SHRM survey & presentation on "Background Checking: Conducting Criminal Background Checks"

I was alerted by a listserv email to a recent study of background check practices among employers, conducted by Society for Human Resource Management (SHRM).  An informative read.  This being a self-reported survey, I'm skeptical how accurately it reflects the reality of employer policy vis-a-vis criminal background check; however, this is the first time I've ever seen any survey of this type, so I'm very much interested in its findings.  Among them are:

1. 73% of all responding ERs conduct background check for everyone, while only 19% do so for selected job candidates;
2. The perceived severity of a crime is highly relevant to whether it will materially impact an applicant's employment prospect (81% finding that it's "very influential" in their decision NOT to extend a job offer);
3. Employee safety (61%) and concern for negligent hiring (55%) are the 2 most important reasons for conducting a background check; and
4. The vast majority of ERs (63%) claim to allow job candidates to explain the results of criminal background check prior to making an adverse decision.

Right away, I notice that, perhaps because of the framing of the survey question, ERs consider the severity of the crime more relevant (81%) than the crime's relevance to the position applied for (73%).  Well, a simple narcotics possession under California Health & Safety § 11360 is a non-reducible straight felony, whereas battery under California Penal Code § 242-243 is most often a misdemeanor.  Well, you tell me who's less likely to be violent just based on the type and "severity" of crime.  To be fair, the existing EEOC guideline on Title VII does mention both the nature and gravity of an offense as a factor to consider.  Still, there's no reason why the increasingly arbitrary felony/misdemeanor distinction in criminal law should dictate employment policies of a private organization.