Child-to-parent Violence Questionnaire (CPV-Q): Validation among Spanish adolescents.

Background/Objective: Child-to-parent violence has generated great interest on the part of professionals and researchers in recent years. Consequently, there is a need to provide an assessment instrument that includes a wide range of child-to-parent violence behaviours and all of elements that define this type of violence. Thus, the purpose of the current study was to analyse the structure, reliability and validity of the CPV-Q, an instrument to assess child-to-parent violence behaviours, including the reasons for aggression against parents. Method: A total of 1,386 Spanish adolescents (55.2% females; M age = 14.7, SD = 1.7) were assessed using the CPV-Q. We obtained descriptive statistics of the items, the internal structure of the scale using both exploratory and confirmatory factor analysis, its reliability, and external evidence of its validity.
Results: Data indicated a matrix of 4 factors with 14 parallel items (for the father and for the mother). Reasons for child-to-parent violence were grouped into instrumental and reactive reasons.
Conclusion: The CPV-Q presents adequate psychometric properties; thus, it is proposed as a useful instrument to assess child-to-parent violence in both professional and research settings.

Violence during adolescence is a phenomenon that extends to all societies and cultures. Thus, research on this topic has investigated this behaviour within different settings, examining, for example, those variables related to aggression at school (Jiménez & Estévez, 2017), as well as the prevalence and assessment of dating violence (Hebert et al., 2017, Rodríguez-Díaz et al., 2017). Another type of adolescent violence that occurs within the family setting is child-to-parent violence (CPV), which has also recently become a matter of concern among professionals and researchers from different countries, who have investigated prevalence rates and factors related to this type of behaviour (e.g., Beckman et al., 2017, Calvete et al., 2013, Contreras and Cano-Lozano, 2014, Contreras and Cano-Lozano, 2015, Contreras and Cano-Lozano, 2016, Margolin and Baucom, 2014). In this respect, Loinaz, Andrés-Pueyo, and Pereira (2017) have recently taken an empirical approach toward identifying risk factors associated with CPV. In Spain, the General Prosecutor Office, in its last report (2018), highlighted that, after experiencing a notable increase over the last decade, the number of CPV cases remains very worrying in recent years. Nonetheless, these data refer only to those cases in which parents report their children's behaviours at Juvenile Court. Thus, many cases of CPV remain unknown.

Even though research on CPV has increased in the last decade, this topic is one of the most understudied types of family violence, and a review of the available literature yields inconsistent information. In this respect, different definitions and the use of varied assessment instruments to measure CPV have contributed to great variability in estimated prevalence (Simmons, McEwan, Purcell, & Ogloff, 2018). Different authors have been adding concrete elements to the definition of CPV since 1979, when Harbin and Madden established the term “Battered Parent Syndrome”, which exclusively included physical aggression and verbal/non-verbal threats of physical harm. The concept has been expanded over time to include other forms of psychological violence together with financial violence, while also incorporating the intention to hurt one's parents to gain control over and dominate them (Cottrell, 2001), the abuser's consciousness of the violent act and the repeated perpetration of the violence over time, excluding isolated acts of violence (Molla-Esparza and Aroca-Montolío, 2017, Pereira et al., 2017).

With respect to the types of CPV, Cottrell (2001) described psychological, physical, and financial abuse. Psychological abuse refers to those behaviours intended to manipulate, control and emotionally hurt parents (intimidating, running away from home, threatening, etc.). Verbal abuse is a type of psychological abuse and includes acts such as shouting, challenging, belittling, etc. Physical abuse refers to acts such as pushing, spitting, kicking, punching, etc. Finally, financial abuse includes behaviours such as stealing money or parents’ belongings, selling parents’ possessions, destroying the home or parents’ belongings, etc. However, the literature reveals some problems arising from the inconsistent operationalization of different types of CPV behaviours (see review by Simmons et al., 2018). Consequently, CPV has been measured on the basis of this inconsistency. In this regard, a relevant source of information about the extent of CPV is provided by studies with community samples, in which children and adolescents report CPV incidents. Nevertheless, variability in results depending on the instrument used to assess CPV makes it difficult to generalize findings across studies. These instruments are briefly described below.

One of the instruments repeatedly used is the Conflict Tactic Scale (CTS; Straus, 1979), adapted in different versions to assess physical, psychological and verbal aggressions against parents (e.g., Beckman et al., 2017, Gámez-Guadix and Calvete, 2012). Elsewhere, Margolin and Baucom (2014) conducted a study with a sample of adolescents from the United States using an ad hoc questionnaire to assess property damage, physical aggression and verbal aggression. In Spain, Ibabe, Jaureguizar, and Bentler (2013) created the Intra-family Violence Scale, which measures physical, psychological, and emotional violence towards parents with 3 parallels items (for the father and for the mother). Besides Calvete et al. (2013) designed the Child-to-parent Aggression Questionnaire, which consists of 10 items referring to the father and 10 items referring to the mother to assess psychological and physical aggression. It also includes an assessment of the reasons for the aggression.

Therefore, there is a need to develop and validate an instrument to assess a wide range of CPV behaviours, including all of the elements that define this concrete type of violence, with consideration that control and domain over parents is a crucial component defining CPV (Cottrell, 2001, Molla-Esparza and Aroca-Montolío, 2017). None of the previous instruments has incorporated all of these aspects. This fact led us to develop an exhaustive instrument to assess different acts of violence towards parents according to the main definitions of CPV (psychological, physical, financial and control/domain over parents), as well as the reasons for the aggressions against parents. This paper is the continuation of a previous study detailing the development of the Child-to-parent Violence Questionnaire (CPV-Q) (Contreras & Cano-Lozano, 2017). Therefore, the purpose of the current study was to analyse the structure, reliability, and validity of the CPV-Q in a sample of Spanish adolescents.

Method

Participants

The sample was made up of 1,386 adolescents (55.2% girls, 44.8% boys) aged between 12 and 18 years old (Mage = 14.7, SD = 1.7) from different high schools from Jaén (80.2%) and Oviedo (19.8%) (Spain). The majority of the parents were married (85%). Their socio-economic levels were as follows: 3.1% low, 75.6% medium, and 21.3% high.

Instruments

The Child-to-parent Violence Questionnaire (CPV-Q) consists of 28 parallel items (for the father and for the mother) measuring different acts of psychological, physical, and financial violence, together with behaviours demonstrating control and domain over parents. Adolescents are asked to indicate how often they have carried out each of the behaviours against their parents in the past year using a Likert scale: 0 (never), 1 (rarely = it has occurred once), 2 (sometimes = 2-3 times), 3 (many times = 4-5 times) and 4 (very often = 6 times or more). The instrument also includes 8 reasons for the aggressions against parents, also measured using a Likert scale: 0 (never), 1 (sometimes), 2 (almost always) and 3 (always).

The Attitudes and Social Cognitive Strategies Questionnaire (ASCS; Moraleda, González, & García-Gallo, 1998), Impulsiveness subscale, which consists of 7 items, with responses captured along a Likert scale ranging from 1 (strongly disagree) to 7 (strongly agree). Cronbach's α for this scale was .71.

The Social Information Processing (SIP) in Child-to-parent Conflicts Questionnaire (Calvete, Gámez-Guadix, & García-Salvador, 2014). This questionnaire assesses some SIP components in three scenarios. Adolescents respond to the items using a Likert scale ranging from 0 (not at all) to 4 (to a great extent). In the current study, the following SIP components were included: Hostile attribution (2 items per scenario) (α = .70); Anger (1 item per scenario) (α = .75); Aggressive response access (2 items per scenario) (α = .70); Anticipation of positive consequences for oneself from the aggressive action (1 item per scenario) (α = .74).

Procedure

First, authorization by the Ethic Committee of the University of Jaén to conduct this study was obtained. Then, authorizations by the Competent Public Administration in Education and the high schools’ directors were also obtained. Parents’ consent for us to assess their children and adolescent's consent were also requested. Parents and adolescents were informed about the aim of this study and the confidentiality of the data. Each participant received an identification code and completed the questionnaires in a group setting in their classrooms. Three evaluators from the research group, who were specifically trained for this protocol, conducted the evaluations. No incentive was offered in exchange for participation.

Data analysis

This is an instrumental study (Montero & León, 2007). Following the recommendation of different authors (Izquierdo et al., 2014, Lloret-Segura et al., 2014), participants were randomly divided into two subsamples in order to conduct exploratory factor analysis (EFA) with the first subsample (n = 693) and confirmatory factor analysis (CFA) with the second subsample (n = 693). As Ferrando and Anguiano-Carrasco (2010) recommended, EFA and CFA are considered to constitute two poles of a continuum, as EFA imposes minimal restrictions to obtain a factorial solution, which can be transformed by applying different criteria, and CFA imposes much stronger restrictions to test the final factor solution. With the aim of verifying the suitability of the data to perform factor analysis, the Kaiser-Meyer-Olkin (KMO) assessment of sampling adequacy and Barlett's test were performed. Given that in a previous development of the instrument, a total of 4 factors were found (Contreras & Cano-Lozano, 2017), a targeted rotation was used. Conducting Horn's Parallel Analysis with 5,000 resamples, a 4-factor structure was confirmed both for the father and mother scales. The extraction of the generalized least squares was used because it allowed us to obtain valid solutions that avoided factor loadings higher than 1. The solutions obtained were iteratively optimized, deleting those items with a different assignation to factors in the father and mother scales and those items with a cross-factorial loading or with an insufficient loading on any factor. The final solution was tested in the confirmatory sample using Structural Equation Modelling (SEM) to examine model fit with Chi squared, Comparative Fit Index (CFI), Tucker-Lewis Index (TLI), and Root Mean Square Error of Approximation (RMSEA). The reliability was analysed as internal consistency using Cronbach's α for the dimensions of the scale, the covariance matrix for the standard alpha and the polychoric correlation matrix to obtain the ordinal alpha (Gadermann, Guhn, Zumbo & Columbia, 2012). In addition, in accordance with Carretero-Dios and Pérez (2007), the search for external evidence of validity (convergent validity) was based on the correlations between the scores for each dimension of the CPV-Q with the scores for the dimensions of the SIP and with the total score for the Impulsiveness subscale, as previous studies have found that CPV is related to some SIP components (Calvete et al., 2014, Contreras and Cano-Lozano, 2016) and to impulsiveness (Rico, Rosado, & Cantón-Cortés, 2017). Regarding the reasons for CPV, the function of aggression has historically been characterized in terms of instrumental aggression and reactive aggression, which are related to some components of the SIP (Crick & Dodge, 1996). Analyses were conducted with R software, using Psych Package for EFA and the Lavaan Package for CFA.

Results

Exploratory factor analysis

Bartlett's test indicated that the correlations between items was not zero, both for the father scale, χ2 (378) = 5987.70, p < .001, and the mother scale, χ2 (378) = 6376.30, p < .001. KMO was equal to .88 for both scales, indicating that the correlation matrices were suitable for factorial analysis. The initial 4-factor solution explained 36% of the variance in the father scale and 38% in the mother scale. A total of 14 items were iteratively deleted to achieve a simple common solution (loadings greater than 0.3 on a single factor) for both scales. The most frequent reason for item deletion was the presence of cross-factorial loadings, but only in the mother scale. The final solution, with 14 items, explained 41% of the variance in both scales (see Table 1).

Table 1

Descriptive statistics of the 28 items of the Child-to-parent Violence Questionnaire (CPV-Q).

	Father				Mother
Item	M	SD	Skew	kurtosis	M	SD	Skew	kurtosis
1. I have shouted at my parents. *	1.33	1.13	0.63	-0.35	1.65	1.19	0.30	-0.85
2. I have run away from home (one or more days) or spent the night out from home without warning my parents. *	0.09	0.43	5.56	33.52	0.10	0.43	5.13	29.46
3. I have intentionally broken things in my house or my parents’ belongings. *	0.16	0.48	3.62	15.10	0.16	0.49	3.53	14.23
4. I have told my parents, “I hate you!” “I wish you were dead”.	0.30	0.69	2.81	8.68	0.32	0.72	2.60	6.98
5. I have insulted my parents.	0.50	0.82	1.81	3.06	0.56	0.86	1.65	2.40
6. I have told my parents that if I want something, they have the duty to provide it for me. *	0.18	0.51	3.09	9.92	0.23	0.57	2.67	7.08
7. I have deliberately not told my parents where I was or what I was doing. *	1.09	1.25	0.87	-0.39	1.17	1.28	0.79	-0.57
8. I have made negative, offensive and/or degrading comments to my parents.	0.48	0.80	1.81	3.23	0.51	0.83	1.76	2.98
9. I have threatened my parents (with hurting them, with hurting myself, with running away from home).	0.15	0.53	4.25	20.04	0.21	0.63	3.62	13.99
10. At home, we watch what I want on TV.	0.44	0.87	2.34	5.43	0.50	0.91	2.08	4.06
11. I have demanded my parents to buy me things even knowing they cannot afford it.	0.19	0.55	3.67	15.72	0.24	0.62	3.13	10.79
12. I have acquired debts that my parents have had to pay.	0.06	0.29	5.88	37.99	0.05	0.30	6.35	44.78
13. I have thrown things at my parents.	0.06	0.35	6.78	53.11	0.08	0.35	5.87	42.77
14. When I argue with my parents, I have the last word.	0.68	1.04	1.54	1.60	0.83	1.10	1.23	0.62
15. I have hit my parents with something that could hurt them.	0.03	0.23	12.00	169.14	0.02	0.19	15.27	289.67
16. I have spit on my parents. *	0.01	0.15	15.50	265.31	0.01	0.17	18.77	404.90
17. I have taken my parents’ belongings without their permission. *	0.75	0.92	1.19	0.93	0.88	1.02	0.98	0.10
18. I have sold my parents’ belongings without their permission. *	0.01	0.11	13.82	211.26	0.02	0.20	10.46	117.70
19. I have kicked, slapped, and/or punched my parents.	0.04	0.26	8.73	85.33	0.03	0.24	9.89	111.53
20. I have stolen money from my parents.	0.36	0.74	2.39	6.02	0.46	0.86	2.10	4.29
21. I have pushed my parents. *	0.11	0.43	5.22	32.70	0.14	0.45	4.15	22.09
22. I have told my parents that at home they have to do what I want.	0.07	0.35	5.87	37.60	0.10	0.38	4.52	22.65
23. I laughed or scoffed in my parents’ face to make them feel bad. *	0.16	0.53	4.20	20.80	0.22	0.62	3.31	11.96
24. I have rejected my parents’ affection with the intention of punishing them. *	0.44	0.84	2.10	4.13	0.49	0.87	1.97	3.58
25. I have done or said things to my parents to make them feel afraid. *	0.06	0.33	6.54	52.52	0.08	0.38	6.43	51.36
26. I have demanded my parents to stop what they are doing to pay attention to me.	0.50	0.81	1.82	3.38	0.62	0.88	1.44	1.61
27. I have lied to my parents. *	1.65	1.14	0.34	-0.60	1.76	1.15	0.25	-0.68
28. I have hurt my parents during an argument. *	0.15	0.51	4.14	19.50	0.19	0.57	3.80	16.57

Note. * Items deleted after the Exploratory Analysis.

Regarding the effect of deleting items on reliability, for the standard alpha, deleting item 13 improved the reliability of physical violence in the father scale (from α = .74 to α = .75) and the mother scale (from α = .70 to α = .74), and deleting item 9 improved the reliability of psychological violence in the mother scale (from α = .76 to α = .77). Using the ordinal alpha, only deleting item 9 slightly improved the reliability of psychological violence in the mother scale (from α = .84 to α = .85). Considering the small increases in reliability with those deletions (within the confidence interval of alpha for each scale), it was decided that all items should be included in the confirmatory analysis.

Confirmatory factor analysis

When performing the CFA for the father scale, a non-valid solution was found due to a correlation greater than 1 between the financial and physical scales. Accordingly, a restriction for this relation was established (the correlation had to be less than .995, the maximum possible value after diverse tests). The CFA showed that model fit was not absolute, χ2 (71) = 158.43, p < .001, with a normalized χ2 = 2.23. The fit indexes obtained were adequate, CFI = .97, TLI = .96, RMSEA = .04, 95% IC [.03, .05]. However, the first three modification indexes presented cross-factorial loadings of item 14 on the psychological dimension, MI = 57.3, the financial dimension, MI = 40.79, and the physical dimension, MI = 35. Table 2 shows the factor loadings for the father scale in the confirmatory sample, ranging from .45 to .98.

Table 2

Factor loading and internal consistency of the Child-to-parent Violence Questionnaire (CPV-Q) dimensions (14 items) in the confirmatory analysis.

	Father				Mother
Item	I	II	III	IV	I	II	III	IV
4.	.83				.83
5.	.85				.88
8.	.84				.84
9.	.75				.72
13.		.80				.81
15.		.92				1
19.		.98				.89
11.			.62				.58
12.			.60				.59
20.			.45				.50
10.				.53				.58
14.				.66				.63
22.				.89				.85
26.				.73				.73
α	.80	.71	.31	.57	.82	.79	.32	.58
α ordinal	.89	.92	.55	.79	.88	.92	.57	.79

Note. I: Psychological; II: Physical; III: Financial; IV: Control/domain.

The CFA for the mother scale showed that model fit was not absolute, χ2 (71) = 177.53, p < .001, with a normalized χ2 = 2.5. The fit indexes obtained were adequate, CFI = .97, TLI = .96, RMSEA = .05, 95% IC [.04, .06]. The main modification index corresponded to the cross-factorial loading of item 14 on the psychological dimension, MI = 39.48, as well as the cross-factorial loading of item 11 on the control/domain dimension, MI = 33.95, and psychological dimension, MI = 24.23. Table 2 shows the factor loadings for the mother scale in the confirmatory sample, ranging from .50 to 1. As shown in Table 2, the internal consistency of the dimensions was adequate for the psychological and physical dimensions but lower for the financial and control/domain dimensions. Correlations among the dimensions are presented in Table 3.

Table 3

Correlations among the dimensions of the Child-to-parent Violence Questionnaire (CPV-Q).

	Father				Mother
Subscales	I	II	III	IV	I	II	III	IV
I. Psychological	1	-	-	-	1	-	-	-
II. Physical	.81	1	-	-	.78	1	-	-
III. Financial	.84	.99	1	-	.79	.83	1	-
IV. Control/domain	.66	.87	.87	1	.61	.71	.82	1

Note. All correlations were significant at p < .001 level.

Exploratory and confirmatory analyses of reasons for CPV

Bartlett's test indicated that the correlations between items was not zero, χ2 (28) = 1303.98, p < .001. Furthermore, KMO was equal to .78, indicating that the correlation matrix was suitable for factorial analysis. Items 5 and 8 presented cross-factorial loadings. The final solution with 8 items and two factors explained 64% of the variance. Factors 1 and 2 were named Instrumental and Reactive Reasons, respectively. The CFA showed that model fit was not absolute, χ2 (18) = 75.20, p < .001, with a normalized χ2 = 4.18. The fit indexes obtained were adequate, CFI = .98, TLI = .96, RMSEA = .07, p < .05, 95% IC [.05, .08]. Cronbach's α for Factor 1 was .74 and .63 for Factor 2. The correlation between Factors 1 and 2 was .48 (see Table 4).

Table 4

Descriptive statistics and factorial loadings of reasons for child-to-parent violence in the confirmatory analysis.

	CFA		Descriptive statistics
Item	IR	RR	M	SD
1.	.65		0.78	0.82
2.	.86		0.42	0.64
3.	.79		0.48	0.65
4.	.61		0.70	0.82
5.	.45	.21	0.27	0.56
6.		.64	0.83	0.96
7.		.87	0.26	0.56
8.		.89	0.38	0.67

Note. IR: Instrumental Reasons; RR: Reactive Reasons.

Evidence of validity

The correlations between the dimensions of the CPV-Q and the dimensions of the SIP and the total score of the Impulsiveness scale were all statistically significant (p < .01) (see Table 5).

Table 5

Bivariate correlations between the dimensions of the Child-to-parent Violence Questionnaire (CPV-Q) and the dimensions of the Social Information Processing and the Impulsiveness Scale.

	Father				Mother				IR	RR
	I	II	III	IV	I	II	III	IV
HA	.34	.17	.25	.30	.39	.20	.26	.35	.41	.36
Anger	.30	.13	.26	.35	.36	.15	.27	.41	.45	.38
ARA	.45	.34	.28	.36	.53	.41	.30	.41	.37	.44
APC	.10	.17	.12	.11	.12	.19	.11	.10	.13	.06
Imp.	.26	.10	.21	.21	.30	.09	.22	.27	.27	.34

Note. All correlations were significant at p < .001 level; I: Psychological; II: Physical; III: Financial; IV: Control/domain; HA: Hostile Attribution; ARA: Aggressive Response Access; APC: Anticipation of Positive Consequences; Imp: Impulsiveness; IR: Instrumental Reasons; RR: Reactive Reasons.

Discussion

The aim of the current study was to analyse the structure, reliability and validity of the CPV-Q in a sample of Spanish adolescents. The EFA results showed a common solution for the father and mother scales, with 14 parallel items in the definitive instrument (see Appendix). Specifically, CFA indicated that the CPV-Q presents a structure consisting of four factors for the father and the mother scales capturing psychological violence, physical violence, financial violence, and control/domain over parents. The psychometric properties obtained were adequate, despite the low reliability of the financial scale. The very high skewness of the items of this scale could explain (Greer, Dunlap, Hunter & Berman, 2006) the difference between alphas based on the covariance matrix, which had very low values (father: α=.31, mother: α=.32), and the ordinal alphas based on polychoric correlations (father: α=.55, mother: α=.57). The ordinal alpha corresponds to a better estimation of the reliability of the scale, as it compensates for the restriction on correlations due to the low prevalence of the behaviours. Therefore, the structure obtained for the CPV-Q is congruent with current definitions of CPV. Moreover, the novelty of this instrument is that, unlike previous questionnaires, control and domain over parents was also assessed (Cottrell, 2001, Molla-Esparza and Aroca-Montolío, 2017). Another contribution of this study is the provision of external validity. The results indicated that the CPV behaviours were related to impulsivity (Rico et al., 2017) and to some SIP components, such as a greater hostile attribution (Calvete et al., 2014, Contreras and Cano-Lozano, 2016), anger, access to aggressive response, and anticipation of positive consequences of the aggression (Calvete et al., 2014).

The CPV-Q also includes 8 reasons for CPV, which were grouped into two factors corresponding to reactive and instrumental reasons, in line with the study by Calvete et al. (2013). On one hand, reactive reasons refer to the use of violence in response to a previous aggression or threat of aggression. This result is consistent with previous studies indicating that CPV is linked to other forms of family violence, such as parent-to-child violence (Beckman et al., 2017, Contreras and Cano-Lozano, 2016, Gámez-Guadix and Calvete, 2012, Margolin and Baucom, 2014). On the other hand, instrumental reasons imply the use of violence to get what the adolescent wants. Previous literature notes that parents who are victims of CPV are perceived by their children as more permissive when compared to parents of non-abusive children (Contreras and Cano-Lozano, 2014, Contreras and Cano-Lozano, 2015). Thus, abusive adolescents may see their parents as weaker than themselves and, consequently, more easily manipulated and intimidated if they use violence towards them.

Some limitations should be noted. First, all of the measures were based on adolescents’ self-reports, so it would be necessary to add reports from parents. Second, future studies should provide additional data, such as the test-retest reliability of the scale. Despite these limitations, the current study has yielded relevant results concerning the assessment of CPV. In this respect, the way in which CPV is defined and evaluated is decisive for both research and professional purposes. On one hand, the results indicate that the CPV-Q is a valid instrument for briefly and easily assessing a wider variety of CPV behaviours according to the current conceptualization of CPV. This aspect is crucial in order to lend consistency to studies of this type of violence and, particularly, to research on the prevalence of these behaviours. On the other hand, regarding the professional context and, with prevention and intervention purposes in mind, it is fundamental to consider several elements, such as specific CPV behaviours and the reasons for the aggressions, which can be measured using the CPV-Q. To know these details, together with identification of the risk and protective factors specific to each particular case (see Loinaz et al., 2017), is crucial for proposing objectives of interventions, and it can provide information for the design of treatment programs and follow-up plans. Finally, this research was conducted with a community sample, so in future studies it would be interesting to validate the CPV-Q in other fields of study, for example, in clinical and forensic samples.

Funding

This work is part of the Project “Child-to-parent violence: An emergent social problem. Study of prevalence and psychosocial variables involved”, funded by the Ministry of Economy and Competitiveness of Spain and the European Union through the European Regional Development Fund. Reference PSI2015-65950-R (MINECO/FEDER).