We’re just 60 days away from Election Day.

It’s the perfect time, then, to address one of our stated missions for Election Insider: Examining the angles of election polls, how they’re conducted and how to read them.

What follows isn’t an exhaustive look at every potential bias in political polls, but it’ll show you what to look for. And what to view with skepticism.

First, let’s define the term oversampling…

Oversampling is deliberately including a larger number of certain demographic groups in a survey sample than their actual representation in the population.

How Oversampling Works

When pollsters oversample, they select respondents so that some groups make up a larger share of the survey sample than they do in the population.

For example, a poll might include a higher percentage of rural respondents than their actual share of the population to get more precise data about rural voters’ opinions.

Correcting for Oversampling

Reputable pollsters account for oversampling through a process called weighting.

After collecting raw data, they adjust the oversampled groups back to their actual proportion of the population.

This ensures that the overall results reflect the population while still benefiting from increased precision.

Now, how do these practices affect polls?

Send your opinions to, feedback@newsyoucanacton.com

Your Rundown for Friday, September 6, 2024...

… But Liars Use Numbers

When done correctly — and that’s key — oversampling and weighting do not bias the overall poll results.

Instead, they allow for more accurate analysis without skewing conclusive findings. But there’s an asterisk…

Margin of Error

The margin of error in polls represents the range of values above and below the sample estimate within which the actual parameter is likely to fall. In English, that means…

• If a poll shows a candidate with 48% approval and a 3-point margin of error, it means the candidate's actual support in the population is likely between 45% and 51%.

Margin of error is typically calculated for a 95% confidence level, meaning that if the survey were repeated multiple times, you’d get the same results 95% of the time.

When weighted appropriately, oversampling can actually help pollsters reduce the margin of error. Which brings us to…

Challenges in Modern Polling

Some recent trends have complicated the polling landscape:

• Response bias: Certain groups — like those with graduate degrees — are more likely to respond to surveys than others.

• Demographic shifts: Changes in voting patterns among different demographic groups can make historical models less reliable.

Those two factors explain why Trump was behind in most Rust Belt polling leading up to the 2016 election, but won Michigan, Pennsylvania and Wisconsin to huge mainstream surprise.

Importance of Methodology

To assess a poll's reliability, reputable polls will provide detailed information about their sampling methods, weighting procedures and margins of error.

An easy way to crosscheck polls you see online — click the links in the fine print. Check the methodology and the disclosures for yourself.

You can also Google the name of the polling agency and check their partisan alliances.

Bottomline: Polls, even those with good data, can be relatively subjective snapshots of a moment in time.

They reveal a trend… but as we saw in 2016, they are not necessarily predictive.

Remember this the next time the mainstream media crows about week-over-week polling numbers like they’ve been sealed with a papal bull.

Market Rundown for Friday, Sept. 6, 2024

The S&P 500 is down 0.30% to 5,503.41.

Oil’s up 0.45% to $69.46 for a barrel of WTI.

Gold is up 0.42% to $2,553.80 per ounce.

And Bitcoin’s up 0.95% to $56,708.54.

Send your comments and questions to: feedback@newsyoucanacton.com