Identifying Families Screened Out of CPS or Referred from other Sources
As noted in Section 1.2, data on the reporter of CA/N reports was received but no categorical source of information was available to the evaluators of referral sources for families referred in other ways.
Intake narratives were available for 84.5 percent of families that accepted PSOP services and on whom other extended need data were available. A content analysis of these narratives after March 2006 was conducted. Inspection of narratives, showed that the term “screened out” was used as a general identifier by intake workers. Screened out families were coded based on the presence of the following in narratives:
screened out, screened-out, child maltreatment, sexual abuse. Of the 2,208 narratives, 54.6 % had at least one of these terms. The accuracy of this coding was confirmed through visual inspection of a sample of cases. The method was conservative and may have yielded an undercount of screened out reports.
The Family Sample
Families were contacted via a letter as soon as the PSOP case was closed in SSIS extracts.
Because of the lag between closing, data entry, data extraction and transfer from Minnesota to IAR, preparation of mailings and reception of the letter, some families had moved with no forwarding address by the time the letter reached their home. Families that responded were paid $20 for the initial family survey.
Responses were received from 608 families. The family sample was compared to the entire population on a variety of variables to determine whether any biases or imbalances occurred due to non-responses. For variables associated with family strengths and needs, no statistically significant differences (p < .05) for the following variables: emotional/mental health, parenting skills, substance use, family relationships, child characteristics (severe/chronic problems), caregiver abuse/neglect history, communication/interpersonal skills, physical health, employment/income management and community resource utilization. Statistically significant differences were found in three areas: 41.1% of the sample vs. 34.7% of the population were considered to have adequate basic needs; 14.6% of the sample vs. 10.7% of the population were considered to have a strong social support network; and, 19.9% of the sample vs. 15.2% of the population were considered to have good life skills. These difference indicate a slight bias toward families that were slightly better off in terms of basic needs and social support. No statistically significant differences (p <
.05) were found in the services offered directly (see Figure 4.3) except transportation which was offered more often to sample families (23.4% to the sample vs. 18.8% to the population, p = .007). No statistically significant differences (p < .05) were found between the family sample and the population in 19 of the 20 community referral categories (See Figure 4.2), the exception being emergency food, where 75.2% of sample were referred vs. 71.1 percent of population (p = .025). Differences in these service and referral categories at these levels would be expected by chance alone when comparing this number of variables.
These indicate relatively small difference in the response of workers to sample families as compared to the entire PSOP population of families served. On this basis, the decision was made to avoid weighting the sample in analyses.
Service Constellations
Service constellations are discussed in Section 5.5. These resulted from a factor analysis of service participation/utilization scores. Scores of 0 indicated that the service was not offered or referred to during PSOP participation and was not in place at the start of PSOP participation. When services were present the level of participation was rated from 1 (low) to 5 (high). Scores were present for 24 service
categories for 2,614 families. A principal components analysis was conducted with Varimax rotation and Kaiser normalization. After inspection, the analysis was limited to five factors. Results are shown in Table A.1, where factor loadings greater .3 are in bold.
Table A.1 Five Service Constellations based on Factor Loadings for Service Participation
Childcare/Daycare 0.14 0.18 -0.04 -0.01 0.72
Respite care/crisis Nursery 0.03 -0.11 -0.01 0.02 0.80
Medical or dental 0.21 0.61 0.01 -0.02 0.04
Marital/family/group counseling -0.02 0.15 0.54 0.25 -0.05
MH/Psych services -0.07 0.34 0.39 0.17 0.07
Drug Abuse treatment 0.09 0.06 0.10 0.79 0.06
Alcohol abuse treatment 0.06 0.07 0.08 0.80 0.01
Domestic violence services 0.29 -0.11 0.66 -0.06 0.02
Emergency shelter 0.43 -0.08 0.36 -0.08 -0.02
Rent/house payments 0.66 0.11 0.05 0.04 0.03
Housing 0.60 0.04 0.15 0.07 0.01
Basic household needs 0.59 0.11 0.05 -0.04 0.14
Emergency food 0.63 0.13 0.09 0.13 0.03
TANF/SSI/FS 0.30 0.58 0.02 0.01 0.02
Transportation 0.44 0.38 -0.06 0.00 0.07
Employment 0.41 0.39 -0.11 0.13 0.00
Voc/skill training 0.13 0.41 0.03 0.12 0.01
Educational services -0.04 0.58 0.13 -0.03 0.03
Legal services 0.20 0.13 0.51 -0.14 -0.12
Parenting classes -0.02 0.34 0.23 0.08 0.36
Homemaker/home management 0.17 0.10 0.26 0.16 0.24
Support groups 0.02 0.08 0.58 0.21 0.15
Disability services -0.07 0.38 0.33 -0.07 0.02
Recreational services 0.05 0.40 0.08 0.07 0.04
The names of the service dimensions were assigned based on the set of services categories with the highest loadings highlighted in the table. For example, the highest loadings for the first factor, poverty-related services, were: emergency shelter, rent/house payments, housing, basic household needs,
emergency food, TANF/SSI/food stamps, transportation and employment. Based on the loadings, service participation scores of each participant for each service were created. In this way a score was generated for each family on each of the five dimensions. While the method used led to largely distinct service
dimensions with very low intercorrelations, some overlap can be seen in the highest factor loadings.
Social Isolation
Items for the social isolation scale are listed in Figure 3.7. Each was scored from 1 (low isolation) to 4 (high isolation). Frequencies of summated scores are shown in Table A.2. Chronbach’s Alpha = .87.
Table A.2 Social Isolation Scores
Score Freq. Cum % Score Freq. Cum %
Quality of Neighborhoods
The 8 items utilized for quality of neighborhoods of PSOP families are listed in Table 3.4.
Frequencies of summated scores are shown in Table A.3. Chronbach’s Alpha = .84.
Table A.3 Neighborhood Quality Scores
Score Freq. Cum % Score Freq. Cum %
Items utilized in the stress measure can be seen in Table 3.5. The frequencies of summated scores are shown in Table A.4 below. Scores ranged from 4 (a lot) to 1 (no stress). Chronbach’s Alpha for the scale = .852.
Mean stress scores for income categories are shown in Table A.5. Statistical tests indicated no relationship (p = .77).
Table A.5 Mean Caregiver Stress by Income Household income during
previous 12 months Mean Less than $4,999 11.8396
$5,000 to $9,999 12.6161
$10,000 to $14,999 12.2255
$15,000 to $19,999 11.5574
$20,000 to $29,999 11.5342
$30,000 to $39,999 12.2927
$40,000 to $49,999 13.1667 Greater then $50,000 10.8571 Total 12.0588
Survival Analysis: Poverty-Related Services
The final variables of the regression equation for the survival analysis described in Section 6.4.1 were:
B SE Wald df Sig. Exp(B)
Need and Services 5.827 3 .120
Adequate to Some Needs/
Services Low to None .121 .164 .543 1 .461 1.128 Adequate to Some Needs/
Services Moderate to High .255 .160 2.539 1 .111 1.290 Serious to Chronic Needs/
Services Low to None .481 .249 3.748 1 .053 1.618
Analysis in Section 6.4.1.1:
B SE Wald df Sig. Exp(B)
Social Isolation .003 .019 .034 1 .855 1.003
Neighborhood Quality .038 .026 2.191 1 .139 1.038 Satisfaction with Worker .217 .124 3.076 1 .079 1.242
Income and Services 8.131 3 .043
Income < $10,000 / Services
Low to None .344 .248 1.922 1 .166 1.410
Income < $10,000 / Services
Moderate to High -.004 .247 .000 1 .989 .996 Income < $10,000 / Services
Low to None -.422 .260 2.643 1 .104 .656
Analysis in Section 6.4.3:
B SE Wald df Sig. Exp(B)
Employment / Welfare-E&T 13.348 3 .004 Employed-No Need / Low
Welfare-E&T -.116 .126 .851 1 .356 .890
Underemployed-Unemployed /
Low Welfare-E&T .304 .122 6.159 1 .013 1.355 Employed-No Need / High
Welfare-E&T .212 .114 3.431 1 .064 1.236
Analysis in Section 6.4.4:
B SE Wald df Sig. Exp(B)
Substance Abuse / Substance
abuse treatment 30.125 3 .000
No SA / No SA treatment -.165 .215 .593 1 .441 .848 SA / Low SA treatment .464 .234 3.927 1 .048 1.590 SA / Moderate SA treatment .192 .287 .448 1 .503 1.211
Appendix 2
The Organization of PSOP
The following flow chart is a representation of the general case-flow for PSOP referrals. Reports and referrals nearly always came through the Central Intake Unit. Any referrals that were screened out of CPS were reviewed further for possible referral to PSOP.
Child Maltreatment
Report
Self-referral or referral from public health, MFIP, other
CP Intake or Central I&R
Non-designated county worker(s) in CP, child welfare,
etc.
FAR or Investigation
Accepted for CP response
Parent Decision Outreach
Reject services
Home visit, Open case, Begin case management Accept services
PSOP Designated Worker
Contracted Community Agency Screened Out
Child Under 10?
When families met the general criteria—a child under 10 was present and at risk—the case was forwarded to the appropriate PSOP worker. These varied according the program model (see Chapter 2) being utilized: 1) a child protection or child welfare worker responsible for PSOP cases in addition to their regular caseload, 2) a designated county worker that handled only PSOP cases, or 3) a contracted community agency worker. After the receiving the case, the worker attempted to secure the family’s participation.
Some counties had slightly modified processes for involving families. For example, in Anoka County, which used three different contracted agencies for
case-management, there were two minor deviations (see the following flow chart). When community referrals to PSOP were received, they were often sent directly to the PSOP coordinator for review bypassing the intake unit. In addition, the county PSOP
Coordinator conducted outreach with all families screened out of CPS and enrolled them in the program before sending the referral on to the contracted agency.
Child Maltreatment Report
Other referrals:
MFIP, public health or self-referral
County Intake
Screened out/Child Welfare Child under 10?
Family Assessment Response or Investigation
Accepted for CP response
Parent Decision
Transfer to Contracted Community Agency Accept services
Outreach
Reject services
County PSOP Coordinator/
Worker
Begin case management