On the morning after Hillary Rodham Clinton's upset victory in New Hampshire, I spoke to MSNBC host Chris Matthews. He said that after anchoring coverage on MSNBC, he had been up all night talking to the NBC pollsters, trying to figure out how the pre-vote polls all got it wrong in projecting a double-digit win for Barack Obama.

Matthews wondered how they had been dead right in Iowa, and on the Republican side, but wrong with the Democrats.

"All I can tell you," Matthews said, "is that people did something inside the voting booth that was different than what they told the pollsters."

The host of


was being diplomatic. Let me be more straightforward: Voters lied to the pollsters, and they did so because of race. I know. I saw it firsthand, in 1987, right here in Philadelphia.

That year, I was Frank Rizzo's political director in his bid to retake City Hall. He had been defeated by W. Wilson Goode four years prior in a Democratic primary and was now taking another shot as a Republican. Marty Weinberg, Rizzo's campaign manager, believed Rizzo could make up the small margin by which he'd lost to Goode among Democrats in 1983 if, in 1987, the city's then 200,000 Republicans were added to the mix.

We in the Rizzo campaign always believed the election was impossible to poll because of race. As with Obama, Goode was what we would now call "the P.C. choice," although I don't know whether

political correctness

was yet an expression we used. By that I mean that Goode was certainly the more publicly acceptable, fashionable choice. In certain quarters, voters were reluctant to admit publicly their desire, much less their willingness, to vote for Rizzo.

On Election Night in 1987, I had the heady experience (for a 25-year-old) of being a spokesman for the Rizzo campaign. Simultaneous with the closing of the polls at 8 p.m., I was scheduled to do a live shot on TV with veteran anchor Larry Kane. Minutes before, Kane told me that the Channel 10 pollster predicted there would be a blowout win for Goode - with 70 percent of the vote! Kane also told me, and informed the Goode campaign, that he was refusing to publicize the poll because he knew it was incorrect.

"When I saw that lopsided tally, I knew people were lying," Kane told me last week.

The polling data were never aired. Goode beat Rizzo by just 2 points. And the pollster was fired.

Kane reminded me that the same thing occurred with former Los Angeles Mayor Tom Bradley, who was the odds-on favorite to win the California gubernatorial election in 1982 but lost, and lost again in 1986. In his honor, the tendency of white voters to say one thing and vote another is sometimes called "the Bradley effect."

I was reflecting on all of this when Chris Matthews gave me his analysis of the New Hampshire vote. Why were polls more accurate about the outcome in Iowa? Because Iowa voted by caucus, meaning, in a public forum. But New Hampshire voted inside a ballot booth. With a curtain! Iowa voters knew they would be publicly accountable for their votes, so they were stuck. New Hampshire voted in anonymity.

Why is what I saw locally in 1987 reemerging nationally in 2008? Obama's campaign has been a juggernaut, in part because he has been the recipient of a free ride by the media, creating a sense of inevitability that he will be the first African American to be nominated by either party. To be opposed to that movement on substantive, issue-oriented grounds is nevertheless to risk being thought a racist. Rather than run that risk, voters choose the easier path of lying to a pollster. Even when anonymity is guaranteed.

The same dynamic makes the media reluctant to put Obama under the microscope.

Just look at what happened to Bill Clinton. In New Hampshire, he talked in substantive terms about what he believes to be inconsistencies in Obama's record pertaining to Iraq: "It is wrong that Senator Obama got to go through 15 debates trumpeting his superior judgment and how he had been against the war in every year, enumerating the years, and never got asked one time, not once, 'Well, how could you say that when you said in 2004 you didn't know how you would have voted with the resolution? You said in 2004 there was no difference between you and George Bush on the war, and you took that speech you are now running on off your Web site in 2004, and there is no difference in your voting record and Hillary's ever since.' Give me a break. This whole thing is the biggest fairy tale I have ever seen."

Reacting to that comment, Donna Brazile told Wolf Blitzer on CNN that "I will tell you, as an African American, I find his tone and his words to be very depressing."

What exactly did she mean by "as an African American"? That had nothing to do with Bill Clinton's substantive comment about Obama and Iraq. Are Obama opponents supposed to be muzzled at this stage, forbidden to use the words

fairy tale

to question his meteoric rise against the backdrop of little media due diligence? One thing is certain: When pundits start speaking as members of a particular race, public discourse will diminish and suffer.

Bill Clinton was right. I hope the media drop their double standard and fully vet Obama's candidacy. This hands-off stuff only adds to an atmosphere in which voters who don't like Obama fear to admit it. It all but forces voters to be hypocritical. As long as Obama gets a free ride, 2008 will be the year of the Bradley effect.

Michael Smerconish's column appears on Thursdays in the Daily News and on Sundays in Currents. He can be heard from 5:30 to 9 a.m. weekdays on "The Big Talker," WPHT-AM (1210). Contact him via the Web at http://www.mastalk.com.