Protesters hold shields as they clash with riot police during ongoing anti-government protests in downtown Kiev. (EPA)

The possibility that Ukraine's crisis could spiral into civil war is getting more and more attention, and it's not hard to see why.

Update, Feb. 20: There are reports that some protesters may be taking up arms. If this trend continues, it may increase the odds that the military would intervene to disperse protests, but decrease the odds that the military would attempt a coup or would divide internally. Unfortunately, it also naturally increases the risk of a prolonged armed conflict, however minor. Read what this all means here.

What's happened that has people so worried?

Increasingly deadly clashes between security forces and protesters have killed at least 25 since Tuesday. Everyone is escalating in action and rhetoric, and no one is showing much sign of backing down, which is only going to get harder to do.

The government appears bent on using crackdowns to "end" the crisis (it's also warning it will launch "anti-terrorist" operations), protesters are using molotov cocktails, Moscow has labeled the protests an "insurgency" and urged the government to disperse them, European leaders are threatening sanctions, protesters occupied government buildings in the city of Lviv and may have seized weapons left by soldiers, the pro-Moscow regional government in Crimea says the protests have become a "civil war," on and on.

Who's talking about a possible civil war breaking out?

Observers worried about the risk of civil war include former Ukrainian president Leonid Kravchuk and the official spokesman for Russia's foreign ministry. Polish Prime Minister Donald Tusk has warned, “We may be witnessing the first hour of a civil war."

Alexei Vlasov, who as director of the Center for the Study of the Post-Soviet Space at Moscow State University is considered something of an expert, has warned, "If the situation goes out of control there, it could lead to civil war." The headline of Newsweek's story from today declares simply, "Ukraine: Heading for Civil War."

So: Is Ukraine really headed for civil war?

What follows are reasons for and against expecting that worst-case outcome. To be clear, anything could happen, predicting outcomes to complex events is notoriously difficult, and it's always possible that this will be decided by factors not currently visible. On net, though, the risk of civil war appears not impossibly low but very low. I explain why below.

How a civil war would happen

As a basic premise, let's stipulate that for Ukraine to become a civil war, the military would most likely have to get involved. So the question of what the army does, or doesn't do, is a really crucial one.

There are two ways for the military to get involved: either it is ordered to do so by Ukrainian President Viktor Yanukovych, most likely to stamp out protests; or it imposes itself, most likely in an attempted coup. In either case, the way that this becomes a civil war is that the military comes in seeking to stabilize the country, but things spiral out of control. Maybe its presence inflames a popular uprising, which it then fights, or the military divides over what to do and fights itself.

Keep in mind that starting a civil war would have to be a two-step process. First the military would have to get involved, either because Yanukovych asked it to step in or because it attempted a coup. Then something else bad – a popular uprising, a split in the military leadership – would have to force the military into widespread armed conflict.

Is Yanukovych going to bring in the military?

Yanukovych has been able to stick with civilian security forces for three months, but the violence is getting worse and the security services have failed to dislodge protesters. So it's possible Yanukovych is getting that desperate.

The government's talk about "anti-terrorist" operations doesn't bode well. And Yanukovych fired his army chief on Wednesday – an extremely bad sign. We don't know why he did it, but speculation has immediately turned to the possibility that the army chief had refused orders to bring the military out into Kiev's streets. If that's the case, then this is worrying both because it implies that Yanukovych may have been pushing for military involvement and because it hints at possible splits within the military leadership. All very bad signs.

The threat of European Union sanctions, already on the table because of the crackdown, makes it less likely that Ukraine might call in the military. But Yanukovych knows he is ultimately more reliant on Russia, which certainly seems open to all options in suppressing the demonstrations. So bringing in the military is not outside the realm of possibility.

Will the military attempt a coup or otherwise intervene on its own?

This seems much less likely. For all its problems, Ukraine is a democracy with a democratically elected government. Yanukovych's crackdown is certainly authoritarian in character, and he's passed some awfully draconian anti-protest laws (which were later rescinded), but there's no indication that the government is going to delay elections or shut down democratic norms entirely.

The next presidential election begins in exactly one year and one week. It seems unlikely that the military would stage a coup to remove Yanukovych, with all the risks that that entails, so soon before voters are probably going to do it anyway.

But rather than just speculate on narrative details, let's look at the preeminent annual study on the risks of coups, which is run by political scientist Jay Ulfelder. Like all model-based forecasting tools, Ulfelder's is about estimating risk, not predicting the future. Still, it's a useful tool. His model predicts a 3.0 percent chance of a coup this year. That's pretty low! There are 49 countries with a higher estimated risk of a coup. Still, it's the most at-risk country in Europe. I asked Ulfelder to help me understand why Ukraine scored how it did. Here's his response:

European countries generally get very low forecasts because most of them are relatively rich (as indicated by low infant mortality rates) and have fairly or very long-standing democratic regimes with no recent history of coup activity. In the past 50 years, coups have almost never happened in countries like that.

Ukraine also has a relatively low infant mortality rate (in global terms) and no coup attempts in the recent past. What pushes its score up relative to the rest of Europe are its "mixed" political regime, the political salience of ethnicity, and its slow economic growth.

You might expect some of the other former Soviet republics with more repressive regimes to be at greater risk — places like Azerbaijan or Kazakhstan or Turkmenistan — but over the past 50 years, coups have occurred very rarely in durable dictatorships like those. It's the "transitional" cases that are generally more susceptible, and that's part of what the forecasts are capturing.

To be clear, by "transitional," Ulfelder means that the government "combines significant features of authoritarian rule and democracy." So it's not exactly American-style liberal democracy, but nor is it Russian-style autocracy, and that matters a lot for the risk of a coup.

Let's say the military gets involved. What are the odds that that leads to civil war?

So, let's focus on the scenario in which Yanukovych asks the military to step in, since that seems significantly more likely than a coup. Will the military succeed in imposing order, or will it fail and thus invite chaos and conflict?

As it happens, my master's thesis addressed this exact question: How can you tell when a military crackdown will succeed and when it will fail? I looked at a wide range of factors across 20 military crackdowns, from the French Revolution through the Arab Spring, and found that there are three factors that best predict the outcome of a military crackdown. Those factors are: troops are demographically distinct from protesters, the military has fought on domestic soil within the past generation, and the civilian government has strong political control over the military.

My research found that, if any one of those factors is present, the crackdown is likely to succeed. In Ukraine, two of them are present. The civilian government does have strong control over the military – the president just fired his military chief, after all. And Ukraine's demographic divide means that it would be hypothetically quite easy for the military to field troops from the country's Russian-speaking half to crack down on the protesters in the Ukrainian-speaking half.

And there are other factors here that tend to correlate with a successful crackdown, or at least that don't correlate with failure. That Yanukovych's most important ally, Russia, approves of a crackdown would make it more likely to go through. Surprisingly, some of the factors that you might expect to aid in the protesters' ability to resist crackdowns – a unified opposition, real democratic institutions – don't seem to have historically predicted outcomes one way or the other.

In other words, the research suggests that, if the Ukrainian military intervenes to put down the protests, it will probably be successful. Probably.

What about if the military divided against itself, though?

That does not tend to happen much. This would mean some number of senior military leaders "defecting," but right now it's not clear that they'd have anything to defect to. There is an outside risk here, given that the military likely reflects the country's demographic divide between Ukrainian- and Russian-speakers. But it's not very likely, particularly since it's hard to see how the military would come to face a crisis serious enough that generals would even have the opportunity to defect if they wanted to.

So what you're saying is that civil war is probably not going to happen.

Right. The most likely outcome is that the military stays out and that Ukraine's crisis, however dire it might become, will not lead to civil war. If the military does get involved, it will probably be to impose order on behalf of Yanukovych, which past case studies suggest would be successful and thus avert civil war.

The scenarios that would lead to civil war seem unlikely but not totally impossible. An attempted coup that leads to a civil war is not that likely since the data suggest that the military has perhaps a 3.0 percent chance of attempting a coup. A failed crackdown that leads to a popular armed uprising is not likely as any military crackdown will likely succeed. An internal military divide is not likely unless the military comes to face a massive internal crisis, which would require one of the above scenarios to happen, which as explained is not too likely.

Still, as remote as all these possible outcomes are, the risk is just real enough that everyone in Ukraine is going to be better off if this crisis is resolved democratic, pluralistically, peacefully and very soon.