I think it's time to step back and take a look at the purpose and value of political polling. The Post reported Feb. 19 on an NBC News exit poll in New Hampshire, in which voters were asked whether they thought Sen. Bob Dole has a ''mean streak.''

What's next? Do we ask voters on Super Tuesday if they think Vice President Bush woke up on the wrong side of the bed? Maybe we should ask them which candidate looks best holding a baby. Perhaps the most important question pollsters should ask -- though it is the least likely to be asked -- is what voters think of the pollsters. CHRIS GIDEZ Rockville

In 1948, shortly after Harry Truman defeated Thomas Dewey for president, George Gallup appeared as a speaker in my college politics class to explain why the polls -- his and most of the others -- had predicted a Dewey victory. Mr. Gallup said that two major reasons for the pollsters' errors were that they had stopped polling too soon before Election Day and had misinterpreted the responses of people who said they had not yet made up their minds.

No doubt polling techniques have improved since then, but the Gallup poll's 17-point error in favor of Bob Dole and the mistaken conclusions of other polls in New Hampshire {front page, Feb. 18} suggest that pollsters are still grappling with some of the same problems they faced 40 years ago. The lesson is for readers and viewers to retain a healthy skepticism about polling results and for the media to exercise caution in building stories around them. GIFFORD D. MALONE McLean