I’m gonna go out on a limb and call this one an outlier. To do otherwise is to assume no connection between a sharp change in direction on voter confidence in the government and presidential job approval. Something seems very off in CNN’s new polling, especially given the trends in other series:

The tentatively positive public outlook that marked the first months of the Biden administration has faded, according to a new CNN poll conducted by SSRS. The survey, released Friday, finds a rising share of Americans who say things in the US are going badly and that the economy is in poor shape, with increased worries about the coronavirus, the economy and crime.

The new poll finds 69% of Americans say things in the country today are going badly, below the pandemic-era high of 77% reached in January just before President Joe Biden took office but well above the 60% who felt that way in a March CNN poll.

And 62% say that economic conditions in the US are poor, up from 45% in April and nearly as high as the pandemic-era peak of 65% reached in May 2020.

And yet …

Biden’s approval rating stands at 52% approve to 48% disapprove, with disapproval up since April. The survey was conducted throughout August and early September, and Biden’s ratings shifted over that time, with his approval rating in the first half of August (55%) more positive than in later interviews (50%).

This survey was conducted over five weeks, beginning on August 3 and ending on September 7. The debacle in Afghanistan took place squarely in the middle of that period. Oddly, despite the long time period, CNN’s pollster SSRS only got 2,119 adults to complete the survey — certainly enough for statistically significant results, but a languorous pace for a sample that size. The latest Economist/YouGov poll collected over 1200 responses in just four days, in comparison. You can bet that some people will claim they were screening responses to get specific results.

Also, only 3% of respondents showed Afghanistan as the most important issue facing the country today, with national security and foreign policy getting only 5% in the aggregate. That seems on the low side in the middle of the American collapse in that country, with Americans abandoned by Biden. It’s tough to say without robust parallel polling during that period that asked the same question, but it looks outlier-ish to me. (Notably, CNN never even polled about Afghanistan in this survey despite the crisis that erupted there.)

Speaking of which, Joe Biden’s job approval in this poll is definitely an outlier. Take a look at the RealClearPolitics aggregation on Biden’s job approval, which so far does not include the CNN results and shows Biden at 45/49. The last poll showing Biden over 50% came just before the fall of Kabul in a Politico/Morning Consult survey (51/46). At that time, RCP’s aggregate average on right/wrong direction was 37/55, and the MC poll had it at 41/59.  The aggregate is now 29/61, and even though it barely budged in MC’s latest survey (39/61), Biden’s job approval in that poll has shifted seven points in the gap to 47/49.

So how did CNN get 50% job approval for Biden in interviews in the latter half of their survey in the middle of the disgrace? According to the crosstabs, they did’t cook the overall sample, at least not demonstrably. The sample distribution was 35% Democrats, 29% Republicans, and 36% independents for a defensible D+6 sample. The demos on the overall approval question show independents more critical than the overall number, however, at 46/54.

That in itself raises a question. If independents are at 46/54 and they comprise 36% of the sample, how did Biden get to 52/48 overall? Republicans (9/91) and Democrats (94/6) were predictably and entirely partisan on this point, but this looks a little off, statistically. If Biden’s underwater by eight in the sample’s largest demo, it seems weird to have him up four points overall.

This pattern keeps repeating throughout the survey as well:

• Economy: Overall 51/49, indies 45/55
• Pandemic: 56/44, indies 53/47
• Kamala Harris approval: 52/48 overall, indies 45/54
• Direction (net): 31/69, indies 28/72

The pandemic and direction numbers are close enough to slough off the MoE differences. On the economy, Harris, and Biden’s overall approval, they’re off enough to call into question the methodology. Is this a mathematical error, or a weighting issue? Or have people disconnected performance from approval entirely?

Biden and Harris have to hope it’s the latter. So far, though, CNN is the only polling sponsor suggesting it.

Update: Via Twitter pal JustKarl, CNN explains its “new methodology” for this survey:

The landscape of political polling is changing, and so too is the way CNN carries out its polls. Beginning with Friday’s newly released CNN poll conducted by SSRS, the network is implementing a new methodology for measuring public opinion.

The new approach relies on both online and telephone interviews, allowing respondents to participate in whichever way is more comfortable for them, and will use different methods for measuring long-term trends and for reactions to breaking news.

The new methodology allows for the researchers behind the CNN poll to have a better understanding of who is and who is not responding to the poll, and to better account for political differences between the two groups, addressing one of the most likely causes of 2020 polling errors.

Election polling in 2020 broadly missed the mark on measuring support for Republican candidates up and down the ballot. Assessments of what went wrong suggest that the problem derived from differential non-response — that is, that the people who took surveys were not the same politically as those who did not.

Looks like this is designed to amplify those problems rather than remediate them. Perhaps a return to the drawing board is in order, CNN.