How to spin poll results... ("Biden's Lead Over Trump Diminishes in New National Poll")

Note that I am posting this in GD, not LBN, because that would be giving the story more credibility than it deserves.

Biden's Lead Over Trump Diminishes in New National Poll

Presumptive Democratic nominee Joe Biden's lead over President Donald Trump has diminished by 6 percentage points, according to a new national poll.

The poll, which was conducted by The Hill/HarrisX, surveyed 933 registered voters from July 3 to July 4. The poll has a margin of error of plus or minus 3.2 percentage points.

According to the poll, 43 percent of respondents said they would vote for Biden if the presidential election were held today, while 39 percent sided with Trump. Among the remaining percent of voters, 5 percent said they would vote for someone else, 5 percent said they wouldn't vote, and 8 percent said they were not sure.

The poll found that Trump has narrowed Biden's lead from previous polls. According to a poll conducted from June 1 to June 4, Biden held a 10-point lead over Trump, when 47 percent of respondents sided with the former vice president and 37 percent chose Trump.

Link to Newsweek story

Uh...no.

You see, while the Newsweek article selectively refers to the early-June poll results from this same firm to make the comparison, it's leaving out another poll from between the two. Here is every single poll on the race from that company in the past year:

7/3 - 7/4 Biden +4
6/22 - 6/23 Biden +4
6/1 - 6/4 Biden +10
5/13 - 5/14 Biden +1
4/19 - 4/20 Biden +2

Two things stand out right away:

First, the "drop" may have been newsworthy, had it been reported in late June, when the poll showing the drop came out. But the fact is that results have been identical over their past two polling cycles. Arbitrarily taking two non-contiguous poll results to make a point is an obvious way of spinning the results. Once could just as easily look at the mid-May poll and breathlessly announce that Biden's lead has quadrupled in less than two months.

Second, this particular outfit has consistently given Biden a far smaller lead than other polls taken at the same time. For example, at the time that their mid-April survey gave Biden a 2-point lead, every other poll was giving him one of between 6 and 10 points. Similarly, their current result contrasts with leads of between 8 and 12 points in every other national poll.

Given this, it seems fair to conclude that The Hill/HarrisX has a substantial "house effect" in favor of Republicans. Ironically, while they seem to be an outlier among polling firms, it would appear that their early-June results were, essentially, an outlier from the outlier.

But the big takeaway from this should be a reminder that, beyond any charge of liberal/conservative bias, the mainstream news media really has a "bias" in favor of excitement that will give people a reason to follow them. It's safe to say that the media really, really wants a "horse race" this year, and will seize on whatever they can to make it seem so - and, by beating the drum for such a narrative, will help to make it conventional wisdom. So don't be surprised if, from now until November 3rd, the media will do all they can to spread the message that "it's not over yet...Trump is catching up!" Don't let it distract us, but let's keep in mind that it means we're going to have to work all the harder, because we aren't going to be receiving any favors from those who shape the media narrative.