The verb “to want” has two different meanings. This difference is linguistically slight but economically weighty, and insensitivity to this difference in meaning causes confusions—including being misled into thinking that polls elicit reliable information on what people want.

Let me illustrate this dual meaning of the verb “to want” with a personal example.

I want a new Lexus automobile. No lie; I really want a new Lexus. As used here, the verb “to want” means nothing more than a fancy or a whim. If I tell you in idle conversation that I want a new Lexus, all this pronouncement means is that if I could acquire a Lexus at little or no cost to myself, I’d happily acquire one.

This first meaning of the verb “to want” refers to unconstrained wanting. It refers to the wishes and dreams that we each have, not only for material possessions, but for all manner of worldly and spiritual outcomes. (In addition to wanting a Lexus, I also genuinely want world peace, universal free trade, an end to human suffering, good weather, a Super Bowl victory for the New Orleans Saints . . . . . my list is endless.) We all want lots of things if they are available to us personally at a zero or near-zero price that is, if there are no constraints on our acquiring these things. Similarly, we all want lots of good things that are simply unobtainable in unlimited quantities.

While linguistically sound, using the word “want” in its unconstrained meaning risks profound economic misunderstanding. The second and more reliable meaning of “to want” is its economically relevant meaning. To want something in an economically relevant sense is to be willing to pay full price for that something. I’m unwilling to pay the $50,000+ necessary to acquire a new Lexus. Under the economically relevant (“constrained”) definition of the verb “to want,” I do not want a Lexus. While I (like most Americans) could afford a Lexus if I dramatically reduced my spending on food, clothing, entertainment, etc., I prefer to drive a less luxurious car so that I can eat something other than cabbage soup at every meal. In fact, I actually want a 1992 Toyota Camry—for that is what I drive. The Camry is not as nice a car as the Lexus, but it is priced right for me and my family.

What has all this parsing of the meaning of “to want” to do with citizen polls? Plenty.

When someone is asked by a pollster if he wants, say, more welfare payments, this person responds free of charge. No matter what response he gives, he bears no material consequences of that response. Of course, the person might understand that greater welfare payments would mean higher taxes for him, but he also understands that his response to a poll question does not alone determine government welfare policy. Because government’s welfare policy is not determined by the opinion of any individual, no individual incurs any personal cost in answering a question about such policies one way or another.

In other words, polls uncover only unconstrained wants—wants that people have independent of the costs of expressing these wants.

Compare the following two scenarios. First, you’re asked in a poll whether or not you want your city to build a subway system. Second, you’re asked by a Lexus dealer if you want to purchase a new Lexus. Your answer to the first question changes nothing. Whether or not the subway system is built is independent of your answer. Therefore, you will likely be much more cavalier in answering such a question. Your answer is without consequence.

But you bear significant personal consequences when answering the question posed by the Lexus dealer. If you say “yes, I’d like to buy this new Lexus,” you commit yourself contractually to buy an expensive automobile. If you say “no,” you do not commit yourself. Either way, your answer has immediate and personal consequences for you. Your answer matters; it determines directly whether or not you purchase the car. As a result, you answer the Lexus dealer’s question carefully and prudently. If you’re not wealthy, you likely will say “no.” You might tell the Lexus dealer that you really like the car, and that if it were less expensive you might buy one. But at its current high price, you do not now want a new Lexus.

So when British citizens claim in a poll to “want” such things as greater welfare disbursements and continued police efforts to prevent adults from smoking pot, these answers do not mean that British citizens “want” these things in the same way that I want my 1992 Toyota or the bag of groceries that I just purchased. Such poll results reveal only Britons’ unconstrainedwants, not their economically relevant (“constrained”) wants.

Because opinion polls do not reveal people’s constrained wants, it is illegitimate to use polls to guide government policy-making.

But what are elections if not polls? In elections as in polls, no individual’s vote determines policy outcomes. Therefore, voters typically vote with far less prudence and shrewdness than they use when buying groceries or choosing a plumber, what people really want—in an economically meaningful sense—is revealed only through private market transactions where each person directly confronts the full costs of expressing his desires. In contrast, no poll or election reveals genuine wants constrained by costs. Only the market reveals genuine wants; therefore, only the market can be trusted to respond to people’s real demands.