Did you know that 83 percent of all quoted statistics are made up on the spot?

It’s (probably) not true, but few things have a longer shelf life than bad numbers, especially when they are repeated by advocacy groups and seem to confirm our fears or expectations.

Statistical factoids are appealing because they seem to neatly encapsulate or summarize a larger and much more complex problem or issue. And they can — if they are valid and people understand how the numbers are calculated. I’ll examine two widely circulated bogus statistics, both of them dealing with threats to youth, a subject rife with alarmist messages.

“One-third of gay teens commit suicide.”

The claim that a third of gay teens kill themselves has circulated widely over the past few weeks in the wake of the suicide of Rutgers student Tyler Clementi.

Joel Best, professor and chair of sociology and criminal justice at the University of Delaware, debunked the statistics behind the gay teen suicide rate in his 2001 book “Damned Lies and Statistics” (pp. 89-93).

“Gay activists invoked this statistic to portray the hardships gay and lesbian youth confront; it suggests that stigma and social isolation are severe enough to drive many adolescents to kill themselves.”

Yet, Best explains, the number was derived from “a chain of bad statistics” using discredited and outdated findings (such as Kinsey’s inflated estimates of incidence of homosexuality from the 1940s), dubious assumptions, math errors, and arbitrarily selecting the highest numbers in estimates of suicide incidence.

“The final figure depends completely on the assumptions used to make the calculations,” Best notes. “Once offered, a statistic such as this one tends to be repeated, to circulate widely, without confronting questions about its validity.”

“One in five children is approached by an online predator.”

That alarming statistic has circulated widely in news stories, but is simply wrong.

The statistic can be traced back to a 2001 Department of Justice study issued by the National Center for Missing and Exploited Children (“The Youth Internet Safety Survey”).

The study did not say that one in five children is approached by an online predator; instead it found that “Almost one in five (19 percent) . . . received an unwanted sexual solicitation in the past year.” (A “sexual solicitation” is defined as a “request to engage in sexual activities or sexual talk or give personal sexual information that were unwanted or, whether wanted or not, made by an adult.” Using this definition, one teen asking another teen if her or she is a virgin would be considered “sexual solicitation.”)

None of the reported solicitations led to any actual sexual contact or assault.

Furthermore, almost half of the “sexual solicitations” came not from “predators” but from other teens (including females) in the course of flirting. When the study examined the type of Internet “solicitation” parents are most concerned about (e.g., someone who asked to meet the teen somewhere, called the teen on the telephone, or sent gifts), the number drops from “one in five” to just 3 percent.

Repeating bogus statistics not only detracts from the credibility of those who repeat the myths (don’t they check their facts?), but can have other consequences as well.

Philip Jenkins, Distinguished Professor of History and Religious Studies at Pennsylvania State University, notes in the 2001 book “You Are Being Lied To” that the “wildly inflated” gay teen suicide rate statistic often heard from gay rights activists actually harms the movement by playing right into the hands of anti-gay zealots: “The belief that homosexuality increases one’s self-destructive tendencies was repeatedly cited in the anti-gay hate literature produced by religious activists like Tim LaHaye and Anita Bryant. Both claimed that approximately half of all American suicides were the direct consequence of homosexuality.”

Suicide, gay teen bullying, and online predators are serious and important issues, and discussions about them deserve real facts instead of fictional statistics.