At first glance, the 2015 health plans offered by the Ohio nonprofit insurer CareSource look a lot like the ones it sold this year, in the Affordable Care Act’s first enrollment season.

The monthly premiums are nearly identical, and the deductibles are the same.

But tucked within the plans’ jargon are changes that could markedly affect how much consumers pay for health care. Generic drugs will soon be free, but the cost of expensive specialty medications will increase. Co-payments for visits to primary-care doctors will go down, but those for emergency room trips will be higher.

Millions of people nationwide bought health insurance this year through the federal government’s health insurance exchange, often through the website Healthcare.gov. Now, as they pick plans for next year, they face a complex battery of choices.

They have until Dec. 15 to select a new plan or they’ll be re-enrolled automatically in the one they currently have. Or, if that plan no longer exists, they’ll be enrolled in another product offered by the same insurer, when available. But even if they get the same plan — of the nearly 2,800 health plans offered in 2014, about 1,700 of them will exist in the same form next year — their benefits may not stay the same.

“You’re getting re-enrolled in the same carrier, but there’s basically no guarantees that your product looks anywhere near the same as it did last year,” said Caroline Pearson, vice president of Avalere Health, a consulting firm.

Much attention has focused on changes to plans’ monthly premiums, but changes to other kinds of benefits — affecting the cost of things like doctors’ visits and prescriptions — can be trickier to understand and make a huge difference in annual health care costs.

A ProPublica analysis of the 2014 and 2015 plans in 34 states being offered on the exchange shows the adjustments taking place. ProPublica has created a tool that allows users to see, quickly and easily, some significant ways the plans have changed from one year to the next.

Customers of more than 900 plans will see their out-of-pocket maximum for medical bills increase, usually to $6,600 for individuals, the most allowed by law for next year. Only about 250 plans are lowering their out-of-pocket maximums. About 180 plans are being discontinued for at least some customers, and the rest are keeping the same limits.

Members of more than 600 plans will see their medical deductibles increase, while those in about 380 will see their deductibles drop. Consumers of one Illinois plan will see their deductible increase by $4,800. Those re-enrolled in plans offered by Florida Blue face deductibles as much as $3,650 higher than those this year, while other customers of the same company will see deductibles decrease by up to $3,000. Florida Blue did not respond to a request for comment.

More than a quarter of the 2,800 health plans altered the costs of specialty medications for conditions like multiple sclerosis and AIDS, mostly increasing the patients’ share.

Some policy changes appear subtle, just a matter of adding or subtracting a few words, but are actually quite significant. This year, many insurers charged members a set fee of a few hundred dollars for emergency room visits. For next year, some of those plans changed the wording of their benefit, adding “co-pay after deductible.” That means the insurers won’t pay for any portion of an emergency room visit until consumers meet their deductible, spending thousands of dollars.

“Everyone has focused on premiums in the press because premiums are at least easy to understand,” Ms. Pearson said. People have a harder time detecting the effect of changes to what’s called a plan’s benefit design. “It’s just incredibly hard to do, but I think it’s really important.”

What ProPublica’s analysis suggests is that even those who would be willing to pay higher premiums to keep their current plan may be surprised to learn that substantial details have changed. They should go back to Healthcare.gov or to ProPublica’s news app to make sure their plan is still the best choice.

Shopping around is essential — and there’s little time to delay.

The open enrollment period continues until Feb. 15, and customers who are automatically renewed in their plans can still make changes until that time, but only changes made by Dec. 15 will take effect on Jan. 1.

The Health and Human Services secretary, Sylvia Burwell, has been encouraging consumers to take an active role in the renewal process. But in the first two weeks of open enrollment, fewer than 400,000 consumers actively re-enrolled. “The first deadline is just a couple of weeks away,” she said in a news release on Wednesday. “We’re encouraging everyone who is already covered through the marketplace to come back and shop because there could be savings.”

Everyone’s health care needs are different. Some people might do best with a plan that has a higher premium and lower out-of-pocket costs for particular services; others might save money by choosing a plan with a lower premium and higher co-payments.

Those earning less than four times the federal poverty rate ($62,920 for a couple) qualify for subsidies to pay their premiums, and those earning even less may qualify for additional help to lower their out-of-pocket costs once enrolled.

Changes to insurance benefits are hardly exclusive to the Affordable Care Act marketplaces. They happen regularly in health plans offered by employers.

Under the law, insurers are somewhat limited in how they can change their plans. Products are grouped by tiers: Bronze plans cover about 60 percent of their members’ overall health services; silver plans 70 percent; gold plans 80 percent. To stay at those levels from year to year, plans can’t just increase all of their charges. If they charge more for some things, that often means charging less for others.

That’s what happened at CareSource, the Ohio nonprofit. Officials there said they changed their benefits based on comments from members and conversations with others who are uninsured. “Many didn’t understand the value of health insurance,” said Scott Streator, vice president of Enterprise Strategy at CareSource. “Therefore, we changed our plan design to make it more simple, more understandable and more preventive, focused on everyday types of health care needs.”

That translated into free generic drugs and lower co-pays for physician office visits, Mr. Streator said. “If you make these changes, there’s trade-offs,” he said. “The costs go up somewhere else.” In contrast with this year, when members pay $250 for emergency room visits, they will need to meet the plan’s deductible next year before their E.R. visits are covered with a co-payment that varies from $250 to $500. And members will now pay 40 percent of the cost of specialty medications, up from 25 percent this year.

CareSource enrolled more than 30,000 people during the 2014 open enrollment cycle and expects to double that amount this time around, Mr. Streator said.

Another insurer whose products are changing is Coventry Health Care. One Coventry silver plan in the Kansas City, Kan., region is decreasing the costs of primary care visits to $5 from $10, but is increasing its medical deductible to $2,750 from $2,000, increasing its out-of-pocket maximum to $6,600 from $6,350, and increasing the cost of generic drugs to $15 from $10, among other changes. Premiums are also going up.

A spokesman said the company tries to balance its benefits and costs.

Vantage Health Plan, based in Louisiana, is increasing the medical deductible in its silver plan to $2,900 from $1,800 and is raising its maximum out-of-pocket costs, too. But the company said most of its members won’t feel the changes much. That’s because about 85 percent of the 8,400 members who enrolled in the last cycle received government subsidies.

Although those without subsidies “are going to get hit, all that was designed so that all those who are getting the subsidy, their blow would be softened because that’s where the majority of our business falls,” said Billy Justice, Vantage’s director of marketing and sales.

Vantage hopes to double its enrollment for next year.

The data analyzed by ProPublica does not include information for states that run their own insurance exchanges, including California and New York. In California, plans are required to offer a standard benefit design, which allows consumers to compare plans more easily. Insurers compete on their brand’s reputation, premiums and on the size of their doctor and hospital networks.

“There can be a big difference in the experience of the consumer in terms of what they pay out of pocket if you don’t have standardized benefits,” said Anthony Wright, executive director of the consumer advocacy group Health Access in California.

The government’s plan to automatically re-enroll consumers for 2015 has come under criticism, with some warning that consumers who don’t make a choice themselves could end up in a plan with higher costs. As a result, the government is considering a different system for 2017 in which consumers who don’t pick their own plan could be shifted to the lowest-cost plan in the market.

Has your insurance company changed your benefits this year? We’d like to hear about it. Email Charles.ornstein@propublica.org.

The divorce rate has been falling for more than three decades. That fact is not news, but it still surprises a lot of people. And so when Claire Cain Miller wrote about the trend for The Upshot this week, several readers asked for more detail, with some citing a Huffington Post article questioning the official Census Bureau data on the trends. As one of the researchers who has studied the issue, I thought it worth digging deeper into the data.

In one sense, divorce is easy to measure, because it leaves a paper trail, in the form of divorce certificates on file at county courthouses. At the end of each year, most states ask each county how many new divorce certificates they’ve issued, and the states report the total number to the federal government. The federal government then calculates a divorce rate, measured as the number of divorces per thousand people.

By this measure, the divorce rate peaked at 5.3 divorces per thousand people in 1981, before falling to 4.7 in 1990, and it has since fallen further to 3.6 in 2011, the most recent year for which data are available. Of course, the marriage rate has also fallen over this period. But even measuring divorces relative to the population that could plausibly get divorced — the number of people who are married — shows that divorce peaked in 1979, and has fallen by about 24 percent since.

Because these data theoretically count all divorces, many researchers regard them as the gold standard for evaluating divorce trends. However, Sheela Kennedy and Steven Ruggles, two researchers at the University of Minnesota, have recently argued that this reliance may be unwarranted. They note that federal funding for states to collect detailed divorce data was cut in 1996, and they suspect that in the time since, some states may have become less vigorous in chasing down divorce numbers from every county courthouse. If the government has grown more lax in counting divorce certificates, the measured divorce rate might fall even if actual divorce rates were rising.

It’s a plausible conjecture, but the bulk of evidence suggests it is not the case. The decline in measured divorce rates is not confined to just a few states. Rather, divorce has fallen since 1990 in every one of the 44 states that report such data. Given that these data come from 44 independent statistical agencies, it seems unlikely that all of them have seen their statistical systems deteriorate in sync. Indeed, the decline in divorce is also evident in those states that have high-quality systems for collecting divorce data. Likewise, the divorce rate within New York State has trended downward over the past 15 years in 56 of its 62 counties, suggesting that the aggregate decline isn’t being driven by underreporting from any specific county courthouse.

It’s even possible that reporting has improved in some states as they have adopted electronic records systems. In those states, the official statistics actually understate the recent decline in divorce, because the data for earlier years underestimated the extent of divorce then. It’s far less likely that the decline in divorce rates reflects a statistical anomaly than that it reflects a fundamental shift in American family life.

Other data point to the same conclusion. An alternative approach to measuring family change comes from asking people about their marital histories. Pollsters ask people if and when they’ve been through a divorce, which allows for a measure that relies on data entirely different from the divorce certificates. Reassuringly, this alternative measure also indicates that divorce has been declining over the past few decades.

These marital histories also offer a useful way of disentangling the extent to which fewer divorces are the result of marriages becoming more stable, rather than simply reflecting fewer marriages. If marriages are becoming more stable, then couples married in recent years should be more likely to celebrate their 10th anniversary together than couples married in earlier decades. And indeed, 76 percent of people whose first marriages occurred in the early 1990s went on to celebrate their 10th anniversary, up from 73 percent for those married in the early 1980s, and 74 percent for those married in the early 1970s. Marriages have become more likely to endure, although they are not yet as stable as those that occurred in the 1960s.

One problem with survey data like this is that if people don’t respond to all of the questions asking about their marital history, researchers are forced to fill in missing items with their best guess. Ms. Kennedy and Mr. Ruggles worry that this statistical guesswork — known as imputation — makes these data less reliable. It is a reasonable concern, for sure, but just as imputation might lead researchers to overstate the decline in divorce, it could also lead them to understate it. The difficulty is that we can never know the divorce histories of those who refused to divulge that information to surveyors.

The crucial point is that our two best ways of estimating divorce trends are imperfect, yet both suggest that it has been falling fairly continuously over recent decades.

A third measure, though only available for recent years, also points in the same direction. For the past few years, the government has asked people in the American Community Survey whether they have gotten divorced in the past 12 months. At any one time, the answers to this question tend to overstate the divorce rate. To see why, imagine that I asked you whether you had gotten a promotion in the past year. You might say yes if your promotion was recently announced, even if you had yet to start working in the new job. Likewise, you might say yes if the promotion was announced 13 months ago but you had started the job, say, 11 months ago. The “past 12 months” formulation tends to lead to overcounting.

With divorces, people who are only partway through the process might say yes, even if their divorce has not been finalized, as might people whose divorce became official 11 months ago. Sure enough, careful studies by the Census Bureau show that around one-tenth of those who say they’ve gotten divorced in the past year did not actually have their divorce granted during that year.

Nonetheless, even if these survey data get the level of divorce wrong, they can get the trend right — so long as we’re careful to compare similarly biased data over time. Consistent with the other data, the American Community Survey suggests that divorce has fallen by about 11 percent since these data were first collected in 2008. This decline is all the more striking given the stress that the recession must have put on so many marriages.

So why do some people worry that all of these data sources are wrong and that divorce isn’t actually declining? Because when you compare one data series today with a different data series from the past — apples and oranges, as the saying goes — the picture can get confusing. In particular, Ms. Kennedy and Mr. Ruggles contrasted the number of divorce certificates issued in 1980 with the number of women roughly 30 years later who told surveyors that they had gotten divorced in the previous 12 months; they then concluded that divorce is rising. You can probably see the problem with this comparison: The higher divorce rates recorded in recent surveys most likely reflect the tendency of the “last 12 months” question to overestimate divorce. The data from 1980, based on counts of divorce certificates, doesn’t have the same bias.

Another common error is to note that the proportion of the population who are currently divorced has risen. But family life is not static, and divorce is often followed only a few years later by remarriage, making current marital status a particularly poor proxy for the risk of divorce. Worse, such comparisons effectively confuse a stock (the number of people divorced at a point in time) and a flow (the number of new divorces within a period of time). It’s a bit like trying to measure the crime rate by the number of people who are in jail. Just as America’s jails have swelled even as crime has declined, so have the ranks of the divorced grown even as divorce has become less common. The stock of people who have been divorced, like the stock of prisoners, reflects not only recent trends, but also decisions made decades ago, like that of the actor Jack Nicholson, who chose never to remarry after his 1968 divorce.

There is no question that the data on divorce, like so much social-science data, are imperfect. It’s a good thing that researchers are pointing out these imperfections, because we will get a clearer picture of reality if we can get better data. But the best evidence all points in the same direction: The old claim that one in two marriages ends in divorce is no longer true.

Of course, we don’t know what will happen in the future. Even if recently married couples are divorcing at lower rates, there is some reason to think that divorce rates among long-married couples will rise. After all, it’s a depressing demographic certainty that every marriage eventually ends; the only question is whether it is in death or divorce. If medical progress continues to succeed at helping fewer marriages end in death, more may end in divorce.

This (mostly happy) pattern largely explains the one clear rise in divorce that we do see: the so-called gray divorce, among couples like Al and Tipper Gore who separate at later ages. As life expectancy continues to increase, gray divorce will most likely continue to rise, even if divorce at earlier ages continues to decline. You can think of the two trends as part of one larger trend, in fact. Gray divorce may rise precisely because so many marriages are surviving longer.

Dec. 4, 2014: Uber, the mobile-phone-based car service, is getting bigger. The company said today that it had raised another $1.2 billion in funding in a round that values the firm at a remarkable $40 billion (thus post-funding the firm is worth $41.2 billion). At 5 years old, the company now has a valuation on par with that of giants like Time Warner Cable and Prudential Insurance.

But will the rich valuation prove justified? Will it one day earn the enormous profits that investors are betting on by putting in money on those terms? While many of the headlines around Uber focus on its questionable customer privacy practices and battles with local taxi regulators, the bigger long-term economic question is one we addressed in June, the last time Uber raised money.

That article is below, updated only to reflect the latest valuation.

Most of the headlines about Uber, the rapidly growing transportation service, involve its battles to do business in more cities around the world. Not surprisingly, cabdrivers who have enjoyed being part of tightly regulated cartels in cities like Madrid, Miami, London and Los Angeles do not much care for the San Francisco-based upstart that brings them new competition.

But whether Uber will ultimately become the kind of wildly profitable company that will justify the valuation of $40 billion reportedly assigned to it by its latest funding round [this was $18.2 billion in June] doesn’t come down to those regulatory battles. If the recent past is a guide, it will eventually win them.

The question for Uber as a business boils down to two words: network effects. That’s the concept in which users of a service benefit from the fact that everybody else uses the service as well. It isn’t much use being the only person to own a fax machine, or the only person to show up at a stock exchange. Things like these become more valuable the more widely they are embraced. Network effects are the key to the wild profitability of a firm like Microsoft; Windows and Office are hard to displace, even if a competitor offers a better, cheaper product, because Microsoft products are entrenched as an industry standard.

And when one company controls a market with strong network effects, it can be one of the few sustainable ways to generate huge profits, holding on to customers and fending off competitors. The billion-dollar question is whether Uber’s model for offering transportation services has some of the same network effects as those of great information industry monopolies (Microsoft, Google), or is more like, say, the travel website business, a brutally competitive industry of middlemen.

Uber is itself a middleman, of course. On one side, it recruits drivers, who typically own or lease their cars. On the other side, it markets to consumers who may want a ride. Then it matches them up; the consumer orders a car, a driver accepts the request, the service is provided, and Uber charges the consumer’s credit card. It keeps a 20 percent commission for itself and pays the rest to the driver.

So where in that exchange are network effects? You can see in some ways how it’s like a stock exchange. Every driver wants to be part of the network with the most consumers, so that they spend more time driving and less time waiting for a call. Every consumer wants to use the service with the most drivers, to minimize the wait time when they need a ride. Uber, because it is first and biggest, will have both more users and better data with which to match up supply and demand.

The bullish case for Uber, then, is that it rapidly becomes entrenched as the biggest, most vibrant marketplace for both buyers and sellers of ride services. Competitors may arise, offering lower commissions and better software, but the fact that everybody already uses Uber will make it impossible for them to get a toehold.

But here’s the negative case: That 20 percent commission is a big, honking target for competitors. So after Uber has done the heavy lifting of hiring lobbyists and fighting the taxi regulators in capitals around the world, acclimating drivers and consumers alike to ride-sharing services and attaining that $40 billion valuation, it will have an enormous bull’s-eye on its back.

Competitors could take a smaller commission, and use the savings to offer higher rates to drivers and/or lower prices for consumers. Uber would face the choice of matching the lower commissions (cutting into its profitability) or risk losing drivers and consumers to more favorable prices offered by others.

You can imagine situations in which both consumers and drivers stick not with Uber but with whichever service offers the best deal at any given moment. Rather than automatically ordering a car through Uber, a person looking for a ride might check several competing services to see which has the closest car at the best price at that moment. Drivers might sign up with multiple services, and take rides that will offer them the highest pay at any given time. Aggregation services like Kayak might even spring up, allowing riders to instantly compare the prices and wait times on offer.

Note that this question assumes that the basic Uber business model will prevail: that car-sharing services, ordered from phones and charged automatically, will become more and more common around the world, overcoming resistance by local authorities and the taxi industry, and expanding the aggregate taxi market.

The task facing Uber is not just to overcome the hurdles and make ride-sharing a multibillion dollar industry. It’s to try to entrench the advantages it has from being first: continually refining its offerings to have the best possible user experience, the best data analytics to ensure that people can get a car when they need one, and not to be greedy with regard to its commission, lest it be all the more inviting a target for rivals. It’s no easy job, but nobody said building a company worth $40 billion is.