Ipsos is one of the most prolific pollsters out there. They do a lot of polls. But I'm starting to see separations from them and other pollsters. And I'm not sure what it means, but let me explain why I'm confused.
The problem is sample size. In every one of the polls I'm about to mention, Ipsos was either the lowest or among the lowest sample size. In most states, they polled half or 1/3 fewer people than the next closest poll.
Take New Hampshire. RealClearPolitics (which does not use Ipsos) has the race at about an 8-point Clinton advantage using an average of the polls. FiveThirtyEight (which includes Ipsos) has three polls by Ipsos in the last 10: One showed a huge Trump lead, and the other showed a one-point lead for each candidate. But every other poll showed Clinton with at least a 5-point lead. Further, the Ipsos sample size (how many people they interviewed) was the smallest at about 130-160 people. The next lowest is 417 people.
In Arizona, the sample size is half of the next smallest pollster, but they're in line with others. In Iowa, they're half the sample size of the next smallest pollster, but nearly a third of most pollsters in the state. They are somewhere in the middle of Iowa polls.
It works both ways for the candidates, though. In Florida polls, they're tied for the smallest sample size, but were 3 of the 4 best polls in the last month for Clinton. In North Carolina, they've got half the sample size of others, but also two of the three best Clinton polls. But in Georgia, the sample size is 1/3 smaller than the next, but they're more pro-Trump lately than others. And in Texas, the Ipsos polls are the most favorable for Trump.
If you listen to Ipsos, the Upper Midwest is close. Of the last 10 polls of Michigan and Wisconsin, two of the three smallest sample sizes were by Ipsos, and those are two of the three best Trump polls of that state — the only ones showing him with a lead.
There's one further complication: Ipsos changed its methodology just before the conventions. They removed the word "Neither" from their choices, so it's now Clinton, Trump or "Other," which used to say "Neither/Other." This was to be more consistent with other poll firms, but I don't know what that means for most of these polls, since they're all from after that switch.
So here's what that means to me: Either Ipsos is right, and the Upper Midwest is suddenly in trouble for Clinton, or Ipsos is wrong and she's in the same place she was about three weeks ago. The small sample size is a concern, but I'm also confused as to whether that's good for Trump (Florida and North Carolina are closer than they appear) or for Clinton (Georgia is in play, and maybe even Texas (!) ). Either way, I'll need evidence from someone other than Ipsos in order to change my mind about a state.
The Muddiness of Political Contributions
Part of the problem for Hillary Clinton is that of perception. "Crooked Hillary" and the word "scandal" get thrown about if it's deserving or not. For instance, at the end of the day, I haven't seen anything alarming about the Clinton Foundation. At worst, there was a $5 million donation by one country to the foundation in a time of negotiations, but that's ultimately a charitable organization. We're saying that Hillary (who gets no salary or compensation from the Clinton Foundation) because of a country's charitable giving? That's a lot of risk for no reward for her.
As NBC News pointed out, there's more proverbial "smoke" in Florida, where Attorney General Pam Bondi met in person with Donald Trump for a political contribution. Weeks later, she announces she's considering joining a lawsuit against Trump University, but hadn't decided yet. Four days later, she gets a donation of $25,000 from the Trump Foundation. Soon thereafter, she decides not to join the lawsuit. Trump later hosts a fundraiser for her. That's more of a direct benefit/quid pro quo than a charitable donation. I thought this quote was choice: "Two things can be true: One, the Clinton Foundation has deserved legitimate scrutiny. Two, the Trump donation to Bondi... is a much clearer case of pay-to-play."
Third Party Woes
In exhibit A of Why Andrew Doesn't Take Third Parties Seriously, there was a "Commander-in-Chief Forum" yesterday. Both major candidates got about 30 minutes to answer questions. From what I've read, Clinton struggled with questions about her emails and Trump was praising Putin, calling the military weak and describing a return to war crimes. None of these are particularly surprising, since it's within our existing expectations of the two.
But the biggest shocker was when Gary Johnson got a lesser spotlight, but still a national interview on MSNBC. I have to reprint this, because the exchange is equal parts shocking and funny.
Host: What would you do if you were elected about Aleppo?
Johnson: About?
Host: Aleppo.
*Pause*
Johnson: And what is Aleppo?
Host: You're kidding.
Johnson: No.
Host: Aleppo is in Syria. It's the epicenter of the refugee crisis.
Johnson: OK. Got it. With regard to Syria, I do think it is a mess.
A few questions later, another host was incredulous at his answer. He asked something like "Don't you think a person running for president should at least know what Aleppo is and where it is?" Johnson responded "No, I do understand Aleppo." as part of a longer answer. I don't expect a President to know more than me on everything. I'm pretty good at Game of Thrones trivia, geography and a few other things. But those aren't useful in running for president. If you're running for president, you'd better know the subject of major issues. Aleppo is a major issue, and apparently I know more about it than Gary Johnson.
Election Update
We'll see if the Commander-in-Chief forum changes anything here. I think Trump came off a bit worse than Clinton, in that it's really hard to imagine him with nuclear codes. Still, the national race is closer, if still a slight edge to Clinton. FiveThirtyEight noted an interesting trend to pay attention to in polls: Registered voter registered voter polls have had better results in swing states for Clinton than "likely voters."
Presidential
Ohio had a close new poll, but it was Ipsos, so I doubt its accuracy. As mentioned, Florida looks more like Clinton territory if you included Ipsos. If you don't, it's a 1-2 point race either way. Virginia had a few polls out with the appearance of a close race, but a closer look shows it's not that close to being a toss-up. More polls show Pennsylvania is firmly blue, as I called about a long time ago. I already mentioned my Ipsos issues with Wisconsin and Michigan. Iowa is really close. Clinton has an edge in Nevada that hasn't gone away, but it's small. Missouri's polls are consistent: It's voting for Trump (That said, the best recent poll for Trump is by... Ipsos). There hasn't been a great poll of Arizona recently. But the ones that have been taken are consistent: They're red. Missouri swings red. Arizona swings red. Toss-up states: Ohio, Arizona, Iowa, Nevada, Georgia, Florida, North Carolina. States on the watch list: Michigan, Wisconsin, Florida,
Senate
Not that many polls have come out, but here's the latest. New Hampshire's latest poll shows Ayotte with a slight lead, so that's back in the "No Clue" section. Two polls in North Carolina show a lead for each candidate, so that's back in the "No Clue" section. Another poll showed Pennsylvania tied, but I don't see enough from that one poll to change the category, especially in a "blue" state. New Hampshire and North Carolina move to No Clue. No Clue: Nevada, North Carolina, New Hampshire. Current outlook +4 Democrat, not enough for a majority.
Summary Judgments
In the only legal outcome you should have expected, Gretchen Carlson settled her lawsuit with Fox News over sexual harassment by Roger Ailes. Reports are she got $20 million, and we'll never hear the details about what she's alleging beyond that. • • • There are people who believe Hillary should have been jailed for her email scandal (which I believe was wrong, but not illegal). You go to jail when you give classified material to those outside the government. She had a private email server, which she used to communicate with those in and out of the government, but no one has alleged she sent classified material or gave it to anyone unauthorized. Further, I give you two counterpoints. The FBI director has said that the case was "not even close" to being prosecuted. The other is this detailed analysis by Mother Jones (so take it with a grain or three of left-leaning salt), which goes more in-depth than most stories on the subject I've read. The long and short of it is that the more I read about the details of this scandal, the more I wonder what the scandal is even about anymore. But I saw a clip of a focus group in Ohio talking about the previously mentioned commander in chief forum. One guy had this to say, which is apt: "When she's talking about the emails, she sounds more like the President's lawyer than the President." • • • I ran for two miles on Sunday, and am now on track for the next race. Only problem: I haven't signed up for that race yet. I need to get on that, since it's a month away. • • • The kids have been really difficult lately. But even in that frustration, there have been moments of happiness. The kids love their "Pack-Packs" that we got them for their birthday. Roland has been singing parts of "Twinkle, Twinkle Little Star" lately. Evie was heard counting to 20 one day last week, which shocked us both.
No comments:
Post a Comment