As noted in the reference section, the PSI has been used in well over 500 studies. It is not possible to review all of them. The research summarized below focuses on looking at the use of the PSI with trauma and diverse populations and as a treatment outcome measure. The focus is also on literature published after 1995, when the manual was published. 1. Validity: This measure has been widely used and validated with a broad variety of populations including: mothers of developmentally delayed children, mothers of infants exposed to cocaine prenatally, clinical samples, parents of children with conduct disorders, parents of hyperactive children, parents of children with attention deficit disorder, depressed mothers, parents of children with various disabilities and physical illnesses, parents who have adopted children, grandparents, adolescents, and parents who have used in-vitro fertilization. It has also been used in attachment studies, language development studies, and treatment-outcome studies. Construct validity has been found in a wide variety of populations including a myriad of developmental issues, behavior problems, disabilities, illnesses, and ethnic backgrounds. The PSI manual, as well as the author’s website, provides an exhaustive list of studies that can be referred to, depending on the relevant population. 2. Examples of treatment outcome studies include: Robbins, Dunlap, & Plienis, (1991), N=12, children and mothers participating in a preschool training project, Acton and During (1992), n=29, parents completing an aggression management training program for children with aggressive problems, Barkley et al. (1988), n=23, parents of children with ADD who got Ritalin dosages. The PSI has been used in numerous randomized trials of treatments for disruptive disorder including Webster-Stratton’s treatment (e.g., Webster- Stratton & Hammond, 1997, Webster-Stratton, Hollinsworth, & Kolpakoff, 1989) and Parent-Child Interaction Therapy (e.g., Bagner & Eyberg, 2003; Nixon, Sweeney, Erickson, & Touyz, 2003; Nixon, Sweeney, Erickson, & Touyz, 2004). Treatment mothers show greater PSI reductions than did mothers of comparison group children, and reductions are maintained at follow-up. In another study, PSI scores were found to be related to participation in treatment for child behavior problems. Mothers who failed to attend a first appointment following referral had higher levels of parenting stress than those who did attend (Calam, Bolton, & Roberts, 2002). 3. PSI scores have been found to predict later child behavior problems in at-risk samples (Goldberg et al., 1997). In a longitudinal analysis, parenting stress in infancy due to the child’s distractibility predicted Child Behavior Problems on the ECBI at age 7 (Benzies, Harrison, & Magill-Evans, 2004). 4. PSI scores differentiate between a number of different groups. Mothers of children with multiple diagnoses (ADHD/ODD or ADHD/ODD/CD) have higher PSI scores than did mothers of ADHD-only children (Ross et al., 1998). In a highrisk sample, PSI scores for mothers with five or more risks were significantly higher than for mothers with four or fewer risks (Nair et al., 2003). 5. Factorial validity was found in three separate analyses, one for each domain: Child Domain, Parent Domain, and overall. In the Child Domain, the 6 factors accounted for 41% of the variance. In the Parent Domain, the 7 factors accounted for 44% of the variance. Overall, the two domains as factors accounted for 58% of the variance. A factor analysis with Chinese (Hong Kong) mothers replicated this structure. However, analyses with African-American and Latina samples have found a 3-factor solution best fit the data (see Notes under “Use With Diverse Populations" for these studies). USE WITH TRAUMA POPULATIONS 1. PSI scores correlated with scores on the Child Abuse and Trauma Scale in a sample of mothers recovering from drug and alcohol addiction (Harmer, Sanderson, & Mertin, 1999). 2. Parents of children in treatment for sexual behaviors displayed high levels of parenting stress (total stress and child domain), with average scores in the 91st percentile. Biological parents scored significantly higher than did foster parents on total stress (Pithers, Gray, Busconi, & Houchens, 1998). 3. Multiple studies have shown correlations between PSI scores and Child Abuse Potential Inventory Scores (e.g., Holden, Willis, & Foltz, 1989, Rodriguez & Green, 1997). 4. Neglecting parents scored significantly lower on PSI scales than physically abusive parents (Holden, Willis, & Foltz, 1989). 5. In a sample of Hong Kong mothers, abusive mothers had higher scores than did nonabusive mothers on the Parent Domain, Child Domain, and Total PSI scores. PSI scores alone correctly classified 62.16% of cases as abusive or nonabusive. 6. A French-Canadian study (Lacharité, Éthier, & Couture, 1999) examined the sensitivity and specificity of the PSI with regard to discriminating maltreating from nonmaltreating mothers. (See next section, #5, last paragraph.) USE WITH DIVERSE POPULATIONS 1. The PSI has been used in numerous studies of adolescent mothers. Interestingly, PSI scores were related to peer support but not to family support in one study of 66 adolescent mothers (Richardson, Barbour, & Bubenzer, 1995). 2. The PSI has also been used in numerous studies with low-income African American mothers. Hutcheson & Black (1996) report acceptable internal consistency and 6-month test-retest reliability. Factor analysis suggests a 3- factor solution best fit the data with Parent, Child, and Parent-Child Interaction factors. Parenting stress has been found to be related to observations of parenting behavior (Chang et al., 2004). 3. A Spanish version of the PSI is available from PAR, and its psychometric properties were investigated by Solis & Abidin (1991). A description of the population involved can be found under “Norms.” They found good internal consistency for domain scores and most subscales (alpha): Child Domain (.94), Adaptability (.65), Acceptability (.74), Demandingness (.58), Mood (.63), Distractibility/Hyperactivity (.65), Reinforces Parent (.76), Parent Domain (.92), Depression (.75), Attachment (.58), Restriction of Role (.74), Sense of Competence (.73), Social Isolation (.74), Relationship with Spouse (.76), Parent Health (.71), Total Stress (.94). Principal components analysis with a varimax rotation for a 2-factor solution (replicating procedures used with the original sample) did not result in a clean solution, and a 3-factor solution was identified based on the scree test and interpretability of the factors. Factor 1 was composed of Depression, Restriction of Role, Social Isolation, Relationship with Spouse, and Parental Health (all are Parent Domain subscales). Factor 2 included Reinforces Parent, Attachment, Acceptability, and Sense of Competence. This factor was identified as the “Parent-Child Interaction Factor.” Factor 3 included Adaptability, Demandingness, Mood, and Distractibility (all are Child Domain subscales). A subsample of mothers with children with a physical or mental handicap was compared to the remaining 200 mothers. They had higher scores on all domains and subscales except Sense of Competence and Attachment. Solís-Cámara et al. (2004) found intervention effects using the PSI with Spanishspeaking parents in Mexico. 4. Tam, Chan, & Wong (1994) examined the psychometrics of the PSI with 2 samples of Chinese mothers in Hong Kong. The first sample included mothers of children with mental retardation, autism, Down Syndrome, and also physically abusive mothers. This sample was considered to have high levels of stress. The second sample was recruited from the community. All mothers spoke Cantonese. A slightly modified version of the PSI (items 59 and 60 “slightly modified to suit the Hong Kong context”) was used. The results suggest good internal consistency for domain scores. Internal consistency ranged from good to poor for subscales. Reliabilities were as follows: Child Domain (.85), Adaptability (.65), Acceptability (.65), Demandingness (.69), Mood (.41), Distractibility/Hyperactivity (.40), Reinforces Parent (.63), Parent Domain (.91), Depression (.75), Attachment (.39), Restriction of Role (.81), Sense of Competence (.74), Social Isolation (.69), Relationship with Spouse (.67), Parent Health (.71), Total Stress (.93). Validity was supported by correlations with the General Health Questionnaire, Langner’s Stress Scale, the Global Assessment of Recent Stress Scale, and single-item measures of self-perceived parenting difficulty and self-perceived parenting pressure. PSI scores differentiated between high-stress and low-stress groups (categorized on the basis of responses to statements regarding stress from child care). Exploratory principal components factor analysis with a varimax rotation identified 2 factors with a structure similar to that found in Abidin (1983). Chan (1994) assessed 50 identified abusive mothers and 37 community sample nonabusive mothers of similar demographic and socioeconomic background. Both samples were recruited in Hong Kong. Abusive mothers had significantly higher scores on all three PSI domains (Child, Parent, and Life Stress). They also had statistically higher scores on the Acceptability, Mood, Reinforces Parent, and Attachment subscales. PSI scores correctly classified 62.2% of the mothers as abusive or nonabusive. 5. The psychometrics of the French-Canadian version of the PSI have also been studied in numerous studies. Bigras, La Freniere, & Dumas (1996) conducted regression analyses using the Dyadic Adjustment Scale, a rating of Insularity, and the Beck Depression Inventory. Each of these measures was moderately correlated with the PSI Parent and Child Domain scores. The Parent Domain score of the PSI was a better predictor of these variables than was the Child Domain. Lacharité, Éthier, & Couture (1999) examined the sensitivity and specificity of the PSI in a sample of 81 maltreating (44.3% neglect, 14.7% abuse, and 41% neglect and abuse) and 81 nonmaltreating mothers of low socioeconomic status. They found an increase in correct classification when the subscales of Adaptability, Hyperactivity, and Competence were included into the analysis. Sensitivity using total scores was 67.9%; using the subscales it was 76.5%. Specificity using total scores was 79%; using subscales it was 75.3%.