(C) Daily Kos
This story was originally published by Daily Kos and is unaltered.
. . . . . . . . . .



Either this CNN poll is pure BS or they are doing a horrible job explaining what they have done. [1]

['This Content Is Not Subject To Review Daily Kos Staff Prior To Publication.']

Date: 2023-09-07

New SRSS/CNN survey appears bad for Biden (CNN Poll: Biden faces negative job ratings and concerns about his age as he gears up for 2024), but what is going on in this thing? I am a professional applied statistician. I am not a survey expert but I have taken courses in survey methodology. So I *should* be able to understand exactly what they have done in this survey and why.

But either I don’t or the survey appears hugely flawed. If we have any survey experts here on DKos I would be happy to be set straight on this if it is legitimate. Otherwise, this is another reason to not believe a damn thing about these polls.

Here’s what it says in the CNN article.

“The CNN Poll was conducted by SSRS from August 25-31 among a random national sample of 1,503 adults drawn from a probability-based panel, including 1,259 registered voters and 391 Democratic and Democratic-leaning independent voters. The survey included an oversample to reach a total of 898 Republicans and Republican-leaning independents; this group has been weighted to its proper size within the population. Surveys were either conducted online or by telephone with a live interviewer. Results among the full sample have a margin of sampling error of plus or minus 3.5 points; among registered voters, the margin of sampling error is 3.6 points, and it is 6.0 for Democratic and Democratic-leaning voters.”

I checked the full SRSS press release and there is even less information there…

So let’s unpack this. first: Random national sample of 1,503 adults drawn from a probability-based panel, including 1,259 registered voters and 391 Democratic and Democratic-leaning independent voters.

So what does “probability-based panel” mean? It means that SRSS has invited a large group of folks to participate regularly in their surveys. These invitations are sent to a group of Americans thought to be representative of the population. But who accepts the invitations? People who are willing to provide a “battery” (SRSS’s term not mine) of information. How large is this group? SRSS doesn’t say. Are they still a representative group? Well, no. But SRSS just says that they are “weighted” and “calibrated” to assure representativeness based on consistency with American Community Survey averages when surveys are taken. I find this mildly suspicious but no huge red flags yet...

Next, they sampled 1,503 adults from this go-to survey group. And they got only “391 Democratic and Democratic-leaning independent voters”??! Here, this thing starts to get really weird. Who are the rest of the people?

Second: The survey included an oversample to reach a total of 898 Republicans and Republican-leaning independents; this group has been weighted to its proper size within the population.

So 898 Republicans make up most of the sampled voters. Why? So this is NOT a random national sample. It is a sample chosen by party affiliation.

But, no worries, supposedly, because the GOP oversampling is “weighted to its proper size within the population.” But how on earth does anyone know what the “proper size” is? National totals of party affiliation registrations? Previously stated opinions from the “battery” of demographic information. Of course, they don’t tell you.

The entire result that Biden and Trump are neck-and-neck hinges on this weight. If you weight R’s more heavily then Trump is the poll winner. If you weight them less heavily then Biden is the winner. Little of this has anything to do with the actual answers by the survey participants.

OK, folks… Am I missing something? Or are polls today worthless?

[END]
---
[1] Url: https://www.dailykos.com/stories/2023/9/7/2191956/-Either-this-CNN-poll-is-pure-BS-or-they-are-doing-a-horrible-job-explaining-what-they-have-done

Published and (C) by Daily Kos
Content appears here under this condition or license: Site content may be used for any purpose without permission unless otherwise specified.

via Magical.Fish Gopher News Feeds:
gopher://magical.fish/1/feeds/news/dailykos/

You are viewing proxied material from magical.fish. The copyright of proxied material belongs to its original authors. Any comments or complaints in relation to proxied material should be directed to the original authors of the content concerned. Please see the disclaimer for more details.