MADISON, Wis. – Democratic presidential nominee Kamala Harris has been widening her lead among Wisconsin voters, according to recent polls. Although all polls were within statistical margins of error, Harris has gained two to six points ahead on Republican Donald Trump. Both candidates, however, still regard Wisconsin as winnable. Harris and running mate Tim Walz have several events scheduled in the state in the remaining 7-1/2 weeks of the campaign, as do Trump and running mate JD Vance. Recent polls, all by organizations that use generally accepted statistical sampling methodology
> September 11 to September 12: Harris 49%, Trump 47%. Sample: 800 likely voters. Conducted by Insider Advantage.
> September 6 to September 9: Harris 49%, Trump 46%. 626 likely voters. By Redfield & Wilton Strategiess. For the London Telegraph newspaper.
> August 30 to September 8: Harris 49%, Trump 46%. 638 likely voters. By Morning Consult.
> September 4 to September 6: Harris 47%, Trump 47% (tie). 917 likely voters. By Americans for In vitro Fertilization.
> September 3 to September 6: Harris 51%, Trump 49%. 946 likely voters. By YouGov. For CBS television network.
> August 28 to September 5. Harris 52%, Trump 48%. 738 likely voters. By Marquette Law School.
> August 28 to September 5. Harris 52%, Trump 48%. 822 registered voters. By Marquette Law School.
Discrepant polls
Why aren’t all poll results the same? Voter preferences are a moving target. No poll pretends to be more than a snapshot. It makes a difference when a survey was conducted. It also makes a difference who was surveyed. In political surveys the most used samples are either likely voters or registered voters. Knowing the preferences of each group can be valuable on campaign strategizing, mostly to plan messaging and to channel advertising buys. Sample size also influences results. A fundamental in probability sampling is that that 384 people, if chosen without favor, will yield 95% confidence that the findings are within five points of any population group, whether the whole U.S. population, or a state, or people who buy laundry detergent. The larger the sample, the greater the confidence level and the narrower the margin of error.
Candidate claims: How trustworthy?
Political parties, interest and candidates need surveys to tap public opinion. They spend millions of dollars to contract polling companies during major campaigns – and they need accurate results. The results are for internal use and understandably carefully guarded, and the survey companies are pledged to confidentiality — with an important exception. If a candidate goes public with poll results and misrepresents them, the polling company will respond publicly to correct the misrepresentation no natter the embarrassment to the client candidate. This contractual exception to privacy is embedded in the polling industry’s code of ethics to protect then industry’s reputation. Candidates who make extravagant claims about polls usually are savvy enough mot to be vague and not to identify the polling company or identify methodology used or the dates or sample size or\other specifics.
News standards
Journalists are wary of polls paid for by candidates or interest groups. The release of poll results is often a campaign tactic or publicity ploy. This wariness has been codified by the Associated Press, the dominant global news-gathering agency.In reporting polls, the AP insists that reporters consider:
> Who did the poll and who paid for it?
> How many people were interviewed? How were they selected?
> Who was interviewed? A poll of business executives, for example, represents only the views of business executives.
> How was the poll conducted – by telephone or some other way? Polls in which computers conduct telephone interviews with a recorded voice are especially problematic.
> When was the poll taken? Opinion can change quickly, especially in response to events.
> What are the sampling error margins for the poll and for subgroups?
> What questions were asked and in what order?
Trumpian claims
Last week at a news conference in Califonia, Donald Trump, for example, made sweeping claims about leading in “all the polls” and “everyone knows that.” He threw ouy margins like 60%, 70% and even 90%. Had Trump been specific – for example, hypothetically — that a Gallup poll gave hm a 92% advantage in Michigan, Gallup immediately would gone on record that Trump was lying. As is frequent with Trump, he avoided anchoring his California claims to any source. The claims unverifiable and made of whole cloth and nothing more than hot air on the stump.