Estimations of the prevalence of behaviors do not necessarily match with reality.
Prevalence itself is the total number of cases divided by the number of people at risk for being cases (Wiki). If you're calculating from measured numbers, this is easy. If you're dealing with perceptions -- not so much.
In particular, it is very easy to over-estimate the prevalence of a behavior or a disease that seems particularly dramatic. Taken one way -- if you see a disease in medical school, you're less likely to think of it as rare (...there are at least enough cases out there that you were able to see it). Taken another way -- when you were in high school, it probably seemed like everyone was having sex.
The American Academy of Pediatrics published a study on the Prevalence and Characteristics of Youth Sexting this week that found the prevalence of sending sexually explicit texts was as low as 1.0% (much lower than previous studies reporting the level somewhere north of 10%).