Troubling study finds rape incidents are far higher than reported by the U.S. Census Bureau

The incidence of sexual violence is likely sharply underreported in a closely-watched barometer of US crime because of the way it's designed, an expert panel said Tuesday.

The National Crime Victimization Survey (NCVS), conducted by the US Census Bureau for the Bureau of Justice Statistics, put the number of rapes and sexual assaults at 346,830 in 2012, up from 244,190 the previous year.

But other surveys suggest the number is far higher, as high as 1.3 million in a year, according to one 2010 estimate by the National Intimate Partner and Sexual Violence Survey from the Centers for Disease Control.

In its report Tuesday, the National Research Council panel said a lack of privacy when conducting oral interviews in households might be a reason for the lower figures in the NCVS.

Since most rapes and sexual assaults are carried out by individuals known by the victim, survey respondents might shy away from disclosing their assault within earshot of family members, the panel said.

It suggested creating a separate survey that would deal with sexual violence in "a neutral context" such as health and well-being, rather than in the narrow context of crime.

Questions could be worded to focus on specific actions -- being forced to have sex while intoxicated, for instance -- which could make it easier for respondents to better express their experiences, it said.

Candace Kruttschnitt, a University of Toronto sociology professor who co-chaired the panel, said quality data was essential for providing support to victims and helping to curtail future attacks.

"To more accurately measure when and how these victimizations occur, we recommend a separate survey that is focused on these specific crimes within a public health context and targets those most at risk for sexual violence," added fellow co-chair William Kalsbeek, a biostatistics professor at the University of North Carolina.