Table 1.

Summary of studies examining parent-youth discrepancies

Study		Sample characteristics		Sample size (N), mean ages (M), age ranges (R)	Sex/gender (male %)		Measures	Procedure	Analyses on parent-youth discrepancies	Key findings with patterns of discrepancies	Correlates of discrepancies	Quality of studies acceptability (%), violation items
Studies that used recommended analysis for discrepancy reports (e.g., polynomial regression with interaction terms, response surface analysis, LCA/LPA)
Goodman (2013)		Parent-youth in the Project on Human Development in Chicago Neighborhoods (PHDCN) comprised of diverse populations (33.2% Hispanics and 37.4% Black)		N = 485 Youth: M = 12.83 (SD = 1.60); R = 10-15	58.1% male youths No report of parents’ gender		6 victimization items in My ETV (Selner-O’Hagan et al. 1998) to assess prevalence of ECV in the last year	One-on-one interviews	Latent Class Analysis	3 classes with (1) “relative agreement” (54.5%) (2) “parents’ overestimation” (24%) (3) “parents’ underestimation” (21.5%) Excluded cases where dyads agreed on absence of victimization	Female youth more likely to be in discrepant classes. Younger youth more likely to be in discrepant classes. Racial differences in classes noted. N.s neighborhood SES differences in classes Mixed findings with youth psychopathology depending on informant for youth psychopathology.	Acceptability: 91% Longitudinal study checklist violation: Item 2 unclear (parents’ gender and age not reported)
Turnidge-Halvorson (2018)		Female caregiver-youth dyads identified as at-risk for child maltreatment from LONGSCAN, with diverse population (55% Black, 6% Hispanic)		N = 854 Youth age 8, 12 timepoints	47.5% male youth at age 8 Female caregivers (primarily biological mothers; 94.5%)		History of Witnessed Violence (LONGSCAN, 1988), Child’s Life Events Scale (LONGSCAN, 1992) at youth age 8 to measure summed total of prevalence of ECV in last year	One-on-one interviews	T tests Polynomial regression, response surface analyses	Female caregivers’ underestimation witnessed violence incidents (t(574) = 21.13, p < .001) Approx. a third of caregivers’ overestimation, a third of caregivers’ underestimation, a third of similar levels of report	Youth gender (boys) and history of caregiver victimization predictor of discrepancies Overall findings: parent-youth agreement on high exposure was associated with highest concurrent and future (4 years) psychopathology. Parent- youth agreement on low exposure was associated with lowest concurrent and future (4 years) psychopathology	Acceptability: 91% Longitudinal study checklist violation: Item 2 unclear (caregiver age not reported)
Studies that used other analysis for discrepancy reports (e.g., difference scores, chi-squared difference, McNemar’s test)
Alers-Rojas et al. (2020)		Latino mothers-adolescents in urban neighborhoods from the Project on Human Development in Chicago Neighborhoods (PHDCN)		N = 287 Youth: T1 (age 11): M = 11.22 (SD = .59); T2 (age 14) M= 13.69 (SD = .55) Mothers T1: M = 35.30 (SD = 6.09)	53% male youth All mothers		8 witnessed violence and 6 victimization items in My exposure to violence (My ETV; Selner-O’Hagan et al. 1998) to assess prevalence of ECV in the last year	One-on-one or phone interviews at two time points	Summed agreement scores (0 = disagree, 1 = agree)	Witnessed T1: M = 13.33 (SD = 2.74), [3-16] Witnessed T2: M = 13.26 (SD = 2.28), [3-16] Victimization T1: M = 11.55 (SD = 1.15), [4, 12] Victimization T2: M = 11.50 (SD = 1.11), [4, 12] Mothers’ overestimation at age 11 Mothers’ underestimation at age 14	General findings: higher agreement significantly correlated with lower concurrent and future (3 years) youth-reported internalizing and externalizing symptoms.	Acceptability: 91% Longitudinal study checklist violations: Item 11 (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)
Ceballo et al. (2001)		Female caregiver-youth with diverse population (53% Hispanic, 13% Black) from predominantly low SES backgrounds (71% welfare history)		N = 104 Youth: M = 10 R = Grades 4–5 Caregivers: M = 36	46% male youths All female caregivers		Witnessed violence and victimization items on Survey of children’s exposure to violence (SCEV; Richters & Martinez, 1990) to assess lifetime prevalence of ECV	Small group interviews for youth, One-on-one interviews for female caregivers	Chi-squared differences in prevalence rate	Witnessed violence: almost all items significant ranging from 5.35 (p < .05; witnessed gun violence) to 27.16 (p < .001; witnessed suicide) Victimization: all items significant ranging 4.24 (p < .05; burglary) to 15.32 (p < .001; threatened) Overall caregiver underestimation	Higher agreement for younger, female youth compared to older, male youth. Other demographics (e.g., marital status, employment) ns. Higher agreement significantly associated with fewer PTSD and internalizing symptoms.	Acceptability: 91% Cross-sectional study checklist violations: Item 11 (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)
Chan (2015)		Parent-adolescent dyads in six cities in China		N = 2624 Youth: M = 16.1 (SD = .8) R = 15 -17 Parents: Primarily age 35–54	52% male youths No report of parents’ gender		Witnessing or indirect victimization (excluding domestic violence) prevalence subscales from the Chinese version of the Juvenile Victimization Questionnaire (JVQ)	Questionnaires	Chi-squared differences, Percentage of agreement, disagreement (dichotomous scores)	χ2 (1, N = 2624) ranges from 37.2 (stolen at home) to 170.4 (witnessed murder) Percentage of agreement: 88.4% - 98.4% Percentage of disagreement: 1.6% - 16.7% Overall parental underestimation	N/A	Acceptability: 72% Cross-sectional study checklist violation: Item 2 unclear (parents’ gender not reported) Item 11 (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)
Hill and Jones (1997)		Parent-youth dyads in high (n = 51) and low-violence (n = 45) communities in Washington, DC, U.S (racial breakdown not reported)		N = 96 Youth: R = 9–13	52% boys		Parent and youth interviews (different questions) on witnessed community violence in past 3 years	One-on-one interviews	Chi-squared differences	Witnessed homicides: χ2 (1, N = 96) = 6.03, p < .05 Witnessed non-fatal shooting: χ2 (1, N = 96) = 5.36, p < .05 Predominantly parents’ underestimation	More agreement on more deadly crimes Youth with more discrepancies had parents perceived less peer support No other demographics differences in discrepant vs. agreement group (e.g., income, maternal edu., child anxiety)	Acceptability: 29% Cross-sectional study checklist violations: Item 2 unclear (parents’ gender, age, youth race not reported) Item 3 (psychometric for interviews on witnessed community violence not sufficiently tested in past studies) Item 4,5 (no other covariates considered in link between discrepancies and peer support) Item 11 (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)
Howard et al. (1999)		Caregiver-youth dyads in low-income public housing developments in the East Coast., U.S. predominantly African Americans (96%) and mostly without two biological parents (71%)		N = 349 Youth: R = 9–15 Parents: M = 36 (reported significant missing data with parental age)	57% male youths 84% female parents		17 witnessed community violence and 12 victimization items on modified SCEV (Richters & Saltzman 1990) to measure frequency of ECV in the past 6 months	Q uestionnaires	Mean difference scores Absolute agreement scores resulting in 3 separate groups with (1) “low (<50%) agreement,” (2) “moderate (50-80%) agreement,” (3) “high (>80%) agreement”	Witnessed violence: Overall caregiver overestimation (24%) and caregivers’ underestimation (53%) Victimization: Overall caregivers’ underestimation (58%) (1) “Low agreement” (26%) (2) “moderate agreement” (56%) (3) “high agreement” (18%)	Significant differences in youth family functioning, psychosocial pathology (violence perpetration, risks), social competence with general trend of worse outcome for “low agreement groups” compared to others	Acceptability: 86% Cross-sectional study checklist violation: Item 11 unclear (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)
Lewis et al. (2012, 2013)		Caregiver-youth dyads identified as at-risk for child maltreatment from Longitudinal Studies of Child Abuse and Neglect (LONGSCAN), with diverse population (55% Black, 6% Hispanic)		N = 766 Youth: M = 12.4 (SD = 0.44) R = 11–13.8	50% male youths 94% female caregivers		History of Witnessed Violence measure (LONGSCAN, 1998) and modified Child’s Life Events Scale (LONGSCAN, 1992) to match History of Witnessed Violence. Questionnaires measure prevalence of ECV in the past year	One-on-one interviews	McNemar’s difference test (1) “neither report” (2) “both report” (3) “youth only report” (4) “parent only report	χ² (1, N = 766) = 129.1, p < .05 (1) 51% “neither report” (2) 8% “both report” (3) 34% “youth only report” (4) 7% “parent only report”	Demographics (e.g., youth gender, race, income) not associated with parent-youth agreement. “Both report” and “youth only report” groups higher on youth-reported trauma symptoms and youth delinquency than “parent only report” “Both report” group higher on counseling service needs than other groups, however ns for the use of counseling services “Youth only report” group reported less youth-reported parental monitoring than “parent only report” group	Acceptability: 57% Cross-sectional study checklist violation: Item 2 unclear (parents’ age not reported) Item 6 unclear measures of service needs and utilization developed by LONGSAN researchers without adapting them from previous findings; however, these are one-item measures with high face validity Item 11 unclear (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)
Raviv et al. (2001)		Mother-youth dyads with 2nd and 4th graders in low (n = 65) and high violence (n = 69) neighborhoods in Tel Aviv, Israel		N = 134 Youth: R = 2nd and 4th graders	50% boys All mothers		VEX-R (Fox & Leavitt, 1995) to measure frequency of ECV in lifetime	Questionnaires	Mean difference (Mother reports – youth reports)	Mean difference scores (mother reports – youth reports) = 019 Mothers’ overestimation of youth mild violence exposure in neighborhood	N/A	Acceptability: 71% Cross-sectional study checklist violation: Item 2 unclear (parents’ age not reported) Item 11 (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)
Richters and Martinez (1993)		Caregivers-youth dyads in low-income, moderately violent urban neighborhood in Southeast Washington D.C., U.S. (97% Black)		N = 165 Youth: R = 6–10 Grades 1–2: n = 111 Grades 5–6: n = 54	51% male youths 79% mothers 9% fathers 11% other relatives		SCEV (Richters & Saltzman, 1990) for parents and 5 – 6th graders Things I have seen and heard (Richters & Martinez, 1990) for 1–2nd grade students to measure frequency in the past 5 years (or lifetime for younger youth)	Small group interviews for youth Questionnaires or one-on-one interviews for parents depending on reading levels	Chi-squared difference of prevalence rates	Older youth lifetime victimization: 69% youth reports vs. 44% parent report, χ² (1, N = 54) = 4.59, p < .05 Younger youth witnessed violence: χ² (1, N = 111) ranging from 17.40 (p < .001; witnessed arrests) – 58.48 (p < .001; witnessed serious accidents) with parental underestimation; Witnessed stabbing and drug deals n.s.	N/A	Acceptability: 71% Cross-sectional study checklist violation: Item 2 unclear (parents’ age not reported) Item 11 (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)
Skar et al. (2021)		Caregiver-youth dyads assessed in Norwegian child and adolescent mental health services between 2012 and 2017		N = 6094 Youth: M = 12.03 (SD = 3.14) R = 6–18	47% boys		Community violence items from trauma screening checklist using DSM-5 (Jensen et al., 2014)	Clinical screenings	Chi-squared difference McNemar test (1) “neither report” (2) “both report” (3) “caregiver only report” (4) “youth only report	Statistics of chi-squared difference test not specified except caregiver report: 35.2%, youth report: 42.6%, p < .001 (1) 46.9% “neither report” (2) 24.7% “both report” (3) 10.5% “caregiver only report” (4) 17.9% “youth only report”	“Both reporting” associated with higher post-traumatic stress symptoms and functional impairment compared to “caregiver only report” or “youth only report”	Acceptability: 57% Cross-sectional study checklist violations: Item 2 unclear (parents’ age and gender not reported) Item 5 covariates (youth age, gender not controlled) Item 11 unclear (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)
Strudler (2003)		Caregiver-youth dyads in residential treatment center in urban Massachusetts, U.S. (18.3% Black, 15% Hispanic)		N = 60 Youth: R = 6–17	51.7% male youths		Things I have seen and heard (Richters & Martinez, 1990), SCEV (Richters & Saltzman, 1990) to measure frequency	One-on-one interviews	Dichotomous scores (1) “Agreement” (2) “Parents’ overestimation” (3) “Parents’ underestimation”	(1) “Agreement” (55%) (2) “Parents’ overestimation” (13%) (3) “Parents’ underestimation” (25%) χ² (2, N = 56) = 17.82, p < 0.001	N/A	Acceptability: 71% Cross-sectional study checklist violation: Item 2 unclear (caregiver age and gender not reported) Item 11 unclear (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)
Zimmerman (2014)		Parent-youth dyads in urban neighborhoods from PHDCN study in age 9, 12, 15 cohorts with diverse population 35.8% Black, 45.8% Hispanic)		N = 2344 Youth: M = 12.0 at baseline (SD = 2.4) R = 7.8–16.9	50.4% male youths 83% biological mothers, 9% biological fathers		9 secondary ECV items in SCEV (Richters & Saltzman, 1990) to measure last year prevalence at T2 interviews	One-on-one interviews	Absolute and directional mean difference scores (youth reports – parent reports)	Absolute mean difference scores (youth reports – parent reports): M = 1.7 (SD = 1.6) [0, 9] Directional mean difference scores (youth reports – parent reports): M = 1.0 (SD = 2.1) [− 7, 9]	More discrepancies associated with demographic factors (e.g., males, non-whites, older youth, 3rd generation immigrants, without two biological parents, without married parents, lower SES) More discrepancies associated with lower family functioning and higher risks at T1 interviews.	Acceptability: 81% Longitudinal study checklist violation: Item 2 unclear (parents’ age not reported) Item 11 (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)
Zimmerman and Farrell (2013), Zimmerman and Pogarsky (2011)		Parent-youth dyads in urban neighborhoods from PHDCN study in age 12 and 15 cohorts with diverse population 33% Black, 51% Hispanic		N = 1517 Youth: M = 13.5 baseline	49.11% male youths 82% biological mothers, 10% biological fathers		10 items on modified SCEV (Richters & Saltzman, 1990) to measure lifetime prevalence of witnessed violence	One-on-one interviews	Overall difference scores (parent-youth), % in (1) “parent-youth agreement” (2) “parents’ overestimation” (combined high and low parents’ overestimation) (3) “parents’ underestimation” (combined high and low parents’ underestimation)	(Parent – male youth) difference scores: − 1.9 (Parent – female youth) difference scores: − 1.6 Overall: (1) 12% “parent-youth agreement” (2) 22% “parents’ overestimation” (3) 66% “parents’ underestimation” Male youth: (1) 11.5% “parent-youth agreement” (2) 19.5% “parents’ overestimation” (3) 68.5% “parents’ underestimation” Female youth: (1) 13.2% “parent-youth agreement” (2) 24% “parents’ overestimation” (3) 62.8% “parents’ underestimation”	Gender differences in parent–youth discrepancies were modest and did not reach significance “High parents’ underestimation” group associated with future (3 years) internalizing, externalizing general offending Weak partial mediation between parents’ underestimation and psychopathology by family support	Acceptability: 81% Longitudinal study checklist violation: Item 2 unclear (parents’ age and gender not reported) Item 11 unclear (recommended analysis for discrepancy reports include LCA/LPA, polynomial regression)