(Bloomberg Opinion) — There are some weird things going on in the coronavirus data. It’s curious that cases dropped so fast, and have stayed pretty low, in the spring hot zones — New York, New Jersey and Connecticut. And why did cases remain so low in Idaho and Hawaii until recently?The mainstream narrative is that it’s all about good behavior when cases go down — mask wearing and giving up our social lives for the greater good. And conversely, bad behavior must be what makes them go up. We talk about certain regions having the virus “under control,” as if falling cases are purely a matter of will-power. A sort of moral reasoning is filling in for evidence.

But why, then, have cases plummeted in Sweden, where mask wearing is a rarity?

This is the time to use scientific methods to understand what’s happening. The pandemic has gone on long enough to reveal patterns in the way it spreads. If it’s all about behavior, that’s a testable hypothesis. If, as a few speculate, dramatic drops in some places have something to do with growing immunity in the population, we can also turn that into a testable hypothesis.

“The issue with data is one can manipulate it to show anything you want if you have an agenda,” says YouYang Gu, an independent data scientist. Cherry picking is easy — prediction is much harder, and Gu is getting some attention for the fact that models he’s been creating since April actually forecast what’s happened with the spread of the disease in the U.S.

He recently took to Twitter to urge public health officials to apply scientific thinking. He pointed to data on Louisiana, where cases were rising earlier in the summer and seemed to level off after various counties issued mask mandates.

But breaking the data down by county, he says, revealed a different story. Mask mandates varied in their timing, but places that implemented them late saw no more cases or deaths than those that did so early. “I don’t think there’s currently enough evidence to support the fact that recent policy interventions (mask mandates, bar closures) were the main drivers behind the recent decrease in cases,” he wrote.

That’s not to say that individual behavior doesn’t matter a lot — and the cancellation of big gatherings and other potential super-spreading events is more important than ever — but there may be more factors than we know driving the bigger picture.

A few scientists are examining the possibility that previously hard-hit areas are now being affected by a buildup of immunity, even if it flies in the face of the widespread understanding that the disease has to run through at least 60% of the population to achieve so-called herd immunity. (So far, antibody tests show only some 10-20% of the U.S. population has had the disease.)

The term herd immunity is a little vague in this context. It was created to characterize the impact of immunization. It refers to the percentage of the population that must get immunized in order for a pathogen to die out — a quantity that depends on the nature of the virus, the efficacy of the vaccine and the behavior of the hosts. If natural immunity is starting to help in some places, that would suggest herd immunity is a reasonable and worthy goal of an immunization program.

But scientists have little experience applying herd immunity to a natural infection, and what understanding they have is changing. Scientists have started to investigate the possibility that there’s another critical factor here — heterogeneity in the way humans interact, and in our inherent, biological susceptibility to this disease.

In a Science paper published in June, University of Stockholm mathematician Tom Britton and colleagues calculated that herd immunity might be reached after as few as 43% of a very heterogenous population becomes infected. People mix unevenly in a way that could lead to  little pockets of immunity, slowing the spread of the virus long before the world achieves herd immunity.

We may also be heterogeneous in our biology. A recent paper in Science suggests that many people who’ve never been infected with SARS-CoV-2 carry a kind of immune cell, called a T-cell, which recognizes this novel virus and may partially mitigate an infection. These cells may be left over from infections with related viruses — the coronaviruses that cause the common cold.

While scientists who authored the paper warn that it doesn’t imply that people with pre-existing T-cells can’t get infected, they leave open the possibility that it might account for some of the vast variability in symptoms.

Whatever the source of this heterogeneity, we know it exists. Most people on the contaminated cruise ship Diamond Princess remained uninfected, while others got asymptomatic infections and still others got severely ill.

Those differences can inform disease models, says statistics professor Gabriela Gomes of the University of Strathclyde in Scotland. “What we see is that infections do not occur at random, but that people who are most susceptible to infection get exposed first,” she says, leaving a pool of ever-less susceptible people behind.

So far, her predictions of the spread in the U.K., Belgium, Spain and Portugal have aligned well with reality. Her models showed small, shallow second peaks that would concentrate away from the places where the pandemic was most rampant last spring. For example, in Spain, the first outbreak was around Madrid, and now a smaller outbreak is happening around Catalonia.

She says her models keep predicting declines after the infection reached between 10% and 35% of the population. That doesn’t mean the virus has gone away — only that by her models, it won’t explode in those same places again. Gu’s models, too, predict no big second waves in New York City or Stockholm, but leave open the possibility of new outbreaks in relatively unaffected areas, just as Hawaii is now fighting outbreaks and New Zealand has imposed a new, short lockdown.

She says she didn’t expect to come up against resistance to her models in the scientific community. While she’s starting to get some attention in the media, she said journal editors told her that her modeling ideas, in preprint, posed the danger of making people feel entitled to relax their vigilance. Maybe the opposite is true, she suggests. Maybe censoring all but the most pessimistic views could discourage action by making the problem seem endless.

The controversy mirrors one that took place a few years ago when renowned cancer researcher Bert Vogelstein dared to suggest that the very nature of cancer had a random element and therefore some people who did everything right would get cancer through bad luck. He was pilloried for the view, not because it was untrue, but because it was deemed a dangerous invitation for people to be bad.

Public health in the United States has a tendency toward moralizing against indulgences. We were told obesity was caused by indulgence in high-fat food even though the evidence pointed elsewhere, and it took years to recognize that opioid addiction is a disease and not a sin. That attitude may be ingrained in the culture, but it shouldn’t get in the way of the search for the truth.

This column does not necessarily reflect the opinion of the editorial board or Bloomberg LP and its owners.

Faye Flam is a Bloomberg Opinion columnist. She has written for the Economist, the New York Times, the Washington Post, Psychology Today, Science and other publications. She has a degree in geophysics from the California Institute of Technology.

For more articles like this, please visit us at bloomberg.com/opinion

©2020 Bloomberg L.P.