The Biden campaign has tried to take solace in the idea that perhaps voters' minds aren't set yet as it wrestles with a deficit at the polls and dismal economic ratings.

What may work in its favor is academic research suggesting that six months before Election Day — a deadline reached earlier this month — is when the countdown really begins.

"The question that politicians ask is, are you better off than you were four years ago?" said Berkeley political science professor Gabriel Lenz. Lenz’s research suggests voters often process the question "a little closer to something like, am I better off than I was six months ago?"

But the research also offers a sizable asterisk: Voters seem to have longer memories when it comes to inflation.

One who says so is Yale professor Ray Fair, who has long run an election model that aims to predict how many votes each party will get depending on the state of the economy.

His model analyzes economic growth over the previous three quarters but stretches back 15 quarters — basically a full presidential term — when looking at inflation.

"People are upset about inflation even though it's coming down now, because the level of prices are much higher now than they were when Biden took office," he noted in a recent interview.

"People are still going to remember" in November, he added.

Overall, both experts agree that Biden could see his economic fortunes tick up marginally if strong stock market returns and economic growth continue. But both also acknowledge there is a persistent gloominess in voters that simply doesn't reflect broader economic data.

A recent Guardian-Harris survey was just one example. It found 56% of respondents wrongly saying the US is in the midst of a recession, 55% saying the economy is shrinking, and 49% saying the S&P 500 index is down for the year.

"The question is why people are somewhat pessimistic about the economy given the growth rate's pretty good," said Fair, "and it's got to be inflation."

What the economic models show

In spite of the many uncertainties, varied election models built around the economy do still project a Biden advantage.

The latest update to Fair's model predicts that the Democratic (i.e. Biden's) share of the November presidential vote will be 51.72%.

Fair's projection does remain within the margin of error, which means the race is basically too close to call. Nevertheless, it shows an ever-so-slight Democratic edge.

Another look at the issue is the Moody's Analytics presidential model. That measure predicts election results in each state based on the economy. It was updated in the past week and found President Biden with a slim lead but a projected margin of victory in the Electoral College that has narrowed in recent months.

For his part, professor Lenz believes "the fundamentals right now favor Biden" and said his research suggests that Biden's outlook will gradually improve if current positive economic trends continue.

"We might expect some forgivingness or forgetting about inflation as Election Day approaches," he said.

But the models paint a strikingly different picture than current polling, which shows a persistent Trump lead. Two new results just in the past week showed Biden's deficits.

One recent survey from the Swing State Project — a collaboration between the Cook Political Report and a GOP and Democratic polling firm — looked at seven swing states and found Trump ahead in six of them. The two candidates are tied in Wisconsin.

Additionally, a national poll from Marquette Law School predicts a 3-point lead for Trump when third-party candidates are included. The race is tied in a head-to-head race.

Other wild cards

There is a range of other factors that could upend the current polling and any predictive models.

One X factor not present in previous races, Lenz noted, is that "people have positive memories of the economy under Trump, which really was, up until the pandemic, pretty good for them."

Another factor, of course, is that the economy could turn south in coming months with either a resurgence in inflation or a drop in GDP growth. Changes in either of those areas would upend the various economic models.

There have also been times in history when voter perceptions ended up not aligning with the economy all the way through Election Day.

Fair cites the 1992 election as a historical example.

That's a year, the professor notes, when George H. W. Bush's economy was actually in decent shape "but people seemed very pessimistic about the economy for some reason, much more than they should have if you just looked at the numbers."

Bill Clinton, of course, won that election and denied Bush a second term.

That's also one of the few elections that Fair's model missed, as it predicted a Bush reelection.

Another notable election his model missed? Biden's win over Trump in 2020 (though he did correctly call Trump's win in 2016).

Ben Werschkul is Washington correspondent for Yahoo Finance.

Click here for politics news related to business and money

Read the latest financial and business news from Yahoo Finance