Parole/probation Files Improved Local Accuracy - Reducing Differential Undercount49

In fact, states and localities demonstrated a significant reduction in state differential undercount. For instance, "Mississippi had a person add rate greater than 40 percent; this state had the highest person add rate as a percentage of its 1989 eligible parolee and probationer population. In other words, 48.6 percent of all Mississippi's eligible parolees/probationers were added to the census from these programs."50 The state of Mississippi's Black undercount, according to the Bureau was 33,990. 51 Without the 5,190 persons added through the parolee/probationer programs, Mississippi’s Black undercount would have been 15 percent higher.52

Given the Bureau’s favorable assessment and the success reducing the differential undercount, it is surprising these programs have been eliminated from the Census 2000 operational plan.53 To support that decision, the Bureau cites subsequent analysis of the local files used in 1990. As the following section describes, this analysis was flawed.

The Bureau’s Evaluation of the
1990 Parole/Probationer Program Was Flawed

In 1990, when a person was not found at an address in the census, but was reported at that address on parole/probation records, that person was added to the census count.

Subsequently, the Bureau compared state and local parole and probation records to a sample of the population – the 1990 Post Enumeration Survey (PES). This comparison estimated an erroneous enumeration rate of 57.3 percent.54 That is, 57.3 percent of the 447,757 parolees and probationers added to the census were not found at the same address by the PES. For the purpose of this comparison, the Bureau assumed the PES was accurate.55

However, in 1992, a high-ranking Bureau committee found the 1990 PES had many errors. The CAPE report concluded about 45 percent of the 1990 PES estimated undercount represented measurement bias in the 1990 PES, not undercount in the census.56 It is also worth noting that one of the reasons the Bureau chose to use parole and probation records was to account for a population that actively avoids government accounting. It is logical to assume that a person with a criminal record who avoids the census would also avoid the PES – or any other government statistical survey.

Clearly, the 1990 PES, full of its own inaccuracies, was inadequate to measure the accuracy of the parolee and probationer programs. The comparison to local records also suggests the 1990 PES was no more effective locating the parolee and probationer population than the census.

In addition, the National Academy of Sciences criticized a similar comparison of administrative records to population samples during the Bureau’s 1995 census test. "Because of the broader coverage of the administrative files, [the Bureau’s] calculations understate the relative frequency with which administrative records could be matched to census records. Without an accompanying explanation, these match rates may be misinterpreted as evidence of poor quality in the administrative records" [emphasis added].57

The parolee and probationer programs added nearly a half-million real persons to the census using state and local sources. That the Bureau has eliminated the program from Census 2000 is troubling, especially in light of current national statistics indicating the potential to find real persons through parole and probation records is even greater for 2000. According to the United States Department of Justice, in 1997 there were over 3.5 million persons on parole and probation in the United States, more than half of whom are African-American, Hispanic, Asian, or members of other minority populations.58 There is still a critical need to use state and local administrative records to help find parolees and probationers.

STUDYING WAYS TO REDUCE

DIFFERENTIAL UNDERCOUNTS

In 1988, the Bureau hired Dr. Jerusa Wilson, professor of psychology at Coppin State College, Maryland, to write Reducing the Undercount of Black Male Persons and Young Black Children in the 1990 Decennial Census. In this report, Dr. Wilson cited Bureau research indicating "blacks accounted for 53% of the total number of persons missed in the 1980 Census."

Dr. Wilson concluded the differential undercount resulted from certain characteristics of a hard-to-count population, such as poverty, high unemployment rates and low education. Dr. Wilson reported "in general, the greater the ‘misery index’ the greater will be the census undercount for a group. The ‘misery index’ is used here to refer to range of factors indicating low level overall adjustments in the American society." He observed "among the major race-sex categories, black men are believed to have the highest rank on the misery scale followed by black women, white men, and then white women. This rank order holds for the census undercount."

Dr. Wilson’s research contributed to the Bureau’s development of the parolee and probationer programs. According to a 1990 Bureau report, "the PPCIP was initiated as a coverage improvement program to help address the differential undercount of Black males … Dr. Jerusa Wilson in 1988 suggested that ‘ . . . parolees and probationers would be highly representative of the hard-to-count Black male group."

Both Dr. Wilson and the Bureau based their conclusions on data regarding the parolee and probationer populations. Based on data from a survey of federal, state and local agencies, the Bureau concluded the 1987 "U.S. probationer population of approximately 2.2 million persons was 72.7 percent male, 22.2 percent Black, and 51.9 percent Nonblack, with 25.9 percent not reporting race." Thus, the Bureau developed a targeted coverage improvement program to attempt to find hard-to-count persons using parole and probation data.