Business Week recently described Frank Luntz by saying, “He persuaded the public to hate the soak-the-rich estate tax by redubbing it ‘the death tax.’” It’s a sentiment I’ve heard echoed by members of Congress who believe the debate over the estate tax was lost on the strength of this phrase.

The same week, George Lakoff argued that by using the word “escalation” to describe Bush’s Iraq plan, the Democratic leadership, now “savvy about the importance of accurate framing … has changed the public’s perception of Republicans’ Iraq agenda.”

Don’t get me wrong. I first read George Orwell’s classic 1946 essay, “Politics and the English Language,” over 30 years ago and wrote on the topic myself. I believe in the power of language. I urged Democrats to talk “escalation” and used that word here. Today, Luntz advises polluters to use the term “climate change” precisely because, over a dozen years ago, we urged the environmental community to call it “global warming.”

To believe in the power of language, though, is not to accept its inerrant transformative power. Sometimes we merely psych ourselves out, convincing each other, without a shred of evidence, that some linguistic maneuver has won or lost the battle for public opinion.

Consider these two examples.

Did the term “death tax” turn voters against the estate tax? Absolutely not. The unambiguous evidence comes in two types.

First, public antipathy to the estate tax predates the use of the term “death tax” by at least 60 years. While the question’s wording is debatable, a 1935 Fortune survey, conducted in the midst of the depression, found 52 percent of respondents opposed to taxes on inheritances of any size and 70 percent opposing taxes on anything less than what today would be a $15 million estate.

In 1982, more than a dozen years before the term “death tax” was coined, 64 percent of Californians actually voted to end what was labeled an “inheritance tax” on the ballot.

A second line of evidence constructs experiments within surveys where half the sample is asked their opinion on the “death tax,” the other half their views on the “estate tax.” If the former were more compelling, “death tax” would engender substantially greater opposition. It does not. In the National Election Study, 68 percent favored repeal of the estate tax, while 70 percent had the same view about the “death tax.” A similar wording experiment on another survey yielded a six-point difference between the two. These differences barely achieve either statistical or substantive significance.

What of “escalation” vs. “surge”?

As good an idea as it was, it is hard to argue from any data that using “escalation” instead of “surge” has changed public perception of the GOP Iraq agenda. Before the word “escalation” was introduced, while commentators and the country were still talking about a “surge,” an Associated Press poll found 70 percent opposed to sending more troops. After “escalation” gained currency, 63 percent opposed the policy. An ABC/Washington Post poll offered a slightly different result. When the country was debating a “surge,” 61 percent opposed additional troops, while as “escalation” was introduced, 67 percent were against sending more soldiers to Iraq.

Thus, while slightly different wording yields slightly different responses and slightly different trend lines, the overwhelming majority of Americans opposed Bush’s plan to send more troops to Iraq irrespective of the label used to describe the policy.

Words have power, but that power is neither infinite nor absolute. Indeed, sometimes they do not matter much at all. Often the impact of words is indirect: they influence public policy not by reshaping public opinion, but by falsely leading decision makers to believe that one side or the other in a debate has captured the public’s imagination by way of vocabulary.

Mark Mellman is president of The Mellman Group and has worked for Democratic candidates and causes since 1982. Current clients include the majority leaders of both the House and Senate.