Unsolvable? Assessing the Accuracy of Missing Person Case Data

Annually, about 100,000 individuals are reported missing to authorities in the US. Fortunately, the majority of these investigations are resolved quickly with the persons being accounted for and reunited with loved ones. Conversely, the number of cases that remain open represent dilemmas for police as the incidents could be connected to unsolved homicides, other violent crimes, or endangered persons. Despite the commonality of missing persons, research on the topic is scant. Predominantly, published studies on missing persons focus on investigations from European countries. In 2008, the US Department of Justice launched the National Missing and Unidentified Person System (NamUs), a publicly accessible repository, to promote sharing of information in these investigations between law enforcement, medical authorities, and the public. The purpose of this study was to examine the accuracy of missing person records from the NamUs database by comparing characteristics to those compiled in arrest reports. The researcher hypothesized that NamUs records contained inaccuracies, and that these mistakes would be more prevalent for incidents involving persons considered as having low socioeconomic status. Theoretically grounded in Black’s Behavior of Law and social exclusion theory, a non-experimental research design was employed using purposive sampling to collect 161 records of missing persons listed in NamUs who had also been previously arrested. From the two sets of data, the heights, ages, hair and eye colors were compared, and an accuracy score was assigned for each case. These scores were then statistically assessed using the variables of SES and the difference in days between date missing and date reported as missing to authorities. Consistent with previous research that found inconsistencies contained in other large criminal justice data, discrepancies in NamUs case listings were noted as well. Specifically, 23.0% of the total cases (n = 37) contained a least one discrepancy. Of these 37 records, 17 (46.0%) contained a very minor conflict between arrest and missing person reports (e.g. a difference of 2+ inches in recorded height). Further, only one record contained conflicts in two of the assessed categories, and no record contained discrepancies in more than two characteristics. Despite these record peculiarities, 6 statistical tests (e.g., t-test, Chi-Square, ANOVA, and regression) determined that the noted discrepancies were not statistically significant. Further, the results of a factor analysis using variables strengthened by linear interpolation to account for missing values produced low Eigenvalues and was unable to explain the variability in the noted discrepancies. This research represents a first empirical look at the NamUs system’s data, as well as one of the few studies to examine missing persons in the US. In addition, this work employed an obscure federal government tool to assign SES based on residential address. Weaknesses of the study included that cases from Southern states were overrepresented, there was a lack of publicly available arrest records, and an unintended recency bias was present in that the cases sampled consisted primarily of missing person records from the past 7 years. Limitations of this research were that the sample consisted exclusively of adults, examined cases involving those with criminal histories only, and the study was hindered by the lack of availability of arrest records to review. Recommendations for future academic research and practitioners were also discussed