Americans’ love for their pets is apparent in dog parks and grooming parlors, on cat videos and pet food price tags. But just how many pets we love is less clear.

Google the U.S. pet population, and you’re quickly confronted with two oft-cited, and contradictory, sources. The American Pet Products Association found that 68 percent of U.S. households owned some sort of pet in 2016 — “equal to the highest level ever reported,” it gushed in the executive summary. Among those pets were about 90 million dogs and 94 million cats, the group said.

The American Veterinary Medical Association, by contrast, recently reported that 57 percent of households had a pet at the end of 2016, a figure the group characterized as having only “inched up” from its previous survey in 2011. The overall pet population, AVMA said, included 77 million dogs and 58 million cats — significantly lower than what the pet trade group cited.

These dueling digits matter nothing to the kitties and doggies that occupy our sofas and our hearts. But they’re important to many humans. Pet companies want to how much they might be able to sell. Veterinarians need to know how many patients to expect, and veterinary schools must determine enrollment numbers and course offerings. Whether people own pets can influence where they decide to live, or even whether they have children.

“For me, as an academic, I’m offended by the fact that everyone is running around using numbers” that vary so much, said Andrew Rowan, a former chief executive of the Humane Society International and a longtime scholar of pet demographics. “You can’t really make public policy decisions in the absence of data.”

It turns out there are other, stronger data from surveys that don’t appear at the top of search engines, and they find far lower pet ownership rates than the pet industry group’s record high figures. The Simmons National Consumer Study, which surveys households annually, found last year that 53 percent owned pets, figures that suggested at least 77 million dogs and 54 million cats, or about the same as the veterinary association.

And the U.S. Census Bureau’s American Housing Survey, which asked about pet ownership in 2013 and 2017, most recently reported that 49 percent of households included pets.

– – –

Why the discrepancies? Methodology matters. Both the industry and veterinary groups publish population figures in enormous tomes that drill deep into spending on pets, where people get pets and what kinds of pets they have. And both recruit respondents using internet “opt-in” panels, an increasingly common method that’s cheaper than randomized mail, telephone or in-person methods. Yet even though responses are typically weighted to match U.S. demographics, experts say such samples may not produce a representative picture. In addition, surveys focused heavily on one topic can also have a bias that’s hard to shake — in this case, toward pet owners, who are more likely to take a survey on pet ownership.

“When it comes to opt-in online surveys, you never really know what you are going to get,” said Robert L. Santos, chief methodologist at the Urban Institute, a think tank. He noted that in 2013, the American Association for Public Opinion Research warned that researchers should avoid this method for estimating populations.

The pet products association, which performs its survey every two years and had 22,000 respondents in 2016, used mail surveys until 2010 — and that’s when its ownership estimate jumped from 62 percent to 68 percent. Although its most recent report advises readers against “direct comparisons” between the mail and online data, its executive summary and news release do just that, touting record levels of pet-owning households and “significant” increases in dog ownership between 1988 and 2016.

Anne Ferrante, APPA’s senior vice president for member relations and business development, said the association believes the switch from mail to web was responsible for the large increase and added that researchers sought to combat bias by tracking which types of people completed the survey. The over time comparisons were highlighted because members wanted them, “so we included it to ensure accuracy in data interpretation,” she said.

The veterinary association, which conducts its survey every five years and had 41,000 respondents in 2016, switched from mail to internet in 2011. But in an interview, the group’s chief economist, Matt Salois, said the most recent report is “more sophisticated” than the previous one because it targeted non-pet owners and weighted better for factors like geography and gender. In 2012, the report points out, about 80 percent of respondents were women, a bias that was not corrected through weighting.

“This is particularly important with cats, because we found that cat ownership is more popular among women than men,” Salois said. But that means that an apparent recent plunge in feline pets, from 74 million in 2011 to 58 million in 2016, might not be real, he said. “We can’t necessarily say that cat numbers are falling,” he said. “They’re not directly comparable.”

Salois also said his association felt confident about its figures for another reason: They’re not too different from those in the Simmons National Consumer Study.

– – –

Simmons, unlike the pet-focused groups, asks about “a little bit of everything,” said chief operations officer Gerry Dirksz. Every year, about 25,000 respondents answer questions about the media they consume, what products they use and their attitudes toward spending. Simmons does this by randomly sampling households from the U.S. Postal Service’s database of addresses and contacting respondents by mail or phone. Simmons, which sells the data to clients and does not release it publicly, provided it to The Washington Post.

Among the questions are a few about pets. And for a dozen years, Simmons has found that ownership is flat, ranging from between 53 percent and 56 percent of households. That suggests that the number of U.S. pets is growing alongside the number of households but that the popularity of pet ownership is steady.

Santos said he viewed the Simmons survey as”more rigorous” and accurate than the ones by AVMA or APPA. But the “gold standard” national survey, Santos said, comes from the U.S. Census Bureau. Its American Housing Survey is conducted in person from a random sample and has a response rate of 81 percent, which is much higher than almost any private survey.

The census first asked more than 30,000 households about pet ownership in 2013 as part of an effort to study respondents’ ability to evacuate during a disaster. That year, it found 48 percent had pets, a figure that barely rose to 49 percent four years later, 19 percentage points lower than what the pet products group found.

The lower Simmons and American Housing Survey figures might seem counterintuitive in a nation where dog-walking is a booming business and emotional support animals regularly make headlines. But Hal Herzog, a researcher of human-animal interaction at Western Carolina University, said flat ownership rates probably reflect opposing trends. Pet acquisition, he argues, is partially a result of “social contagion,” and ubiquitous images of dogs and cats on social and traditional media are probably fueling it in some communities. But other trends, including increasing urbanization, numbers of people living alone and growing minority populations, are likely fueling declines in other communities, Herzog said, because all are associated with lower levels of pet ownership.

Wouldn’t pet industry associations want to know this? Maybe not that much, said Stephen Zawistowski, science adviser emeritus at the American Society for the Prevention of Cruelty to Animals and a founder of the National Council on Pet Population Study and Policy. AVMA and APPA “are basically businesses,” he said, and “data they collect on the numbers of dogs and cats are less important to them than how much do people buy to feed their dog and cat.”

Opt-in online surveys may not be as reliable, but they can reflect what types of people have pets and their characteristics, Santos said. “Not everyone needs a gold-standard survey,” he said.

John de Jong, the president of AVMA, said his group’s research shows “pets continue to have a vital and important role in people’s lives.” But of particular interest to him, he said, were results indicating rising ownership of exotic animals and backyard poultry. “Vet schools and educational institutions should take note of these changes and perhaps even develop curriculum for growing needs of care,” de Jong said. “It’s an opportunity.”

But solid population numbers do matter, others say. Zawistowski, who spent much of his career focused on animal shelter demographics, said such figures are important for putting steadily dropping shelter euthanasia rates in context, for example.

Robust data can also strengthen veterinary epidemiological studies, Rowan said, and perhaps inform feral cat population estimates, which are important for assessing their impact on wildlife. The numbers could also help us know whether those declining euthanasia rates mean that deliberate puppy breeding — a target of many animal welfare groups and local and state laws — might be necessary to keep up with pet demand.

“There are a whole lot of questions that require us to at least have that baseline demographic number,” Rowan said.