THE figures are staggering. In 1984, 84.9 million American households (98 percent of them) had television sets. They kept them in use 7 hours and 8 minutes a day, making television-watching far and away the most popular leisure-time activity ever. Even if those households had only one person watching for those seven-plus hours -- a very conservative estimate -- the American public would have invested 217 billion hours before its sets last year. That's more than double the number of hours spent by the American armed forces in fighting World War II for a full year.
But how accurate are the numbers, and how are they generated?
The primary source of audience statistics is the A. C. Nielsen Company in Northbrook, Ill., a market-research firm founded in 1923. In 1940, it began issuing radio-program ratings generated by an ``Audimeter'' -- which, when attached to a radio, recorded whether it was on or off and to which station it was tuned.
Today, Audimeters are wired to the television sets of 1,700 ``Nielsen'' families around the nation (for national ratings) and to up to 500 families in each of the 11 major markets that Nielsen also follows. The samples are carefully selected, says the company, to reflect the demographic profile of the nation. Each day, the results of households' viewing are sent from the Audimeter to the company's central computer in Dunedin, Fla., for processing. In addition, during each of the four ``sweeps'' months (November, February, May, and July), the company sends out 200,000 diaries in which household members record the kind of who-watched-what information unavailable through the Audimeters.
On the assumption that these samples accurately reflect the nation as a whole -- an assumption often challenged, but defensible according to standard statistical sampling theory -- Nielsen's computer generates estimates of the total audience for each program.
And what do the results look like? Ratings for the major markets are electronically available overnight. Every two weeks the company sends to subscribers the so-called ``pocket piece,'' which lists the national ratings of programs shown by the networks. Beside each program appears:
Total audience: The number, in millions, of the nation's 84.9 million TV households tuning in to that program for at least six minutes, and the percentage that figure represents.
Average audience: The number tuning in for an average minute. Expressed as a percentage of all TV households, this number is known as the show's ``rating.''
Share of audience: The percentage of all TV sets in use at the time tuned to the program. This number is known as the show's ``share.'' This number is always larger than the rating, since the proportion of television households actually using their sets, even in prime time during winter (when viewing is substantially higher than in summer), peaks at less than 70 percent.
Average audience by half hour. Watching this figure, network executives can tell whether a show held its audience or lost it to channel switching.
In addition to Nielsen, the New York-based Arbitron Rating Company provides detailed profiles of television households in various cities and towns. The information on the 211 television ``Areas of Dominant Influence'' which it defines are particularly useful to advertisers.
Both Nielsen and Arbitron look primarily at audience size. Various qualitative measures have also been devised. One of the newest, from the Boston-based Television Audience Assessment Inc., samples audience response to a program's appeal and impact. Recognizing that many households use television as little more than a ``talking lamp,'' this service questions viewers to gauge the amount of attention they give to a program. Among its findings:
Some 40 to 50 percent of the TV audience is doing something else -- reading, telephoning, eating, doing the dishes -- while watching TV.
Half the audience leaves the room at least once during the course of a show.
Less than two-thirds of the audience for an average hour-long program watches to the end.
Viewers who give a show a high rating are far more apt to remain in the room during commercials than those who rate it low.