METHODOLOGY

The survey was conducted through Survey Monkey, with the link distributed primarily through the IAHSS and IAHSS Foundation contact lists, the IAHSS website, and personal outreach by IAHSS Foundation board members.

The received data were vetted by the IAHSS Foundation Board of Directors. Multiple responses were discarded for reasons including, but not necessarily limited to, missing bed counts, outlier data that could not be confirmed, and responses coming from clinics rather than inpatient hospital facilities.

All submissions are kept confidential, and only aggregate data is reported.

LIMITATIONS

There were several limitations associated with the 2022 Crime Survey, including, but not necessarily limited to, the following:

  • The 192 responses leave about 97 percent of hospitals in the United States unaccounted for. Respondents were a self-selected group, and it is possible that sampling bias resulted in this group not being a representative sample of the nation’s hospitals.
  • Outreach targeted IAHSS members, so hospitals with no association members on staff are unlikely to have been included.
  • In several instances, judgment calls had to be made regarding what constituted illegitimate, outlier data, raising the possibility of both Type 1 and Type 2 errors. Rejected responses included, but were not limited to, one reporting 20 murders at a 400-bed hospital and three reporting very large security staffs (1,200, 2,032 and 50,000 employees). Not all responses were dismissed because of high numbers, though. A response citing 11 rapes and 112 robberies, for example, was included, in part because it was reported with a bed count of 2,797 and a security staff of 405, so it appeared likely that it represented multiple sites.
  • Some responses were non-specific. For example, the word “approximately” was used several times, as was a plus sign (+) following a reported number. In these cases, the number was used as if it was exact. Even in responses without such qualifiers, rounding and/or estimation appears to have been not uncommon, given the fairly frequent appearance of round numbers in both the incident counts and bed counts. Where a range was given (e.g., “10 to 16” security employees), the midpoint was used in calculations. One respondent cautioned that aggravated assaults were “grossly underreported,” but the reported numbers were used as submitted.
  • Some responses to the questions about psychiatric/behavioral units, threat management teams and visitor management programs noted that they operated for certain parts of the year. If they were in place for six months or more, the response was regarded as a “Yes.” Otherwise, it was considered a “No.” Three submissions that reported the use of visitor management programs in “certain areas” were counted as “Yes” responses.
  • Since responses were for individual hospitals, some hospital systems were represented multiple times.
  • The use of bed counts may not be the best indicator of hospital size and population. For example, number of Emergency Department visits, number of employees, hospital square footage, average daily census, and adjusted patient days can also be used to calculate crime rates. Bed counts, however, were the most consistently reported indicator of size and/or population and allow for continuity with Crime Surveys from previous years.
  • Data may have been mis-entered by respondents.
  • Notwithstanding the inclusion of UCR definitions in the survey, respondents may have compiled data using different definitions of crimes. There may also be variation in the definitions used for “elopements” and other terms.
  • With 192 responses, a small number of hospitals reporting a large number of incidents could significantly affect the overall rates of certain crimes. For example, four responses accounted for 84 percent of all rapes reported (37 of 44), and the 112 robberies in a single report noted above represented 46 percent of the survey total.