ON Jan. 28, following President Bush's State of the Union address, CBS News conducted a telephone call-in "poll" of viewer reactions. It involved what John Tierney of the New York Times correctly called "probably the largest biased sample in the history of instant polling."
Actually there has been a larger biased sample. It was the infamous 1936 election "poll" done by The Literary Digest magazine, which tabulated nearly 2.4 million "ballots" mailed in by readers - and which showed Republican Alfred Landon poised to win a victory over incumbent Democrat Franklin Roosevelt.
To measure public opinion accurately a representative sampling of the populace is needed. That's why so much time and money are spent on sample selection, and why poll-takers are concerned about viewer call-ins. For all the disclaimers by Dan Rather and Connie Chung - that the "poll" was "unscientific" - the data were presented as though the sampling was scientific.
If the responses are unrepresentative of what Americans at large are thinking - about the economy, the president, Governor Bill Clinton - why are they being shown? In this diverse country all manner of different judgments are held by people.
What do we know about the 314,786 callers whose answers were recorded following the president's speech? That they were unusually persistent. A friend who's concerned about CBS's much publicized pseudo-poll, and interested in learning more about how it operated, called in 83 times, hitting his automatic redial button, before finally getting through.
AT&T computers recorded nearly 25 million attempts to get through - though only 315,000 did. Since Nielsen ratings show that in an average minute roughly 9 million households were tuned to the CBS program, and since it may safely be assumed that many of these viewers never tried to call, we have a suggestion at least of how widespread "multiple determined call-ins" probably were.
Who besides polling experts and journalists might have ridden their redial buttons? Since CBS News publicized its plans for the program heavily for two weeks, all manner of interested groups had time to prepare, to try to tip responses in the direction of their interests. CBS didn't give out the precise 800-number until the program was aired, but groups could have had everything else in place.
No one can know how heavy or how minimal were the efforts to load the dice, or what the implications are. That's one of the problems - highly publicized call-ins on politically important subjects and events invite manipulation of results.
In general, viewer call-ins, especially when it's difficult to get through, are likely to yield results weighted toward those with a more intense concern or interest. Perhaps this is why the CBS call-in tally Jan. 28 found its respondents markedly more worried about economic problems than the public-at-large has been shown to be.
We know one thing about this recorded call-in: Westerners were notably underrepresented, Northeasterners overrepresented. While the l990 census shows 21 percent of Americans live in the Western states, CBS reports that only 12 percent of its recorded calls came from that region. On the other hand, 20 percent of the public, but 26 percent of the calls, were from the Northeast. The skewing is probably explained by CBS's post-speech program not being broadcast live in the West.
A final note. When it aired its viewer call-in tallies, CBS also presented responses to the same questions from what it billed as a parallel "scientific" poll. The latter was in fact based on phoned-in answers of 1,241 adults out of 2,897 whom CBS had interviewed earlier, had asked to watch the president's speech, and had subsequently given a separate 800-number. This latter approach is less open to bias than try-and-persist-if-you-will call ins, though this method has its own problems.
CBS News has a proud history as a leader in opinion research. Why squander that now? Why risk giving many Americans a distorted view of public sentiment, by engaging in ballyhooed call-in "polls" that are only political theater?