

There's- I don't know how many polls on the Internet daily...let's call it 100, just to have a number.
Do you participate in ALL those 100 polls, or do you participate in only those in which you have an interest and some information?

Bring it nearer...there's polls daily here on DU and 100,980 users...
but on the ones I've seen there's NEVER been anything even close to 1% participation, let alone full. Some I participate in, some I don't.

A poll where you are randomly selected by an outside source would have greater statistical validity, since this particular poll might not be something you would have chosen on your own.
Self-selected polls would seem to be more like taking a vote among a closed group, whereas a randomly selected population for a poll has a greater chance of a more varied population.

Did that make sense?

[donsu]

10. yes

nt

[boobooday]

11. think you meant to reply to Donsu

But that is a very clear explanation.

I'm studying for my PhD comprehensive exams, so I'm wading through sampling techniques and validity tests until I'm purple in the face!

:hi:

[The Magistrate]

12. Polls Where The Respondent Contacts The Poll, Ma'am

Such as the site polls of news shows on various questions, are polls where the respondents choose themselves. Polls put up here on Democratic Underground are the same sort. People decide themselves to respond, and so by definition, the respondents are not a random sampling. They reflect only the views of people who are aware of the poll, and who are moved to make some response to it. There is no reason at all to suspect that these people are representative of the population as a whole. Terms like "freeped" enter the argot to refdlect the fact that interest groups will sometimes use their grapevines to rally people of one view to respond to such a poll, which can further distort the pattern of responses. Again, that a few thousand activists may be rallied to answer in one way or another, and so tilt a result, cannot be taken to indicate anything about the views of the uninvolved public. Such polls may serve here and there as a crude indicator of the fervor of one group of activists, or of their relative strength versus other groups of activists: this may have some influence on the popular mood, but does not reflect its actual state.

[VA_08er]

2. Cook is fine...

His thoughts on Internet polls are true. They are bogus and should not be relied on. He doesn't lean right; he's fairly objective. Just because he disses Internet polls (and I guess by extension, the Netroots) does not make him a rightie.

[donsu]

3. his opinion on internet polls is not why I thought he leaned right

it's the things he said and the words he chose to say it about the various people running for prez.

[salin]

5. Nah - it is more a recognition of polling/selection bias per the internet

The only polling outfit that does net polling that seems to try to build a legitimate/reliable (reflective of the public) polling method is Zogby - and Zogby doesn't do his polling by putting a poll on a site and depending on the representativeness of those visiting the site for a valid sample. They do their polling via email (eg let their sample folks know when a new poll is being conducted with an invite to participate) and it is somewhat more reflective of traditional phone polls. However, of course there is an inherent bias towards folks who are savvy with the internet and email (and have access) even on Zogby.

I think Cook is a purist on polling - and with decent arguments for being so. I don't think that his criticism is an indictment on the 'internet' or its impact on campaigns. Just on polling data.

[Captain Hilts]

7. Zogby is doing the Internet polls to test demographics of who would respond...

and how to work/compare them to the traditional methods of polling.

[Captain Hilts]

6. The Internet still skews young and male. 17% of eligible voters under 30 voted in 2004. nt