THE MEDIATING ROLE OF PARENTING
BEHAVIOURS IN THE RELATION BETWEEN
PARENT AND INFANT ANXIETY
FACULTY OF SOCIAL AND BEHAVIOURAL SCIENCES
Graduate School of Childhood Development and Education
Masterscriptie Orthopedagogiek Pedagogische en Onderwijskundige Wetenschappen Universiteit van Amsterdam L.D. Willems 6097901 1st Supervisor Dr. E.L. Möller / M. Metz, MSc. 2nd Supervisor Dr. E.I. De Bruin Amsterdam, April 2015
Abstract ... 4
Samenvatting ... 5
Introduction ... 6
Parenting behaviour ... 7
Paternal and maternal parenting differences ... 9
Social referencing and modeling ... 11
This study ... 14
Method ... 16
Participants ... 16
Procedure ... 17
Measures ... 18
Plan for data-‐analysis ... 22
Data analysis and results ... 23
Preliminary checks and descriptive statistics ... 23
Association of parenting behaviours with parental anxiety and infant anxiety. ... 24
Discussion ... 27
References ... 32
Abstract
The mediating role of parenting behaviours was assessed in the relation between parent and infant anxiety for mothers and fathers separately. Eighty-‐one infants aged 10-‐15 months were encouraged to cross the deep side of the visual cliff by their mother (N = 40) or father (N = 41). During this task parent and infant anxiety were observed. During two free play tasks (one with and one without toys), challenging, overprotective and overcontrolling parenting behaviours were coded. Additionally, the Comprehensive Parenting Behaviour Questionnaire (CPBQ-‐1) was filled-‐out by parents to assess
challenging, overprotective and overcontrolling parenting. More parental anxiety was assumed to relate to more infant anxiety because of a decrease in challenging and increases in overcontrolling and overprotective parenting. Mediation was assessed with PROCESS for SPSS (Preacher & Hayes, 2004), but no evidence for mediation was found. A borderline positive association between paternal and infant anxiety was found. A slight trend for a relation between for more paternal challenging parenting and less infant anxiety was found. Both results were in the expected direction. Some evidence emerged that infants of anxious fathers are more anxious; paternal challenging
parenting may form a buffer for infant anxiety. Mediation might emerge when children are older.
Samenvatting
De mediërende rol van opvoedingsgedrag is onderzocht in de relatie tussen angst van ouders en baby’s voor vaders en moeders apart. Eenentachtig baby’s van 10-‐15 maanden werden aangemoedigd door vader (N = 41) of moeder (N = 40) om de diepe zijde van de visual cliff over te steken. Angst van ouder en kind werden geobserveerd. Middels twee vrij spel taken (één met en één zonder speelgoed) werden uitdagend, overcontrolerend en overbeschermend opvoedgedraging gecodeerd. Ook werden uitdagend, overcontrolerend en overbeschermend opvoedgedraging gemeten middels de Comprehensive Parenting Behaviour Questionnaire (CPBQ-‐1). Verondersteld werd dat angstige ouders angstigere baby’s hebben door een afname van uitdagend
opvoedgedrag en een toename van overcontrolerend en overbetrokken opvoedgedrag. Mediatie werd getest door middel van PROCESS voor SPSS (Preacher & Hayes, 2004), maar niet gevonden. Een borderline positief verband werd gevonden voor angst van vader en die van kind. Een trend was zichtbaar dat meer uitdagend opvoedgedrag van vaders geassocieerd was met minder angst van het kind. Beide resultaten waren in de verwachte richting. Zeer voorzichtig kan gesteld worden dat angstige vaders baby’s hebben met meer angst; uitdagend gedrag van vaders zou hier een buffer voor kunnen vormen. Mogelijk vindt mediatie plaats wanneer kinderen ouder zijn.
Introduction
Anxiety disorders amongst children and adults in the Netherlands and other western societies are highly prevalent and can be severe. In children and adolescents 5 to 10 percent suffer from anxiety disorders (Carr, 2006; Costello et al., 2011; Ferdinand & Dieleman, 2007; Verhulst, 2001). A recent meta-‐analysis by Costello et al. (2011) reviewing datasets using a DSM-‐III-‐R, DSM-‐IV or ICD-‐10 diagnosis of anxiety disorders found the following prevalence rates for all anxiety disorders combined: 12,3% in elementary-‐school-‐age children, 11% in 13-‐18 year old children and a prevalence of 10,2% using all data sets with age ranges between 2-‐21 years of age. Children with an anxiety disorder are at high risk of becoming adults with an anxiety disorder (Rapee, Schniering, & Hudson, 2009).
A raised vulnerability to anxiety has been found in children of parents with an anxiety disorder (Turner, Beidel, & Costello, 1987). In a study where children aged 7 to 12 years of anxious parents, depressed parents, a combination of anxious/depressed parents and those of non-‐diagnosed parents were compared, it was found that a significantly larger amount of children of parents with a psychiatric diagnosis had an increased risk of psychopathology (Beidel & Turner, 1997). Offspring of parents with an anxiety disorder alone showed to have more anxiety disorders, whereas children of depressed and anxious/depressed parents showed a broader range of psychopathology. From a different point of view, Cooper, Fearn, Willets, Seabrooke and Parkinson (2006) assessed the anxiety disorders of parents of 85 children diagnosed with an anxiety disorder and compared them to the parents of 45 control children with no current anxiety disorder. A significantly higher rate of anxiety disorders was found in parents of children with an anxiety disorder themselves.
This intergenerational transmission of anxiety can be explained by two factors: vulnerability factors such as genetic factors or having an anxious temperament, and environmental factors (Murray, Creswell, & Cooper, 2009). The latter includes adverse life events, modelling, information transfer and parenting behaviours. This study will explore the possibility of a mediating role of parenting behaviours in the relation between parent and infant anxiety. Parenting behaviours and modelling will be discussed.
Parenting behaviour
Many studies report on the association between parenting behaviours and child anxiety (Chorpita & Barlow, 1998; McLeod, Wood, & Weisz, 2007; Rapee, 1997; Wood, McLeod, Sigman, Hwang, & Chu, 2003). The most studied parenting behaviours in association with child anxiety are parental control and rejection (McLeod et al., 2007; Wood et al., 2003).
Parental overcontrol, overprotection, overinvolvement and intrusiveness have been used in the literature interchangeably, generally under the umbrella term: control. These restrictive parenting behaviours are characterized by excessive regulation of children’s activities and/or routines, high levels of intrusiveness, encouragement of the child’s dependency on parents and discouragement of the child’s independent problem solving (Barber, 1996; Chorpita, Brown, & Barlow, 1998; Maccoby, 1992; Majdandžić, De Vente, & Bögels, 2010; Van Der Bruggen, Stams, & Bögels, 2008). According to some theoretical models (Chorpita & Barlow, 1998; McLeod et al., 2007; Wood et al., 2003), overcontrol can increase child anxiety by preventing children to explore their surroundings independently, by reducing their feelings of competence, and by increasing children’s interpretation of their environment as threatening. Autonomy granting can be seen as the opposite of parental control and therefore promotes child independence. Autonomy granting is often studied instead of, or in addition to, parental control (Clark & Ladd, 2000; Ginsburg, Grover & Ialongo, 2008; Majdandžić et al., 2010; Steinberg, Elmen & Mounts, 1989).
In a meta-‐analysis by Möller, Nikolić, Majdandžić, De Vente, & Bögels (in press), a distinction is made between overcontrolling and overprotective parenting and it is assumed that they affect child anxiety in different ways. Overcontrolling parenting restricts the child by needlessly helping or interfering with the child’s behaviour without taking their needs, interests or feelings into account. Overprotective parenting limits the child’s exposure to new objects, people or situations by taking into account the child’s health and safety in an excessively cautious and protective manner. In the meta-‐analysis, paternal and maternal differential parenting behaviours are both related to child anxiety. Results showed that while there were no significant correlations between maternal or paternal overcontrol and child anxiety, for maternal overprotective parenting and child anxiety a significant relation was found and for paternal overprotective parenting and child anxiety a borderline significant but
stronger correlation was found, indicating that overcontrolling and overprotective parenting behaviours indeed relate differently to child anxiety.
Parental rejection is characterized by hostility and negativity towards the child, lack of warmth, lack of responsiveness and criticising reactions towards the child (McLeod et al., 2007; Rapee, 1997; Wood et al., 2003). Rejecting parenting behaviours enhance child anxiety by continuous negative responses, thereby increasing the child’s feelings of incompetence and worthlessness (McLeod et al., 2007; Rapee, 1997; Wood et al., 2003).
The meta-‐analytic review conducted by McLeod et al. (2007) illustrated that the overall link between parenting and child anxiety is small. However, when looking merely at observational data instead of questionnaires, a medium effect size was found for the association between parental overcontrol and child anxiety while only a small effect size was found for the relation between parental rejection and child anxiety. Similar results have been found in a review by Wood et al. (2003). Parental rejection could be more strongly related to child depression, while parental overcontrol could be more related to child anxiety (Rapee, 1997). Furthermore, in the meta-‐analysis by Van der Bruggen et al. (2008) examining the association of parent and child anxiety and parental overcontrol, a medium effect size between child anxiety and parental overcontrol was also found.
A fairly new parenting construct is challenging parenting behaviour. Challenging parenting behaviour entails behaviours of the parent towards the child where, during playful interaction, the child is challenged, teased, excited, stimulated or encouraged towards competition and risk-‐taking behaviour (Bögels & Perotti, 2011; Bögels & Phares, 2008). In a recent study by Majdandžić, Möller, De Vente, Bögels, & Van den Boom (2014) a distinction is made between more physical forms of challenging behaviours, such as rough-‐and-‐tumble play, and more verbal forms, such as teasing or stimulating risk-‐taking behaviour, which both play a role in the construct. So far, hypotheses have been made that challenging parenting behaviour buffers for child anxiety, exposing the child to arousal at an early age and in a safe and comfortable manner (Bögels & Perotti, 2011; Bögels & Phares, 2008). However to this point, little research had been done (Majdandžić et al., 2014; Möller, Majdandžić, & Bögels, in press). In both of these studies (Majdandžić et al., 2014; Möller et al, in press), significant relationships have been found to child anxiety, although in different
directions for fathers and mothers. For fathers a negative relationship was found, while a positive relationship for mothers was found. This suggests that the more challenging behaviour fathers display, the less anxious their children are, while the more challenging behaviour mothers display, the more anxious their children are (although in the study by Möller et al. (in press) the positive relation for mothers was only borderline significant).
It is possible that mothers and fathers differ in the extent to which they show different types of parenting behaviours (Bögels & Perotti, 2011). This will be further described in the next paragraph.
Paternal and maternal parenting differences
In a review on the evolutionary basis of sex differences in parenting (Möller, Majdandžić, De Vente, & Bögels, 2013) it is described how fathers and mothers differ in their parental roles due to assumed different specializations developed during evolution. Where males are characterized by more risk-‐taking behaviour, are more strongly orientated to social dominance and are more competitive, females tend to be more risk-‐aversive, caring and empathic or sensitive to the needs of others (Bögels & Phares, 2008; Möller et al., 2013). These specializations are expected to be reflected in the way parents parent, as fathers and mothers are likely to model and actively teach their offspring their own gender-‐roles. According to Paquette (2004) for instance, the father-‐child relationship can be characterized by an activation relationship as opposed to the mother-‐child attachment relationship. The father-‐child activation relationship “satisfies the child’s need to be stimulated, to overcome limits, and to learn to take chances in contexts in which the child is confident of being protected from potential dangers” (Paquette, 2004, p. 202). This is put into practice during the before mentioned
rough-‐and-‐tumble play (RTP) in which fathers surprise their children, momentarily
destabilize them and encourage children to take risks (Paquette, 2004). The attachment relationship (in reference to Bowlby, 1969 in Paquette, 2004) entails the child’s confidence in the parent’s response to the child’s basic needs, calming and comforting in times of stress, and offering warmth.
As described in the review by Möller et al. (2013), several studies have found that fathers spend more time playing with their children than mothers, that paternal
play is more often more physical (RTP) than the play of mothers, and compared to fathers, mothers’ play is more often characterized by pretend play. For example, a study by Lamb (1977) revealed that infants as young as 7 to 12 months are equally attached to both parents and that both parents hold their babies for equal amounts of time. Interestingly, the reason parents pick up their babies differed from each other. Mothers picked up and held their babies for caretaking and soothing practices, whilst fathers did this mainly for play interactions (Lamb, 1977). A side note in the review by Möller et al. (2013) is that some studies have found very small to no difference in the type of play fathers and mothers engage in, or in the sensitivity and responsiveness of fathers and mothers. However, none of the studies described found that fathers are more sensitive than mothers.
It is hypothesized that mothers, based on their evolutionary acquired skills, are specialized in the child’s internal world providing warmth and protection, providing in the child’s basic needs (e.g., feeding, changing diapers) and caring and soothing when the child is in stress. Fathers on the other hand are thought to be specialized in the external world encouraging the child to push and overcome limits, engaging in more competitive and challenging play (Bögels & Perotti, 2011; Bögels & Phares, 2008). It is also hypothesized that paternal challenging behaviour is a protective factor for the development of child anxiety because it lets children gain experience in dealing with ambiguous situations in a safe manner (Bögels & Perotti, 2011; Bögels & Phares, 2008). In line with this, Bögels and Perotti (2011) pose that fathers who stimulate their children for risk taking and competitive behaviours through RTP, enhance the coping skills of their children through positive experiences (Möller et al., 2013) and prepare them for competing in the outside world later in life.
Some researchers suggest that what is a protective factor in paternal parenting behaviour, could be a risk factor in maternal behaviour and vice versa (Bögels & Perotti, 2011; Bögels & Phares, 2008, Möller et al., 2013). In a longitudinal study conducted to examine the independent influence of paternal and maternal parenting behaviour on child social anxiety it was found that paternal challenging behaviour predicted less social anxiety at 4,5 years, whereas maternal challenging behaviour predicted more social anxiety, suggesting that the same parenting behaviours can be either a risk factor or a protective factor in the development of child social anxiety (Majdandžić et al. 2014).
Regarding the intergenerational transmission of anxiety, Bögels and Phares (2008) describe a model in which parental anxiety affects the paternal and maternal role in different ways. This hypothesis is further elaborated in a formal model (Bögels & Perotti, 2011). For instance, because women are more risk-‐aversive, caring, empathic and sensitive to the needs of others, being anxious does not necessarily change their habits, only strengthens them. On the other hand, men are more risk-‐taking, competitive and seek out social dominance. Being anxious could possibly alter this behaviour very much. Therefore, an anxious mother would still be capable of fulfilling her soothing, caring and protecting role but an anxious father is less capable of stimulating, competitive, risk taking and social dominant behaviour.
Social referencing and modeling
One way to investigate the intergenerational transmission of anxiety is through social referencing studies. Social referencing is a skill infants develop around the age of ten months (Feinman & Lewis, 1983; Walden & Ogan, 1988). When infants look towards the emotional expressions of someone else in an ambiguous situation, usually an important figure like a parent, and use this information to understand the unfamiliar situation, this is called social referencing (Feinman, 1982). When infants reference and subsequently use the new understanding of an ambiguous situation to alter their behaviour, this is called modeling (Murray et al., 2009).
Many studies have investigated this phenomenon with young children to see whether they are influenced by their parents’ reaction to a non-‐social stimulus or social stimulus (De Rosnay, Cooper, Tsigaris, & Murray, 2006; Feinman & Lewis, 1983; Mumme, Fernald, & Herrera, 1996; Murray et al., 2008; Walden & Ogan, 1988). An example of a social referencing study with a non-‐social stimulus is the study of Gerull and Rapee (2002). Here, 30 toddlers were presented with two fear-‐provoking animal toys, which were alternately combined with negative or positive maternal reactions to the toy, such as vocal, facial, and bodily gestures. The key aim of this study was to see if fear/avoidance behaviour of the child in reaction to the toy would persist when mother stopped her negative emotional signalling after a 1-‐minute and 10-‐minute delay. Results indicated that children indeed show significantly increased fear and avoidance to a toy after negative maternal signalling as opposed to positive maternal signalling. These
differences were also significant after the 1-‐minute and 10-‐minute delay, suggesting that the children formed an association between the toy and mothers’ reaction. In a similar study (Engliston & Rapee, 2007), positive maternal modelling towards a fear-‐ relevant stimulus decreased the fear acquisition of toddlers in an experimental design. Here, 77 toddlers and their mothers where assigned to one of three pre-‐exposure conditions: a modelling group, a stimulus-‐only group and a control group. The control group was not presented with a fear-‐relevant stimulus but was allowed free play time. Results revealed that toddlers in the positive modelling condition significantly differed from the two other groups: these toddlers showed increased positive reactions to the toy in both the post-‐test and follow-‐up trial compared to the baseline trial. Toddlers in the stimulus only and control condition showed decreased positive reactions towards the toy during post-‐test and follow-‐up trial. These results suggest that positive maternal modelling can be seen as a protective factor in the acquisition of fear in toddlers (Engliston & Rapee, 2007).
Social referencing studies using a social stimulus examine infants’ response to an unfamiliar person. For example, De Rosnay et al. (2006) examined infants’ response to a stranger in an experimental design. First the stranger approached and interacted with the mother in full view of the infant. Next, the stranger approached the infant while the mother was asked to disregard the child and read a magazine. This experiment was conducted with each mother-‐infant pair twice, each time with a different male stranger. On one occasion, mothers were asked to respond to the stranger in a socially anxious way, and on the other occasion, mothers were asked to respond in a non-‐anxious way. Mothers were previously trained to act socially anxious and non-‐socially anxious during the interaction. Infants’ responses towards the stranger showed significantly more fearfulness and avoidance, and showed less positive emotional tone after the socially anxious mother-‐stranger interaction, compared to their responses to the stranger after the non-‐socially anxious mother-‐stranger interaction.
Another social referencing study using a social stimulus (Murray et al., 2008) examined the intergenerational transmission of social anxiety using a longitudinal design. The response given to a stranger by infants of mothers with (N = 79) and without social phobia (N = 77), were assessed at 10 months and 14 months of age. At each time, a female stranger first approached the mother for a short conversation in full view of the infant. Next, the stranger approached and interacted shortly with the infant,
and subsequently picked-‐up the child, while the mother could intervene whichever way she pleased. Results revealed that, at 10 and at 14 months, socially anxious mothers where more anxious towards the stranger in both mother-‐stranger and infant-‐stranger interactions, and these mothers where less engaged with the stranger during the mother-‐stranger interaction and less encouraging during the infant-‐stranger interaction. Infants of socially anxious mothers showed increased avoidance and fearfulness towards the stranger over the four-‐month period, compared to infants of non-‐socially anxious mothers. These two studies have found evidence for infants’ capability of using maternal emotional signals in a direct and indirect manner and subsequently altering their own behaviour during their infant-‐stranger interaction.
Another type of stimulus used in social referencing studies is the visual cliff. The visual cliff is a large table with a variable drop-‐off in the middle, both sides covered in chequered linoleum. The entire table is covered with a Plexiglas surface. The shallow side has the Plexiglas surface directly on top of the chequered surface, and on the deep side the Plexiglas hangs some distance above the chequered surface. The visual cliff was first used in studies examining depth perception and fear of heights in animals and infants, and later on used to examine social referencing behaviours in infants (Adolph & Kretch, 2012). In a renowned social referencing study by Sorce, Emde, Campos and Klinnert (1985), infants were placed on the shallow side of the visual cliff while mother stood behind the deep side of the cliff, on the other side. Mother placed an attractive toy in front of her on the deep side of the cliff and was instructed to show a previously trained happy, fearful, interested, angry or sad facial expression when her infant approached the edge in the middle of the visual cliff. In the happy and interested condition, most infants crossed the cliff (74% and 73%), while none of the infants crossed in the fearful condition and only few (11%) crossed in the anger condition. In the sadness condition 11% of the infants crossed the cliff. During pilot-‐testing, different cliff heights were used to define an appropriate height in which infants were hesitant but not completely avoidant of the cliff, and where they referenced their mothers frequently. This was a drop of 30 cm. In addition, a different group of infants was placed on the table with no drop-‐off on the other side, and mothers were instructed to pose a fearful expression when their infant reached the middle of the table. All infants crossed the table, of which 83% did so without referencing their mothers at all. This study
suggests that infants only reference when the situation is ambiguous enough and are influenced in their behaviour by maternal emotional signalling (Sorce et al., 1985).
A different study using the visual cliff in a social referencing paradigm, examined the strength of different types of cues coming from mothers (Vaish & Striano, 2004). Expressing only positive emotional signals, mothers encouraged their infants to cross the visual cliff using three different types of cues: facial expressions only, vocal expressions only and a third condition of facial and vocal cues simultaneously. Infants crossed the cliff in all three conditions, but crossed the cliff faster in the voice plus vocal condition and vocal only condition compared to the voice only condition, suggesting that a vocal cue is more potent than a facial cue alone.
This study
In the current study it was hypothesized that there is a relation between parental anxiety and infant anxiety and that parental anxiety is related to parenting behaviours. It was furthermore hypothesized that parenting behaviours have different effects on infant anxiety in the visual cliff paradigm depending on parent gender due to evolutionary based roles. According to evolutionary hypotheses, fathers are specialized in being more challenging whilst mothers tend to be more caregiving (Möller et al., 2013). When parents follow up on these models, their role in the socialization of their offspring can be a protective factor in their child’s development of anxiety. When they do not, possibly due to their own anxiety, this may become a risk factor for the development of child anxiety (Bögels & Perotti, 2010; Bögels & Phares, 2008).
In this study, an explanation for the intergenerational transmission of anxiety through a possible mediating role of parenting behaviours is sought. Challenging, overprotective and overcontrolling parenting behaviours of fathers and mothers are linked to the anxiety displayed by parent and infant during the visual cliff experiment. Parenting behaviours of fathers and mothers were assessed during a free play task conducted prior to the visual cliff task. Parenting behaviours were also assessed by means of questionnaires. Although the parenting behaviour measures were retrieved from a different task than the anxiety measures, it was assumed that the parenting behaviours displayed during the free play task were comparable to the behaviours the
infants have grown accustomed to from their parents in their home-‐setting (Gardner, 2000).
For mediation to occur, four conditions should hold (Baron & Kenny, 1986). First, the predictor variable parental anxiety should significantly predict the mediator variable parenting behaviour. Second, the predictor variable should also significantly predict the proposed outcome variable infant anxiety. Third, the mediator variable should significantly predict the outcome variable and fourth; the predictor variable should significantly predict the outcome variable less strongly with the mediator variable present in the model. These conditions are tested in the following research questions.
The first question addressed is the following. Is parental anxiety associated with
challenging, overprotective and overcontrolling parenting behaviours? It was
hypothesized that anxiety of the parent is associated with the challenging, overprotective and overcontrolling parenting behaviours they might show. Parents with higher rates of anxiety might show more overprotective and overcontrolling behaviours and less challenging parenting behaviours than parents with lower levels of anxiety.
The second question is: Is parental anxiety associated with infant anxiety? It was hypothesized that anxiety of the parent is positively associated with infant anxiety. Thus, more parental anxiety was expected to be related to more infant anxiety and less parental anxiety was expected to be related to less infant anxiety.
Thirdly: Is parenting behaviour associated with infant anxiety? It was hypothesized that challenging, overprotective and overcontrolling parenting behaviours are associated with the anxiety of the child. Challenging parenting behaviours were expected to relate negatively to infant anxiety: more challenging parenting behaviours were thought to be associated with less infant anxiety. Overprotective and overcontrolling parenting behaviours were expected to relate positively with infant anxiety: more overprotective and overcontrolling parenting behaviours were thought to be associated with more infant anxiety.
Fourth: Does parenting behaviour mediate the relationship between parental
anxiety and infant anxiety? It was hypothesized that the three different parenting
behaviours (challenging, overprotective and overcontrolling parenting), mediate this relationship. In other words, the positive relationship between parental and infant anxiety is better explained by its relationship to parenting behaviours. More anxiety in
parents is associated with a decrease in their challenging behaviours and increase in their overcontrolling and overprotective behaviours. In turn, less challenging and more overcontrolling and overprotective parenting behaviours are expected to be associated with infants showing more anxiety.
Fifth: Are these relations different for fathers and mothers? It was hypothesized that these mechanisms work in different ways for fathers and mothers due to evolutionary specializations of men and women (Bögels & Phares, 2008; Möller et al., 2013). Paternal challenging behaviours were expected to be associated with less infant anxiety, while maternal challenging behaviours were expected to not be associated with infant anxiety or even, in accordance to Majdandžić et al. (2014), in the opposite way. Paternal overprotective and overcontrolling parenting behaviours were expected to be associated with more infant anxiety and so were maternal overprotective parenting behaviours. Therefore, all analyses will be done separately for fathers and mothers and challenging, overprotective and overcontrolling parenting behaviours (see Figure 1 and 2). In accordance with these hypotheses, paternal anxiety was expected to have a stronger effect on infant anxiety than mothers, through the mediating role of parenting behaviours. Paternal parenting behaviour Maternal parenting behaviour Paternal
anxiety anxiety Infant Maternal anxiety anxiety Infant
Figure 1. Mediation by paternal challenging
behaviour. Figure 2. Mediation by maternal challenging behaviour.
Method Participants
Participants were 81 infants (40 girls and 41 boys) between 10 and 15 months who were randomly invited to the lab once, with father (N = 41) or mother (N = 40). In table 1, characteristics of infants and their parents that visited the lab are illustrated. First, total and mean scores are presented of all participants. Next, total and mean scores are divided for infants participating with mother or with father. Visiting fathers and
mothers differed only in their fulltime work status, with a significantly larger amount of fathers working fulltime than mothers, which is characteristic for the situation in the Netherlands (Wielers & Raven, 2013).
Table 1.
Characteristics of participants
Infants Total N = 81 With mother N = 40 With father N = 41 t (df) or Χ2 (df) Girls – N (%) 40 (49,4%) 19 (47,5%) 21 (51,2%) .11 (1)
Age infant; in months – M (SD) 11,88 (1,25) 11,88 (1,30) 11,87 (1,21) .03 (79)
Parent Total N = 81 Mothers N = 40 Fathers N = 41 t (df) or Χ2 (df)
Age parent; in years – M (SD) 35,66 (4,82) 34,84 (3,96) 36,39 (5,49) -‐1.42 (76) Number of children – M (SD) 1,45 (.77) 1,36 (.67) 1,54 (.85) -‐1.03 (76) Married/living together – N (%) 75 (92,6%) 37 (92,5%) 38 (92,7%) .00 (1) Born in Netherlands – N (%) 61 (75,3) 29 (72,5%) 32 (78,0%) .34 (1) Education level – M (SD) a 7,23 (1.08) 7,23 (1.20) 7,23 (1,27) .00 (76)
Working fulltime – N (%) 31 (38,3%) 5 (12,5%) 26 (63,4%) 22.22 (1)*
Note: Chi square tests were performed for proportions, independent samples t-‐tests for means. a On a scale from 0
(primary education) to 8 (university). * p < .001.
Procedure
All participants were recruited by means of handing out flyers by childcare centres, and an information letter distributed by the Amsterdam municipality to parents who just had a baby. Once the infant was able to crawl at 10 to 15 months of age, parents were contacted to schedule an appointment to visit the lab. Parents were sent a confirmation letter of the scheduled appointment, information about the lab visit and the first set of questionnaires (one for each parent) via mail. Both parents filled out questionnaires, but only one parent visited the lab due to memory effects of the infant on the cliff. The visiting parent was asked to bring both individually filled out questionnaire booklets to the lab. The lab visit took place at the Research centre of Child Development and Education where, after a short introduction and signing of the informed consent, the following tasks where conducted in the observation room: five minutes of free play with toys, five minutes of free play without toys and the visual cliff experiment. All tasks where filmed with three video cameras, which were operated from the registration room behind a one-‐way screen. The tasks in the lab lasted approximately 30 minutes. After the experimental tasks the second and final set of questionnaire booklets (one for
home and send back once filled out. The visiting parent was reimbursed all travel expenses, received a copy of all video recordings of the experimental tasks and could choose between a compensation of €10,-‐ or a donation of the same amount to Orange Babies, an organisation helping HIV infected mothers and their babies in Africa. The infant received a small toy.
Measures
Parenting behaviours – self-‐reported
For assessment of self-‐reported parenting behaviours several scales of the Comprehensive Parenting Behaviour Questionnaire 1-‐year version (CPBQ-‐1; Majdandžić, De Vente, & Bogels, 2013) were used. All items were rated on a 5-‐point Likert-‐scale, ranging from 1 (totally not applicable) to 5 (Completely applicable). The questionnaire consists of the following scales: Challenging behaviour, Overprotective behaviour, Overcontrolling behaviour, Warmth, Negativity, Negative discipline and Positive discipline. In this study, only Challenging behaviour (46 items), Overprotective behaviour (18 items) and Overcontrolling behaviour (10 items) were used. The scale for Challenging behaviour consisted of items regarding: teasing, rough-‐and-‐tumble play, encouragement of risk taking, social daring, competition, encouragement of assertiveness, and challenging modeling. The scale for Overprotective behaviour consisted of items regarding limiting exposure of new objects, people or situations to the infant and the scale for Overcontrolling behaviour consisted of items regarding needlessly helping the infant or interfering with the infants behaviour regardless of their needs. Reliability ranged from acceptable to good for most variables. α’s for Challenging behaviour were .82 for mothers and .79 for fathers. For overprotective parenting, α’s were .75 for mothers and .63 for fathers. For overcontrolling parenting α’s were 48 for mothers and .50 for fathers. This means internal consistency was quite low but used for further analysis regardless.
Parenting behaviours – observed
Procedure. During two sessions of free play, lasting five minutes per session, parenting
behaviour towards their infant was scored in an unstructured setting while parent and infant could get used to the laboratory setting in a comfortable manner. The first
session was free play with toys, entailing a box with hand puppets, a small book that makes sounds, a soft ball and a set of magnet blocks. Parent and infant were free to choose which toys they played with. The second session was free play without toys. Here, parent and infant where given only a set of cushions to play with. Each session started the moment the test leader closed the door after leaving the observation room. Instructions by the test leader entailed parent and child could play with/without toys as they were used to. Each session finished after 5 minutes.
Coding. A total of 8 different parenting dimensions where coded in intervals of one
minute using a 5-‐point Likert scale. The coding protocol is based on the Meso Behavioural Rating System for Families with young children (MeBRF, Mahoney et al., 2000) but adapted for use in this particular study. In this study only the parenting dimensions challenging parenting behaviour, overcontrolling and overprotecting
parenting behaviour where used.
The dimension of challenging parenting behaviour is new and based on the theory of Paquette (2004), elaborately described earlier in this paper. Examples of challenging parenting behaviour during both free play tasks are: teasing the infant with toys or cushions, mildly scaring the infant with or without toys, hide-‐and-‐seek games, lifting the infant up high, challenging the infant to walk, tickling and biting games. These behaviours are given scores indicating levels of frequency and intensity, ranging from 1 (none or minimal) to 5 (very often or very high).
The dimension of overcontrolling parenting behaviour is partly based on the Intrusiveness scale from Erickson (Egeland, Erickson, Clemenhagen-‐Moon, Hiester, & Korfmacher, 1990; Erickson, Sroufe, & Egeland, 1985), and measures the extent to which parents help or intervene in the child’s activity when this is not necessary. Examples of overcontrolling parenting behaviour are: choosing which toy to play with for the infant, taking away toys from the infant to play with the toys themselves, offering too many toys to the child at once, feeding or hugging the child when he/she doesn’t want to. These behaviours are given scores indicating levels of frequency and intensity, ranging from 1 (none or minimal) to 5 (very often or very high).
The dimension of overprotective parenting behaviour is partly based on the Involvement scale by Hudson & Rapee (2001), and measures the extent to which parents protect their child in an overly manner. This can be seen in how the parent physically handles his/her infant and remarks the parent makes. Examples are: cleaning
the infants face when this is not important, making remarks about tiredness of the child, excessively holding the child when trying to walk or crawl to prevent him/her from falling, excessively making remarks like ‘watch out’, ‘be careful’, ‘not so fast’. These behaviours are given scores indicating levels of frequency and intensity, ranging from 1 (none or minimal) to 5 (very often or very high).
For the three different parenting dimensions (for both tasks separately), a final score was computed by averaging across the five interval scores. Then, a final mean score for challenging behaviour was made by averaging the means of challenging behaviour during both tasks. The same was done for a final score of overcontrolling and overprotective parenting behaviour. Internal consistency of the two variables for challenging parenting was .25, for the two variables of overcontrolling parenting .12 and for the two variables for overcontrolling behaviour .42. Internal consistencies for the observed parenting behaviours are very low, but both tasks can elicit different types of parenting behaviours to so they form a robust measure together.
Coding was carried out by two trained students. Of the observed data, 20% was double coded to obtain inter-‐rater reliability (IRR). IRR for challenging parenting behaviour was .97 during both tasks, .81 for overcontrolling parenting during free play with toys and .78 during free play without toys and IRR was 1 for overprotection during both free play with toys and without toys, indicating high levels of agreement.
Parental and infant anxiety – observed
Procedure. The visual cliff is a table covered in a plexiglass surface, split in two sides. A
shallow side, where the plexiglass rests directly on a red and white chequered surface, and a shallow side, where the chequered surface drops 30 cm down while the plexiglass surface continues on the same height. A 20 cm high border, also made of plexiglass, surrounds the visual cliff table. The test leader introduced the task with an instruction. The parent was explained to place their infant on the table with their feet touching the back border of the table and subsequently walk over to the other side of the table to stand on the marked spot on the floor. The parent was asked not to make any remarks about the table until he or she was standing on the spot and the child had looked at him/her. After the child had looked at the parent, which the test leader cued to the parent by raising her hand, the parent could start encouraging their child to crawl over to them. The parent was not allowed to cross his/her arms over the table or tap the
surface to indicate the safety of the table. The test leader was present in the room and stood behind a pillar during the task, making no eye contact with parent or child but keeping the safety of the situation in check. The task ended when one of the following situations occurred: 1) the child had crossed the deep side and touched the border, 2) the parent picked up the child at the end of the deep side, 3) when 10 minutes had passed without the child reaching the end of the cliff, or 4) when the child became too upset or the parent indicated no longer wanting to proceed with the task.
Coding parental anxiety. The measure for parental anxiety consists of a
combination of three measures of observed anxiety. For the coding of facial expression of anxiety the AFFEX system (Izard, Dougherty & Hembree, 1983) was used. Facial expressions (e.g. big eyes, lifted eyebrows, stiff lips), bodily expressions (e.g. tense posture, raised shoulders, nervous movement) and verbal expressions of anxiety (tone and content where both taken into account, for example ‘oh oh’, ‘watch out!’) were scored every 10-‐second interval on a scale from 0 to 3. Higher scores indicated a higher frequency or intensity of each expression of anxiety (0 = absence of the behaviour; 1 = light expression of the behaviour; 2 = clear expression of the behaviour; 3 = clear and intense expression of the behaviour). The scoring started the moment the parent stood on the marked spot behind the deep side of the cliff and the infant had looked at the parent, when social referencing could commence. All interval scores on each of the three expressions of anxiety were averaged for a single final score for each variable. Internal consistency of the three variables (α = .72) was good; a mean score for parental anxiety was made from the three averaged variables.
Coding infant anxiety. For the coding of facial expression of anxiety the AFFEX
system (Izard, Dougherty & Hembree, 1983) was used. Infant anxiety is similarly a combined measure of three separate measures of observed anxiety. Facial expressions (e.g. wide eyes, stiff lips), bodily expression (e.g. stiff muscles, head ducked in shoulders, holding breath) and vocal expressions of anxiety (e.g. whining, crying) were scored every 10-‐second interval on a scale from 0 to 3. Higher scores indicated a higher frequency or intensity of each expression of anxiety (0 = absence of that behaviour; 1 = light expression of that behaviour; 2 = clear expression of that behaviour; 3 = clear and intense expression of that behaviour). The coding of the infant variables started the moment the parent placed the infant on the cliff and the infant touched the cliff in any way. All interval scores on each of the three expressions of anxiety were averaged for a