While reporting on next year's governors' races and writing the gubernatorial chapters for the Almanac of American Politics 2018, I often struggled to find polling data. Whenever I looked for an approval rating for a particular governor, it seemed like I would come up with only one option: the periodic 50-state surveys by Morning Consult, a survey research and media company.
It led me to wonder: Have newspapers and other pollsters cut back on approval-rating polls? I don't have any hard data to answer that question, but it appears to be the case, judging by the observations of nearly a dozen political experts I spoke with.
"Yes, this is a real and problematic trend," says Keith Gaddie, a University of Oklahoma political scientist.
"We are seeing this pattern in Texas," says Mark P. Jones, a political scientist at Rice University. "Texas' mainstream media outlets have for all intents and purposes ceased doing polls of any type, except maybe one or two during the election year."
And Jennifer Duffy, who tracks gubernatorial races for the nonpartisan Cook Political Report, says that "looking around for numbers for this cycle's incumbents is largely a needle-in-a-haystack proposition."
If this is accurate, then the pattern is occurring hand in hand with concerns about state-based polling during the 2016 presidential election. While national polls generally ended up fairly close to the final popular vote margin between Donald Trump and Hillary Clinton, surveys in key battleground states were in many cases significantly off base.
One suspected factor in the decline is cutbacks among media companies, particularly by major newspapers in the states. "The loss of local papers and the consolidation of ownership has reduced spending on local polls," says Steven S. Smith, a political scientist at Washington University in St. Louis and director of the American Panel Survey.
One journalist at a midwestern newspaper who requested anonymity said that his executive editor has told him that "he has to balance the value of learning more about our readers and how we can serve them, against the value of a poll that is a one-day story. He'd like to poll and hasn't ruled it out, but the cost is the big factor."
Politicos told us of similar patterns in locales as far-flung as Colorado, Minnesota, Missouri and Oregon.
What's more, cheaper polls are replacing "more traditional, and in my view, higher-quality polls, leading to a variant of Gresham's Law -- bad polls driving out good ones," says Andrew Smith, director of the University of New Hampshire Survey Center.
While cost is one issue, it's not the only one.
Polling has been undergoing a difficult transition as public response rates fall and Americans continue to shed their landlines. Replacing calls made to landlines with calls made to cellphone numbers is crucial for keeping polls accurate, but by law, that means using human survey questioners.
"I think polling in general is down, as the methodology is under fire," says Mark Weaver, a Republican strategist in Ohio. "The decrease of landline usage and the difficulty of identifying likely voters are more challenging than ever before."
Experts say the decline in state-based polls, especially gubernatorial approval polls, is causing a range of problems.
One potential issue is that "people with an agenda fill in the void with their polls," says Louisiana pollster John Couvillon. "The media needs to demand certain things [to avoid this problem] -- documentation of methodology, including sample composition and question wording."
Duffy of the Cook Political Report agrees. "There is a lot of amateurish polling going on out there," she says. "The result is that I trust very little of the public data."
Another problem is that national polls are unlikely to dig deeply into state policy questions the way a local newspaper poll would.
"This is really the value of public opinion surveys -- not just the score of how one politician is doing," says Benjamin Toff, an assistant professor at the University of Minnesota's Hubbard School of Journalism and Mass Communication. "What are the things citizens care about? Is the government addressing it or not?"
Bob Beatty, a political scientist at Washburn University in Kansas, echoes Toff's sentiments. "Polls are second to elections as a way to gauge citizen public opinion."
Beatty says that during the 2016 election year, Kansas had 11 different gubernatorial approval polls. But during 2015, an off year, the state had only had two, and wouldn't have had any if Ft. Hays University hadn't done its regular, twice-a-year poll.
Karlyn Bowman, a polling specialist at the American Enterprise Institute, has particularly mourned the loss of California's Field Poll, which closed down in late 2016 after decades of survey work. "It was a tragedy, because they had so many good questions on issues such as Proposition 13 [which dealt with property taxes] and Proposition 128 [environmental protection], probably never to be repeated again," she says.
The shrinkage of locally based media polling has left the field open to national outfits. Morning Consult has stepped into that breach.
It regularly conducts state interviews, selecting people from a wide range of ages, races and ethnicities, genders, educational attainment, and regions. In November 2015, editors realized that they "were sitting on huge batch of information" on gubernatorial approval that could be of interest to readers, says Jeff Cartwright, a Morning Consult spokesman. So they began releasing the data collected every quarter.
It's essentially a nationwide online poll, with enough participants to provide reliable data for every state. Overall, a typical survey will involve 195,000 respondents, with the smallest states having about 400 participants and the largest having about 10,000.
The results are released all at once, most recently on Oct. 31. The poll asks respondents if they approve of the job their governor is doing, among other things. Future releases might also include questions on whether respondents would support the re-election of their incumbent governor, Cartwright says.
"Morning Consult is a quality operation with young, but good, social scientists working on design and analysis," says Smith of Washington University. Charles Franklin, director of the Marquette Law School Poll, agrees, saying they are "as serious about getting the statistics right as anyone in the online sphere."
One potentially good outcome of having a national outfit conducting these 50-state polls is that small-population states that might not have been big enough to get rated at all in the past now get approval polls every quarter.
That said, some experts worry that substituting Morning Consult approval polls for local ones has drawbacks.
The biggest methodological concern is the time period the gubernatorial approval polls are taken over -- three months. "While we appreciate the Morning Consult polls, they are taken over too long a time to properly reflect recent events," says Al Cross, director of the University of Kentucky's Institute for Rural Journalism and Community Issues.
Cartwright says that Morning Consult's polling results have generally mirrored those of more traditional polls. With the exception of those enmeshed in scandals, governors "are not in the news every day, so minds aren't changing every day," he says.
Despite his generally warm feelings toward the media company, Marquette's Franklin says that "online polls are still proving themselves, so Morning Consult inherits all those methodological and sampling concerns."
There are some signs of life for gubernatorial approval ratings beyond Morning Consult, though. For instance, there are indications that TV stations and universities have picked up some of the slack.
In North Carolina, "we've had a nice little run of polls testing [Democratic] Gov. Roy Cooper's approval rating, but they are not coming from newspapers, which seem to conduct polls with decreasing frequency," says Jonathan Kappler, executive director of the free market North Caroline FreeEnterprise Foundation. Kappler cites polls by Meredith College, Elon University, and High Point University.
And in Texas, Jones of Rice University says he's pleased that The Texas Tribune, a nonprofit news site, has become a major player in political polling. Tribune polls "provide us with regular data with which to gauge gubernatorial approval ratings, as well as a variety of other political metrics every four months," he says.
Unfortunately, Jones says, "most states are too small to sustain an expensive, professionally run operation like the Tribune that has the vision and resources to conduct regular polls."
*This story has been updated.