Results of Ukrainian Adaptation of the Pediatric Symptom Checklist with Abused Children

questionnaires. Methods of classical test theory, confirmatory factor analysis and ROC-analysis were used for tests validation. Adapted versions appeared reliable and valid but had unusually low cut-offs and two-factor structure (internalizing and externalizing symptoms without separate attention deficit factor). Weak economy, military actions and upbringing peculiarities can be the reasons of unusual cut-off and factor structure of adapted tests. This test will be useful for practitioners and researchers in mental health and social work areas. PSC-Uke Y-PSC-Ukr can be recommended for use with problematic populations like children from families with domestic abuse as these methods reveal forms and degrees of children’s psychosocial dysfunction through negative growth conditions.

Introduction.Psychosocial functioning reflects a person's ability to perform the activities of daily living and to engage in relationships with other people in ways that are gratifying to him and others, and that meets the demands of the community in which the individual lives (Mehta, Mittal, Swami, 2014).Pediatric Symptom Checklist in two versions: PSC (parent version), Youth Pediatric Symptom Checklist (Y-PSC, youth version) of M. S. Jellinek and J. M. Murphy is a very popular effective questionnaire for children's psychosocial functioning assessment in the world, which can provide both a general assessment of psychosocial functioning and assessment of certain components, such as internalization, externalization, attention deficit, school problems (Jellinek et al, 1988) This questionnaire is adapted in more than 25 languages (Pediatric Symptom Checklist, 2021).The availability of such a tool can make the work of practical psychologists, social workers, educators, pediatricians more effective and useful for children, families, and society.Therefore, we asked and obtained permission from the authors to adapt these tests in Ukraine.
So, the aim of the study was to adapt both PSC questionnaire versions in Ukraine.The additional purpose was to investigate how these questionnaires would work with children from families with domestic abuse that may reveal their criterion validity and check the opportunity to use for problematic groups of families.It is already known that children who had bad family experience (e.g.experienced sexual harassment, had substantial exposure to violence, lived with drug abusing fathers, had low-income, urban mothers suffered from intimate partner violence, had parental wartime military deployment) were significantly more likely to score higher by PSC (Y-PSC) than those who had not experienced such negative impact (Aranda, Middleton, Flake & Davis, 2011).
Method.It was taken full version (35 items) of PSC and Y-PSC for adaptation.Methods of Classical Test Theory were used for adaptation these tools in Ukraine (Kline, 2016) Translation was done by professional psychologist with certified English B2 level.During the translation, the necessary cultural adaptation of the test was carried out.An examination of the translation was conducted by four experts who were highly qualified English language professors.The experts approved the translation and provided comments on the improvements that were implemented.Backtranslation from Ukrainian to English was done by professional interpreter and Ph.D. in Psychology who has not seen the original version of the tests.Subsequently, a double-test procedure was conducted for Ukrainian schoolchildren who speak English, aged 11-14, 63 children (27 girls among them).The children performed both English versions of the test: the original and back-translation ones.
Approbation.Pilot research was conducted with 30 children of 10-11 years old in Kharkiv region.
Validation.It was taken three psychological methods for analyzing of construct convergent validity.
1. Kinetic Family Drawings (K-F-D) of Burns, R. C., & Kaufman, S. H. (1970).Adapted version included such sсales: a favorable family situation, anxiety, conflict in the family, a sense of inferiority in a family situation, hostility in a family situation.Adaptation of K-F-D was made by R. F. Beliauskaite (1987).
3. Clinical interview with 7 situations of closed questions and 23 opened questions designed by psychiatrists.In the first part (7 closed questions) the child was asked to transfer stressful psychosocial situations of other children to their own situation.The second part of the interview was presented with open questions which related to emotional abuse against the child, lack of care, the child's experiences about the situation in the family, behavioral, attention and emotional problems; neurotic, vegetative (related to increased reactions of autonomic nervous system) and somatic symptoms (23 questions).The interview ended with questions about the dreams and expectations of the child.During the interview, nonverbal signs of psychological and emotional (psychophysiological) stress were noted.Qualitative and quantitative analysis was conducted with the interview.
Participants.It was examined 321 children in total: ordinary children (197, including 121 girls); children from families with domestic abuse (110, including 45 girls); children on inpatient treatment in the department of psychiatry (14, including 6 girls).
The age range of children was 6-18 years.As we adapted both versions of the test -the self-assessment version for children from 11 years and the parent version, the total number of study participants was 532: 281 parents fulfilled PSC-Ukr and 251 children fulfilled Y-PSC-Ukr.

Results and Discussion
«Back-translation».The correlation analysis of results of schoolchildren fulfilled original and "back translated" test versions showed a very high significant correlation between them: ρ = .94at p < .0001.There were not found significant differences between these test versions as well: Wilcoxon Matched Pairs Test T = 660.5 at p = .98.So translated test does not distort content and meaning of original test.

Item analysis and discriminative capacity.
All tests items correlated by Spearman analysis with the total test index at significant level: for PSC-Ukr ρ = from .36 to .67 at p < .0001,for Y-PSC-Ukr from ρ = .25 to .61 at p < .0001 to p = .017.There were no items with only one or two types of response in our sample, that is, all types of item responses (Never, Sometimes, Often) were met with different frequencies.The discriminatory check was carried out according to the formula of Ferguson's δ (Kline, 2016).
For PSC-Ukr and Y-PSC-Ukr δ = 1.02 that means high discriminatory (informative) capacity of them.
It was checked a correlation between PSC-Ukr and Y-PSC-Ukr in the pairs of children and their parents: ρ = .63,р < .0001.This procedure is similar to the parallel form's reliability.Not very high but sufficient correlation can be explained by observation that children not enough aware about or admit their disruptive behavior, and parents not enough aware about children's internalizing symptoms.
The test-retest reliability of Y-PSC-Ukr was performed with 49 children (aged 10-11, 24 girls) after three weeks interval.The correlation coefficient of the test / retest was: ρ = .91,p < .0001.

Convergent validity. It was found moderate
significant correlations between almost all scales of other methods and indexes of the PSC-Ukr and Y-PSC-Ukr: ρ was from .59 to .13,p was from < .0001 to = .046.Only one scale had no correlations with both PSC-Ukr and Y-PSC-Ukr indexes -Anxiety Scale from K-F-D.We suppose that in our culture many children have anxiety to draw anything for psychologist, they percept this activity as their drawing performance assessment, not psychological testing.In Ukraine we have this lesson (drawing) at different kinds of schools till 8 th grade.So, children's personal anxiety may mix with anxiety to draw not enough beautiful picture.

Concurrent validity (criterion-related validity).
It is used criterion groups for validation by criterion -that is, groups, which exactly have the manifestation of the criterion -for example, any traumatic factor, that impair psychosocial functioning.We used the impact of domestic abuse in our study.Therefore, to validate test by the criterion, we included in the study children who are registered at the social services that help children form families with domestic abuse.
The differences between the criterion and control groups in both versions of the test were significant (Kolmogorov-Smirnov test): the mean scores of the results were almost twice as high in the group of abused children (for PSC-Ukr M = 30.18and for Y -PSC-Ukr M = 27.00)than in the group of ordinary children (for PSC-Ukr M = 15.67 and for Y-PSC-Ukr M = 16.67),p < .001.
Another way of the criterion validation is conducting an experts' survey -questioning people who are aware about studied subjects' psychological states and life situation.We interviewed the teachers of the studied children as experts.We designed special form with Likert scale for assessment of teachers' opinion about studied children' psychological problems.The significant coefficients of Spearman rank-order correlation were obtained for the teacher's assessment of the children psychological problems level and the results of the PSC-Ukr method (ρ = .63,p < .0001)and Y-PSC-Ukr (ρ = .64,p < .0001),which proved the criterion validity of the adapted techniques.
The analysis shows that the adapted PSC-Ukr and Y-PSC-Ukr are valid tools for measuring the level of children psychosocial dysfunction.
Factor Analysis.Firstly, we performed an EFA to identify possible factor models of the questionnaire construct, and then a CFA to test the quality of obtained models.
EFA was performed by the Principal Components method with Direct Oblimin rotation, which is recommended for the selection of correlating factors (Kline, 2016) with a hierarchical Schmid-Leiman's solution.Since these tests were measured in the rank scale with three response gradations (0; 1; 2), therefore, in factorization we did not use the Pearson correlation coefficient, but the polychoric one.The adequacy of the sample for factor analysis was very good by Kaiser-Meyer-Olkin statistics (KMO = .89for PSC-Ukr and KMO = .84for Y-PSC-Ukr).
According to the most variants of PSC adaptations in different countries (Pediatric Symptom Checklist, 2021) three factor models could be considered for this questionnaire using EFA: with one factor, two factors and three factors.Two-factor and three-factor solutions were well interpreted in our sample.There were internalizing and externalizing factors in the twofactor solution known from the works of T.M. Aachenbach et al (2016).Except these two factors third factor in the three-factor structure can be interpreted as "problems in school" or "school maladaptation".
There are several ways to check a decision about how many factors to leave -the Cattell criteria, the parallel analysis using the Monte Carlo method and the Velicer's minimum average partial (MAP) test (Kline, 2016).By the criterion of Cattell's scree plot we could leave 3, 2, or 1 factor for our data.Parallel analysis showed that 3-factor structure is recommended for the PSC-Ukr, and for the Y-PSC-Ukr can be appropriate a 2-factor or 3-factor structure (the latter with less probability).It was recommended to leave 2 factors for each version of the test by MAP criterion.In the threefactor model the third factor of "school problems" included somewhat different points in the parent and adolescent versions, reflecting different meanings and different perceptions of school problems of parents and children.We decided to choose a more stable twofactor hierarchical model with two factors of first order (internalizing and externalizing behaviour) and one factor of the second order, which reflected general psychosocial dysfunction of the child.This made the parent and adolescent test versions equivalent in administrating.
The first order factors have explained 42.7% of the measured variable variance (psychosocial dysfunction of the child) after the EFA with PSC-Ukr.
The first factor that explained 33.1% of the variable variance included 15 items of the questionnaire.This factor included signs of anxiety and depression, which according to Achenbach, is one of the two main forms of psychosocial disturbances of children -internalizing behavior.
The second factor that explained 9.6% of the variable variance included 20 items of the method.These items relate to aggressive, indifferent behaviors, reduced attention, predisposition to risk, traumatism, conflict, regression, which, according to T. Achenbach, is the second of the two main forms of child's pathological psychosocial development -the externalizing behavior.
The general factor of the second order included all items of the questionnaire.There were significant correlations between all factors: the factors of the first order (F1, F2) correlated ρ = .42p < .05; the factors of the first order correlated with the general factor (G1): F1 with G1 ρ = .50,F2 with G1 ρ = .84. p < .05.
The EFA of the Y-PSC-Ukr allowed to extract two similar factors with the same items as in the parent version, which explained 33.7% of the variance of the measured variable.The first externalizing factor explains 26.8% of the variance, and the second internalizing factor explains 6.9% of the variance.Factors of the first order (F1, F2) correlated ρ = .41p < .05; the factors of the first order correlated with the general factor (G1): F1 with G1 ρ = .49,F2 with G1 ρ = .84. p < .05.
CFA was performed for ordinal / categorical data.One-factor and two-factor models were compared using CFA and indicators for the two-factor model were better.See table 1.According to the results of the CFA, the two-factor hierarchical model of the test (with two first-order factors and a general factor of the second order) can be considered as relevant to the empirical data, that is, acceptable and optimal, unlike the one-factor model, which does not meet the criteria of a good model.Authors present their 35-items PSC as one-scale questionnaire and extracted three factors in short 17items version -Externalizing, Internalizing and Attention (Pediatric Symptom Checklist, 2021).We can explain the presence of only two factors in the Ukrainian questionnaire version by the peculiarities of upbringing culture and childish behavior perception in Ukraine.Ukraine is one of the countries with a low economic level (Worldometers, 2021) and child's obedience in Ukraine is highly valuable like in other low-income countries (Park & Lau, 2016).So, distractions are perceived here mostly as behavioral problems, not as cognitive problems.The child is criticized and punished for not doing homework, for mistakes, late performance, and distraction in the same way as if the child does it on purpose, that is, as behavioral problems.Therefore, these factors, which in American and some other cultures, are manifested separately -externalizing behavior and attention deficit (Pediatric Symptom Checklist, 2021), in our culture are manifested as one factor.This can be the root of not enough good psychosocial functioning of Ukrainian children because as shown in some studies, the high value of parenting obedience as opposed to the values of independence, freedom, trust, and respect leads to low feelings of happiness in people, children's resistance to learning, and low economic situation in general (Park & Lau, 2016;Conzo, Aassve, Fuochi & Mencarini, 2017;Karakul, 2016).

Standardization of Y-PSC-Ukr and PSC-Ukr.
It was calculated norms for both adapted versions.The level of psychosocial dysfunction measured by parent version in girls and boys had trend to be different: M girls = 19.67,M boys = 23.24(Kolmogorov-Smirnov such children is unusually higher.This can be explained by the problems that produce stress in Ukrainian families: low economic level of the country, difficult transition from an authoritarian to a democratic society, military actions that going on in the country from 2014.Studies have shown (Murphy & Jellinek,1988) that the level of psychosocial functioning decreases significantly in economically poor populations.Conclusion.The validity and reliability of PSC and Y-PSC were found excellent in Ukrainian children.But the factor structure and cut-offs for these questionnaires were rather different which connected with cultural specifics.These questionnaires can be successfully used with problematic populations like children from families with domestic abuse as they clearly reveal forms and degrees of children's psychosocial dysfunction through negative growth conditions.As proved by Achenbach (2016), internal (internalizing) symptoms turn into avoidance problems, somatic complaints (including sleep disturbances, eating disorders, pain) and anxiety and depression, and external (externalizing) symptoms appear in delinquent and aggressive behavior, attention deficit hyperactivity disorder.Such changes in the behavior and mental state of children are not accidental.From the evolutionary psychology point of view, externalizing and internalizing behavior are two strategies of human adaptation to the expected negative conditions of life.Namely, if a child in the first years of life encounters an unfriendly, dangerous or problematic environment, it is updated forecast that it will continue to be so, so to survive in such conditions you should be depressed, careful and inactive (to survive at the bottom of the hierarchy, not to irritate stronger individuals, save energy, wait), or you should be aggressive, antisocial and impulsive (to get at least some resources in conditions of competition and shortage of the most necessary) (Bjorklund, Sellers, Roberts, 2011).Funding.Financial aid for empirical studies conducting was given by Deutsche Gesellschaft für Internationale Zusammenarbeit (GIZ) GmbH.Пункти, які дитина залишила незаповненими не підраховуються (їм присвоюється значення 0 балів).Якщо таких пунктів 4 та більше -опитувальник вважається недійсним.
Approval /Patient consent.Plan of this study was approved by Commission on Ethics of Psychological Research, School of Psychology, V.N.Karazin National University.Parents of all participants-children and participants-parents gave informed consent about their voluntary participation in the study.Declaration of Conflicting Interests.The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
. Most part of statistical analysis was done using program STATISTICA 7.0, Stat Soft, Inc.The Receiver operating characteristic (ROC) analysis was done by SPSS 16.0, SPSS Inc. Exploratory Factor Analysis (EFA) was performed with the FACTOR 10.3.01 program of U. Lorenzo-Seva & P. J. Ferrando, and Confirmatory Factor Analysis (CFA) was conducted by Mplus program (Muthén & Muthén).

Table 1 .
The goodness-of-fit indexes of the factor models

Table 2 .
Selection of cut-offs for the tests -most appropriate values are marked in bold