Upload
others
View
0
Download
0
Embed Size (px)
Citation preview
PERCEIVBD CONTROL AND PARENT-CHILD COERCTVE EXCHANGESt A DYNAMIC SYSTBMS ANAI,YSIS
Christine Yvette Tardif
A thesis submitted in conformity with the requirements for the degree of Master of Arts
Department of Human Development and Applied Psychology Ontario Institute for Studies in Education of the
University of Toronto
@copyright by Christine Yvette Tardif 1997
National Library Biblioth' ue nationale du Cana 7 a
Acquisitions and Acquisitions et Bibliogrephic Services sewices bibliographiques 395 Wellington Street 395, nie Wellington ûttawaON K1AON4 ûttawa ON K1A ON4 Canada Canade
The author has granted a non- exclusive licence dowing the National Library of Canada to reproduce, loan, distribute or seil copies of this thesis in microform, paper or electronic formats.
The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts fiom it may be printed or otherwise reproduced without the author's permission.
L'auteur a accordé une licence non exclusive permettant à la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfiche/nlm, de reproduction sur papier ou sur format électronique.
L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation.
Perceived Control and Parent-Child Coercive Exchanges:
A Dynamic Systems Analysis
Masters of Arts 1997
Christine Yvette Tardif
Human Development and Applied Psychology
ontario Institute for Studies in Education
University of Toronto
To expand on paît inquires regarding parent-child coercive
exchanges this study adopted a dynamic systems approach. Parent
and child were regarded as a self-organizing system, and the
emergence of communicative patterns in real time interaction was
highlighted. Specifically, this study was designed to explore
differences in communicative patterns among parent-child dyads
during a stressful situation, and to explore the relation of
these differences to both parental and child perceptions of
control. Eight parent-preadolescent dyads were videotaped while
discussing three topics: (1) a pleasant topic, (2) an unpleasant
topic ( L e . , stressful situation), (3) a pleasant topic. Al1
verbal utterances were coded for affect and content. Results are
consistent with the expectation that the onset of a stressful
situation reveals systematic differences in the intensity, timing,
and stability of negative parent-child communication patterns, as
a function of perceptions of control. The theoretical and clinical
implications of these findings are discussed.
ACKNOWLEDGEMENTS
As I contemplate the many steps taken in order to
successfully complete this research ptoject, my thoughts steer
invariably toward those who have offered their support and
assistance throughout my endeavour. 1 now embrace the
opportunity to thank these individuals. First, 1 would like to
thank al1 the parents and children who participated in this
study. Their contributions have made this research project
possible, and their enthusiasm has made it an interesting and
exciting venture. 1 would like to express my appreciation for
the academic support offered to me by Dr. Marc Lewis's research
group, and 1 am also thankful for the assistance that Simon
Williams offered regarding the development of appropriate coding
systems. Moreover, 1 would like to thank Dr. Marc Lewis for his
insightful suggestions, guidance, and support while completing
this research project. 1 would also like to express my gratitude
toward Dr. Daniel Keating for extending his insightful comments
on an earlier version of this report. Lastly and perhaps most
importantly, 1 would like to thank my family and friends for
offering their support, patience, and understanding. Thank you
for believing in me, and for teaching me t h a t with a lot of
perseverance and hard work 1 can realize my dreams and attain my
goals.
iii
TABLE OF CONTENTS
LIST OF FIGURES ............................................... vii Chapter :
I INTRODUCTION* ............................................. 1 parent-Child Coercive Exchanges: The Role of ........................................ Perceived Control 3 A Look at the Conceptual and Methodological Limitations Inherent in Previous Investigations .......... 6 Dynamic Systems Theory: A Brief ~ntroduction ............. 9 Perceived Control and Parent-Child Coercive Exchanges: A Dynamic Systems Analysis .................. A 2 The Present Study ....................................... 16
................................................... II . METHOD 19 Subjects ................................................ 19 Procedure ..............................m................ 19 Measures ................................................ 20
Parenting Locus of Control Scale ..................... 20 Nowicki-Strickland Internal-External Control Scale for Children ................................... 21 Defining External and Interna1 Parent-Child Dyads .... 22 Frames of Conversation ............................... 24 Transcribing and Coding Videotaped Conversations ..... 25 Coding Verbal Content of Parent-Child Conversations .....................................me. 26 Coding Affect Associated with Parent-Child Conversations ........................................ 28
Data Analysis Procedures ................................ 28 Conventional Analysis ................................... 30
Question #l Exploring Differences in Level of Coerciveness ............... 30
Dynamic Systems Analyses ................................ 30 Question 12 Detecting Phase Transitions ............ 30 Graphical and Visual Report ........................ 30 Exploratory Report ,. ........................ ....... 31 Questions #3. and #4 Measuring Latency ..................................... and Hysteresis 33 .................................................. . III mSULTS 34
Conventional Analysis ................................... 34 Question #1 ......................................... *34
Interna1 and External Parents ................... 34 Interna1 and External Children .................. 34
Dynamic Systems Analyses ................................ 40 Question 82 .......................................... 40 Graphical and Visual Report .......................... 40
Verbal Content .................................. 40 .......................................... Affect 44
Exploratory Report e . . . . . . ; e e . . e e ~ e ~ ~ e e e e e e e e e e ~ ~ ~ e e e ~ 4 8 ............ Interna1 Dyads ...................... 50 ................................. Dyads #l and #4 50
Dyad #8--Child ..............e......e...ee*.4.. *.53 External Dyads ...................... .......... .*55 Dyads #3, 15, and #7 ............................ 55
Exploring the Correspondence Between Changes in Affect and Phase Transitions in Verbal Content .... 60 .................................. Interna1 Dyads 60 ............ External Dyads ......ee.....~.....4. * 6 4 Question $3 .......................................... 68 Question #4 ......................................... *70
IVm D I S C U S S I O N . e e a e e e . e e e o e e e e o e m m e . e e e . e c . e . - o e . e o o o o o m e . o e o e 7 2 Perceptions of Control and Parent-Child Coercive Communication. .......................................... 77
Dynamic Systems Theory as an Enpirical Guide ......... 77 ........................... A Dynamic Systems Analysis 78
Some Clinical Implications of This Study ................ 81 Some Limitations of This Study and Future Directions .... 83 Concl~sion.~. .............m............................. 85
APPENDICES ..................................................... 96 (A) Parental Letter ............................m..e..... *.96 (B) fnfomed Consent F o m .......................m.e....... 97
............................ (C) Demographic Questionnaire - 9 8 .................. (D) The Parenting Locus of Control Scale 99 (E) The Nowicki-Strickland Internal-Externa1 Control
Scale for Children ...............m................... 102 (F) 8-Item Subscale of the Nowicki-Strickland
Internal-External Control Scale for Children ......... 107 (G) Gottman's (1979) Couples Interaction Scoring System .......................................... (C.I.S.S.) .IO8 (H) Floating Variance Graphs f o r External Dyad #2 and
................................... Interna1 Dyad #6 ..log (1) Floating Variance Graphs for the Internal Parent #8 .. 110
LIST OP TLEBLES --
Table #:
Distribution of Perceived Control Scores for Parent-Child Dyads Categorized as External................23
~istribution of Perceived Control Scores for Parent-Child Dyads Categorized as Internal................24
Coding Scheme For Verbal Content.............. ....m....m . . 2 6 Coding Scheme for Verbal Content: Some Examples ........... 27 Coding Scheme for Affect (Nonverbal Behaviour) .........* ..29 Total Number of Verbal Utterances From the Beginning of Frame B to the Onset of Phase Transitions for Interna1 and External Dyads. .............. ......O.e. .*69 Total Number of Verbal Utterances From the Beginning of Frame C to the Second Phase Transition for Internal and External Dyads ...........m.... .............m..........71
LIST OB BZGüRES
Figure #:
1. Histogram Depicting the Average Affect Scores of Internal and External Parents Across Conversation Frames A, 8, and C........................*...............36
2 . Histogram Depicting the Average Verbal Content Scores of Internal and External Parents Across Conversation Frames A, B, and C . . . . . . . . . . . . . . . . . . . ..... ..J7
3. Histogram Depicting the Average Affect Scores of Internal and External Children Across Conversation Frames A, B, and C..........................-......t......38
4. Histogram Depicting the Average Verbal Content Scores of Internal and External Children Across Conversation Frames A, B, and C . . . . . . . . . . . . . . . . . . ......... 39
5. Time Series Graphs Depicting Verbal Content and Affect Scores for Internal Dyads #1 and #4................42
6. Time Series Graphs Depicting Verbal Content and Affect Scores for Internal Dyads #6 and #8 . . . . . . .......... 43
7. Time Series Graphs Depicting Verbal Content and Affect Scores for External Dyads #3 and # 5 . . . . . . . . . ....... 46
8 . Time Series Graphs Depicting Verbal Content and Affect Scores for External Dyads #7 and #2................47
9. Dyad #1: Floating Variance Graph and Phase Space Maps Of Verbal Content Scores.............. .....m.. .51
10. Dyad #4: Floating Variance Graph and Phase Space Maps Of Verbal Content Scores.............. .....-.. -52
11. Dyad #8 Child: Floating Variance Graph and Phase Space Maps Of Verbal Content Scores................ ....... 54
12. Dyad #3: Floating Variance Graph and Phase Space Maps Of Verbal Content Scores.......................56
13. Dyad #5: Floating Variance Graph and Phase Space Maps Of Verbal Content Scores.......................57
14. Dyad #7: Floating Variance Graph and Phase Space Maps Of Verbal Content Scores.. ..................... 58
15. Dyad #1: Floating Average Graphs of Affect Scores and Floating Variance Graphs of Verbal Content Scores............................................61
16. Dyad $4: Floating Average Graphs of Affect Scores and Floating Variance Graphs of Verbal Content S~ores........................~...................62
17. Dyad 1 8 Child: Floating Average Graphs of Affect Scores and Floating Variance Graphs of Verbal Content S~ores..........................................~.63
18. Dyad #3: Floating Average Graphs of Affect Scores and Floating Variance Graphs of Verbal Content Scores....................~.~.~.~.................65
19. Dyad #5: Floating Average Graphs of Affect Scores and Floating Variance Graphs of Verbal Content Scores.......s....................................66
20. Dyad 17: Floating Average Graphs of Affect Scores and Floating Variance Graphs of Verbal Content Scores . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 7
v i i i
CHAPTER 1
Introduction
During the past decade, interest concerning the phenomenon
of child maltreatment has resulted in an abundance of inquiries
investigating the intricate nature of, and issues related to,
parent-child conflict (Blos, 1989 ; MacKinnon, Lamb, Belsky &
Baum, 1990; Marini, 1993; Minuchin, 1992; Snyder, Edwards,
McGraw, Kilgore & Holton, 1994; Steinberg, 1990). Despite recent
focus on the negative aspects of parent-child conflict, and its
contribution to child maltreatment, theorists contend that
conflictual episodes among parents and their children can be
dealt with constructively or destructively, hence leading to
either functional or dysfunctional outcomes (Deutsch, 1973).
Contemporary views emphasize that, when dealt with
constructively, conflict is a normal and inevitable process often
giving sise to positive states and outcomes. (Blos, 1989; Cooper,
1988; Cooper, Grotevant & Condon, 1983; Marini, 1993).
Nevertheless, it rernains critical to note that a substantial
degree of destructive conflict continues to delineate some
parent-child relationships (Montemayor, 1982). It has been
estimated that one in five families is characterized by parents
and their children engaging in prolonged, intense, and repeated
conflict (Montemayor, 1982). Also, it has been reported that
over 10% of children are targets of parental aggression, ranging
from severe spankings to attacks with knives and guns (Gelles &
Straus, 1988) .
2
Moreover, a number of reviews have indicated that prolonged
parent-child conflict can have adverse consequences for the
developing child (Minuchin, 1992). Such negative consequences
may include arrested social development, low self-esteem, and
aggressive behaviour directed at both peers and adults (Mueller &
ilv ver man, 1989; George & Main, 1979).
Accordingly, it remains particularly important to explore
those circumstances in which parent-child conflict becomes
typically dysfunctional and destructive in nature.
Investigations have suggested that, although parent-child dyads
may experience similar levels of conflict, these dyads may differ
vastly and with profound consequences, regarding the strategies
they employ when attempting to resolve conflict (Snyder et al.,
1994). For example, Snyder et al. (1994) demonstrated that
aggressive children and theit mothers were more likely to employ
escalation as a means to resolve conflict, whereas non-aggressive
children and their mothers were more likely to employ de-
escalation as a means to resolve conflict.
From a social learning perspective, researchers (Snyder et
al., 1994) have concluded that the frequency and functional
utility of escalation in terminating conflict serve to
differentiate aggressive and non-aggressive dyads. For instance,
when mothers gave in to their children as the child was
escalating the level of aversiveness during a conflict, they were
negatively reinforcing their chilàrenrs behaviour, thus making it
more likely that their child would engage in escalation in
subsequent conflicts. While Snyder et ale's (1994) mode1
illustrates how the use of escalation within a parent-child
serves to influence coercive exchanges, it does not explore
3
dyad
the
factors predisposing individual parent-child dyads to frequently
engage in coercive exchanges from the outset.
parent-~hild ~oercive Exchanaes: The Role of Perceived Contra1
In recent attempts to differentiate dyads susceptible to
engaging in coercive exchanges, numerous inquiries have examined
parental traits, parenting styles, as well as parental and child
causal attributions (Dix & Grusec, 1985; Dodge, 1980; Janssens,
1994; MacKinnon et al., 1990; Newberger & White, 1989; Patterson,
1986). Of particular interest in the present thesis is the
recent work exploring the role of perceived control, or locus of
control, as a moderator of adult reactivity to child behaviour
(Bugental, Blue & Cruzcosa, 1989) . Locus of control has been defined by Rotter (1966) as a
generalized expectancy or belief for internal versus external
control of reinforcement. When individuals believe events to be
contingent on their own behaviour, they are said to hold interna1
expectancies for control of reinforcement. Conversely, when
individuals believe events to be the result of luck, chance,
fa te , or I1powerful othersm (Le., physicians) they are said to
hold external expectancies for control of reinforcement.
The importance of perceived control over one's environment
has been demonstrated by many researchers ( e . g . , Langer & Rodin,
1976; Krause, 1987). Numerous investigations have demonstrated
4
that individuals with low perceptions of control (i.e., external
locus of control) over the outcome of events are more reactive to
hard-to-control (i. e. , stressful or Wif f icultmr) situations, and are more prone to the debilitating effects of Vearned
helplessness~ (Dweck & Repucci, 1973; Lefcourt, Hogg, Struthers,
& Holmes, 1975) . In light of these findings, Bugental and Shennum (1984) have
argued that when parents believe they are victims (Le., external
locus of control) of a difficult or stressful event (e.g., child
misbehaviour), they are likely to develop feelings of
helplessness. Further, in believing their efforts are futile,
these parents may develop apathetic feelings about child-rearing
and may come to feel threatened by child uuncontrollability.~
Hence, in an attempt to re-establish and maintain a sense of
control with respect to their role as a parent, these adults may
be more likely to make use of coercive discipline tactics and may
be less likely to use cognitive-rational appeals, such as
providing explanations and negotiating.
Alternately, Bugental and Shennum (1984) have argued that
when parents hold high perceptions of control over the outcome of
events (Le., interna1 locus of control), they will be less
likely to experience feelings of helplessness when disciplining
child misbehaviour. Further, they rnay feel less threatened by
child uncontrollability, and hence feel less compelled to re-
establish and maintain a sense of control and power with respect
to their role as a parent. Consequently, these parents will be
5
less likely to use coercive discipline tactics and will be more
likely to make use of cognitive-rational appeals.
Recently, investigators have proved successful in
demonstrating a link between parental perceptions of control and
parental discipline strategies, For example, Loeb (1975) found
that "externalW parents used highly controlling strategies when
teaching their children, in contrast with Vnternaln parents who
made use of less controlling and more persuasive styles. Also,
abusive mothers have been found more likely than non-abusive
mothers to believe they can do little to prevent negative
caregiving outcomes (Bugental et al., 1989). Furthemore,
external parents, as opposed to interna1 parents, were found to
be more authoritarian when controlling their child, especially
when they perceived their child as an internaliser (Le.,
withdrawing from problems) or an externaliser ( e , acting out
to resolve problems, e.g., aggressiveness) (Janssens, 1994).
To demonstrate the importance of adult perceptions of
control as a moderator of the effects that child behaviour has on
adults, researchers have experimentally varied child
llresponsivenessf~ and ~unresponsivenessl~ (Bugental, Caporael, &
Shennum, 1980; Bugental & Shennum, 1984; Bugental & Cortez,
1988). It was demonstrated that adults with low perceived
control over caregiving outcomes respond with greater negative
affect, greater elevated arousal, and higher levels of
helplessness to difficult or unresponsive children, than do
adults with high perceived control. Adults with high perceived
control over caregiving outcomes have been found to display
little differential reactivity to responsive, versus
unreçponsive, children (Bugental & Cortez, 1988).
In yet another investigation, mothers of seriously il1
infants have been found to cope in a less adaptive manner if they
believed their childrs condition was outside of their control and
was under the control of other individuals in the environment
( e . g . , physicians, nurses, etc.) (Affleck, Allen, McGrade, &
McQuenney, 1982). Thus it appears that low perceptions of
control act in a sensitizing role, whereas high perceptions of
control act in a buffering role.
A Look a t the Conceptual and Methodoloaical Limitations
Inherent i n Previous fnvesticrations
The preceding findings suggest that parental perceptions of
control over the outcome of stressful or difficult events (Le.,
negative caregiving outcomes) play a critical role in influencing
the quality of parental discipline strategies. While these
findings are both revealing and meaningful, it is noteworthy that
the methodology employed in previous studies has been limited in
a number of important respects.
For the most part, past researchers have failed to
acknowledge that parent-child coercive exchanges arise as a
result of, and should be examined as, a dyadic process. For
instance, these researchers ( e . g . , Affleck et al., 1982; Bugental
et al., 1989; Janssens, 1994; Loeb, 1975) have often implicity,
if not explicitly, assumed a direct linear causal relationship
7
between parental perceptions of contxol and parental discipline
strategies. By not considering parent-child coercive exchanges
as a dyadic event, this relationship has traditionally been
measured in isolation with little regard to the possibility that
children's perceptions of control may act in a dynamic,
reciprocal fashion to shape parent-child coercive exchanges
(Affleck et al., 1982; Bugental et al., 1989; Janssens, 1994;
Loeb, 1975).
Additionally, previous inquiries have often relied on
correlational methods, thus allowing for an investigation of
parent-child coercive exchanges following their formation in real
time (Le., outcome). However, little consideration has been
accorded to the unfolding of parent-child coercive exchanges, or
to the processes by which these exchanges emerge and stabilize in
real time (Le., immediate moment to moment interactions).
Moreover, past inquiries have typically accorded
insufficient importance to the context in which parent-child
exchanges are embedded in real time (e.g., stressful or calm
event). When context has been considered, it has once again for
the most part been correlated retrospectively with reports of
parental discipline sttategies, with little consideration granted
to the question of how contextual influences serve to alter the
quality of parent-child exchanges in real time (Janssens, 1994;
Loeb, 1975).
Implicit within previous studies is the opinion that
retrospective correlational approaches are a useful means of
8
addressing the study of individual differences in parent-child
exchanges. However, by ignoring the role of contextual
influences in real tirne, previous inquiries often overlook
individual differences in the quality of exchanges among
particular parent-child dyads. For instance, even parents who
hold low versus high perceptions of control do not engage in
coercive exchanges with their children at al1 times and in al1
contexts; rather, they do so within particular contexts.
While it has been a common approach for investigators to
either control for context, or to Ifround off1' the enormous
variability in parent-child exchanges, the behaviour of most
parent-child dyads is nevertheless highly variable and context
specific. Hence, it should be a primary goal of investigators to
begin identifying those contextual factors leaving parent-child
dyads susceptible to engaging in coercive exchanges at a given
moment in time.
As a consequence of these conceptual and methodological
limitations, I feel that several meaningful questions remain
unanswered. For instance, how does the on-line experience of a
stressful or difficult event combine with perceptions of control
( i . , parental and child) to promote coercive exchanges among
parents and their children in real tirne? Moreover, once coercive
exchanges are initiated, do perceptions of control influence the
stability or strength of these exchanges?
To address these questions and to improve on past conceptual
and methodological shortcomings, an alternative approach is
9
needed, which has as its goal the specification of the processes
by which parent-child coercive exchanges emerge and stabilize in
real time. More specifically, what is needed is a theoretical
approach that recognizes the dyadic nature of parent-child
exchanges, explicitly acknowledges the importance of reciprocal
qualities in dyadic processes, and is sensitive to the importance
of contextual influences on the quality of parent-child
exchanges. Here, 1 introduce a recent and alternative
theoretical approach, known as dynamic systems theory, which may
be better suited to addressing the aforementioned theoretical
requirements.
Dvnamic Svstems the or^: A Brief Introduction
Recently, many psychological theorists are recognizing and
embracing principles of dynamic systems theory as a resourceful
means of conceptualizing the appearance of novel structures or
behaviours in complex systems (see Barton, 1994; Cooney & Troyer,
1994; Fogel, 1990; Ford & Lerner, 1992; Howe & Rabinowitz, 1994;
Lewis, 1995, 1996; Thelen, 1989, 1990; Thelen & Ulrich, 1991).
Inherently, complex systems are multicausal, nonlinear,
nonstationary, and are often found to be in a state of
thermodynamic nonequilibrium. It is noteworthy that complex
systems may extract energy, or information, from sources within
or outside the system. Thus, for example, neither genes or
environment hold priority in effecting change (Thelen, 1989).
Complex systems are comprised of many components or subsystems
free to interact with each other in many ways (Thelen, 1989;
10
Thelen, 1990). Complex systems are thus said to have many
"degrees of freedom" (Thelen, 1989).
Nevertheless, complex systems are often self-organizing.
That is without prior prescription and undet certain contextual
or task-specific constraints, the many components of a complex
system spontaneously interact, producing order and patterned
behaviour, thus reducing the system's original degrees of freedom
(Thelen, 1989; Thelen & Ulrich, 1991) . Lewis (1995) has outlined two main premises concerning self-organizing systems. First,
novel fonns emerging through the processes of self-organization
are qualitatively different from the lower order components
interacting within the system. Second, self-organizing systems
tend to move toward coherence and stability.
Stability in a self-organizing system emerges over time.
Self-organization relies on feedback between coupled elements
within a system, and stability in a self-organizing system is the
result of the ongoing reciprocal adjustment, or covariation,
among these related components (Lewis, 1995; 1996). Under
certain contextual or task-specific constraints, the self-
organized behaviour of a complex system tends to prefer certain
attractor states in the state space ( e , a topographical mode1
(landscape) of al1 the possible states a system can gravitate
toward). (Lewis & Douglas, in press; Thelen, 1990). That is,
without prior prescription, or llblueprint,lt a system's components
will autonomously gravitate toward attractor states, and will
return to these attractors following perturbation. Here, the
11
system is said to be in a steady state of dynamic equilibrium, or
dynamic stability (Thelen, 1990).
Although systems acquire dynamic stability, they remain
sensitive to their environments, and are continually subject to
disruptions or perturbations of their attractor states (Ford &
Lerner, 1992). When a change in one or more of the systemfs
parameters, or components, crosses a critical threshold, the
system may undergo a phase transition. A phase transition is
characterized by the spontaneous and abrupt reorganization of the
systemfs components, causing the destabilization of one or more
attractors, and the emergence of one or more new attractors
(Kelso, Ding & Schoner, 1993; Thelen, 1989). Those components
that drive a system across a critical threshold from one
attractor to another are known, in dynamic terminology, as
control parameters.
It is important to note that, when in a state of dynamic
stability, or an attractor state, perturbations to the system
produce few variations from the stable state. However, during
points of instability, perturbations are extremely influential in
initiating a phase transition (Kelso et al., 1993; Thelen, 1989).
Moreover, when a control parameter crosses a critical threshold
and the system is driven from one attractor to another, it is
possible to measure the stability of this new attractor by
rneasuring the time it takes the system to return to the previous
attractor once the control parameter returns below the critical
threshold. When the system requires a substantial amount of time
12
to return to its previous attractor, a hysteresis effect is s a i d
to have occurred (Kelso et al., 1993). This means that the new
attractor is particularly stable.
Developmental dynamic systems theorists conceptualize human
development as the stabilization and destabilization of
attractors over time (Lewis & Douglas, in press; Thelen & Ulrich,
1991). They diçtinguish two time scales in which self-organized
behaviour can occur. Developmental self-organization occurs with
the emergence and change in attractors over months or years,
whereas the convergence of behaviour to attractors over seconds
or minutes constitutes self-organization in real time (Thelen &
Ulrich, 1991) . Perceived Control and Parent-Child Coercive Exchanaes:
A Dvnamic Svstems Analvsis
Recall that, for the most part, past investigators have
failed to recognize the reciprocal and dyadic nature of parent-
child exchanges, and have emphasized disproportionately the
outcorne, rather than the processes, characterizing these
exchanges. However, more recent theorists have emphasized the
importance of recognizing and exploring the reciprocal nature of
dyadic processes (Fogel, 1990, 1992, 1993; Fogel & Hannan, 1985;
Ford & Lerner, 1992; Reimers t Fogel, in press). More
specifically, parents and their children are regarded as a CO-
acting dyad, or a dynamic and self-organizing system, whereby
differences among dyadic relationships are created in the process
of interaction (Fogel, 1990, 1993). Hence, during interaction,
13
both parent and child characteristics, as well as contextual
factors, act in a dynamic fashion to mutually constrain one
another's behaviour (Fogel, 1993).
Applying dynamic systems principles to his own work, Fogel
(1993) has described al1 dyadic interactions as a process of
continuous and mutual adaptation, that is, as a process of co-
regula t ion . He defines co-regulation as a %ocial process by
which individuals dynamically alter their actions with respect to
the ongoing and anticipated actions of their partnersw (Fogel,
1993, p. 34). Also, during interaction an individual's behaviour
is emergent from the constraints imposed by their own body, by
their expectations, as well as by the actions of the partner.
Through CO-regulated interaction, consensual social patterns are
created and elaborated over the, and these CO-regulated patterns
can be recognized by their repetition and coherence over t h e
(e. g. , social games) (Fogel, 1993) . Moreover, in describing dyadic communication Fogel (1993)
has introduced the concept of consensual frames. A frame is a
CO-regulated consensual agreement (Le., attractor) about the
main topic of discourse, about the limits of what is to be
communicated between a dyad, and about when and how interactions
take place and for how long. Note that frames must exist before
partners can actually engage in facused communication, and like
the ensuing discourse, frames are consensual and CO-regulated.
Fogel distinguishes two types of consensual frames: (1)
stable consensual frames for creativity and innovation, and (2)
24
stable consensual frames for rigidity and dissolution. First,
soma stable consensual frames in relationships are characterized
by inventiveness and creativity. For instance, dyadic
communication is characterized by themes, and each tirne a theme
is mutually engaged in something new emerges. Here, both
partners are rnotivated by the potential for creativity in their
relationship about a particular theme. Second, some stable
consensual frames are marked by rigidity, and by a motivation to
avoid creativity around particular themes (e.g., ritualized
greetings). Rigid consensual frames rnay persist for long periods
of time ( e . g . , a child may insist on introducing a topic that the
parent wishes to avoid), and may become destructive (e.g.,
continually denying a partner emotional support for an
information theme that relates to adaptive functioning) (Fogel,
1993).
In line with both Fogelrs (1993) theorizing and principles
of dynamic systems, we can imagine a landscape (Le., state
space) delineating the many potential frames or patterns of
communication (Le., attractor states), that a particular parent-
child dyad may tend toward at a given point in the. For
instance, when faced with a stressful situation, there are a
number of potential approaches (Le., many degrees of freedom)
for a dyad to resolve the inunediate situation. However, as their
behaviour self-organizes, it will tend toward certain patterns of
communication, or attractor states, over others, thus reducing
the systemfs many degrees of freedom.
The question remains why the communication of particular
parent-child dyads ( e . , dyads with external or low perceptions
of control) tends to prefer, or converge toward, coercive
attractor patterns in real tirne? What factors are responsible
for driving the system from an attractor pattern characterized by
open and positive communication (i.e., creative frame) toward
another characterized by coercive communication (i.e., rigid
frame)? Moreover, once coercive communication patterns are
established, why do these become stable and rigid patterns for
some parent-child dyads, and not others?
By reconceptualizing parent-child interactions as dynamic,
self-organizing, and CO-regulated, we can begin by graphically
mapping the many communication patterns in the state space for a
particular dyad. Further, those patterns that are particularly
stable for a dyad can be identified as attractor states, that is
communication patterns toward which the dyad tends to converge.
Next, identifying attractor patterns for a dyad allows for the
detection of phase shifts in dyadic communication (e.g., a shift
from positive to coercive communication patterns). Moreover,
once a phase shift has been detected we can begin to identify and
possibly manipulate the control pararneter(s) (e.g., individual or
contextual) responsible for inducing this transition (Thelen &
Ulrich, 1991) , Lastly, once particular attractor patterns (such
as coercive frames) are established, the dynamic systems concept
of hyçteresis facilitates the investigation of the stability and
resiliency of these patterns. It is noteworthy that, in contrast
- 16
with past correlational and retrospective approaches, a dynamic
systems analysis is properly suited to the consideration of the
dynamic and reciprocal nature of parent-child interactions.
Also, a dynarnic systems approach can permit us to directly
outline the factors ( e . g . , perceptions of control) responsible
for inducing and maintaining coercive communication among parents
and their children in real tirne.
The Present Studv
Drawing upon the aforementioned principles of dynamic
systems, it was the purpose of the present study to expand
further on previous inquiries regarding the relationship between
perceptions of control and parent-child coercive communication.
Due to a marked increase in parent-child conflict as the child
reaches the pubertal stage (Montemayor, 1983), the present study
focused on relations among parents and preadolescent children.
Four exploratory research questions were proposed. Note that the
first research question is of the conventional sort, whereas the
second, third, and fourth research questions are premised
specifically on principles of dynamic systems.
First, 1 explored differences in the overall level of
coercive communication among dyads identified with external (low)
versus internal (high) perceptions of control (over the outcome
of family events). In line with past inquiries (Bugental et al.,
1980; Bugental & Shennum, 1984; Janssens, 1994; Loeb, 1975), 1
expected that parents and children identified with external
versus internal perceptions of control might evidence higher
17
levels of coercive communication,
Second, 1 explored one likely context in which parent-child
dyads might engage in coercive communication patterns in real
the. Specifically, I investigated whether experimentally
manipulating ( e . , increasing) the level of perceived stress by
parent-child dyads might act as a control parameter inducing a
phase transition, leading to a pattern of communication
characterized by coercive exchanges. Also, the synchrony of
communication patterns within parent-child dyads was observed,
prior to and following the experimental manipulation.
Third, it was expected that most parent-child dyads could
encounter a level of stress that would induce a phase transition
toward coercive communication. Consequently, 1 explored
differences in the timing (Le., latency) of this phase
transition between internal and external dyads. Recall from
earlier findings that external parents were found to react
sensitively to stressfui situations (Bugental et al., 1980;
Bugental & Shennum, 1984; Bugental & Cortez, 1988), Therefore, 1
expected that, for external dyads, lower levels of perceived
stress might be needed to induce a phase transition, and hence,
less tirne might elapse before this transition occurred.
Fourth, once the behaviour of a parent-child dyad converged
toward a pattern that was characterized by coercive
communication, 1 explored differences between internal and
external dyads regarding the stability of this communicative
pattern. Because external parents have been found to react
18
sensitively to stressful events, I expected that communication
with their child has had more occasion to become negative and
coercive. That is, for these dyads, the recurrence of coercive
communication might result in this type of communication becoming
a stable pattern. Consequently, when levels of stress return to
a lfmanageable" level (i. e., below the critical threshold) , it might take longer for their communication to return to its
original pattern (Le., baseline). Simply put, coercive
communication in these dyads might be a more stable pattern of
communication, and thus evidence of a hysteresis effect might be
apparent.
CHAPTEII II
Methoil
Subi - ects
Participants included eight parent-preadolescent dyads. The
mean age for children was 10 years 8 months (SD=0.998), ranging
from 9 years 5 months to 12 years 11 months (six girls and two
boys). The mean age for parents was 37.25 years (SD=3.92),
ranging from age 33 to 44 years (seven mothers and one father).
Of these parents, five indicated they were rnarried and three
indicated they were divorced. Reports of occupational status
included: three homemakers, one taxi driver, one transit driver,
one data entry clerk, and two business owners. Education ranged
from some high school to a completed university degree at the
Master's level. The sample consisted of six mother/daughter
dyads, one mother/son dyad, and one father/son dyad. There were
no father/daughter dyads included in this study.
Al1 dyads were recruited from the community through I1word of
mouth". Participants were provided with a letter describing the
nature of the study (see Appendix A), and asked to indicate
voluntary participation by signing an informed consent form (see
Appendix B). Participants were informed that al1 information
would be kept strictly confidential, and they were compensated
$20 for their participation.
Procedure
Al1 dyads were seen in their homes. Complete participation
was conducted in two sessions, and lasted approximately 40
2 0
minutes. The first session involved the completion of standard
questionnaires. The second session was conducted
observationally, and involved the videotaping of parent-child
conversations. In the firçt session, both parents and their
children were asked to complete paper-and-pencil questionnaires.
Parents were asked to complete a basic demographic questionnaire
(see Appendix C), followed by the Parenting Locus of Control
Scale (Campis, Lyman & Prentice-Dunn, 1986). Children were asked
to complete the Nowicki-Strickland Internal-External Control
Scale for children (Nowicki & Strickland, 1973). Dyads were
asked to complete their respective questionnaires independently.
This ensured the privacy of their responses, particularly the
responses of the children. Completion of these questionnaires
required approximately 15 minutes. In the second session,
parent-child dyads were videotaped while they collaboratively
discussed and tried to resolve three separate issues.
Altogether, videotaped conversations lasted 18 minutes, with
discussion for each individual issue lasting 6 minutes.
Measures
Parentina Locus of Control Scale
Parents' beliefs or perceptions regarding interna1 versus
external control were assessed using the Parenting Locus of
Control Scale (Campis et al., 1986). The Parenting Locus of
Control Scale assesspn parental perceptions of control concerning
child rearing outcomes. This scale consists of 47 items (for
example, When my child is well-behaved, it is because helshe is
21
responding to my efforts") ta which participants indicate their
beliefs on a five-point Likert-type scale ranging from strongly
disagree (1) to strongly agree (5). Also, this scale consists of
five factors or subcategories. These include: Parental Efficacy,
Parental Responsibility, Child Control of Parents' Life, Parental
Belief in Fate/Chance, and Parental Control of Childts Behaviour
(see Appendix D). Scores on this scale range from -73 to +115,
with negative scores reflecting increasing perceptions of
internal control and positive scores reflecting increasing
perceptions of external control.
Nowicki-Strickland Xnternal-External Control Scale for Children
Childrens* beliefs or perceptions regarding internal versus
external control were assessed using the Nowicki-Strickland
Internal-External Control Scale for Children (Nowicki &
Strickland, 1973). This scale consists of forty items (for
example, "Are some kids just born lucky?I1) to which participants
indicate their beliefs with a yes or no answer (see Appendix E).
Scores on this scale range from O to 40 with lower scores
reflecting increasing perceptions of internal control and higher
scores reflecting increasing perceptions of external control.
Note that a standardized measure to assess childrents
perceptions of control, pertaining specifically to the outcomes
of events with parents, does not exist. However, the Nowicki-
Strickland Internai-External Control Scale for Children is
comprised of 8 items addressing children's perceptions of control
regarding the outcomes of events with their parents. Hence, in
22
this study, these 8 items were extracted to fashion a subscale.
When the word llparentsm appeared in these questions, it was
changed to %omw or "dadlI depending upon whether the child was
participating with their mother or father (see Appendix F).
Definina External and Interna1 Parent-Child Dvads
To select dyads in which both parental and child perceptions
of control were the same ( e , both high internal or both high
external), a larger sample of 21 dyads was tested. An
examination of parents scores on the Parenting Locus of Control
Scale revealed few scores falling within the regions of above +15
and below -15. Therefore, these values were used to mark the
boundaries for defining external and internal groups,
respectively. Parents were categorized as external when their
overall score on the Parenting Locus of Control Scale exceeded
+15, whereas parents were categorized as internal when their
overall score exceeded -15.
For children, two separate measures were employed to
distinguish external versus internal groups. First, an
examination of children scores on the Nowicki-Strickland
fnternal-External Control Scale for Children revealed few scores
falling within the regions of above +18 and below -13. Thus,
these values were used to mark the boundaries for defining
external and internal groups, respectively. Moreover, responses
on the 8 items i e , extracted subscale) relating to chilàrenOs
beliefs regarding the outcome of events with their parents were
considered. Children were categorized as external when their
23
overall score exceeded +18, and when a minimum of five external
responses on the 8-item subscale were indicated. Conversely,
children were categorized as internal when their overall score
was lower than +13, and when a minimum of five internal responses
on the 8-item subscale were indicated.
To be classified as an internal or external dyad, both
parents and children needed to be identified as meeting the
previously mentioned demarcations for their respective groups.
Consequently, of the larger sample of 21 dyads, only 8 dyads were
identified as external or internal. Of these eight dyads, four
were categorized as external and four were categorized as
internal. See Table 1 and Table 2 for a distribution of the
scores in each group.
Table 1
Distribution of Perceived Control Scores for Parent-Child Dvads
~ateaorized as External
Dyad #3 1 +24 I le
Dyad #
Dyad #5
Dyad 52 I l8
Childrensf scores on 8- item subscale
Parental scores
+16 6/8 external responses
Childrensf scores (overall)
19
5/8 external responses
5/8 external responses
5/8 external resmonses
Brames of Conversation
Parent-child dyads engaged in three separate
conversations, each lasting 6 minutes. Al1 three Vramesll of
conversation shared the same dialogical format, that is
collaborative discussion and resolution. Dyads were first
instructed to choose collaboratively two topics they considered
pleasant, and one topic they considered unpleasant (Le.,
stressful task and experimental manipulation). Parents and their
children were instructed to choose their own conversation topics
so as to ensure that topics were meaningful and relevant to them.
Table 2
Distribution of Perceived Control Scores for Parent-Child Dyads Cateaoriaed as Interna1
Dyad #8 1 -18 1
l
Dyad #
Dyad fi
Dyad #6 1 -22 1 l2 Dyad #4 1 -35 I l2
Parental scores
-4 9
Childrenst scores on 8- item subscale
Childrensf scores (overall)
12 6/8 internal responses - --
7/8 internal responses
7/8 internal responses
6/8 internal responses
- --
Examples of pleasant topics included: deciding what to do for a
birthday, deciding where to go on a vacation, and deciding what
kind of new house to buy. Examples of unpleasant topics
included: division of chores, deciding an appropriate bedtime,
and deciding an appropriate age to Wear makeup and high heel
shoes. Once a topic was selected, al1 dyads were assured that it
25
was fine if they were unable to resolve these topics within six
minutes, and rather that it was the process whereby they
attempted to resolve them that was of interest. The videotaped
conversations of parent-child dyads always took place in the
evening, and in the same location: the kitchen table.
The order of the conversations was as follows: (A), (B), and
( C ) . Conversation frame A was designed to procure a base rate
regarding the style of communication characterizing each dyad,
and hence dyads discussed and tried to resolve a pleasant topic.
Conversation frame B was designed to experimentally manipulate
the control parameter (i.e., increase the level of perceived
stress) after a baseline pattern was observed, and to capture the
shift ( i . e . , latency) in communication that dyads might exhibit.
Therefore, dyads discussed and tried to resolve an unpleasant
topic (i.e., stressful task). Conversation frame C was designed
to allow the level of stress perceived by the dyad to decrease,
and to test for the occurrence of a hysteresis effect. Hence,
dyads once again discussed and tried to resolve a pleasant topic.
Note that, elapsed time between conversations frames A, B, and C
was kept to a minimum, and generally included only the time taken
to stop the video camera and to prepared it again for the next
frame of conversation.
Transcribinq and Codina Videotamed Conversations
Al1 conversations were transcribed verbatim. Time (or
event) sampling was not conducted; rather, for both parent and
child, each verbal utterance served as a unit of analysis which
26
captured the sequential flow of parent-child conversations. For
both parent and child, each verbal utterance was coded on two
dimensions: verbal content and affect.
Codina Verbal Content of Parent-Child Conversations
Soth parent and preadolescent verbal utterances were coded
for content. The coding scheme used for scoring the content of
each verbal utterance was an adapted version of Gottman8s (1979)
Couples Interaction Scoring Systern-(see Appendix G). The coding
scheme employed in this study consisted of ten codes ranging from
the most tacilitative (1) to the rnost aversive (10) (see Table
3). Each verbal utterance was assigned one code. When an
extended verbal utterance consisted of two separate thought units
(this occurred relatively infrequently), each thought unit was
coded separately.
Table 3
Codincr Scheme For Verbal Content
Facilitative behaviours
Approving, or supportive comment Direct agreement/Compliance Problem Solving/Offering or eliciting an opinion or suggestion Assent/Affirmative
Neutra1 behaviours
Exchange of information--Question/Clarification Summarizing self or other/lindreading
Aversive behaviours
Disagreement with rationale supplied "Yes buttt agreement Direct disagreement/Noncomp1iance Command/criticize/complain/accusa/or obstinate-antagonistic
Al1 verbal utterances were coded by the author. A second
individual, blind to the categorization of dyads as interna1
versus external, independently coded 20% of the total responses
and agreement between coders was 81%. See Table 4 for examples
of al1 ten codes constituting the coding scheme used in this
study . Table 4
Codinu Scheme for Verbal Content: 8ome Exam~les
Approving or supportive
II Direct agreement/ Cornpliance Problem solving/Offering or eliciting an opinion
II Exchange of information/ Question/Clarification II
- - - - - - - - -
Summarizing self or other/ Mindreadina
Disagreement with rationale supplied
"Yes buttt agreement. . .
II Direct disagreement/ Noncornpliance Command/Criticize/Accuse/ Complain/or obstinate antagonistic
IuYeah, 1 am impressed with how well youfre taking care of your f ish, good jobuu!
ItYeah , you ' re righttu . %O ok that sounds like it's settled, October starts you so to bed at 9 t 0 0 rishtuf?
What s the Rideau Riveruu?
"Oh 1 see now you said a sleepover would be goodm.
InNo 1 donr t think so, 1 doubt any of your friends will be allowed to go unsu~ervised~~ . ItYeah but it doesn8t really feel like you want me to QO" . "No way corne onit.
V f l l just go buy a gun now and go shoot some peoplew. ItWell 1 think that youfre being a bit selfishN.
Codincr Affect Associated with Parent-Child Conversations
To code the "affective climatew of each verbal utterance, a
second scale from Gottman's (1979) Couples Interaction Scoring
System was employed (see Table 5). For both parent and child, a
code was assigned for the affective climate of each verbal
utterance. Affect codes were given ratings of +, O, or - (positive, neutral, or negative) . To determine this rating, three separate cues could be used: facial cues, voice tone cues,
body cues. Using both the transcript and the videotape, the
coder first considers each of the positive and negative facial
cues. If an affective code of positive or negative is unable to
be made, the coder then considers each of the positive and
negative voice tone cues. If an affective code of positive or
negative still cannot be made, the coder then considers each of
the positive and negative body cues. At this point, if the
verbal utterance still cannot be coded as positive or negative,
the coder is to assign a neutral code to the verbal utterance.
The affective climate of al1 verbal utterances were coded by the
author. Once again, a second individual, blind to the
categorization of dyads as interna1 versus external,
independently coded 20% of the total responses for affective
climate, and agreement between coders was 84%.
Data Analvsis Procedures
Due to the exploratory nature of this study, and to the
small sample size, statistical analyses were not conducted.
Rather, to address the research questions in this study,
Conventional Analvsis
Question #X Expiering Differences in Level of Coerciveness
To explore my first research question histograms were
employed. For each frame of conversation (A, B, and C), averages
were computed for the verbal content and affect scores of both
internal and external groups, and this analysis was completed
separately for parents and children. Histograms allowed for a
descriptive analysis of difierences in parents' and childrens'
affect and verbal content scores, as a function of internal
versus external perceptions of control.
Dvnamic Systems Analyses
Question #2 petacting Phase Transitions
Graphical and Visual Report
To obtain a preliminary look at the communication patterns
of both internal and external dyads, time series graphs were used
(see Thelen, 1991; Van Geert, 1994). On these time series graphs
both verbal content and affect scores were plotted as they
unfolded in real tirne, that is, for each verbal utterance (1
to...) across a l 1 three frames of conversation. First, these
graphs facilitated the visual detection of synchrony within the
communication patterns of parent-child dyads. Second, these
graphs permitted the visual detection of periodicity, and
fluctuations in variance, within the data. It was then possible
to visually inspect the data for evidence of a phase transition
in the communicative patterns of parents and their children.
31
Exploratory Report
Upon cursory inspection of the data, it became apparent that
although there were notable changes in the level and variance of
verbal content scores, there was very little fluctuation in the
variance of affect scores. Thus, it was decided that verbal
content scores would be used to demarcate the onset of phase
transitions in communication. As a secondary evaluation, the
correspondence between the occurrence-of phase transitions in
verbal content, and changes in affect, were observed.
A floating variance graph was created for al1 verbal content
scores across conversation frames A, B, and C. This facilitated
the visual detection of large fluctuations, or inflection points,
in the variance. The occurrence of large inflection points was
considered to signal the onset of an abrupt reorganization, or
phase transition, in a pattern of communication within the state
space. Floating variance graphs were created using two separate
lîwindowsll: a window of 20 verbal utterances (e.g., 1 to 20, 2 to
21, ...), and a window of 10 verbal utterances (e.g., 1 to 10, 2
to 11, ...). First, windows of 20 were used to detect robust
transitions in variance because they were less sensitive to minor
fluctuations than windows of 10. Second, because windows of 10
were more sensitive to changes in the variance they were used to
determine the precise timing of transitions.
An inflection point was considered to signal a phase
transition only if the change in variance lasted a minimum of 20
verbal utterances (half the duration of frarne B on average,
32
across al1 individuals). In al1 cases inflection points were
delineated as ranging across a number of verbal utterances (e.g.,
52 to 56); that is, the sudden change in the variance of verbal
content scores lasted a minimum of a few utterances. Therefore,
phase transitions were denoted as occurring at the middle point
of this change (e.g., verbal utterance # 5 4 ) .
Once phase transitions were identitied, a phase space map
(see Van Geert, 1994) for verbal content scores was created for
intervals immediately preceding and following the transition.
Using phase space maps, it was possible to observe the patterns
characterizing verbal content scores, and to monitor structural
differences in these patterns (e.g., variance, and stability).
Because there was no periodicity in the data, these maps were
created using a lag of 1 ( [ (x, y) = (VC,, , VC,) , (VC,, VC,) , (VG, V C ) , ) Windows of 20 verbal utterances were used to create
phase space maps. Note that the interval, or number of
utterances, delineating inflection points was excluded in the
creation of phase space maps. This ensured the precise and
smooth depiction of cornmunication patterns prior to, and
following, the occurrence of phase transitions.
Recall that there was very little fluctuation in the
variance of affect scores. Therefore, changes in the average of
affect scores were examined. A floating average graph was
created for al1 affect scores across conversation frames. A
window of 10 verbal utterances was used to create these graphs.
Floating average graphs facilitated the visual detection of
33
global changes in the level of affect scores, It was then
possible to observe the correspondence between these changes and
the occurrence of phase transitions in verbal content.
Questions #3, and #4 Measuring Latency and Hysteresis
To rneasure the latency of phase transitions (leading to
coercive communication), the total number of verbal utterances
from the beginning of frame B ( e t stressful task) to the onset
of phase transitions were counted: - Similarly, to measure
hysteresis associated with coercive communication patterns, the
total number of verbal utterances from the beginning of frame C
( . e , pleasant discussion) to the next phase transition were
counted (when communication returned to baseline patterns). It
was then possible to make cornparisons between interna1 and
external dyads regarding the latency of phase transitions, and
the degree of hysteresis associated with coercive communication
patterns.
revealed. Moreover, only inconsequential differences in the
verbal content scores of internal and external children were
revealed (see Figure 4) . For the child group, these results are inconsistent with my expectation.
It is noteworthy that for both parents and children
(internal and external) affect changed from positive (frame A) to
negative during frame B, and subsequently returned back to
positive during frame C. Moreover, for both parents and children
(internal and external), higher levels of negative verbal content
scores occurred during frame B. These results offer support for
the effectiveness of the experimental manipulation employed in
this study ( i . e . , the stressful conversation task in frame B) .
Figure 1,
Eistoaram Depictina the Averacre Affect Scores of Internal and
External Parents Acrosa Conversation Frames A, B, and C
Frame C
Frame A
Frame B
F i g u r e 2
H i a t o c r r a m D e p i c t i n c r the A v e r a c r e V e r b a l C o n t e n t Scores of Interna1
and External Parents Across C o n v e r s a t i o n Frames A, B, and C
Frame A Frame B Frame C
Figure 3
Histoqram D e m i c t i n c r the Averaae A f f e c t Scores of Interna1 and
Extemal Children Across Conversation Frames A, B, and C
.O5
.O4
.O4
. O3
.O31
. O21 ,021
. Ol!
.01i
. OO! O
- . O O ! - . Ol( -. O P - . OS( - ,025 - .03(
-.03!
-. 04( - .O45 -. OSC
Frame C Frame A
Frame B
Internal
Externa1
Figure 4
Hiatoqram Depictincr the Averacre Verbal Content Scores of Internal
and External Children Across Conversation Frames A, B, and C
Frame A
111111111 III Frame B Frame C
9 t U al n b
t-i ci
H S U
4 JJ s Pi W
tn dl k O U tn U d al U E: O U
41
child fluctuated mostly between neutral and negative across a l 1
three frames (see Figure 6) . For this dyad, a simple visual inspection of t h e time series graphs was insufficient to detect
the occurrence of a phase transition.
Figure 5
Time Ser ies Graphs Depicting Verbal Content and Affect Scores - f o r Interna1 D~ads #1 and # 4
Time Series Parent (1)-lnt
- -
Seriesl
Series2 - - - - - - - - - - - I=
# of Utterances
Time Series Child (1)-tnt
# of Utterances
Time Series Parent (4)-lnt
Y of Utterances
Time Series Child (4)-lnt
Seriesl
# of Utterances
Note. Series 2 represents verbal content scores. Verbal content ranges from 1 t o 10, with higher scores ref lec t ing higher Ievels of coerciveness. Ser ies 1 represents a f fec t scores. Affect ranges from -1 t o +1, with lower scores re f l ec t ing more negative a f fec t . A red l i ne is inserted t o separate frames A, 0, and C.
Figure 6
Time Series Graphs Depicting Verbal Content and Affect Scores
for Interna1 Dyads #6 and #8
Time Series Parent (6)-lnt 8 I
r O Q ) a l b C 0 4 ) * ( 3 O J C V P 3 L n b U J r P ' ) r C ) b
r r r r
# of Utterances
Time Series Child (6)-lnt
# of Utterances
- Serlesl 1- . - Series2
Time Series Parent (8)-lnt
# of Utterances
Time Series Child (8)-lnt
# of Utterances
44
For three of the external dyads ( # 3 , # S I and #7), the time
series graphs (see Figures 7 and 8) also revealed a notable
degree of synchrony between the verbal content scores of parents
and their children. For these dyads, patterns of verbal content
scores for parents and children were similar to those of interna1
dyads. That is, verbal content scores fluctuated mostly between
positive and neutral during frames A and C; however, following
the experimental manipulation (i.e., frarne B) there was an
increase in negative verbal content, and scores fluctuated mostly
between neutral and negative. For the fourth external dyad (#2),
the experimental manipulation did not appear to elicit a phase
transition. Communication was marked by higher levels of
negative verbal content, and scores fluctuated mostly between
neutral and negative across al1 three frames (see Figure 8 ) . Note that for this dyad there was less synchrony in the verbal
content scores between parent and child, as there was more
fluctuation in the child's verbal content scores than the
parent ' S.
Affect
The time series graphs revealed that, following the onset of
the experimental manipulation (Le., frame BI, there appeared to
be a notable change in the patterns of affect scores for al1
dyads (see Figures 5, 6, 7, and 8) . In al1 cases, the tirne series graphs revealed that communication appeared to be marked
by more negative affect during frame B than during frames A and
C. Also, in al1 cases, the time series graphs revealed a notable
degree of synchrony between the patterns of affect scores of
parents and their children.
L L L
O0 L
68
8L
L9
9 s
SV
PB
€ 2
e t
L
LPL
LZ L
€CL
66
Ç8
CL
L Ç
€ 9
62
ÇL
C
L L L
O0 C
68
8L
L9
9s
SP
PB
€ 2
Z L
L
LZC
6 0 5
L6
Ci8
E t
C9
6t
LE
ÇZ
E C
C
CZC
60 1
L6
Ç 8
EL
L 9
6P
LE
SZ
6 C
L
I S L
9E L
LZL
1ÇL
9B L
CZC
Exploratory Report
~ecall that phase transitions in communication were assurned
to be denoted by a large change in the variance of verbal content
scores; therefore, al1 inflection points in the floating variance
graphs were marked. Next, phase space maps were used to observe,
and monitor structural differences ( e . g . , variance, and
stability) regarding. the patterns characterizing verbal content
scores prior to and following the marked inflection points.
A l s o , floating average graphs were used to examine the
correspondence between changes in affect and the noted phase
transitions in verbal content.
First, for two dyads and one parent, evidence of a phase
transition in verbal content scores was not revealed (see
Appendices H and 1). For external dyad # 2 , there were no
discernable inflection points in the variance of verbal content
scores. For both parent and child, verbal content scores were
marked by consistently high levels of coerciveness, across al1
three frames. For internal dyad #6, an inflection point in the
variance was not noted for the child. An inflection point was
noted for the parent; however, the variance did not last a
minimum of 20 verbal utterances (criterion for phase transition).
The verbal content scores of this parent and child fluctuated
mostly between positive and neutral. For internal parent #8, an
inflection point in the variance was noted during frame C, but
this transition could not be considered a direct result of the .
experimental manipulation (i . e . , f rame B) . Because the data for
49
the preceding dyads did not reveal a phase transition during
frame 8, these dyads were excluded £ r o m subsequent analyses.
Interna1 Dyads
Dvads #l and #4
For dyads #i and #4, similar results were found. Figures 9
and 10 show that, for these dyads, there was evidence of a phase
transition in the communicative patterns of both parent and
child. It can be seen that, for parents and their children,
there is a considerable increase in the variance of verbal
content scores during frame B (inflection points are rnarked).
As shown in figures 9 and 10, prior to the first inflection
point there was little variance in the verbal content scores of
parents and their children. The phase space maps reveal that
communication was characterized mostly by positive and neutral
verbal content, and this was a stable pattern. However,
following the inflection point there was a high degree of
fluctuation in verbal content scores ( L e . , increase in
variance). A look at the phase space maps show that
communication was marked by higher levels of coerciveness, and
that this pattern remained stable until the second inflection
point, when there occurred a decrease in the variance of verbal
content scores, for both parents and their children. Here, as
revealed by the phase space maps, communication settled back to
its baseline pattern, and verbal content scores fluctuated mostly
between positive and neutral.
Figure 9
Dyad I I : Floating Variance Graph and Phase Space Maps of Verbal Content Scores
-- Note. Series 1 represents a window of SU, and S e n e s L represents a window of - A red l ine i r inserted to separate frames A, B, and C. Blue l ines are inserted to mark inflection points.
Figure 10
Dyad t 4 : Floating Variance Graph and Phase Space Maps of Verbal Content Scores
53
Dvad #8--Child
While the verbal content scores of both parent and child
fluctuated considerably between neutral and negative, and were
for the most part synchronous, evidence of a phase transition in
communication was revealed for the child only. For the child, a
notable decrease in the variance of verbal content scores was
observed during frame B, as shown in figure 11 (inflection points
are marked) . As seen in figure 11, prior to the first inflection point
the phase space map revealed a high degree of variance in the
verbal content scores of this child. Verbal content scores
fluctuated mostly between neutral and negative, and this appeared
to be a stable pattern of communication for this child. However,
following the inflection point verbal content scores ceased to
fluctuate (i.e., decrease in variance), and rather they remained
consistently negative for a period of tirne. A look at the phase
space rnap shows that this pattern of communication remained
stable until the second inflection point, when there was once
again an increase in the variance of verbal content scores.
Here, the phase space map shows that the communication pattern
for this child settled back to its baseline pattern, as verbal
content scores, once again, began to fluctuate mostly between
neutral and negative (see Figure 11) .
Dyad #8 Child: Floating Variance Graph and Phase Space Maps of
Verbal Content Scores
External Dyads
Dvads #3, # 5 , and #7
The communicative patterns of dyads # 3 , #5, and #7 were
quite sirnilar. As shown in Figures 12, 13, and 14, there was
evidence of a phase transition in the communicative patterns of
parents and children. For parents and their children, the
variance of verbal content scores increased considerably during
frame B (inflection points are marked) l.
As seen in figures 12, 1 3 , and 14, prior to the f irst
inflection point there was little variance in the verbal content
scores of parents and their children. The phase space maps
reveal that communication was characterized mostly by positive
and neutral verbal content, and this was a stable pattern.
Following the inflection point however, a high degree of
fluctuation in verbal content scores was observed (i.e., increase
in variance), and communication was marked by higher levels of
coerciveness. The phase space maps show that this pattern of
communication remained stable until the second inflection point,
when there occurred a decrease in the variance of verbal content
scores for both parents and their children. Here, as revealed by
the phase space maps, communication settled back to its previous
pattern, and verbal content scores fluctuated rnostly between
positive and neutral (see Figures 12, 13, and 14).
Note. The first inflection point for the child in dyad #S was less obvious. Therefore, using both the variance windows of 10 and 20, a conservative appraisal was applied and the inflection. point was determined to occur at verbal utterance # 39.
Dyad It3: Floating Variance Graph and Phase Space Maps of Verbal Content Scores
Dyad # S : Floating Variance Graph and Phase Space Maps of Verbal Content Scores
Figure 14
Dyad (7: Floating Variance Graph and Phase Space Maps o f Verbal Content Scores
Patent
I'
59
The preceding results suggest that the experimental
manipulation i . . , frame B) was successful in eliciting a phase
transition in a pattern of communication (Le., verbal content
scores) in approximately 70% of the cases. In a l 1 cases, phase
transitions were marked by a considerable increase or decrease in
the variance of verbal content scores, and communication patterns
were characterized by higher levels of negative, or coercive,
verbal content. These results are consistent with the
expectation guiding rny second research question.
60
lorin nu the Corres~ondence Between Chancres in Affect and Phase
Transitions in Verbal Content
For the previously mentioned dyads, floating average graphs
revealed notable changes in affect across frames A, B, and C.
Depending on the individual histories of dyads, communication
patterns were marked by more negative affect during frame B than
during frames A and C. Floating average graphs w e r e used,
alongside floating variance graphs, to explore visually the
correspondence between changes in affect and phase transitions in
verbal content. Recall that comparisons between averages and
variances were made because these best described the data for
affect and verbal content scores, respectively.
Interna1 Dyada
For dyads #1 and # 4 , the floating average graphs revealed a
change in affect during frame B. During frames A and C, the
communication patterns of parents and children were characterized
by positive affect; whereas during frame B affect was more
negative. A look at figures 15 and 16 shows that, in
correspondence with the timing of phase transitions in verbal
content, average affect scores were at their lowest levels for
both parents and their children. For child #8 , there is only a
slight change in the average of affect scores during frame B.
However. as show in figure 17, this change coincided with the
timing of the phase transition in verbal content, and here,
average affect scores were at their l o w e s t ,
L9L ,
9B C
LZC
99 1
LSC
96 L
LZC QI
90 Lg C .
L6 Q 9 ~ f : s 9 = ' * 9P
LE ;
9 L
1
Dyad #8: Floating Average Graph of Affect Scores and
Floating Variance Graph o f Verbal Content Scores
V. Content
External Dyads
A look at the floating average graphs for dyads # 3 , # S . and
#7, revealed a notable change in the average of affect scores
during frame B. As shown in figures 18, 19, and 20, the
communication patterns of parents and children were characterizet
mostly by positive and/or neutral affect during frarnes A and C;
whereas, during frame B affect was more negative. Moreover, a
look at figures 18, 19, and 20 revealed that, in correspondence
with the timing of phase transitions in verbal content, average
affect scores were at their lowest levels for parents and their
children alike.
For both interna1 and external dyads, the floating average
graphs showed that changes in the average of affect scores,
during frarne B, were characterized by higher levels of neutral
and/or negative affect. Moreover, changes in affect appeared to
be synchronous, or coupled, with phase transitions in verbal
content. That is, following the phase transition, when verbal
content scores were characterized by higher levels of
coerciveness, communication was also characterized by higher
levels of negative affect.
LP L
LZL : €CL j
LPC 1
LZ L
CLL
m o p u r ~ rad a 6 e r a ~ ~
: z Y ' a l al 1; al z i al
LÇL
9E L
CZC
90 1
16
9 L
19
9P
L E
9 L
L
L 9 Y . O 9 d , ? * 7 I 0 0 0 ? ? O ? 1
m o p u ~ rad a6wa~v I
!
68
Question #3
My third research question was concerned with exploring
differences between internal and external dyads in the latency of
the previously mentioned phase transitions. 1 suggested that for
external dyads lower levels of perceived stress might be needed
to induce a phase transition in a pattern of communication, and
hance, less time might elapse (following the onset of frame B)
before this transition occurs.
Table 6 reveals only a minimal difference between internal
and external dyads regarding the latency of phase transitions.
Overall, these results are inconsistent with my expectation.
However, upon closer inspection it is noted that, while the
difference in latency between internal and external children was
minimal, a more substantial difference in latency was obaerved
between internal and external parents. Hence, the preceding
results are partially consistent with rny expectation.
Table 6
Total Number of Verbal Utterances From the Becrinnincr of Frame B
to the Onset of Phase Transitions for Internai and External Dvads
1 ~nternala Dyad #l
Dyad #4
Dyad #8
Parent
17
Dyad #3
Dyad #5
9
- - -
Dyad #7
Chiid
9
O
4
Total
2 6
3
6
5
9
6
6
14
6
18
3
23
8
32
70
Queetion #4
MY fourth research question was concexned with exploring
differences between internal and external dyads regarding the
stability of the communicative patterns (coercive patterns) that
aroçe during frarne B. I suggested that coercive communication
might be a more stable pattern for external, than internal,
dyads. Thus, when decreasing the level of perceived stress
( i . e . , with the onset of frame C), it might take longer for their
communication patterns to return to baseline level; that is, a
hysteresis effect would be more likely to occur.
Table 7 reveals a considerable difference in hysteresis
among internal and external dyads. These results are consistent
with my fourth research question. Note that the observed
difference in hysteresis between internal and external dyads is
somewhat more substantial for the parent group, than for the
child group.
Table 7
Total Number of Verbal Utterances From the Beainnins of Frame C
to t h e Second Phase Transition f o r Interna1 and External Dvads
II internais 1 Parant 1 Child 1 Total
Dyad #3 25 2 1 4 6
Dyad #5 13 11 24
II Dvad #7 1 O 1 O I D
73
relationship. That is, most parents assume a controlling and
authoritarian stance vis-a-vis their children, and overall,
children may be less likely to display excessively high levels of
coercive communication while being observed. Also, the video
camera may have acted to rninirnize any differences between
interna1 and external children. Generally, most of the children,
upon my arrival, informed me that they were very excited about
being "on camera, and it is possible they may have tried (or
were asked by their parents) to put on a "good impression.It
My second research question investigated whether
experimentally increasing the level of perceived stress ( L e . ,
control parameter) would induce a phase transition in the
communicative patterns of parent-child dyads. Moreover, I
expected that phase transitions would result in patterns of
communication characterized by coercive exchanges. Results were
consistent with this expectation. In 70% of the cases, such a
phase transition was observed. In particular, higher levels of
negative verbal content characterized their communication
following the experimental manipulation ( L e . , increasing level
of perceived stress). Moreover, the phase space maps revealed
that this pattern of communication remained stable until levels
of perceived stress were decreased. These results are congruent
with the findings of other investigations highlighting the
relevance of contextual stressors in initiating and maintaining
coercive exchanges among parents and their children (see Belsky,
1984; Campbell, Pierce, March, & Ewing, 1991) .
74
Interestingly, the results also revealed that changes in the
affective quality of parent-child communication appeared to be
coupled, or synchronouç, with phase transitions in verbal
content, for al1 dyads. That is, when communication between
parents and their children was characterized by higher levels of
negative verbal content, affect was also more negative. These
results are consistent with my expectation, and with other
studies demonstrating the importance of affect and arousal in the
initiation and maintenance of coercive exchanges among parents
and children (Bugental, Blue & Lewis, 1990; Dix, Ruble, Grusec &
Nixon, 1986; MacKimon et al., 1990). Zt is notable that the
above results may offer some support for appraisal theories that
point to the reciprocal influences between cognition and emotion
(Frijda, 1993; Lewis, 1996) . My third research question explored differences in the
latency of phase transitions among internal and external dyads.
1 expected that for external (versus internal) dyads, lower
levels of perceived stress might be needed to induce a phase
transition. Consequently, less time would elapse before this
transition occurred, and their