Child Psychiatry & Human Development

, Volume 42, Issue 3, pp 270–290

An Investigation of Control Among Parents of Selectively Mute, Anxious, and Non-Anxious Children

Authors

    • Department of Psychiatry & Behavioral NeurosciencesMcMaster University
    • Child and Youth Mental Health Program, Chedoke SiteMcMaster Children’s Hospital, Hamilton Health Sciences
  • Mary Ann Evans
    • Department of PsychologyUniversity of Guelph
  • Angela E. McHolm
    • Department of Psychiatry & Behavioral NeurosciencesMcMaster University
  • Charles E. Cunningham
    • Department of Psychiatry & Behavioral NeurosciencesMcMaster University
  • Matilda E. Nowakowski
    • Department of Psychology, Neuroscience & BehaviorMcMaster University
  • Michael Boyle
    • Department of Psychiatry & Behavioral NeurosciencesMcMaster University
  • Louis A. Schmidt
    • Department of Psychology, Neuroscience & BehaviorMcMaster University
Original Article

DOI: 10.1007/s10578-010-0214-1

Cite this article as:
Edison, S.C., Evans, M.A., McHolm, A.E. et al. Child Psychiatry Hum Dev (2011) 42: 270. doi:10.1007/s10578-010-0214-1

Abstract

The authors examined parent–child interactions among three groups: selectively mute, anxious, and non-anxious children in different contexts. The relation between parental control (granting autonomy and high power remarks), child factors (i.e., age, anxiety, verbal participation), and parent anxiety was investigated. Parental control varied by context but parents of children with SM were more controlling than parents in the comparison groups in all contexts. Regression analyses indicated that child and parent anxiety predicted parental control, with increased anxiety associated with increased control. Older child age predicted less parent control. Group categorization moderated the relation between age and high power remarks, such that age was not a significant predictor for children with SM. Finally child-initiated speaking predicted high power remarks over and above other variables. These results support previous theories that parents take over for their children when they fail to meet performance demands, especially when the child or parent is anxious.

Keywords

Selective mutismAnxietyParental controlParent–child interactions

Introduction

Selective mutism (SM) is characterized by the failure to speak in certain situations, while speaking normally in other situations [1]. Parents of children with SM describe them as typical children in the home with regard to their comfort with speaking and their communication skills, but as reticent and avoidant of speaking in other social situations such as parties, and school. Since conversation with others is an everyday occurrence, parents of children with SM are often placed in distressing situations when their child is unable to conform to conversational expectations.

Relatively little is known about the development and perseverance of SM or its relation to parenting behaviors. In recent years, increasing attention has been given to conceptualizing and investigating the family as a transactional system, influenced by and influencing each other. Rapee and Spence [2] proposed that children elicit protective controlling behaviors from their parents that can act to maintain or increase a child’s withdrawal or avoidance of the feared situation. Wood and colleagues [3] summarized the literature on parental control and reported that observational findings consistently demonstrate that anxious children tend to have parents (primarily mothers) who are relatively more controlling than mothers of non-anxious children [for example see 48]. To further support the perspective of the transactional nature between parents and children, Rubin and Burgess [9] suggest that expressions of social fearfulness among socially withdrawn children evoke parental feelings, such as concern and frustration, and that these may be exacerbated when parents are anxious themselves.

The present study examined whether parents of children with SM demonstrate greater control with their children than parents of a clinical sample of anxious children without SM and of non-anxious children. Secondly, we examined the extent to which parent anxiety and child characteristics would predict parental control. The two types of control groups have been included in previous research on SM [e.g., 10]. Inclusion of a clinical control sample of anxious children allowed for a comparison with children who share similar temperamental characteristics and with parents who are also likely to experience distress regarding their child. Inclusion of this control group was also thought to guard against common differentiating factors between clinical and normal samples (e.g., SES, parental interest in study participation). Primary caregivers and their children were observed in four contexts during a laboratory visit: (1) free-play, (2) discussion of the child’s most recent birthday, (3) preparation for the child to do a birthday speech, and 4) presentation of the child’s birthday speech.

The Role of Child Characteristics

Within the field of developmental psychology, it has long been recognized that children are active participants in their environment, both reacting to and eliciting other’s behavior [11]. Thus, it seems likely that the behaviors exhibited by an anxious child when in an anxiety provoking situation will elicit different parenting behaviors than would the behaviors of a non-anxious child. Rubin and Burgess [9] proposed that expressions of social fearfulness exhibited by a child will naturally evoke parental feelings of concern, sympathy, and perhaps frustration. Thus, in response to a child’s distress, a parent may be drawn to solve the child’s social dilemma or take over to relieve both their own and their child’s distress. Such ideas have received empirical support. A longitudinal study by Rubin, Nelson, Hastings, and Asendorpf [12], for instance, indicated that parents’ perceptions of their child’s social wariness at age two predicted their lack of encouragement of independence with their child at age four. Thus, parents’ past experiences with their children in anxiety-provoking situations appeared to guide their current behaviors.

It is also the case that parenting behaviors change in accordance with the developmental level of the child [13]. Research suggests that parental beliefs and attitudes toward social withdrawal change as children grow, as does their behavioral response. For instance, in a study on maternal control among withdrawn children, Mills and Rubin [14] found that mothers demonstrated greater control with younger children than with older children.

The Role of Parental Characteristics

One should also consider characteristics of parents that may be associated with their own behavior. The heritability of anxiety ranges from 25 to 50% in the literature [15, 16]. As such, parents of children with anxiety have an increased likelihood of exhibiting anxiety themselves. Developmental models of anxiety, such as that by Manassis and Bradley [17], suggest that parent anxiety may interfere with parenting practices and lead to behaviors that promote anxiety in children. Anxious parents may be especially prone to becoming over-concerned with their child’s anxious behaviors, leading them to overprotection.

Some research exists to support the association between parental anxiety and parental control. After observing parent–child conversations, Whaley, Pinto, and Sigman [18] found that compared to non-anxious mothers, anxious mothers catastrophized more, criticized more, granted less autonomy, and displayed significantly less warmth and positivity with their children, regardless of the child’s anxiety level. However, the authors also reported that maternal anxiety status was more predictive of low maternal warmth and positivity, whereas child anxiety status was more predictive of maternal granting of autonomy. Thus, both child and parent anxiety warrant consideration in understanding parent behavior.

The Role of Context

The context in which the parent–child interaction occurs may elicit different responses from parents. Specifically, Dennis [19] suggests that situations involving challenges versus situations involving play will place distinct demands on parents and on the child. Situations can also elicit different responses by nature of the comfort and familiarity for the parent and child. Such variation has been observed within the laboratory. For instance, in a study by Ginsburg, Grover, Cord, and Ialongo [20], higher rates of controlling behavior, anxious behavior, and criticism were observed during a structured task (i.e., copying designs using an Etch-A-Sketch) in comparison to during free play. Ginsburg and colleagues speculated that the increases in the aforementioned domains were associated with the experience of stress, as it related to the situational components of time pressure, difficulty of the task, and the evaluation component. The authors postulated that variation in parental behavior between different observational contexts may help explain contradictory findings on parental behavior within the literature.

The Present Study

We investigated the relation in laboratory tasks between child and parent characteristics and two indices of parental control: the extent to which they grant autonomy and extent to which they engage in high power remarks demanding a verbal response. These indices of control have been used separately in previous research but not examined together. The primary child factors explored were: child verbal participation, child anxiety (parent report and observed anxiety in the laboratory), and age. Parent anxiety was indexed by self-report and observed ratings. Of the four observational contexts in this study, the free play and birthday speech situations were chosen for analysis, given the contrast in demands on the children and parents, with the free play being an unstructured situation and the speech having specific verbal participation and performance demands.

Based on relevant research and its application to the current study, the following hypotheses were tested:
  1. 1.

    Parents of children with SM would employ more control with their children than would parents of anxious children or non-anxious control children, given that parents of anxious children and of shy children demonstrated greater control in comparison to parents of non-clinical children [3].

     
  2. 2.

    Parental behaviors would vary in accordance with the laboratory situation. Specifically, parents would engage in greater levels of control in the birthday speech context in comparison to the free play context. Rubin, Cheah, and Fox [21], for example, reported that parents engaged in a greater frequency of control behavior during a stressful situation in comparison to a free play situation.

     
  3. 3.

    Select parent and child characteristics would predict parental control based on the premise that parents’ behaviors are not independent of their children. Specifically, parents would engage in greater levels of control when children were anxious, when parents themselves were anxious, and when children demonstrated less verbal participation. We also hypothesized that parents would engage in increasingly greater levels of control with younger children. The group categorization of SM versus control children (including both anxious and non-anxious children) was also explored as a moderating variable for the aforementioned factors. Given that context differences were expected in parental control, the association between child and parent characteristics and control was explored separately for free play and birthday speech situations.

     

Method

Participants

A sample of 63 children (28 males, 35 females, ages 4–13 years) and their primary caregivers participated. Children were referred to a participating children’s mental health referral site. Families from the Department of Psychology’s Child Database at McMaster University were also recruited. This database contains contact information for mothers and their healthy infants.

Of the caregivers who participated, 57 (90%) were mothers and 6 (10%) were fathers. In all cases, with the exception of one, the caregivers were biologically related to the child. All families reported English as the primary language spoken in the home. English was the first language for 58 (92%) of the primary caregivers. Parents reported gross annual family incomes from less than $15 000 (3% of the sample) to above $100 000 (24% of the sample). Five (8%) of the parents reported earnings of $15 000–$30 000, 5 (8%) earnings of $30 000–$45 000, 4 (6%) earnings of $45 000–$60 000, 16 (25%) earnings of $60 000–$75 000, and 16 (25%) earnings of $75 000–$90 000.

SM Group

Children were placed into the SM group if they were reported as failing to speak in at least two situations on the parent or teacher Speech Situations Questionnaire (SSQ) [22] or in two situations on the SM Brief Child and Family Phone Interview (BCFPI) [23] Screener. The aforementioned questionnaires must also have indicated that these children spoke in other situations and evidenced mutism for at least 1 month. No cases met exclusion criteria for speech and language disorders (according to parent report) or limited familiarity with English. These criteria to classify children with SM have also been reported elsewhere [24].

Twenty-one participants (13 female, 8 male) met criteria for the SM group. Primary caregiver report on the internalizing section of the Computerized Diagnostic Interview Schedule for Child (C-DISC IV) [25] was used to examine the number of internalizing disorders for which the participants met criteria. Children were not excluded from the SM group if the C-DISC-IV revealed comorbid diagnoses. The C-DISC-IV was not available for 2 (9%) of the children in the SM group. Of the 21 children with SM for which the C-DISC-IV was available, 7 (33%) had one or more comorbid internalizing disorders. These included: specific phobia (n = 3), social phobia (n = 3), and separation anxiety disorder (n = 3).

Anxiety Group

Children in the anxiety group met criteria for one or more anxiety disorders according to the C-DISC-IV. Seventeen participants (8 female, 9 male) met criteria for this group. Anxiety diagnoses included specific phobia (n = 13), generalized anxiety disorder (n = 3), panic disorder (n = 1), social phobia (n = 1), separation anxiety disorder (n = 3), obsessive compulsive disorder (n = 3), and posttraumatic stress disorder (n = 1). Six (35%) of the children met criteria for more than one anxiety diagnosis. None of these children met criteria for SM.

No-anxiety Group

Children from the Department of Psychology’s Child Database at McMaster University were placed into the no-anxiety group if they did not meet criteria for any internalizing disorders according to the C-DISC-IV. Twenty-five (14 female, 11 male) participants met criteria for the no-anxiety group.

Procedure

Following arrival at the Child Emotion Laboratory, the parent(s) and child were briefed about the procedures. Written consent was obtained from the parent(s) and written assent was obtained from the child. All procedures were approved by the McMaster University Health Sciences Research Ethics Board.

For the laboratory observation, the experimenter explained to parents that instructions would appear on a laptop screen for them to read, each preceded by a chime. There was a set of instructions for each of four observation segments. In the first 5 min, free play, parents were instructed that they could do anything they liked. In the second context of 2 min, parents were to talk with their child about the child’s last birthday in order to cue the child’s memory. In the third context of 5 min parents were to tell their child that they would have to stand in front of a camera in a few minutes and talk about their last birthday so that other children could later watch the tape and hear all about it. In the last segment of 3 min, the birthday speech, parents were instructed to have their child stand in front of the camera and talk about their last birthday. The first and last segments were of particular interest in that the first was perhaps closest to free play interactions with the parent at home, and the last closest to social situations where the parent is present and the child is expected to talk. These two situations have different demands, with free play being an unstructured context and the speech having verbal performance demands. During the laboratory visit, parents also completed a set of questionnaires.

Measures

Caregiver and Child Background Information

Parents reported their demographic characteristics (e.g., family composition, languages spoken in the home, family status, employment, and income level). They also reported on their child’s age, gender, and grade, currently prescribed medications, history of SM (e.g., age of onset) and treatment history.

Child Characteristics

Parents also completed the Screen for Child Anxiety Related Emotional Disorders (SCARED): Parent version [26]. This questionnaire is a screening tool for childhood anxiety disorders. The total score on the SCARED has an internal consistency reliability of .90 [27]. Total scores on the SCARED were calculated for each participant.

After the laboratory visit, the computerized version of the C-DISC-IV [25] was completed by the primary caregiver over the phone with graduate level psychology students trained by a Ph.D. level clinical psychologist and whose reports were verified by the clinical staff. This interview includes diagnostic criteria as specified in the DSM-IV and ICD-10. Two diagnostic sections of the interview were administered to determine whether the children met diagnostic criteria of any anxiety disorder or mood disorder. The diagnostic reliability (K statistics) of the C-DISV as published in the manual was 0.56 for anxiety disorders and 0.55 for mood disorders [25].

Parent Self-report Measures

Parents completed the Social Phobia and Anxiety Inventory (SPAI) [28], a 43-item self-report scale that measures the severity of symptoms of social anxiety for adolescents and adults. This questionnaire has an internal reliability of 0.96 for the social phobia scale and a test–retest reliability of 0.86. Total scores on the SPAI were calculated for each participant. Higher scores represented greater social anxiety.

Parents also completed the Beck Anxiety Inventory (BAI) [29], a 21-item self-report scale that measures the severity of anxiety in adolescents and adults. The BAI has an internal reliability of 0.92 and a test–retest reliability of 0.75. Total scores on the BAI were calculated for each participant. Higher scores represented greater anxiety.

Observational Coding of Control

Wood and colleagues [3] defined parental control as a pattern of excessive regulation of children’s activities and routines, autocratic parental decision-making, overprotection, or instruction on how to think or feel. This form of behavioral control then essentially focuses on granting of autonomy. More specific to speech situations is conversational control [see 3032], reflected in the use of high power remarks that demand a response on the part of the child. The relation of child and parent characteristics to both forms of control was coded here. First, global ratings of parental granting of autonomy and child and parent anxiety were made following Siqueland, Kendall, and Steinberg’s [8] previously published coding schemes (also employed by Whaley, Pinto, and Sigman [18] and Moore, Whaley, and Sigman [33]. Second, the nature of the verbal interaction was coded by drawing upon two previously published coding schemes: (1) Evans and Bienert’s [34] coding scheme for high and low power t-units in parent–child interactions and (2) Mash, Terdal, and Anderson’s [35] response class matrix coding system [used by 36].

Global Ratings

Laboratory observations were segmented into one-minute intervals, and behavior within segments was rated for parental granting of autonomy, parental anxiety, and child anxiety. Average ratings across all intervals and for each of the two contexts were calculated for each group.

Ratings of the granting of autonomy were intended to capture the extent to which the parent constrained or encouraged the child’s individuality through the use of inductive disciplinary techniques [see 8]. A global rating in each one-minute interval was made using a 5-point Likert scale, from 1 (discourages autonomy) to 5 (encourages autonomy) and was based on how much the parent: (a) solicited the child’s own opinion, (b) tolerated differences of opinion, (c) acknowledged and demonstrated respect for the child’s views, (d) avoided judgmental reactions to the child’s views, (e) encouraged the child to think independently, and (f) used explanation and other inductive techniques.

Global ratings of parent anxiety and of child anxiety in the dyadic interaction were based on the following behaviors: (a) fidgeting, (b) nervous smiling or giggling, (c) tense posture, (d) tense facial expression, (e) leg shaking, (f) change in voice, (g) crying, (h) decreased eye contact, and (i) avoidance behaviors. The 5-point scale ranged from 1 (very relaxed) to 5 (very anxious).

The first author and two Honours-level undergraduate students, blind to group membership, rated the videotapes. The raters were trained on the definitions of the identified constructs based upon written descriptions, group discussions, and observations of videotape examples. The observers viewed the videotaped interactions and independently rated the parents and children at 1 min intervals. Reliability values were calculated on 25% of the tapes. The interrater reliabilities ranged from 0.71 to 0.85, using intraclass correlations (ICCs).

Coding of Conversational Control

The verbal interactions between children and their parents were first segmented into minimal terminable units (t-units) [37], defined by Hunt as a main clause and any subordinate clause attached to it. Each parent t-unit was then coded as either a high or low power remark. High power remarks included: (1) commands/directives (e.g., “Say your name”), (2) choice questions (e.g., yes/no or choice between two alternative answers offered), and (3) product/process questions (e.g., wh-type questions and how questions). Low power remarks included: (1) personal contributions (e.g., “I am going to color the sky blue”) and (2) acknowledgments or phatics by which the parent interestedly acknowledged the child’s contribution (e.g., “Uh hmmm;” “What a beautiful picture”). The proportion of high power t-units out of the total parent t-units was then calculated. Each child t-unit was coded as either 1) spontaneous or 2) solicited by the parent. Both the total number of child t-units and the total number of spontaneous t-units were calculated for each participant.

The first author and one graduate student rated the videotapes. The raters were trained on the definitions of the identified constructs based upon written descriptions, group discussions, and observations of videotape examples. The observers viewed the videotaped interactions and independently rated the parents and children at one-minute intervals. The interrater reliabilities, calculated on 25% of the tapes using Cohen’s kappa ranged from 0.88 to 0.90 for identifying a t-unit, for categorizing a parent remark as high vs. low power and for categorizing child remarks as, solicited or unsolicited by the parent).

Results

Preliminary Analyses

Prior to the analyses, standard univariate and multivariate tests of statistical assumptions were conducted. Extreme values resulting from univariate tests for outliers were retained for examination as multivariate outliers. Values were suspected of being multivariate outliers if influence values neared or exceeded 1.0. No influence values within the current data set approached this cut-off value. However, audiovisual problems and incomplete questionnaires resulted in some missing data. Thus, the sample size varies slightly in the analyses below.

Preliminary analyses revealed no significant group differences in child age or gender. There was a significant group difference in gross annual family income F(2, 60) = 7.22, P < 0.01. Post-hoc comparisons using the Tukey HSD test indicated that the mean income for the no-anxiety group was higher than for the SM and anxiety groups, which did not differ.1 Table 1 provides means and standard deviations for the study’s variables. Distributions on both the BAI (M = 4.51, SD = 5.21) and SPAI (M = 208.26, SD = 138.24) were positively skewed and reflected very low self-reported anxiety. As such, data from these measures were not used. SCARED scores were not significantly correlated with granting of autonomy in the free play or birthday speech situation or high power remarks in the free play or birthday speech situation. As such, these scores were not included in the regression analyses.
Table 1

Descriptive statistics of the observed variables for the SM, anxiety, and no-anxiety groups and the total sample

 

Free play

Birthday speech

SM

Anxiety

No-anxiety

All

SM

Anxiety

No-anxiety

All

Granting of autonomy

 M

3.81

4.08

4.24

4.06

2.51

3.60

3.91

3.36

 SD

0.90

0.89

0.71

0.83

0.91

1.02

1.07

1.17

High power T-Units

 M

0.47

0.39

0.35

0.41

0.68

0.59

0.58

0.62

 SD

0.11

0.11

0.08

0.11

0.10

0.11

0.20

0.15

Total speaking

 M

4.80

7.09

7.68

6.51

5.18

8.61

9.06

7.57

 SD

3.75

1.73

3.04

3.25

4.21

2.53

3.48

3.90

Spontaneous speaking

 M

2.16

2.72

3.50

2.80

1.15

3.98

4.98

3.40

 SD

1.99

1.37

1.92

1.86

1.72

2.18

3.82

3.22

Observed child anxiety

 M

1.81

1.45

1.31

1.51

3.35

2.65

2.07

2.64

 SD

1.18

0.79

0.66

0.91

1.38

1.06

1.07

1.28

Parent anxiety

 M

1.46

1.18

1.06

1.22

1.82

1.27

1.19

1.41

 SD

0.69

0.56

0.17

0.52

0.98

0.63

0.50

0.76

Parental Control

Observed parental control, measured by observed granting of autonomy and the proportion of high power remarks, was compared among the three groups. Context (i.e., free play, birthday discussion, birthday speech preparation, and birthday speech) was a repeated measure variable. Therefore, a Split-Plot ANOVA was conducted and the sphericity violations were corrected for using the Greenhouse-Geisser correction. The ANOVA homogeneity of variance assumption was fulfilled in all cases, with the exception of granting of autonomy in the birthday discussion and the proportion of high-power remarks in the birthday speech. Given the near equal sample sizes, this was not thought to have invalidated the results.

With regard to observed granting of autonomy, there was a significant effect for group, F(2, 60) = 13.97, P < .01, η2 = 0.32. Pairwise post-hoc comparisons indicated a difference in observed granting of autonomy between the SM and anxiety groups, F(3, 180) = 3.53, P < 0.05 and between the SM and no-anxiety groups, F(3, 180) = 4.89, P < 0.01. There was no difference between the anxiety and no-anxiety group. There was also a main effect for context, F(2, 159) = 14.13, P < 0.01, η2 = 0.19. Post-hoc Tukey comparisons indicated that parents granted more autonomy in the free play context than in the birthday speech preparation context, q (4, 180) = 4.74, P < 0.01 and than in the birthday speech context, q = 3.68, P < 0.05. There was no context by group interaction, F(5, 159) = 1.43, P > 0.05, η2 = 0.05.

With regard to the proportion of high-power remarks, there was a significant effect for group, F(2, 52) = 11.26, P < 0.01, η2 = 0.30. Pairwise post-hoc comparisons indicated a difference in the proportion of high power remarks between the SM and anxiety groups, F(3, 156) = 5.5, P < 0.01 and between the SM and no-anxiety groups, F(3, 156) = 7.00, P < 0.01. There was no difference between the anxiety and no-anxiety groups, F(3, 156) = 4.50, P > 0.05. Again, there was an effect for context, F(2,147) = 44.41, P < 0.01, η2 = 0.46. Post-hoc comparisons using the Tukey’s HSD test indicated that there was a significantly lower proportion of high power remarks by parents in the free play context than the birthday discussion context, q (4, 156) = 4, P < 0.01, which was in turn lower than that for the birthday speech context, q = 6, P < 0.01. High power t-units were also significantly more common in the birthday discussion context than in the birthday speech preparation context, q (4, 156) = 4, P < 0.05. Finally, there was a lower proportion of high power remarks in the birthday speech preparation context than in the birthday speech context, q (4, 156) = 10, P < 0.01. There was no interaction effect, F(5, 147) = 0.83, P > 0.05, η2 = 0.03.

Child and Parent Factors

Because no differences in parental control were found between the anxious and non-anxious groups, these groups were collapsed to form a single comparison group for the next set of analyses. There was no correlation between the extent to which parents granted autonomy and spoke high power remarks in the free play context, but a negative correlation appeared in the birthday speech context. Bivariate correlations also explored the relation between each of the child variables (i.e., total speaking, spontaneous speaking, observed child anxiety, age, observed parent anxiety), and the parental control indices. Table 2 provides the intercorrelations of child and parent variables with granting of autonomy and high power remarks.
Table 2

Correlation tables for SM group (above the diagonal) and control group (below the diagonal)

Variables

Variables

1

2

3

4

5

6

7

8

9

10

11

12

13

1. Child’s age (months)

−0.28

0.27

0.14

0.17

0.17

0.24

0.17

0.07

0.16

−0.06

0.24

−0.05

2. Mean of granting of autonomy in FP

0.01

−0.03

−0.01

−0.17

0.12

0.37

0.13

0.15

−0.32

−0.25

−0.34

−0.1

3. Mean of granting of autonomy in BS

0.44**

0.05

−0.61**

−0.51*

0.48*

0.59**

0.43

0.22

−0.28

−0.45*

0.1

−0.26

4. Proportion of high power t-units in FP

−0.41*

−0.19

−0.36*

0.4

−0.52*

−0.31

−0.62**

−0.34

0.66**

0.42

0.05

0.26

5. Proportion of high power t-units in BS

−0.40*

−0.14

−0.36*

0.48**

−0.43

−0.3

−0.02

−0.01

−0.05

−0.09

−0.11

0.14

6. Spontaneous speech in FP

0.09

0.12

0.06

−0.35*

−0.32

0.47*

0.64**

0.29

−0.24

−0.12

0.43

0.11

7. Spontaneous speech in BS

0.44**

0.08

0.45**

−0.41*

−0.37*

0.04

0.47*

0.51*

−0.32

−0.61**

−0.11

−0.21

8. Total speech in FP

0.01

0.29

−0.01

−0.01

−0.17

0.64**

0.25

0.70**

0.55**

−0.49*

−0.03

−0.19

9. Total speech in BS

0.19

0.22

0.01

−0.30

0.29

0.18

0.67**

0.39*

−0.52*

−0.55*

0.05

−0.09

10. Mean child anxiety in FP

0.26

−0.34*

0.09

0.15

0.08

−0.15

−0.05

−0.19

−0.26

0.55*

0.36

0.29

11. Mean child anxiety in BS

0.19

0.04

−0.13

0.04

0.08

0.16

−0.47**

0.03

−0.26

0.36*

0.23

0.45*

12. Mean parental anxiety in FP

0.24

−0.41**

0.21

−0.03

0.09

0.05

−0.04

0.13

−0.20

0.50**

0.17

0.68**

13. Mean parental anxiety in BS

0.19

−0.09

0.04

−0.19

0.05

0.13

−0.08

0.06

−0.14

0.42**

0.37*

0.38*

P < 0.05, ** P < 0.01, *** P < 0.001

In general, correlations indicated that increasing parental control was associated with less self-initiated speaking, greater observed child anxiety, lower child age, and greater observed parent anxiety. The association between the variables differed between the SM group and the comparison groups and between the situations. Age was significantly associated with granting of autonomy and high power remarks for the comparison group but not the SM group. As such, group categorization was explored as a moderating variable for the relation between age and both indices of parental control in the regression analyses.

Parent and Child Characteristics as Predictors of Parental Control

We conducted multiple regression analyses to evaluate the contribution of child and parent characteristics in predicting parental control and to explore moderating effects of group. We standarized scale scores for the predictor and moderator variables to reduce problems associated with multicollinearity between the main effects and the interaction terms (as recommended by Frazier, [38]. Product terms representing the interaction between the predictor and moderator variables were created by multiplying together the standardized predictor and moderator variables. The order of variable entry into the regression model was as follows: Step 1, observed child and parent anxiety; Step 2, age, group, age × group; Step 3, spontaneous child remarks. Spontaneous child remarks were entered last as this variable is closely associated with the group category of SM (speaking inhibition is the hallmark characteristic of SM). A moderating effect was identified at the second step by a statistically significant change in R2 (i.e., ∆R2 = 0.01) and a significant beta weight. Table 3 displays the results of regressions predicting granting of autonomy and high power remarks in the free play situation.
Table 3

Hierarchical moderated regression analyses predicting four outcomes

 

Step 1

Step 2

Step 3

B

SE B

β

t

B

SE B

β

t

B

SE B

β

t

Outcome 1: autonomy in free play

Child anxiety

−0.24

0.12

−0.26

−1.98

−0.23

0.12

−0.26

−1.88

−0.17

0.13

−0.19

−1.33

Parental anxiety

−0.45

0.20

−0.29

−2.19*

−0.41

0.22

−0.27

−1.90

−0.51

0.23

−0.34

−2.24*

Age

    

0.00

0.00

0.10

0.64

0.00

0.00

0.09

0.55

SM groups

    

−0.14

0.23

−0.08

−0.61

−0.07

0.24

−0.04

−0.28

Age × SM groups

    

−0.01

0.01

−0.16

−1.04

−0.01

0.01

−0.17

−1.10

Spontaneous speech

        

0.78

0.06

0.18

1.29

Adjusted R2

0.20

 

0.17

 

0.18

 

R2

0.23

 

0.02

 

0.02

 

F

8.1**

 

0.45

 

1.68

 

Cumulative R2 at step 3 = 0.27, overall F(6, 52) = 3.17*

Outcome 2: high power in free play

Child anxiety

0.06

0.02

0.48

3.59**

0.06

0.01

0.46

3.98***

0.04

0.01

0.35

3.02**

Parental anxiety

−0.02

0.03

−0.08

−0.61

−0.03

0.03

−0.16

−1.29

−0.01

0.03

−0.04

−0.34

Age

    

−0.00

0.00

−0.41

−3.14**

−0.00

0.00

−0.39

−3.13**

SM groups

    

0.08

0.03

0.35

3.02**

0.07

0.03

0.28

2.44*

Age × SM groups

    

0.00

0.00

0.29

2.21*

0.00

0.00

0.30

2.44*

Spontaneous speech

        

−0.02

0.01

−0.30

−2.71**

Adjusted R2

0.18

 

0.40

 

0.46

 

R2

0.20

 

0.25

 

0.07

 

F

7.15**

 

7.99***

 

7.34**

 

Cumulative R2 at step 3 = 0.52, overall F(6, 52) = 9.36***

Outcome 3: autonomy in birthday speech

Child anxiety

−0.35

0.13

−0.37

−2.72**

−0.25

0.11

−0.27

−2.29*

−0.13

0.13

−0.14

−1.01

Parental anxiety

−0.18

0.21

−0.12

−0.85

−0.06

0.18

−0.04

−0.34

−0.09

0.18

−0.06

−0.48

Age

    

0.02

0.01

0.38

2.98**

0.01

0.01

0.25

1.71

SM groups

    

−0.86

0.29

−0.35

−2.97**

−0.68

0.30

−0.28

−2.26*

Age × SM groups

    

−0.01

0.01

−0.11

−0.82

−0.00

0.01

−0.05

−0.35

Spontaneous speech

        

0.10

0.06

0.27

1.78

Adjusted R2

0.17

 

0.41

 

0.43

 

R2

0.20

 

0.27

 

0.03

 

F

6.85**

 

8.67***

 

3.16

 

Cumulative R2 at step 3 = 0.49, overall. F(6, 52) = 8.40***

Outcome 4: high power in birthday speech

Child anxiety

0.01

0.02

0.07

0.44

0.00

0.02

0.02

0.14

−0.02

0.02

−0.16

−0.94

Parental anxiety

0.03

0.03

0.14

0.94

0.02

0.03

0.10

0.73

0.02

0.03

0.12

0.89

Age

    

−0.00

0.00

−0.46

−2.96**

−0.00

0.00

−0.28

−1.61

SM groups

    

0.07

0.05

0.23

1.59

0.04

0.05

0.13

0.88

Age × SM groups

    

0.00

0.00

0.33

2.11*

0.00

0.00

0.25

1.59

Spontaneous speech

        

−0.02

0.01

−0.36

−1.98

Adjusted R2

−0.00

 

0.16

 

0.20

 

R2

0.03

 

0.19

 

0.05

 

F

0.99

 

4.44**

 

3.93

 

Cumulative R2 at step 3 = 0.28, overall F (6, 52) = 3.41**

P < 0.05, ** P < 0.01, *** P < 0.001

Granting of Autonomy

Free Play

At Step 1, child and parent anxiety were related significantly to granting of autonomy (adjusted R2 = 0.20, P < 0.01). Age, group, and the interaction of age and group at Step 2 did not account for a significant amount of variance, and the interaction term was not significant (∆R2 = 0.02). Spontaneous speaking,2 added at Step 3 of the model, also did not account for a significant change in variance accounted for (∆R2 = 0.02). Parent anxiety was the strongest predictor when all of the variables were entered into the model.

Birthday Speech

At Step 1, child and parent anxiety accounted for 20% of the variance (adjusted R2 = 0.41, P < 0.01). Specifically, higher levels of observed child anxiety were associated with less granting of autonomy. Once we controlled for child and parent anxiety, age, group, and the interaction of age by group accounted for an additional 27% of variance (P < 0.001). Lower age and being selectively mute were associated with less granting of autonomy. The interaction term, however, was not significant, suggesting that group categorization did not moderate the relation between age and granting of autonomy. At Step 3, the addition of spontaneous speaking did not account for a significant change in variance accounted for (∆R2 = 0.27). The contribution of child anxiety and group were no longer significant when spontaneous speaking was entered. This pattern may have been due to the significant correlations of child anxiety and age with spontaneous speaking, or due to a loss in power when a new variable was added.

High Power t-units

Free Play

At Step 1, child and parent anxiety were related significantly to high power remarks (adjusted R2 = 0.175, P < 0.01). Specifically, higher observed child anxiety was associated with more high power remarks. Once we controlled for child and parent anxiety, age, group, and the interaction of age by group accounted for an additional 25% of variance (P < 0.001). Younger age and being selectively mute were associated with more high power remarks. The interaction term was also significant (∆R2 = 0.02, P < 0.001), suggesting that group categorization moderated the relation between age and high power remarks (see Fig. 1). At Step 3, the addition of spontaneous speaking accounted for an additional 7%. (P < 0.01). Fewer self-initiated remarks were associated with more high power remarks. The interaction term remained significant. There was a significant, negative relation between age and high power remarks in the comparison group but not in the SM group. Hence, parents in the comparison group made more high power remarks with younger children but age was not associated with high power remarks among the SM group.
https://static-content.springer.com/image/art%3A10.1007%2Fs10578-010-0214-1/MediaObjects/10578_2010_214_Fig1_HTML.gif
Fig. 1

Group as a moderator of the association between age and high power t-units in the free play

Birthday Speech

At Step 1, child and parent anxiety were not significantly associated with high power remarks. At Step 2, the addition of age and group accounted for an additional 19% of the variance and the interaction term was significant (∆R2 = 0.19, P < 0.01), suggesting that group categorization moderated the relation between age and high power remarks (see Fig. 2). Similar to the free play situation, there was a significant, negative relation between age and high power t-units in the comparison group but not the SM group. After age and group were controlled, parents made more high power remarks when children made fewer self-initiated remarks. At Step 3, the addition of spontaneous speaking accounted for an additional 5% (P < 0.05). The interaction term was no longer significant with the addition of spontaneous speaking.
https://static-content.springer.com/image/art%3A10.1007%2Fs10578-010-0214-1/MediaObjects/10578_2010_214_Fig2_HTML.gif
Fig. 2

Group as a moderator of the association between age and high power t-units in the birthday speech

Discussion

The present study had three objectives: (1) to determine if parents of children with SM engage in greater control than parents of anxious and of non-anxious children, as indexed by two different behaviors: degree of granting of autonomy and degree of power in conversational remarks, (2) to determine if parental control would vary in accordance with the situation in which dyads were observed, and (3) to increase our understanding of what contributes to demonstrations of control among parents. Three populations were investigated in the study: children with selective mutism (SM), anxious children (without SM), and non-anxious children.

Parental Control

With regard to group differences, parents of children with SM were rated by observers to be more controlling according to both forms of measured control. Specifically, parents of children with SM were rated as granting less autonomy than parents of anxious and non-anxious children. Parents of children with SM also made a higher proportion of high power remarks (i.e., directives, choice questions, and product questions) and, therefore, a smaller proportion of low power remarks (i.e., personal and phatic comments) than the parents of the other two groups. In fact, half of the remarks made by parents of children with SM were of a high power style compared to only a third of those made by parents of non-anxious controls.

Unexpectedly, there was not a statistically significant difference between the parents of anxious children and parents of non-anxious children in the extent to which parents granted autonomy or in their conversational style. The comparable scores of the anxious and non-anxious groups are likely due, in part, to the fact that several of the anxious children in the present study were from a community sample. Thus, their anxiety may have been of lesser severity than that found in a clinically referred sample. On the other hand, it may be that it was restricted child speech that principally drove parent behavior in the interactions observed in the study, an idea that was explored through the regressions.

Several explanations can be put forward to explain why parents of children with SM demonstrated greater control. Rubin and colleagues [21] proposed that parents experience the desire to protect their socially wary children from anxiety provoking situations and feel compelled to take control of the situations for them. Children with SM are typically very shy and inhibited in social situations, particularly when they involve speaking demands. This response by these children may lead the parents to want to decrease their child’s discomfort and anxiety, causing them to speak for the child (e.g., “Her name is Lisa”). As suggested by Bell and Harper [39], parents may also employ more directive strategies as a means to simply increase their child’s speaking participation. A third explanation may be that parents experience substantial embarrassment or frustration when in a social situation in which their child does not respond to another individual. Hence, the parent may direct the child as a reflection of his or her desire for the child to comply in a socially appropriate fashion.

The Role of Context

As suggested by Dennis [19], task-oriented parent–child interactions place distinct demands on parents in comparison to play interactions. In the present study, parental control varied in accordance with the context in which the dyad was observed. As predicted, parents demonstrated less control during the free play context than in the birthday speech context. This finding was consistent across all three groups and is consistent with that of Ginsburg and colleagues [20]. Ginsburg suggested that parents demonstrated greater control during a structured task than during free play as a result of increased difficulty, time pressure, and the evaluation component of that task. Similar conclusions can be drawn about the findings from the present study. In the birthday speech context, children had to perform a speech that the parents had helped them to prepare. They may have exhibited greater control to ensure that the child succeeded in stating the various components of the speech discussed in preparing for it.

While group differences were found in the degree of control demonstrated by parents according to both indices, change in control across the laboratory contexts did not differ among groups. Hence, when children were expected to give a birthday speech, parents in all groups demonstrated greater control. The finding that parental control varied by observational segment helps explain the inconsistent findings in the literature on parenting behaviors with anxious children, in which parent–child dyads have been observed in different contexts from study to study (see Ginsburg and colleagues [20]).

Child and Parent Characteristics as Predictors of Parental Control

Granting of autonomy was a construct that included parental judgment of their children and their encouragement and acceptance of different or independent opinions and actions of their children. Higher observed child anxiety predicted less autonomy granting in both free play and the birthday speech situations. This association is in accordance with developmental psychopathology models of anxiety [e.g., 9], which propose that increased parental control in the presence of visible anxiety is a means by which parents attempt to lower their child’s distress. Moreover, witnessing their child in discomfort during the interaction may lead parents to presume that he or she will not succeed without guidance. The association between child age and autonomy further supports this, in that when both child and parent anxiety were controlled for, parents granted less autonomy during the birthday speech with younger children whose discourse skill was likely less developed and thus were less certain to excel.

Previous research has indicated that anxious mothers grant less autonomy to their children in comparison to non-anxious mothers [18]. In the present study, parent anxiety was associated with granting of autonomy during the free play situation but not the birthday speech. Evans [40] suggested that parents who are shy themselves may be uncomfortable when being observed. In the present study, it may be that parents experienced heightened anxiety at the beginning of the laboratory observation but relaxed as time progressed. Further, free play was the least structured context in that parents were not given any direction. The lack of structure may have further increased the parents’ anxiety and demonstrations of control. Finally, during free play the parent and child were equally visible, placing increased pressure on the parents in comparison to the birthday speech where the focus was more on the child.

With regard to conversational dominance, observed child anxiety in free play, but not the birthday speech, predicted more high power parent remarks, partially replicating the findings with respect to the granting of autonomy. These high power remarks principally entailed asking the child questions. However, parent anxiety had no predictive value for the parents’ verbal style in this context nor during the birthday speech. Whatever parents may have experienced in the unstructured free play task that led them to grant less autonomy to the child and make judgmental comments did not apply to conversational control in this same context. In addition, child anxiety was not associated with high power remarks in the birthday speech. It is likely that in this context, parents were more focused on the task itself and what the child said. Moreover, the preparatory work for the speech had already been done, leaving the child to stand and deliver.

In both free play and the birthday speech, after controlling for child and parent anxiety, age, and group categorization, parents made more high power remarks when children made fewer self-initiated comments. The more children made self-initiated comments, the less conversational control parents exerted. Decreased speech among even normally chatty children may have been interpreted as discomfort, leading parents to try to engage them. As noted earlier, Rubin and Burgess [9] hold that a child’s inhibition elicits parental concern, leading the parents to try to solve the dilemma or take over to relieve their child’s distress. Parents of children who offered fewer spontaneous remarks may have responded to that lack of verbal participation by trying to increase their child’s speaking through high power remarks to reduce their child’s perceived discomfort and maybe even a bit of their own over their child’s relative silence.

The influence of the child’s spontaneous speech on parent behavior also explains why parents made more high power remarks with younger children in both situations, and granted less autonomy with younger children during free play. Parent beliefs and attitudes toward child behavior and competence change as children mature [41]. However, in addition to having a main effect such that SM predicted parental control, group categorization also moderated the relation between age and high power remarks: the association of less of this form of control with older children did not hold for the parents of the children with SM. One explanation is that it is normative for parents to decrease conversational control with children as they grow older, based upon the premise that their children are increasingly competent and successful in a variety of tasks and situations as they mature. This sense of increasing competence, however, may be compromised when children have significant anxiety about speaking outside of the home. Thus, for children with SM who have a history of speaking inhibition, parents continue to engage in conversational control, regardless of the age of the child. That spontaneous speaking accounted for a significant change in variance after controlling for group categorization further suggests that it is not just previous speaking inhibition but speaking during the interaction that influences parental control above and beyond parents’ previous experiences with their children.

In sum, the results indicate that child and parent anxiety, child age, and children’s speaking during the interaction were predictive of both forms of control measured in the present study. The variation in the relative strength of the predictor variables between granting of autonomy and high power remarks highlights that the different coding schemes capture different aspects of control. Finally, while the results highlight the relative strength of select factors in predicting parental control, they also show that there was much variance remaining to be accounted for. Future research investigating additional factors such as adults’ interpretations of the observational situation and beliefs about parenting may be of value in further understanding the interactions between children with SM and their parents.

Limitations

Information about the child’s interactions with both parents was not obtained in the present study, limiting the results mainly to mothers and limiting generalizations on parent–child interactions among the populations studied. Moreover, as highlighted earlier, parents in all three groups rated their anxiety to be low on self-report questionnaires, yet some parents were observed to be anxious during the laboratory tasks. The present study may have been strengthened if parents had rated their anxiety before and after the tasks in order to explore their state and trait anxiety. The lack of counterbalancing the free play and birthday speech impedes our ability to determine whether results are specific to the situational demands, order of presentation, or increasing/decreasing comfort with time in the laboratory.

The present study is also limited by sample size and as such, the respective power of the analyses to detect statistically significant effects. Finally, as with many laboratory studies, it is unknown to what extent the behaviors displayed and relationships observed would generalize to naturalistic situations in everyday life.

Summary

The present study has contributed knowledge on parental control among parents of children with SM, anxious children, and non-anxious children. Parents of children with SM exerted more control than did parents in the comparison groups and parental control varied by situational context. What predicted parental control was also investigated. Results indicated that child anxiety, parent anxiety and child age were predictors of parental control but that group membership moderated the relation of age to high power remarks. Finally, self-initiated speaking was found to predict control over and above being categorized as selectively mute, highlighting the importance of child behaviors in the present moment with parents.

While the results are essentially correlational and dynamic effects cannot be determined, they support previous theories that parents take over for their anxious children, and that this may be especially true when parents themselves are anxious, as though to reduce their and their child’s distress. In addition, the weak correlations between the ratings of granting of autonomy and high power remarks suggest that control is not unidimensional. As such, different aspects of control and their potentially different impact on child development should be examined in future research. Certain forms of control may help children to manage their anxiety, whereas other forms of control may increase the child’s anxiety and avoidance.

The findings also have relevance to methods of intervention for SM. A review by Cohan, Chavira, and Stein [29] indicates that behavioral and cognitive-behavioral interventions, each incorporating gradual exposure (i.e., practice speaking in select situations) have proven effective in helping children become more comfortable speaking. Findings from the present study suggest when establishing the exposure hierarchy, it is of value to consider not only the specific contexts that are associated with a fear of speaking but also the degree of control demonstrated by parents in commonly encountered situations (e.g., the playground, the school hallway). In addition to gradually exposing children to speaking in increasingly anxiety provoking contexts, one could also gradually decrease the degree of control exerted by parents or teachers in these situations.

Implications can also be drawn from the reported association between observed parent anxiety and parental control. This association suggests that parents would potentially benefit from learning strategies to manage their own distress in situations where their child fails to speak. Furthermore, educating parents about the potential associations between child, parent, and contextual factors would increase their awareness of how these factors may elicit controlling behaviors and make for more effective intervention programs.

Footnotes
1

With the highest income earners removed from the no-anxiety group (such that there were no significant differences in income levels between groups), all analyses were re-calculated. No differences were found in the pattern or significance of results.

 
2

Spontaneous speaking was chosen as the verbal participation variable, as correlations were stronger with this variable in comparison to total speaking.

 

Acknowledgments

This paper is based on the first author’s Ph.D. thesis submitted to the School of Graduate Studies at the University of Guelph. This research was supported by a grant from the Ontario Mental Health Foundation. We thank Lindsay Bennett, Jenny Macsween, Sue McKee, Renee Nossal, and Chantal Regis for their assistance with data collection; Cristina Andreescu, Nila Mulpuru, and Gillian Stanley for their assistance with observational coding; and Yvonne Chen and Diana Urajnik for their assistance with statistical analyses. Finally, we thank the families who participated in this research.

Copyright information

© Springer Science+Business Media, LLC 2010