IT'S hardly the kind of news marriage-minded women want to believe: Only 5 percent of bachelors over the age of 40 will ever make it to the altar. Yet that gloomy statistic is being trumpeted in headlines across the country this month as the latest authoritative evidence of changing marital patterns.But wait. Just where is this official-sounding information coming from? The Census Bureau? Marriage certificates? Not quite. It stems from a study of only 30 white, never-married men between the ages of 40 and 50, conducted by a psychologist at the University of Akron. How can 30 men offer a definitive portrait of confirmed bachelorhood? Obviously they can't. But when the findings appear in the morning paper, they quickly take on a legitimacy that dwarfs the tiny size of the sampling. Figures, after all, don't lie. These percentages will take their place along with the high and low temperatures of the day, last night's baseball scores, and the closing stock quotations as verifiable numbers. The mind may balk and become suspicious in the presence of an opinion, but it opens and swallows uncritically in the presence of a statistic. Only confirmed skeptics dare to question the hallowed authority of Gallup, Roper, and Yankelovich, for instance, when they calculate - right down to the decimal point - Americans' views on everything from the serious ("Which presidential candidate are you voting for?") to the silly ("How many hours do you sleep?How often do you change channels to avoid commercials? "). Polls may be full of pseudo-information, but poll-watchers take them seriously and behave as if they're true. Squeeze a statistic right, they believe, and you'll get some great truth out of it. Who, though, can find the truth when statistics vary widely? The bachelor survey recalls a 1986 Harvard-Yale study that struck terror in the hearts of single women by claiming that a woman over 40 has only a 1.3 percent chance of marrying. Later that year a Census Bureau study restored a measure of hope by stating that a 40-year-old woman has a 17- to 23-percent chance. Which survey do you believe? Even when bachelors and career women do wed, statisticians predict that half the marriages will fail. If that happens, no one can quite agree on the economic consequences. Several years ago Lenore Weitzman, a Stanford sociologist, reported that after divorce a woman's standard of living falls 73 percent, while a man's rises 42 percent. But a forthcoming book by Susan Faludi, "Backlash, the Undeclared War on American Women," states that a woman's post-divorce standard of living declines 33 percent, and a man's increases 15 percent. Who is right? This statistical overload in a number-crazy culture gets carried to the extreme in the September issue of Money, which features the magazine's annual list of the best places to live. Editors asked 252 subscribers ("median age: 46; median household income: $61,000") what they valued in a community. Then they fed information on the 300 largest metropolitan areas into a computer and presto! Instant rankings. What else but a computer could shuffle the numbers quite so cavalierly, upgrading Bryan, Tex., from 100th place last year to 3 this year and demoting Portland, Maine, from 78 to 271? Sometimes, of course, the vaunted precision of numbers plays a useful and necessary role. Certain statistics - the ones journalists call "sobering serve as little red warning lights. The number of children who go hungry, the percentage of college women who say they have been date raped, the billions of dollars in uncollected child support - all these figures bring an awareness that words alone cannot. Even so, the bigger and more ominous the statistics, the easier it can be to look through them. Buried under mega-statistics like the deficit, people can be forgiven if they run for Harper's Index, which treats statistics like canapes and hors d'oeuvres. ("Percentage of American mothers who say they have searched their child's room: 43.Percentage of Americans who say their best friend is their father: 0.") Just one more relevant number and all the equations of life will add up. This is the statisticians' dream, increasingly imposed upon the rest of us, who may be entitled at this point to say, "Count me out."