Every election year Catholics are asked, by bishops or political activists, to uphold Catholic values when they vote. They are told they must vote with a well-formed conscience, put on the seamless garment of the church’s social teaching, and not try to renegotiate the nonnegotiables.

These are all good rules as far as they go—and they go farther together than separately—but they must be applied in complicated circumstances. They do not by themselves yield any conclusions about how a Catholic should vote in a particular election. In addition to even the best rules, voters also need prudence.

If politics is the art of the possible, prudence is the virtue of the possible. It is not just caution or moderation or tact, though those good things all require it. Prudence is the virtue that translates every other virtue into particular decisions, the link between conviction and action. The word “prudence” may lack the heft and grandeur of “wisdom,” but without prudence all wisdom is impotent. The real dilemmas we face in life, including political dilemmas, do not neatly conform to any set of models or list of rules. The rules must always be applied by someone, and the application is never automatic: there aren’t—there can’t be—rules for how to interpret every rule.

That’s why it isn’t easy to vote well, with or without good voters’ guides. The difficulty becomes clear once one understands what voting shouldn’t be. To begin with, voting shouldn’t be a tribalistic exercise in which we express our sympathy for the candidate who most reminds us of ourselves. Character is an important consideration, of course, but mere personality shouldn’t be. If a candidate is as tough as John Wayne or as cool as Steve McQueen, so much the better, but being cool or tough (or beautiful) isn’t a qualification. The so-called intangibles, so important to campaign operatives and so fascinating to commentators, should be regarded with distrust by voters. In political jargon, “intangible” usually means frivolous or irrational; when it does, voters should do their best to ignore it. What pundits call a “good narrative” does not guarantee good policies or compensate for bad ones. If a voter’s main answer to the question “Why do you support this candidate?” is “I just like him” or “She’s one of us,” the voter isn’t yet prepared to vote prudently.

This may all seem obvious enough to the conscientious voter, but there’s another error that’s less obvious and just as common. If voting should not be mainly about a voter’s attraction to someone’s personality or “narrative,” neither should it be a simple expression of agreement or a declaration of one’s general principles. In fact, voting is not the right occasion for gestures or declarations of any kind—if you have something to declare, write a letter to the editor or run for office yourself. You are wasting your time behind the curtain. Many “single issue” voters treat presidential elections as if they were a single-issue referendum: because issue A matters more than any other issue, no other issue matters. Those who argue thus are fooling themselves. Political problems do not appear singly, and voters must be able to consider more than one issue at a time.

So much for what voting isn’t. It’s a harder thing to say what voting should be, and a much harder thing to vote well, in full awareness of both how little any one vote counts and how much may be at stake in an election. To vote well is to understand one’s choice as involving a kind of prediction: If the candidate I’m voting for is elected, the consequences will be better overall than if his opponent is elected. To make such a prediction responsibly, one must consider not only the candidates’ positions but also the circumstances they’ll face if elected. A presidential candidate may be powerless to pursue his—and your—favorite part of his agenda. Or your favorite part of his agenda may mean very little to him. A Republican president who must work with a Congress controlled by Democrats will face certain constraints, and those constraints will not bear equally on all issues. He may have trouble getting an anti-Roe Supreme Court appointment confirmed, while a Democratic Congress may have even greater trouble overriding his veto on health-care legislation.

Calculating the likelihood of such events, weighing the importance of one likelihood against another, gauging the relationship of a politician’s promises to his real priorities—these are all difficult tasks. They require careful attention to details of policy and an understanding of what is possible at a given moment. They also require some humility, since such calculations are always provisional. Above all, they require prudence, which proves itself only in retrospect. In politics, even the safest predictions aren’t quite certainties.