FACULTEIT DER MAATSCHAPPIJ- EN GEDRAGSWETENSCHAPPEN
Graduate School of Child Development and Education
MASTER
ORTHOPEDAGOGIEK
2017-2018
MASTERTHESIS
Name: Anniek Wolfs
Address and city: Rietmeent 258, Almere
Phone and email address: 0634486492, Anniekwolfs@hotmail.com Student number: 10739254
Topic: The Happy Victimizer Phenomenon: Age-Related Development and the Influence of
Parenting Practices
Supervisor: dr. Milica Nikolić
Second reviewer: dr. Cristina Colonnesi Time schedule: January 2018-November 2018
Allocation group: Ontwikkeling- en Opvoedingsproblemen
The Happy Victimizer Phenomenon: Age-Related Development and the Influence of Parenting Practices
Masterthesis Orthopedagogiek Pedagogische en Onderwijskunde Wetenschappen University of Amsterdam Student: A. Wolfs Supervisor: dr. M. Nikolić Second reviewer: dr. C. Colonnesi Amsterdam, November 2018
Contents Abstract 4 Introduction 5 Methods 11 Participants 11 Materials 11
Interview measuring children’s moral knowledge, moral emotion 11 attributions and moral reasoning
Coding system for the children’s moral knowledge 12 Coding system for the children’s moral emotion attributions 12 Adjusted Alabama Parenting Questionnaire for children and parents 12 measuring parenting practices
Procedure 14
Data analysis 15
Results 16
Preliminary analyses 16
Age-differences in the attributed emotions to the victimizer and the self 19 Age-related differences and the influence of parenting practices on children’s 19 moral emotion attributions
Post-hoc analyses of type of punishment and age-group influencing 25 children’s moral emotion attributions
Discussion 26
References 34
Appendix 1: Protocol stories interviews 37
Appendix 2: Questionnaire for children 39
Appendix 3: Questionnaire for parents 41
Abstract
The Happy Victimizer Phenomenon reflects the empirical finding that young children more often attribute positive moral emotions to the victimizer than older children. In the present study, differences in children’s moral emotion attributions and the influence of parenting practices on these emotion attributions was investigated in a sample of 64 normally developing Dutch children aged 4 to 9 (𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 = 6.28, 𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎𝑎𝑎 = 1.52). Children were divided into a younger (𝑛𝑛 = 28, 𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 = 4.82, 𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎𝑎𝑎 = .67) and older age-group (𝑛𝑛 = 36, 𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 = 7.42, 𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎𝑎𝑎 = .87) and were interviewed on their emotion attributions to the victimizer and the self. Additionally, questionnaires on parenting practices were filled in by parents and children. Results indicated that older children more often attributed mixed and negative emotions and less often attributed no emotions to the self than younger children. Furthermore, in young children high amounts of positive parenting were related to more reports of negative emotions, while high amounts of inductive discipline and punishments were related to fewer reports of negative emotions. Last, for both younger and older children, high amounts of inconsistent discipline were related to fewer reports of negative emotions. The results indicate the importance of parenting practices in promoting young children’s negative emotion attributions, which in turn promotes children’s prosocial behaviour.
Keywords: Happy Victimizer Phenomenon, moral emotion attributions, parenting
The Happy Victimizer Phenomenon: Age-Related Development and the Influence of Parenting Practices
Emotions play an important role in children’s behaviour. It is argued that the emotions children expect to experience when pursuing certain behaviours influence their actual behaviour (Arsenio & Kramer, 1992). For example, when a child expects to experience happiness following a certain behaviour, it is likely that the child will engage in this behaviour. Additionally, Nunner-Winkler and Sodian (1988) stated that children’s moral emotions also play an important role in their behaviour, namely their moral behaviour. Moral behaviour is defined as behaviour that is according to one’s moral values and standards and it reflects children’s capability to care for others (Talwar, 2011).
When considering how a child will behave in a situation such as a moral transgression, the expected moral emotions, for instance shame, following the transgression play an important role in the actual moral behaviour children show (Nunner-Winkler & Sodian, 1988). This is also supported by the study of Malti and Ongley (2014) who found that moral emotions, together with moral cognition, influence children’s moral behaviour. The combination of moral emotions and moral cognition promote children to perform reparative behaviours such as apologies and restricts immoral behaviours like aggression. Thus, moral cognition and moral emotions both influence children’s moral behaviour. However, they are different from one another. Malti and Ongley (2014) defined moral cognition as the understanding that certain acts are wrong and labelled immoral, whereas moral emotions, such as guilt, are both self-conscious and self-evaluative emotions, since it requires children to understand the situation and to have learned moral standards to attribute a certain moral emotion. Therefore, moral cognition is a prerequisite of moral emotions. For example, a child may understand that a transgression is wrong, thus the child possesses moral cognition, but does not feel the appropriate moral emotion, such as guilt or shame, following the transgression.
It is found that children from the age of 3 are already able to make moral judgments about immoral acts (Malti & Ongley, 2014). This means that 3-year-olds already have moral knowledge and know that immoral acts are wrong, not because of possible sanctions but because they inflict harm upon the victim (Krettenauer, Malti, & Sokol, 2008). They are however not yet able to attribute the expected moral emotions in reaction to these immoral acts (Krettenauer et al., 2008). Only around the ages 6 to 8, do children start attributing the expected moral emotions in reaction to immoral acts, because they can make mixed moral judgments (both positive and negative) about immoral acts (Malti & Ongley, 2014).
When considering children’s moral emotion attributions to someone who has harmed another, there are three patterns to be distinguished (Keller, Lourenço, Malti, & Saalbach, 2003). First, there are children who attribute only positive emotions towards the victimizer, the so-called happy victimizer. Second, there are children who attribute only negative emotions towards the victimizer, which is called the unhappy victimizer. Last, there are children who attribute both positive and negative emotions towards the victimizer, called the mixed victimizer. Especially the first pattern, the so-called Happy Victimizer Phenomenon is interesting. This phenomenon reveals that although children know the immoral acts from the victimizer are wrong and report them as wrong, those same children do attribute positive or happy emotions towards the victimizer (Krettenauer et al., 2008).
The most common way to test the Happy Victimizer Phenomenon is to interview children about their emotions after being told a story about an immoral act (Krettenauer et al., 2008). In their review, Krettenauer et al. (2008) discussed the different ways the Happy Victimizer Phenomenon has been investigated and its results. First, most research on the Happy Victimizer Phenomenon has only one question regarding the emotions experienced. These interviews revealed that young children consistently reported happy emotions. However, according to Krettenauer et al. (2008) it should be noted that young children are both more likely to select positive emotions, instead of negative emotions and unlikely to exhaust their memories for alternatives. When children were asked for alternative emotions, by providing extra questions, it was found that younger 4-year-olds still attributed more happy emotions than older 6- and 8-year-olds when asked how the victimizer would feel after an immoral act (Krettenauer et al., 2008). This means that young children continue to report positive emotions, even after being made aware that the victimizer could also feel an opposite emotion.
Second, Krettenauer et al. (2008) stated that most research on the Happy Victimizer Phenomenon asks for an evaluation of the emotions felt by the victimizer (emotions to the victimizer) and not about the emotions the child would feel him- or herself (emotions to the self). However, most fictional acts in the happy victimizer research concern intentional acts, which could explain why children expect the victimizer to feel happy about the act. Keller et al. (2003) examined whether it makes a difference to ask for emotions to the self in comparison to emotions to the victimizer. Keller et al. (2003) found that 85% of the older 8- and 9-year-olds reported negative emotions to the self in comparison to only 50% to the victimizer. For younger 5- to 6-year olds this result was not found, younger children
continued to report lower amounts (30%) of negative emotions when asked about emotion attributions to the self.
This age-trend from attributing mostly positive emotions towards mostly negative emotions is also found by Smith, Chen and Harris (2010) who examined the moral emotion attributions among 4- to 9-year-olds and how an apology given by the victimizer might change children’s moral emotion attributions. First, Smith et al. (2010) found that children in the condition where no apology was given, demonstrated the pattern of the Happy Victimizer Phenomenon, in which younger children more often attributed positive emotions to the victimizer than older children. However, Smith et al. (2010) found in the apology-condition that both younger and older children attributed negative emotions towards the victimizer, even when the apology was not specifically marked by the word ‘sorry’. It was concluded that children already understand the emotional effects of an apology. In addition, Chaparro, Kim, Fernández and Malti (2013) examined the difference in moral emotion attributions between Chilean and Swiss children. They found a similar trend for both Chilean and Swiss children from attributing mostly positive emotions to the self at 6 years old towards mostly negative emotions at 9 years old. Chaparro et al. (2013) concluded that the age-related development of moral emotion attributions to the self is similar in different cultures.
Children’s moral emotion attributions are part of their moral development, which is influenced by emotional development. This is based on Dunn (2005) who concluded that both cognitive and emotional development influences whether a child can internalise moral behaviours and behave morally. Moral behaviour, like moral emotions is part of children’s moral development. Based on findings from Colonnesi, Engelhard and Bögels (2010) regarding children’s emotional development and how language development influences this, it might be argued that differences found in children’s attributions of moral emotions may be influenced by differences in language development rather than differences in emotional development. Colonnesi et al. (2010) concluded that the difference found in the understanding of embarrassment between younger and older children, might be due to differences in language development rather than the presumed differences in emotional development. This was based on the result that younger children had a lower understanding of embarrassment than older children when verbal methods to measure the understanding of embarrassment were used, but not when non-verbal methods were used. Based on children’s differences in language development found by Colonnesi et al. (2010) and how this might distort children’s emotional- and therefore moral development, it can be stated that young children might implicitly but not explicitly know about moral emotions.
It is found that older children do not only show higher amounts of negative emotion attributions than younger children. Older children also demonstrate higher amounts of moral reasoning than younger children (Malti, Gasser, & Buchmann, 2009). Like moral emotions, moral reasoning is an indicator of children’s moral development. Janssens and Deković (1997) found that among 6- to 11-year-olds, children demonstrated higher amounts of moral reasoning at an earlier age when parents were supportive, authoritarian and less restrictive (controlling). It might therefore be argued that parenting practices, such as positive parenting (warmth) influence children’s moral development, especially so at an early age. This is also supported by Spinrad et al. (1999), who found that parents who displayed higher amounts of positive affect (smiling, laughing, positive comments) and encouragement had children (aged 4 to 8 years old) who displayed higher amounts of the moral emotion sympathy. While parent’s negative affect (irritation, frowning, negative comments) was negatively related to the moral emotion of sympathy. Additionally, Stormshak, Bierman, McMahon and Lengua (2000) concluded that among 6-year-olds, high amounts of parental warmth were related to less behaviour problems such as oppositional, aggressive, hyperactive and internalizing behaviour problems. These behaviour problems are an indicator of children’s moral development.
The importance of parental warmth on children’s moral development is also described by Kerr, Lopez, Olson and Sameroff (2004) alongside the importance of inductive discipline. Kerr et al. (2004) found that among 3.5-year-olds, parents who showed less warmth and inductive discipline, had children with lower levels of moral regulation (ability to abstain from misbehaviour) and higher levels of externalising behaviour. Likewise, in their review Grusec and Goodnow (1994) discussed that reasoning and induction are the most effective forms of parental discipline. Especially reasoning in the form of other-oriented induction was discussed as an effective technique to promote internalisation. Internalisation was defined as taking over the values and attitudes of society as one’s own, which allows one to show socially acceptable behaviour not because one expects external consequences but because of motivation by internal factors (Grusec & Goodnow, 1994). Thus, when children internalise society’s standards, it is expected that moral emotions will also occur.
Additionally, Stormshak et al. (2000) emphasized the importance of consistent discipline for children’s moral development in the form of less behaviour problems. It was found that high amounts of consistent discipline were related to less oppositional, aggressive, hyperactive and internalizing behaviour problems among 6-year-olds. Similar results were also found among younger children aged 1.5- to 4 years old, by Arnold, O’Leary, Wolff and
Acker (1993). Arnold et al. (1993) found that parents who showed high amounts of laxness, thus inconsistency in their discipline, reported higher levels of externalising behaviour problems in their children.
Furthermore, in their review of how parenting practices influence children’s internalisation of societal standards, Grusec and Goodnow (1994) discussed that power assertion (punishments) arouses both anger and hostility in children who will become unwilling and opposing to the parent’s wishes. It was argued that children with parents who rely on punishment as form of discipline develop fear of punishment as form of motivation to act accordingly to societal standards. Thus, when parents use punishments, children may not be internally motivated to behave according to societal standards and therefore will not express moral emotions, because they fear external consequences such as punishments in reaction to their behaviour. This is also supported by Spinrad et al. (1999) who found that when parents gave sanctions or negative responses to their child’s negative emotions these children were less likely to react with sympathy to other’s distress, for they feared sanctions from their parents.
Additionally, Grusec & Goodnow (1994) also discussed different types of punishments and its influence on children’s internalisation of societal standards. It was concluded that parents who relied on object-oriented techniques (materialistic rewards, taking away material objects and physical punishments) had children with lower levels of internalised parental values, than children whose parents used love-oriented techniques (praise, social isolation and withdrawal of affection). Although it seems that withdrawal of love might be beneficial to children’s internalisation, other research argues that love withdrawal alone does not influence children’s moral development (Hoffman & Saltzstein, 1967; Krevans & Gibbs, 1996). In fact, most research states that the parenting practices power assertion and induction have the largest effect on children’s moral development (Hoffman & Saltzstein, 1967; Krevans & Gibbs, 1996; Eisenberg & Valiente, 2002). Along with this, Grusec and Goodnow (1994) argued that a small amount of power assertion in combination with reasoning can sometimes be effective to promote children’s internalization of societal standards, because it allows the parent’s message to be heard.
Furthermore, in their studies Stormshak et al. (2000) and Kerr et al. (2004) also investigated the influence of physical punishment on children’s moral development of moral behaviour. First, it was found by Stormshak et al. (2000) that high amounts of physical punishment (e.g. spanking and physical aggression) were related to more oppositional, aggressive, hyperactive and internalising behaviour problems among 6-year-olds. Likewise,
Kerr et al. (2004) found that among younger 3.5-year-olds, parents who more frequently used physical punishments (spanking, grabbing, shaking), had children with lower levels of moral regulation and higher levels of externalising behaviour.
Although parenting practices influence children’s moral development, there is no research about how parenting practices may influence children’s moral emotion attributions to the victimizer and the self at a younger age when happy victimizer may occur and at a later age, when children attribute negative emotions in reaction to immoral acts. Research on this topic is important. In their meta-analysis, Malti and Krettenauer (2013) found a positive relation between children’s moral emotion attributions and prosocial behaviour and negative relations between children’s moral emotion attributions and antisocial behaviour. This means that to promote children’s prosocial behaviour and reduce antisocial behaviour, children’s moral emotion attributions can be stimulated. Thus, it is important to understand how and which parenting practices may stimulate children’s negative moral emotion attributions. Therefore, in a sample of normally developing Dutch 4- to 9- year-olds, this research investigated whether age influenced how children attributed moral emotions to the victimizer and the self in reaction to fictional immoral acts and how parenting practices in combination with age effected children’s emotion attributions.
To investigate this, one-on-one interviews were conducted with children and both children and one of their parents received a questionnaire about parenting practices. First, it was expected that younger 4- to 6.5-year-olds would more often report positive moral emotions to the victimizer and the self than older 6.5- to year-olds and that older 6.5- to 9-year olds would more often report negative moral emotions to the victimizer and the self than younger 4- to 6.5-year-olds. Additionally, it was expected that the parenting practices positive parenting, inductive discipline, inconsistent discipline and punishments influenced children’s moral emotion attributions both to the victimizer and the self, especially for younger children based on the past studies from Janssens and Deković (1997), Spinrad et al. (1999), Kerr et al. (2004) and Stormshak et al. (2000). Children who gave negative emotion attributions were expected to have parents who showed more positive parenting, used higher amounts of inductive discipline, were more consistent in their discipline and punished less often, than children who gave positive emotion attributions. Additionally, children who more often reported positive moral emotion attributions were expected to have parents who more often gave punishments and were inconsistent in their discipline, than children who gave negative emotion attributions. All these effects of parenting were expected to appear especially in younger children, in whom Happy Victimizer Phenomenon can be observed.
Methods Participants
Participants in this research were 64 normally developing Dutch children aged 4-9 (𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 = 6.28, 𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎𝑎𝑎 = 1.52) and one of their parents, consisting of 40 mothers, 19 fathers and 5 remaining caretakers (e.g. mother-father couples, grandparents or unknown). Furthermore, children participating in this research were divided into two age groups, one younger group (𝑛𝑛 = 28, 43.80%) and one older group (𝑛𝑛 = 36, 56.30%). The younger group consisted of children between the ages 4 to 6.5 years old (𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 = 4.82, 𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎𝑎𝑎 = .67). The older group consisted of children from 6.5 to 9 years old (𝑀𝑀𝑎𝑎𝑎𝑎𝑎𝑎 = 7.42, 𝑆𝑆𝑆𝑆𝑎𝑎𝑎𝑎𝑎𝑎 = .87). The sample consisted of 32 girls (𝑛𝑛𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑎𝑎𝑎𝑎𝑦𝑦 = 14, 𝑛𝑛𝑦𝑦𝑜𝑜𝑜𝑜𝑎𝑎𝑦𝑦 = 18) and 32 boys (𝑛𝑛𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑦𝑎𝑎𝑎𝑎𝑦𝑦 = 14, 𝑛𝑛𝑦𝑦𝑜𝑜𝑜𝑜𝑎𝑎𝑦𝑦 = 18). The participating children lived in a city or town in the Netherlands. Children participating in this research were recruited through acquaintances, messages on social media, playgrounds, schools and outside of school care. The children who participated in this research received no reward for their participation. Before children were interviewed, the parent signed a consent form. The research design was approved by the Ethics Committee of the Research Institute of Child Development and Education at the University of Amsterdam. Materials
Interview measuring children’s moral knowledge and moral emotion attributions. The participating children were interviewed by a trained master student. The child was told four stories in random order, in which a child from the same gender as the interviewed child displayed a moral transgression. These moral transgressions consisted of two different acts: two stories described a theft and two stories described violence towards another child.
After each story the child was first asked one question about his or her moral knowledge. An example question about the child’s moral knowledge is “Is Robin allowed to steal the cake? Why/why not?”.
Following the question about the child’s moral knowledge, two questions about the child’s moral emotion attributions were asked, both emotion attributions the victimizer and the self were included in these questions. An example question of emotion attributions to the victimizer is “How do you think Robin feels after stealing the cake? Why?”. An example question of emotion attributions to the self is “How would you feel if you stole the cake? Why?”. The questions regarding the child’s moral emotion attributions also had possible probing questions to get more information about whether the emotions were more positive or
negative. If the child answered only positive or negative emotions, an example from a possible probing question was “Could Robin feel more emotions? Why?”. If the child answered an equal amount of positive and negative emotions, a possible probing question was “Which emotion do you think Robin feels more, more positive or negative? Why?” These probing questions are in accordance to the procedure of Krettenauer et al. (2008). The stories that were told during the interviews and the questions asked after each story can be found in Appendix 1.
Coding system for children’s moral knowledge. The questions about children’s moral knowledge were measured on a nominal scale which was allocated into four different categories. These categories were (1) allowed (e.g. “Yes. Because she likes it.”), (2) mixed
judgment (e.g. “A little in the middle.”), (3) not allowed (e.g. “No. It belongs to someone
else.”) and (4) no knowledge (e.g. “I don’t know.”). Part of the answers (34%) were double-coded by another trained independent rater and reliability using intra-class correlations (total agreement) was calculated and revealed excellent agreement between the two raters, MeanICC: 1.00. Classifications of reliability were based on the distribution from Koo and Li (2016) and Ciccheti (1994).
Coding system for children’s moral emotion attributions. The answers to the questions about the child’s moral emotion attributions both to the victimizer and to the self were allocated on a scale into six possible answers. These answers were (1) only positive (e.g. “Good, a boy was naughty and he did something back.”), (2) mostly positive and little
negative (e.g. He could feel two ways, he could feel relieved that the boy left. But he could
also feel a little guilty because he threw the ball in someone else's face.”), (3) mostly negative
and little positive (e.g. “The same, a little happyish and very sad.”), (4) only negative (e.g.
“Stupid. Because he has stolen the pencil.”), (5) equal amount of positive and negative, thus
mixed judgment (e.g. “A little happy because she has that pencil, but I think she would also
feel guilty. Because she stole something.”) and (6) do not know (e.g. “I don’t know.”). Again, part of the answers (34%) were double-coded by an independent rater and reliability using intra-class correlations (total agreement) was calculated and ranged from good to excellent, MeanICC: 0.96, 𝑟𝑟𝑟𝑟𝑛𝑛𝑟𝑟𝑟𝑟: 0.87 − 1.00. Classifications of reliability were based on Koo and Li (2016) and Ciccheti (1994). Disagreements between the two raters were discussed and the consensus between the two raters was reached for the cases in which the coding was not the same.
Adjusted Alabama Parenting Questionnaire for children and parents measuring parenting practices. An adapted version of the long and short version of the Alabama
Parenting Questionnaire (APQ) was used (Frick, 1991; Elgar, Waschbusch, Dadds, & Sigvaldason, 2007). Additionally, some items on the questionnaires were added in line with the hypotheses and the definition of inductive discipline given by Kerr et al. (2004). There were identical parent-reported and child-reported questionnaires, which can be found in Appendices 2 and 3. Parents were instructed to fill in the parent-reported questionnaire. The child-reported questionnaires were administered by the experimenter with help of a paper-thermometer which indicated the different answer options through colours: the brightest colour indicating the answer option ‘always’ and the least bright colour indicating the answer option ‘never’, with the other categories (‘almost never’, ‘sometimes’ and ‘often’) being the intermediate colours. Additionally, the questions on the questionnaires were formulated from either the child’s or the parent’s perspective.
Both questionnaires consisted of four scales of parenting practices with a total of 11 items. Each of these items had answers on a 5-point-likertscale (1= never, 2= almost never, 3= sometimes, 4= often, 5= always) with minimum scores of 1 and maximum scores of 5 for each item and scale-average. First, three items measured ‘positive parenting/warmth’ (e.g. “Your parent lets you know when you are doing a good job with something.”). Second, three items measured ‘inductive discipline’ (e.g. “When you misbehave your parent calmly explains why your behaviour was wrong.”). Third, three items measured ‘inconsistent discipline’ (e.g. “Your parent threatens to punish you and then does not actually punish you when you misbehave.”). Last, two items measured ‘parental punishment’. One item measured ‘overall amount of punishment’: “How often does your parent punish you when you misbehave?” and one item measured ‘types of punishment’ and how often these occurred: “What form or punishment does your parent use when you misbehave (e.g. use time-out, give reprimand etc.) (Please state all the forms you can remember and how often this form of punishment takes place)”. Answers on the ‘types of punishment’ question were coded into three different types of punishment: ‘physical punishment’ (e.g. spanking), ‘taking away privileges’ (e.g. no electronics) and ‘social isolation’ (e.g. go to room). All forms of punishment, including ‘overall amount of punishment’ were summed up to calculate total frequency of parental punishment.
To make the scales on parenting practices, analyses to calculate reliability were performed. For some parenting practices the reliability was lower than allowed based on Tavakol and Dennick (2011), 𝛼𝛼 < .70. This can partly be due to the small number of items per parenting scale. To check how items from the same scale correlate between each other, for each parenting scale, inter-item correlations were also calculated. As described by Tavakol
and Dennick (2011) when the inter-item correlations were found to be too low, 𝑟𝑟 < .30, these items were removed to improve reliability. This means that in each scale the items with low correlations to the other items in the scale were excluded from the scale to improve the reliability of the scale. To check which items from the original questionnaires were removed see Appendix 4 for the corresponding items. First, in the original study Elgar et al. (2007) found a reliability of 𝛼𝛼 = .57 for positive parenting. In this study, the reliability of parent-reported positive parenting was 𝛼𝛼𝑝𝑝𝑎𝑎𝑦𝑦𝑎𝑎𝑦𝑦𝑝𝑝 = .80 and the inter-item correlation was 𝑟𝑟𝑦𝑦𝑎𝑎𝑦𝑦𝑎𝑎𝑎𝑎 = .62 − 72. Reliability and inter-item correlations were high enough and therefore no items were removed from the scale of parent-reported positive parenting. For child-reported positive parenting the reliability and inter-item correlations were 𝛼𝛼𝑐𝑐ℎ𝑖𝑖𝑜𝑜𝑜𝑜 = .45, 𝑟𝑟𝑦𝑦𝑎𝑎𝑦𝑦𝑎𝑎𝑎𝑎 = .13 − .38. To improve the child-reported positive parenting scale item 1 was removed from the scale and the new reliability found was 𝛼𝛼𝑐𝑐ℎ𝑖𝑖𝑜𝑜𝑜𝑜= .59 and the inter-item correlation was 𝑟𝑟 = .42. Second, in the original study Elgar et al. (2007) found a reliability of 𝛼𝛼 = .62 for the three items on inconsistent discipline. In this study, the reliability and inter-item correlations of parent-reported inconsistent discipline were 𝛼𝛼𝑝𝑝𝑎𝑎𝑦𝑦𝑎𝑎𝑦𝑦𝑝𝑝 = .54, 𝑟𝑟𝑦𝑦𝑎𝑎𝑦𝑦𝑎𝑎𝑎𝑎 = .19 − .48. To improve the scale, item 2 was removed, the new reliability and inter-item correlation found were 𝛼𝛼𝑝𝑝𝑎𝑎𝑦𝑦𝑎𝑎𝑦𝑦𝑝𝑝 = .74, 𝑟𝑟 = .58. For child-reported inconsistent discipline the reliability and the inter-item correlations were 𝛼𝛼𝑐𝑐ℎ𝑖𝑖𝑜𝑜𝑜𝑜 = .45, 𝑟𝑟𝑦𝑦𝑎𝑎𝑦𝑦𝑎𝑎𝑎𝑎 = .21 − .36. To improve reliability item 1 was removed, the new reliability and inter-item correlation found were 𝛼𝛼𝑐𝑐ℎ𝑖𝑖𝑜𝑜𝑜𝑜 = .47, 𝑟𝑟 = .31. Last, the parenting scale inductive discipline was formed using the definition given by Kerr et al. (2004), therefore no information about the original reliability was available. In this study, the reliability and inter-item correlations of parent-reported inductive discipline were 𝛼𝛼𝑝𝑝𝑎𝑎𝑦𝑦𝑎𝑎𝑦𝑦𝑝𝑝 = .49, 𝑟𝑟𝑦𝑦𝑎𝑎𝑦𝑦𝑎𝑎𝑎𝑎 = .30 − .32. Inter-item correlations were high enough and therefore no items were removed from the scale of parent-reported inductive discipline. For child-reported inductive discipline the reliability and inter-item correlations were 𝛼𝛼𝑐𝑐ℎ𝑖𝑖𝑜𝑜𝑜𝑜= .50, 𝑟𝑟𝑦𝑦𝑎𝑎𝑦𝑦𝑎𝑎𝑎𝑎 = .24 − .36. To improve reliability on child-reported inductive discipline item 1 was removed, the new reliability and inter-item correlation were 𝛼𝛼𝑐𝑐ℎ𝑖𝑖𝑜𝑜𝑜𝑜 = .52, 𝑟𝑟 = .36.
Procedure
Children participating in this research were either interviewed at school, home, outside of school care or at playgrounds by a trained master student. When the interviews were conducted with the parent present, that parent filled in the parent-reported questionnaire during the interview. When the interviews were conducted at school, parents received the questionnaire when they signed the permission letter, after which the child was interviewed at
school. Each interview lasted approximately 20 minutes. During the interviews the children were given an explanation about the interview, were asked about the four stories and the items from the child-reported parenting questionnaire. It was emphasized to the child that there were no right or wrong answers and there was room for questions if the child had any.
After the interview, the experimenter thanked the child for participating, asked if the child had any questions and how the child experienced the interview. If the child had questions or the experimenter concluded the child was upset about the interview, a debriefing procedure was introduced. If the child had any questions the experimenter explained to the child that the aim of the study was to investigate what children of different ages think of a social situation in which a social rule is broken. If the experimenter concluded the child to be upset after the interview because of the answers the child gave during the interview, the experimenter reminded the child that there were no right or wrong answers and the interview was about the child’s own opinion. The experimenter also informed the child of the aim of the study.
Data analysis
First, children’s moral knowledge was not included in the analyses, because there was little variance in children’s answers. Almost all 64 children mentioned that the moral transgressions in the stories were ‘not allowed’ (story 1: 𝑛𝑛 = 61 (95.30%); story 2: 𝑛𝑛 = 63 (93.80%); story 3: 𝑛𝑛 = 60 (98.40%) and story 4: 𝑛𝑛 = 59 (92.20%)). Before performing analyses, the data obtained from the questionnaires filled in by parents and children was checked for outliers (M ±3 SD’s). Outliers found in the data were winsorized by bringing the outlier value to the next observed value in the range of M ±3 SD’s. The data was also checked for normality, using skewness and kurtosis and their standard errors. Last, possible sex influences on emotion attributions and the parenting scales positive parenting, inductive discipline and parental punishment were tested using an independent t-test for the questionnaires and chi-square tests for children’s moral emotion attributions. Sex was then added as a covariate to the analyses if sex differences were found for the variables in question.
Second, to investigate whether there were age differences in the attributed emotions chi-square tests were performed. Chi-square tests allow to investigate if younger and older children differed in the attributed moral emotions. Furthermore, to investigate whether parenting practices, age group and their combination influence emotion attributions, multi-nomial regression analyses were performed. These analyses allow to investigate the effects of age, parenting practices and their interaction effects on a categorical outcome, such as
children’s moral emotion attributions, while also accounting for all the other variables in the model. The six original categories of children’s moral emotion attributions (1= only positive, 2= mostly positive and little negative, 3= mostly negative and little positive, 4= only negative, 5= equal amount of positive and negative and 6= do not know) were altered to four new categories (1= positive emotions, 2= mixed judgment, 3= negative emotions and 4= no
emotions) to increase power due to the limited sample size.
The four separate stories measuring emotion attributions with eight different questions (four questions about emotion attributions to the victimizer and four about emotion attributions to the self) were combined into two scales of emotion attributions, one measuring emotion attributions to the victimizer with four different stories and the other measuring emotion attributions to the self with four different stories. In the chi-square tests children’s age-group acted as predictor and children’s moral emotion attributions, both to the victimizer and the self, acted as dependent variables, resulting in two dependent variables. Furthermore, in the multi-nomial regression analyses, children’s moral emotion attributions both to the victimizer and the self, again acted as dependent variables, also resulting in two different dependent variables. Additionally, the age-group acted as an independent variable in the analysis alongside parenting practices and the interaction between age-group and parenting practices.
Results Preliminary Analyses
Data from all 64 children who participated in this research were used for the current analyses. There were two cases missing child-reported questionnaires, however there were parent-reported questionnaires and these cases were therefore still included in the analyses. First, data was checked for outliers (M ±3 SD’s) and revealed four outliers, one on the child-reported punishment, one on the child-reported inductive discipline and two on parent-reported physical punishment. These outliers were winsorized by bringing the outlier value to the next observed value in the range of ±3 SD’s. Furthermore, data was checked for normal distribution of the parenting practices and the types of punishments separately, based on skewness and kurtosis. First, analyses of skewness, 𝑟𝑟𝑟𝑟𝑛𝑛𝑟𝑟𝑟𝑟: −.57 − .66, 𝑆𝑆𝑆𝑆: . 30 − .31, and kurtosis, 𝑟𝑟𝑟𝑟𝑛𝑛𝑟𝑟𝑟𝑟: −.76 − 1.05, 𝑆𝑆𝑆𝑆:. 59 − .61, were performed for parenting practices. Additionally, analyses of skewness, 𝑟𝑟𝑟𝑟𝑛𝑛𝑟𝑟𝑟𝑟: −.43 − 2.86, 𝑆𝑆𝑆𝑆: . 30 − .31, and kurtosis, 𝑟𝑟𝑟𝑟𝑛𝑛𝑟𝑟𝑟𝑟: −1.06 − 6.35, 𝑆𝑆𝑆𝑆: . 59 − .61, were performed for types of punishments. The analyses showed that the data from the questionnaires were normally distributed.
The four stories measuring children’s moral emotion attributions to the victimizer and to the self were combined to form one scale of children’s moral emotion attributions to the victimizer and one to the self. This was done by classifying the emotion attributions of all four separate stories into one of the four emotion categories: (1) positive emotions, (2) mixed
emotions, (3) negative emotions or (4) no emotions. First, children were sorted into the
category ‘positive emotions’ when the emotion attributions on all four stories were positive, or when the child mentioned no emotions on one or more stories, but when an emotion was mentioned for one or more stories this was positive (and not negative). Second, children were sorted into the category ‘mixed emotions’ when the child responded to the stories with at least one or more positive emotions and one or more negative emotions. Third, children were sorted into the category ‘negative emotions’ when children attributed negative emotions on all four stories or mentioned no emotions on one or several stories, but when they did mention an emotion this was negative (and not positive). Last, children were sorted into the category ‘no emotions’ when they attributed no emotions on all four stories. This division into four categories of moral emotion attributions was done both for emotion attributions to the victimizer and emotion attributions to the self, resulting into two different scales of emotion attributions. Table 1 shows the frequencies and percentages of children for each category of children’s moral emotion attributions, both to the victimizer and to the self. Table 1 also shows how many children reported the different emotion attributions per category of children’s age-group. As seen in Table 1, very few children reported positive emotion attributions to the victimizer and the self.
Table 1
Frequencies (%) of Children’s Moral Emotion Attributions to the Victimizer and the Self by Age-Group
Emotions to the victimizer Emotions to the self Variable Younga Oldb Total Younga Oldb Total
Positive emotions 1 (3.60%) (2.80%) 1 (3.10%) 2 (3.60%) 1 (0.00%) 0 (1.60%) 1 Mixed emotions 5 (17.90%) (16.70%) 6 (17.20%) 11 (7.10%) 2 (22.20%) 8 (15.60%) 10 Negative emotions 19 (67.90%) (72.20%) 26 (70.30%) 45 (60.70%) 17 (77.80%) 28 (70.30%) 45 No emotions 3 (10.70%) (8.30%) 3 (9.40%) 6 (28.60%) 8 (0.00%) 0 (12.50%) 8 Total 28 (100%) (100%) 36 (100%) 64 (100%) 28 (100%) 36 (100%) 64
Table 2 shows the descriptive statistics for the different parenting practices measured by the questionnaires from parents and children. Calculations using two-tailed Pearson correlations between child- and parent-reported questionnaires were also performed, the results are showed in Table 2. As seen in Table 2, parenting practices positive parenting, inductive discipline and inconsistent discipline as reported by children and parents were uncorrelated. Child- and parent-reported parental punishment, overall amount of punishment and social isolation were correlated. However, since physical punishment and taking away privileges were uncorrelated, parent- and child-reported punishments were not combined to form one scale of parental punishment. Since most correlations were non-significant all parent- and child-reported parenting practices and punishments were analysed separately. Table 2
Means, SD’s, Range and Pearson Correlations on Parent- and Child-Reported Questionnaires
Variable N M SD Minimum Maximum 𝑟𝑟
Parenting practices
Positive Parenting Parent 64 4.18 .44 3.33 5.00 .14
Child 62 3.55 1.06 1.00 5.00
Inductive discipline Parent 64 4.04 .46 2.99 5.00 -.10
Child 62 2.73 1.09 1.00 5.00
Inconsistent discipline Parent 64 2.35 .68 1.00 4.00 -.05
Child 60 2.43 .97 1.00 4.50
Parental punishmenta Parent 64 2.33 .55 1.00 3.25 .35***
Child 62 2.17 .60 1.00 3.95
Types of punishment Overall amount of
punishment Parent Child 64 62 3.09 2.82 1.08 .85 1.00 1.00 5.00 5.00 .25* Physical punishment Parent 64 1.09 .29 1.00 2.00 -
Child 60 1.00 .00 1.00 1.00
Taking away privileges Parent 64 2.53 1.15 1.00 5.00 .11
Child 60 2.04 1.35 1.00 5.00
Social isolation Parent 64 2.58 1.05 1.00 4.00 .26**
Child 60 2.75 1.22 1.00 5.00
Note. a Parenting-scale for parental punishment, consisting of the overall amount of punishment, physical
punishment, taking away privileges and social isolation. *𝑝𝑝 ≤ .100, **𝑝𝑝 ≤ .050, ***𝑝𝑝 ≤ .010.
To test if there were sex differences between boys and girls in the attributed moral emotions, chi-square tests were performed. Children’s gender acted as independent variable and children’s moral emotion attributed to the victimizer and the self as dependent variables. First, results revealed no sex differences between boys and girls on the reported emotion attributions to the victimizer, 𝑋𝑋2(3) = 2.84, 𝑝𝑝 = .417. Also, results revealed no sex
differences between boys and girls on the reported emotion attributions to the self, 𝑋𝑋2(3) = 1.92, 𝑝𝑝 = .589.
To test if there were sex differences on reported parenting practices between boys and girls, independent t-tests were performed. Sex acted as predictor and the parent- and child-reported parenting practices ‘positive parenting’, ‘inductive discipline’, ‘inconsistent discipline’ and ‘parental punishment’ as dependent variables. The independent t-tests, with equal variances assumed, found one significant sex difference between boys and girls on child-reported inductive discipline, 𝑡𝑡(60) = −2.75, 𝑝𝑝 = .008. Girls (𝑀𝑀 = 2.38, 𝑆𝑆𝑆𝑆 = 1.07) reported lower amounts of inductive discipline than boys (𝑀𝑀 = 3.10, 𝑆𝑆𝑆𝑆 = 1.00). Furthermore, independent t-tests were also performed to test if there were sex differences on reported punishments between boys and girls. Again, sex acted as predictor and the parent- and child-reported punishments ‘total amount of punishment’, ‘physical punishment’, ‘taking away privileges’ and ‘social isolation’ acted as dependent variables. Results found one significant sex difference between boys and girls on the parent-reported punishment of taking away privileges, 𝑡𝑡(62) = −2.00, 𝑝𝑝 = .050. Parents reported to more often take away privileges from boys (𝑀𝑀 = 2.81, 𝑆𝑆𝑆𝑆 = 1.26) than girls (𝑀𝑀 = 2.25, 𝑆𝑆𝑆𝑆 = .98). When significant sex differences were found, sex was added as a covariate in the multi-nomial regression analyses.
Age-differences in the attributed emotions to the victimizer and the self
To investigate whether there are age-differences in children’s moral emotion attributions, two chi-square tests were performed. Children’s age-group acted as predictor and children’s moral emotion attributions acted as dependent variables, in which there was a division in emotion attributions to the victimizer and the self, resulting in two separate analyses. First, there were no significant differences between younger and older children in their emotions attributions to the victimizer, 𝑋𝑋2(3) = .18, 𝑝𝑝 = .980. Second, there were significant differences between younger and older children in their emotion attributions to the self, 𝑋𝑋2(3) = 14.52, 𝑝𝑝 = .002. Younger children less often and older children more often attributed both mixed and negative emotions to the self than expected. Also, younger children more often and older children less often attributed no emotions to the self than expected. Age-related differences and the influence of parenting practices on children’s moral emotion attributions
To investigate whether different age groups, parenting practices, and the interaction between age-group and parenting practices influence whether children attribute positive,
negative, mixed, and no emotions to the victimizer and to the self, separate multi-nomial regression analyses were performed. There were two models of emotion attributions to the victimizer and two models for the self. In one model, child-reported parenting practices were predictors next to the age-group and their interaction and in the other model parent-reported parenting practices were predictors, next to the age-group and their interaction. Table 3 shows the model-fit for each model of multi-nomial regression analysis. Since two separate models were used to predict the same dependent variable, Bonferroni correction that corrects for multiple testing was applied. This means the significance level was altered from 𝑝𝑝 = .050 to 𝑝𝑝 = .025. This new significance level was taken in account when interpreting the results.
First, Table 3 shows in which model at least one of the predictors age-group, parenting practices and their interaction differed from zero and thus significantly improved the model, compared to an empty intercept model, in predicting children’s moral emotion attributions, both to the victimizer and the self. As seen in Table 3, there were two models where one of the predictors significantly improved the model, this was for the child-reported model on emotion attributions to the victimizer and the parent-reported model on children’s emotion attributions to the self. Also, there was one model indicating a trend where one of the predictors improved the models, this was for the child-reported model on children’s moral emotion attributions to the self. Last, none of the predictors significantly improved the parent-reported model for emotions to the victimizer.
Table 3
Results from Likelihood Ratio (LR) Chi-Square Tests for Model-Fit on Adding Age-Group, Parenting Practices and Their Interaction for Predicting Children’s Moral Emotion Attributions
Model LR 𝒳𝒳2 N df Nagelkerke 𝑅𝑅2 p
Model of child-reported parenting practices
on emotion attributions to the victimizer 55.62 60 27 .74 .001*** Model of child-reported parenting practices
on emotion attributions to the self 27.85 60 18 .46 .064* Model of parent-reported parenting practices
on emotion attributions to the victimizer 30.67 64 27 .46 .285 Model of parent-reported parenting practices
on emotion attributions to the self 46.34 64 27 .63 .012**
Note. 𝑑𝑑𝑑𝑑 = 27.
*𝑝𝑝 ≤ .100, **𝑝𝑝 ≤ .050, ***𝑝𝑝 ≤ .010.
Parameter estimates for the effects of age-group, child-reported parenting practices and their interaction on children’s moral emotion attributions are reported in Table 4. Negative moral emotion attributions and the older age-group were used as reference categories. As seen in Table 4, there were no significant effects of age-group, child-reported
parenting practices or their interaction on children’s moral emotion attributions, both to the victimizer and the self when negative emotion attributions were used as a reference. This means that age, child-reported parenting practices, and their interactions did not influence whether children attributed negative vs. other categories of emotions to the victimizer and the self.
In addition to using negative moral emotion attributions as reference category, calculations using positive moral emotion attributions as reference category were also performed. These analyses resulted in one significant interaction effect of children’s age-group and child-reported parental punishment on children’s moral emotion attributions to the victimizer. Young children, relative to older children, who reported their parents often use parental punishments, less often reported mixed emotions, relative to positive emotions to the victimizer, 𝐵𝐵 = −456.93, 𝑆𝑆𝑆𝑆 = 1.79, Wald = 65322.01, 𝑝𝑝 < .001. No other effects of children’s age-group and child-reported parenting practices on children’s moral emotion attributions to the victimizer were found. Also, no effects of children’s age-group, child-reported parenting practices or their interaction on children’s moral emotion attribution to the self were found when positive emotions were used as reference category.
Likewise, calculations using mixed emotion attributions as reference category were also performed. These calculations revealed no significant effects of age-group, child-reported parenting practices or their interaction on children’s moral emotion attributions, both to the victimizer and the self.
Table 4
Influence of Age-Group, Child-Reported Parenting Practices and Their Interaction on Children’s Emotion Attributions
Outcome Predictor Negative emotions vs. 𝐵𝐵 𝑆𝑆𝑆𝑆 Wald 𝑝𝑝 Emotion attributions to the victimizer
Age Positive emotions -2804.18 25529.56 .01 .913 Mixed emotions 1.20 5.22 .05 .818 No emotions -2683.68 30342.84 .01 .930 Positive parenting Positive emotions -24.10 4575.30 .00 .996 Mixed emotions 1.05 .93 1.27 .261 No emotions -114.95 2926.94 .00 .969 Inductive discipline Positive emotions 41.70 3503.25 .00 .991 Mixed emotions -1.30 1.17 1.22 .269
No emotions -8.21 6222.93 .00 .999
Inconsistent
discipline Positive emotions Mixed emotions -16.92 .44 2340.31 .50 .00 .77 .994 .382 No emotions -10.87 2601.40 .00 .997 Parental
No emotions 152.48 3371.21 .00 .964 Age* Positive
parenting Positive emotions Mixed emotions 258.58 -.03 4803.92 1.16 .00 .00 .957 .981 No emotions 347.93 3739.64 .00 .926 Age* Inductive
discipline Positive emotions Mixed emotions 36.84 1.40 4251.55 1.30 1.16 .00 .993 .282
No emotions 90.68 6632.70 .00 .989
Age* Inconsistent
discipline Positive emotions Mixed emotions 232.02 -2.33 3332.28 1.46 2.57 .01 .944 .109 No emotions 222.97 3899.54 .00 .954 Age* Parental
punishment Positive emotions Mixed emotions 457.01 .08 4696.23 1.79 .01 .00 .922 .966
No emotions 260.75 .00 - -
Emotion attributions to the self
Age Mixed emotions -.45 5.77 .01 .938
No emotions 13.22 4881.14 .00 .998
Positive parenting Mixed emotions 1.55 .96 2.59 .107
No emotions .17 1149.48 .00 1.000
Inductive discipline Mixed emotions .06 .86 .00 .948
No emotions -.02 1154.27 .00 1.000
Inconsistent
discipline Mixed emotions No emotions -.08 .00 959.52 .46 .03 .00 1.000 .865 Parental
punishment Mixed emotions No emotions -1.19 -.12 1.22 .78 .95 .02 .329 .882 Age* Positive
parenting Mixed emotions No emotions -1.69 .28 1149.48 1.41 1.45 .00 1.000 .229 Age* Inductive
discipline Mixed emotions No emotions -.21 .18 1154.27 1.10 .03 .00 1.000 .868 Age* Inconsistent
discipline Mixed emotions No emotions 1.24 .42 959.25 1.27 .96 .00 1.000 .327 Age* Parental
punishment Mixed emotions No emotions 1.32 .78 1.80 .00 .53 - .465 -
Note. 𝑑𝑑𝑑𝑑 = 1.
*𝑝𝑝 ≤ .100, **𝑝𝑝 ≤ .050, ***𝑝𝑝 ≤ .010.
Furthermore, Table 5 shows the results from parameter estimates for the effects of age-group, parent-reported parenting practices and their interactions on children’s moral emotion attributions. In Table 5 negative emotion attributions from children and the older age-group were used as reference categories. As seen in Table 5 there were interaction effects of children’s age-group and parent-reported positive parenting on children’s moral emotion attributions to the victimizer. Young children relative to older children, with parents who reported high amounts of positive parenting, more often reported both mixed emotions and no emotions, relative to negative emotions to the victimizer. Furthermore, after accounting for the interaction effect, effects of parent-reported positive parenting and inconsistent discipline on children’s moral emotion attributions to the victimizer were also found. Children with parents who reported high amounts of positive parenting, less often reported no emotions relative to negative emotion attributions to the victimizer. Also, children with parents who
reported high amounts of inconsistent discipline, more often reported mixed emotions, relative to negative emotion attributions to the victimizer. However, after applying Bonferroni correction, these results were no longer significant.
In addition, calculations using positive emotion attributions as reference category were also performed. These analyses only revealed interaction effects between children’s age-group and parent-reported inductive discipline on childreln’s moral emotion attributions to the victimizer. Young children relative to older children, with parents who reported high amounts of inductive discipline, less often reported mixed emotions, relative to positive emotions to the victimizer, 𝐵𝐵 = −86.00, 𝑆𝑆𝑆𝑆 = 4.79, Wald = 322.69, 𝑝𝑝 < .001. Also, young children relative to older children, with parents who reported high amounts of inductive discipline, less often reported negative emotions, relative to positive emotions to the victimizer, 𝐵𝐵 = −82.03, 𝑆𝑆𝑆𝑆 = 4.48, Wald = 334.89, 𝑝𝑝 < .001. No effects of age-group, parent-reported parenting practices or their interaction on children’s moral emotion attributions to the self were found.
Likewise, calculations using mixed emotion attributions as reference category were also performed. First, an interaction effect between children’s age-group and parent-reported positive parenting on children’s emotion attributions to the victimizer was found. Young children relative to older children, with parents who reported high amounts of positive parenting, less often reported negative emotions, relative to mixed emotions to the victimizer, 𝐵𝐵 = −6.14, 𝑆𝑆𝑆𝑆 = 2.82, Wald = 4.75, 𝑝𝑝 = .029. Second, after accounting for the interaction effects, an effect of parent-reported inconsistent discipline on children’s moral emotion attributions was found. Children with parents who reported high amounts of inconsistent discipline less often reported negative emotions relative to mixed emotion attributions to the victimizer, 𝐵𝐵 = −1.61, 𝑆𝑆𝑆𝑆 = .96, Wald = 2.79, 𝑝𝑝 = .095. However, after applying Bonferroni correction, both results were no longer significant. Last, no effects of age-group, parent-reported parenting practices or their interaction on children’s moral emotion attributions to the self were found.
Table 5
Influence of Age-Group, Parent-Reported Parenting Practices and Their Interaction on Children’s Emotion Attributions
Outcome Predictor Negative emotions vs. 𝐵𝐵 𝑆𝑆𝑆𝑆 Wald 𝑝𝑝 Emotion attributions to the victimizer
Age Positive emotions -549.01 .00 - -
Mixed emotions -3.28 10.24 .10 .749 No emotions -20.31 17.94 1.28 .258 Positive parenting Positive emotions -6.53 6.37 1.05 .305 Mixed emotions -3.33 2.35 2.00 .157
No emotions -5.47 3.22 2.88 .090* Inductive
discipline Positive emotions Mixed emotions 5.78 2.42 4.97 2.02 1.35 1.42 .245 .233
No emotions -2.78 4.22 .44 .509
Inconsistent
discipline Positive emotions Mixed emotions 1.61 -.15 1.66 .96 2.79 .01 .095* .927
No emotions 2.65 2.13 1.54 .214
Parental
punishment Positive emotions Mixed emotions 1.65 .19 3.04 1.17 .30 .03 .586 .869
No emotions -.07 1.77 .00 .968
Age* Positive
parenting Positive emotions Mixed emotions -56.60 6.14 95817.42 2.82 4.75 .00 .029** 1.000
No emotions 6.52 3.83 2.90 .089*
Age* Inductive
discipline Positive emotions Mixed emotions 137.10 -3.96 90145.84 2.49 2.53 .00 .999 .111
No emotions 1.73 4.48 .15 .699
Age* Inconsistent
discipline Positive emotions Mixed emotions 28.25 -1.22 26375.37 1.32 .00 .85 .999 .358
No emotions -2.71 2.46 1.21 .271
Age* Parental
punishment Positive emotions Mixed emotions 36.79 -1.29 63057.60 1.67 .00 .59 1.000 .442
No emotions -2.05 2.26 .83 .364
Emotion attributions to the self
Age Positive emotions -267.44 14766.58 .00 .986 Mixed emotions 367.69 2713.06 .02 .892 Positive parenting No emotions Positive emotions 7.20 -.06 4270.76 3338.14 .00 .00 1.000 .999
Mixed emotions -.34 1.05 .11 .746
No emotions -.06 1027.68 .00 1.000
Inductive
discipline Positive emotions Mixed emotions -.09 -.79 3117.30 1.02 .00 .59 1.000 .441
No emotions -.09 1.64 .00 .959
Inconsistent
discipline Positive emotions Mixed emotions -.07 -.66 2147.12 .75 .00 .78 1.000 .378
No emotions -.07 759.12 .00 1.000
Parental
punishment Positive emotions Mixed emotions .07 .44 2710.45 .92 .00 .23 1.000 .632 Age* Positive parenting No emotions .07 957.50 .00 1.000 Positive emotions 20.58 3360.75 .00 .995 Mixed emotions 110.84 797.06 .02 .889 No emotions -.07 1027.68 .00 1.000 Age* Inductive
discipline Positive emotions Mixed emotions -272.26 6.08 3288.20 1952.99 .00 .02 .999 .889
No emotions .55 .00 - -
Age* Inconsistent
discipline Positive emotions Mixed emotions 11.25 70.90 2200.22 545.59 .00 .02 .996 .897
No emotions 1.03 759.12 .00 .999
Age* Parental
punishment Positive emotions Mixed emotions 43.84 -6.46 2763.08 298.30 .00 .00 .987 .983
No emotions 1.25 957.50 .00 .999
Note. 𝑑𝑑𝑑𝑑 = 1.
Post-hoc analyses of type of punishment and age-group influencing children’s moral emotion attributions
Calculations of parameter estimates revealed one significant effect of parental punishment on children’s moral emotion attributions to the victimizer. Parental punishment as formulated in Table 4 and Table 5 consisted of four types of punishments: ‘overall amount of punishment’, ‘physical punishment’, ‘taking away privileges’ and ‘social isolation’, for which the frequencies were taken together to form one total parental punishment variable. To see if any type of parental punishment specifically predicted children’s moral emotion attributions, multi-nomial regression analyses for testing the influence of age-group, type of punishment and their interaction on children’s moral emotion attributions were performed.
Table 6 shows in which model at least one of the predictors age-group, punishments and their interaction differed from zero and thus significantly improved the model, compared to an empty intercept model in predicting children’s moral emotion attributions. Physical punishment was not included as a predictor in the models where child-reported predictors were included, because children never reported on this. As seen in Table 6 there were two models where at least one of the predictors significantly improved the model, this is for the models of emotion attribution to the self, both for child- and parent-reported punishments. It should be noted that two models were used to predict the same dependent variable. Therefore, Bonferroni correction that corrects for multiple testing was again applied. This means the significance level was again altered from 𝑝𝑝 = .050 to 𝑝𝑝 = .025. This new significance level was considered when interpreting the results.
Table 6
Results from Likelihood Ratio (LR) Chi-Square Tests for Model-Fit on Adding Age-Group, Punishments and Their Interaction for Predicting Children’s Moral Emotion Attributions
Model LR 𝒳𝒳2 N df Nagelkerke 𝑅𝑅2 𝑝𝑝
Model of child-reported punishments on
emotion attributions to the victimizer 27.321 60 21 .454 .160 Model of child-reported punishments on
emotion attributions to the self 22.755 60 14 .392 .064* Model of parent-reported punishments on
emotion attributions to the victimizer 32.549 64 27 .481 .212 Model of parent-reported punishments on
emotion attributions to the self 42.081 64 27 .586 .032**
Note. 𝑑𝑑𝑑𝑑 = 14.
*𝑝𝑝 ≤ .100, **𝑝𝑝 ≤ .050, ***𝑝𝑝 ≤ .010.
Parameter estimates were calculated for the models in which child-reported punishments were used as predictors of moral emotion attribution to the victimizer and to the self, next to age-group and the interaction between age-group and child-reported punishments.
Negative emotion attributions and the older age-group were used as reference categories. These analyses revealed no significant effects of age-group, child-reported punishments or their interaction on children’s moral emotion attributions, both to the victimizer and the self.
Additionally, calculations using positive emotion attributions as reference category were also performed. First, a significant interaction effect of children’s age-group and punishment in the form of taking away privileges on emotions attributions to the victimizer occurred. Young children relative to older children, who reported their parents often take away privileges, less often reported mixed emotions, relative to positive emotions, 𝐵𝐵 = −59.20, 𝑆𝑆𝑆𝑆 = .58, Wald = 10511.82, 𝑝𝑝 < .001. After accounting for the interaction effects, two effects of child-reported overall amount of punishment on children’s moral emotion attributions to the victimizer occurred. First, children who reported their parents often give punishments less often reported negative emotions, relative to positive emotions to the victimizer, 𝐵𝐵 = −42.42, 𝑆𝑆𝑆𝑆 = 1.89, Wald = 505.49, 𝑝𝑝 < .001. Also, children who reported their parents often give punishments less often reported mixed emotions, relative to positive emotions to the victimizer, 𝐵𝐵 = −42.46, 𝑆𝑆𝑆𝑆 = 1.92, Wald = 484.58, 𝑝𝑝 < .001. No effects of age-group, child-reported punishments or their interaction on children’s moral emotion attributions to the self were found.
Likewise, mixed emotion attributions as reference categories were also performed. These calculations revealed no significant effects of age-group, child-reported punishments or their interaction on children’s moral emotion attributions, both to the victimizer and the self.
Furthermore, parameter estimates were also calculated for the models in which parent-reported punishments were used as predictors of moral emotion attribution to the victimizer and to the self, next to age-group and the interaction between age-group and parent-reported punishments. Calculations using either negative, positive or mixed emotion attributions as reference category revealed no significant effects of age-group, parent-reported punishments or their interaction on children’s moral emotion attributions, both to the victimizer and the self.
Discussion
In the present study, it was investigated if there are differences between younger and older children in the attributed moral emotions to the self and the victimizer. First, only one or two children reported positive emotion attributions to the victimizer or the self on all four stories. Most children reported on negative emotion attributions on all four stories. Mixed (both positive and negative) and no emotion attributions were also reported more often than positive emotion attributions. When younger and older children were compared in the
attributed emotions, no difference was found for emotions to the victimizer. However, there were age-differences in the attributed emotions to the self. Older children more often reported mixed and negative emotions to the self than younger children. Also, younger children more often reported no emotions to the self than older children.
Additionally, this study investigated the influence of parenting practices in combination with the age group on children’s moral emotion attributions both to the victimizer and the self. In contrast to our hypotheses effects of age, parenting practices or their interaction were only found for children’s moral emotion attributions to the victimizer, not to the self. First, results on positive parenting were both in agreement and disagreement with the hypotheses. On the one hand, as hypothesized it was found that both younger and older children with parents who reported high amounts of positive parenting less often reported no emotions than negative emotions to the victimizer, which indicates that children who are positively parented have knowledge of the victimizer’s emotions and report the emotions to be negative. On the other hand, contrary to the hypotheses, it was found that younger children compared to older children, with parents who reported high amounts of positive parenting more often attributed both mixed and no emotions than negative emotions to the victimizer. Additionally, younger children, compared to older children with parents who reported high amounts of positive parenting less often reported negative emotions relative to mixed emotions to the victimizer. These results thus indicated that children, especially young children, more often reported on happy victimizer in the form of mixed emotions (consisting of both positive and negative emotions) or had no knowledge of the victimizer’s emotions when they were positively parented. However, results on positive parenting were not robust and disappeared after including the correction for multiple testing and should therefore be interpreted with some reservation.
Second, it was hypothesized that high amounts of parental inductive discipline were related to more reports of negative emotions. This was not found in the results. In fact, it was found that younger children, when compared to older children, with parents who reported high amounts of inductive discipline less often reported both negative and mixed emotions than positive emotions to the victimizer. This result remained robust after applying correction for multiple testing. This thus indicates that high amounts of inductive discipline were related to more reports on happy victimizer, especially in young children.
Contrarily, results on parent’s inconsistent discipline were in line with our hypotheses. It was found that both younger and older children with parents who reported high amounts of inconsistent discipline more often reported mixed emotions than negative emotions to the
