child temp and parenting_sanson-rothbart

Upload: mada-b

Post on 13-Apr-2018

218 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    1/29

    CHILD TEMPERAMENT AND PARENTING

    Ann SansonUniversity of Melbourne

    and

    Mary K. RothbartUniversity of Oregon

    Ann Sanson, Department of Psychology, School of Behavioural Science,

    University of Melbourne, Parkville 3052, Australia

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    2/29

    Running Head: TEMPERAMENT AND PARENTING 2

    INTRODUCTION

    Parents often do not become believers in temperament until after the birth of their secondchild. Before this time, their child's behavior may be seen as a simple and direct outcome of theirupbringing, "a tribute to" or "the fault of" the parents. With the second child, management strategiesthat worked well with the first child may not be effective. Problems experienced with the first child

    (in feeding, sleeping, coping with strangers) may not exist with the second, but new problems mayarise. Such experiences suggest strongly that "nature" as well as "nurture" influences childdevelopment, that children differ from each other from very early in life, and that these differenceshave important implications for parent-child interaction. A number of these individual differences fallunder the rubric of child temperament, the subject of this chapter. Here, we define temperament asindividual differences in reactivity to internal and external stimulation, and in patterns of motor andattentional self-regulation.

    The modern understanding that children make important contributions to their socialinteractions has two roots. The first is temperament research initiated by Thomas and Chess and theircolleagues in their pioneering New York Longitudinal Study (NYLS, Thomas, Chess, Birch, Hertzig, &Korn, 1963). The second is Bell's (1968, 1974) reconceptualization of socialization as a mutuallyinteractive process, with both child and caregiver seeking to redirect, reduce, stimulate, or augmentthe behavior of the other. These insights together have led to the recognition that children differ in

    such qualities as responsiveness to parents' soothing strategies, capacity to control their own emotionalresponsivity, and capacity to bring pleasure or distress to their parents. As Rothbart (1989a, p. 195)put it, "the infant's temperament regulates and is regulated by the actions of others from the earliesthours."

    This chapter explores some of the important mutual influences of parenting and temperament,drawing on both empirical data and clinical insights. We begin by describing the current state ofthinking about temperament, as it has developed from its ancient beginnings, and from Thomas andChess' NYLS study begun in the 1950s. Major dimensions of temperament, their stability overchildhood, and relation to other variables are summarized. We then review empirical evidence forrelations between temperament and parenting. Theoretical and methodological problems arediscussed, and future directions for research are suggested. We also briefly discuss some of theimplications of temperament theory and research for parenting.

    THE NATURE OF TEMPERAMENT

    Historical BackgroundViews of adult temperament as linked to the physiology of the individual were found in the

    ideas of early Greco-Roman physicians; these ideas persisted throughout the Middle Ages (Diamond,1974). In this approach, a fourfold typology was described and related to a balance of the bodilyhumours. The melancholic person, negative and prone to sentimentality and sadness, was seen ashaving a predominance of black bile; the choleric individual, explosive and anger-prone, apredominance of yellow bile; the sanguine person, positive and outgoing, a predominance of blood, andthe phlegmatic person, slow to react, a predominance of phlegm.

    Research on individual differences following this tradition or closely related to it has continuedto the present day. In the twentieth century, the ideas directly influenced the work of Pavlov, Wundt,Ebbinghaus, and Eysenck (Eysenck & Eysenck, 1985). The term temperament, however, was usedprimarily to describe research carried out in the Soviet Union and eastern Europe, whereas personalitywas the term used to describe the mostly factor analytic research done in Britain, western Europe andthe United States. (See Eysenck & Eysenck, 1985; Rothbart, 1989b; and Strelau, 1983). Research onindividuality in infants and young children was originally not closely related to the adult tradition (butsee Rothbart & Derryberry, 1981). This situation is now beginning to change, as similarities betweenbasic dimensions of temperament and personality in children and adults are identified (Halverson,Kohnstamm, & Martin, in press). These are discussed below.

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    3/29

    Running Head: TEMPERAMENT AND PARENTING 3

    In some of the earliest research on temperament in childhood, the NYLS (Thomas, Chess, Birch,Hertzig, & Korn, 1963) took a clinically oriented approach that was strongly related to parenting issues.The NYLS reacted against a tradition that saw parents as responsible for their children's problems.Chess and Thomas noted instances of child psychopathology that occurred with healthy and committedparenting, and other cases where children showed a consistently adaptive developmental course, eveninto adulthood, despite severe parental disturbance, family disorganization, and social stress (Chess &Thomas, 1989). A major starting point for Thomas and Chess was the idea that the child'stemperament must be considered in any discussion of appropriate parenting.

    In addition to its practical emphasis, the NYLS was linked to important theoretical issues, as inthis observation by Chess, Thomas, and Birch (1965, p. 21), "Events in themselves can have nodevelopmental meaning...the environment is first filtered by the child's own characteristics. Childrenwith different characteristics, therefore, will be affected differently by the same objectiveoccurrence. Not only does the child screen his environment, he also influences it...The child, by hisown nature, "conditions" his environment, at the same time that the social and cultural environmentaffects him." The NYLS concept of goodness-of-fit between characteristics of the child andrequirements of the child's environment has been influential in guiding later research, including that onparenting-temperament interactions. We discuss these constructs in greater detail below.

    Temperament DimensionsOnly in recent years have adult temperament/personality and child development traditions of

    temperament begun to come together. Although debate and discussion concerning underlyingelements of temperament continue (Goldsmith, Buss, Plomin, Rothbart, Thomas, Chess, Hinde, &McCall, 1987), recent research on children's temperament has identified a limited number ofdimensions that appear to have parallels in the higher-order factors extracted in studies of adultpersonality (Digman, 1990; Halverson, Kohnstamm, & Martin, in press; Rothbart, 1989a). The threebroad factors that appear to consistently cut across child and adult studies include a positive affectand approach factor (called variously extraversion, surgency or sociability), a negative affect factor,and a control or constraint factor.

    In the early work of the NYLS, Thomas, Chess, and their colleagues analyzed the content ofinterviews of 22 parents of infants 2-3 months and older about their infant's reactions to everydaysituations (Thomas et al., 1963). This analysis produced a set of nine temperament categories:

    Activity Level, Rhythmicity, Approach versus Withdrawal, Adaptability, Intensity, Threshold, Quality ofMood, Distractibility, and Attention Span/Persistence. They also identified behavioral patternsincluding "difficult" and "easy" infants.

    "Difficultness" describes one pole of a cluster identified in the early NYLS work, includingnegative mood, withdrawal, low adaptability, high intensity, and low regularity (Thomas et al., 1963).The opposite pole of this measure was described as "easy." The "difficult child" construct has had astrong influence on the field, and many studies of temperament and parenting have employedmeasures of child difficultness. In subsequent research in the area, however, dimensions making upthis "difficultness" construct have not been found to cluster together (Bates, 1989). This has led someresearchers to develop their own "difficultness" measures. Thus, a "difficultness" measure in one studymay employ the NYLS definition. In another, it may simply include irritability or negative emotionality,and in another, it may reflect the actual empirical clustering of temperament variables. This createsproblems for consolidating findings using the construct.

    The difficultness construct has been criticized in the literature for other reasons (Plomin, 1982;Rothbart, 1982). It adds a value connotation to temperament that may not correspond to the actualfeelings of the parent, and ignores the fact that any temperament characteristic (e.g., high or lowapproach, high or low attentional control) may be "difficult" or "easy," depending on the requirementsof the situation. Despite our reservations about the usefulness of the construct, a number of theresearch studies reviewed in this chapter rely upon "difficulty" because they have been so widelyemployed.

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    4/29

    Running Head: TEMPERAMENT AND PARENTING 4

    The nine more specific NYLS dimensions have also been widely used in research on childhoodtemperament. However, questions have arisen about the extent to which scales measuring thedimensions show conceptual overlap with one another, or are not internally consistent. Because ofthese problems, factor analyses of questionnaire items have been carried out on parent report scales ofinfant temperament derived from the NYLS categories (Bohlin, Hagekull, & Lindhagen, 1981; Sanson,Prior, Garino, Oberklaid, & Sewell, 1987). A review of results from these analyses, related to analysesof scales derived from other theoretical frameworks, suggests that fewer than nine dimensions canadequately account for infant temperamental variability (Rothbart & Mauro, 1990). This "shorter list"of temperament dimensions in infancy includes Fear, Irritability/Anger, Positive Affect (includingapproach), Activity Level, and Attentional Persistence. A sixth dimension of Rhythmicity has beenreliably extracted, but tends to account for only a relatively small portion of the variance.

    Factor analyses of questionnaire items based on the NYLS for older children have revealedsimilar broad factors. Data on a large sample of toddlers in the Australian Temperament Project (ATP)identified factors labeled Irritability, Approach, Cooperation-Manageability, Activity-Reactivity,Rhythmicity, Persistence, Threshold and Distractibility (Prior, Sanson, & Oberklaid, 1989a; Sanson,Prior, & Oberklaid, 1991a). When a second factor analysis was performed on these scales, three broaddimensions emerged, labeled Negative Emotionality, Self-Regulation, and Sociability. The majorfactors emerging from factor analyses of the Thomas and Chess (1977) Childhood TemperamentQuestionnaire, completed by mothers of children in the ATP at the ages of 3 to 8 years, were

    Inflexibility (irritability and uncooperativeness), Persistence, Sociability and Rhythmicity (Sanson,Smart, Prior, Oberklaid, & Pedlow, in press b); these factors remained constant over three age groupswithin this period (Pedlow, Sanson, Prior, & Oberklaid, 1993).

    Using the Children's Behavior Questionnaire, an extensive parent report measure oftemperament for 3- to 8-year-olds (Ahadi & Rothbart, in press; Ahadi, Rothbart, & Ye, in press;Kochanska, DeVet, Goldman, Murray, & Putnam, 1993), three broad factors have consistently beenfound. The first, called Surgency, is defined primarily by the scales of Approach, High IntensityPleasure, Activity Level, and a negative contribution from Shyness. The second, called NegativeAffectivity, is defined by the scales of Discomfort, Fear, Anger/Frustration, Sadness, and loadingnegatively, Soothability. The third factor, labeled Effortful Control, is defined by the scales ofInhibitory Control, Attentional Focusing, Low Intensity Pleasure and Perceptual Sensitivity. The firstthree factors emerging from a recent factor analysis of the NYLS-inspired Middle ChildhoodTemperament Questionnaire items (Hegvik, McDevitt, & Carey, 1982) for 8- to 12-year-olds (McClowry,

    Hegvik, & Teglasi, 1993) also shows striking similarity to these three factors: Approach/Withdrawal,Negative Reactivity, and Task Persistence. Their two smaller factors, Activity and Responsiveness, alsoparallel smaller factors in the ATP, Activity and Threshold (Sanson, Smart, Prior, Oberklaid, & Pedlow,in press b).

    Considering variability in item content, age of children, and whether analyses were based onitems or scale scores, these levels of comparability among factors is notable (Martin, Wisenbaker, &Hutunen, in press). Further, these factors show strong similarities with the "Big Three" factors andthree of the "Big Five" factors that have emerged from analyses of self and peer reports describingpersonality in adult subjects (Goldberg, 1993; McCrae & Costa, 1987; Tellegen, 1985). The NegativeAffectivity factor from childhood measures appears to map on the broad adult dimension ofNeuroticism or Negative Emotionality. The Surgency, Sociability, or Approach/Withdrawal factors mapon the broad adult dimension of Extraversion or Positive Emotionality. The Persistence or Effortful

    Control factors can be seen to map upon the adult dimension of Control/Constraint (see Ahadi &Rothbart, in press).

    Temperament and BiologyMuch data has now accumulated in twin, family, and adoption studies on genetic bases for

    individual differences in temperament and personality (see discussions of Bornstein, Gaughran &Homel, 1986; Goldsmith, 1989; Plomin & Stocker, 1989). Current research and thinking in neurosciencealso continues to make links between temperamental variation and neural substrates (Gray, 1987a,1987b; LeDoux, 1989; Panksepp, 1986; Zuckerman, 1991; reviews by Gunnar, 1990, Rothbart, 1989b;

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    5/29

    Running Head: TEMPERAMENT AND PARENTING 5

    Rothbart, Derryberry & Posner, in press). Just as ancient students of temperament related behavior tothe human body as it was understood at that time, so current research suggests relations betweenindividual differences in behavior and physiology.

    In current frameworks of affective neuroscience, positive and negative affective processesunderlying tendencies to approach (i.e., Surgency or Extraversion) and avoid (i.e., Negative Affectivityor Anxiety) have been related to systems labeled Behavior Activation/Facilitation and Behavior

    Inhibition, respectively (Gray, 1987a, 1987b). These systems have in turn been related to the activityof distributed neural circuits linking cortical brain regions to limbic and brainstem regions. The limbiccomponents of these circuits are proposed to recognize evolutionarily significant information such asthe presence of reward (for Approach) or threat (for Fear). Outputs of these evaluations contactcircuits for motor reactions and the autonomic activity that supports motor action.

    In this view, individual differences in temperament reflect variability in the informationprocessing of the value or significance or events or objects to the individual. This emotionalinformation can be further influenced by attentional systems of the brain (Posner & Petersen, 1990),and individual differences in attentional systems can also be seen to support behavioral persistenceand effortful control (Posner & Rothbart, 1991). Although research linking temperament andphysiology is at an early stage, developments in this area will allow a more complete understanding ofbasic dimensions of temperament in the future.

    Stability of TemperamentIt has generally been assumed that, to be meaningful or important, temperament must show

    substantial stability across time. Typically, however, modest to moderate stability across age has beenfound, with correlations ranging from 0.2 to 0.4 (see Hubert, Wachs, Peters-Martin, & Gandour, 1982;McDevitt, 1986; Rothbart, 1989a; Slabach, Morrow, & Wachs, 1991). Two recent developments,however, may serve to modify conclusions about temperamental stability. First, it has been noted thateven genetic underpinnings do not imply immutability over time (Hinde, 1989). Second, whenconceptual and methodological problems are controlled, higher levels of stability are found.

    Given major changes in a child's behavioral repertoire, manifestations of temperament willlikely change over time. To assess relative stability of an individual's temperamental characteristics, itis therefore necessary to establish continuity in the temperament constructs studied across time(Sanson et al., 1991a). Early work, including research based on the NYLS conceptualization, did notattend to this issue with rigor. Apparent instability may therefore have been due to discontinuity inthe underlying constructs. Another possible source of instability is likely to be error of measurement,also rarely taken into account to date.

    A recent study by Pedlow et al. (1993) on the ATP sample from infancy to 7 to 8 years of ageillustrates the consequences of dealing with these issues. By using structural equation modeling, a setof factors that apply either across the whole age range (Approach/Sociability, Rhythmicity) or acrossthree or more time intervals (Irritability, Persistence, Cooperation-Manageability, and Inflexibility) wasidentified. The model, which corrects for attenuation of correlations due to error of measurement,was then used to assess individual stability on these factors from year to year. Estimates wereconsiderably higher than those previously reported, mostly in the range of 0.7 to 0.8. Even with theselevels of stability, however, there is considerable room for individuals to change in their relativecharacteristics. When children in the ATP were placed in four categories from lowest to highest on the

    basis of their temperament scores at each age, few remained in the same category over all years frominfancy to 8 years (Sanson et al., 1991a). On the other hand, very few changed from one extremecategory to the other extreme.

    Functional SignificanceIt is now generally agreed that individual differences in child temperament are predictive of

    later development and psychiatric risk (Rutter, 1987). Both concurrent and prospective relationsbetween temperament and behavior problems have been documented (e.g., Barron & Earls, 1984;Garrison & Earls, 1987; Kyrios & Prior, 1990; Maziade, 1989; Prior et al., 1989a), but prediction from

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    6/29

    Running Head: TEMPERAMENT AND PARENTING 6

    infancy is relatively weak (Bates, Maslin, & Frankel, 1985; Cameron, 1978; Maziade, 1989; Sanson,Oberklaid, Pedlow, & Prior, 1991c). Early temperament is also associated with more specificbehavioral outcomes such as aggression (Sanson, Smart, Prior, & Oberklaid, in press a). Otheroutcomes studied include physical health, intellectual development, rate of development, child abuse,and reaction to stressful life events such as parental divorce. There is at least tentative evidence forrelations with temperament in all these areas, although the effects are generally small (see Prior,1991). Many of these studies have employed some version of the "difficult temperament" construct;others have tapped specific temperamental characteristics such as negative affect, distress proneness,high activity or intensity, and low adaptability, and have found associations between them andbehavioral problems.

    Several methodological issues must be considered in interpreting this literature on prediction,however. Most of the studies are based on maternal reports of both temperament and outcome, so themeasures are not independent of each other. A concern also arises regarding potential confounding ofthe concepts of extreme or "difficult" temperament and behavior problems. For example, thequestions asked to assess temperament dimensions such as activity, approach/withdrawal, andadaptability may be very similar to those used to assess hyperactivity, shyness and oppositionalbehavior, respectively. For example, the question, "How wary is your child with strangers?" may tapapproach/withdrawal (temperament) or extreme shyness (a behavior problem). A study by Sanson,Prior and Kyrios (1990) suggested that measures of externalizing (acting out) behavior disorders are

    relatively unconfounded with temperament measures, but some confounding occurs for measures ofinternalizing problems (shyness, anxiety). Factors of Activity/Intensity and Irritability showedconfounding with both internalizing and externalizing problems. Although some conceptual overlapmight be expected if temperament contributes to the development of behavior problems (Bates, 1990),caution is required in interpreting the association. This problem may increase with age--"difficult"temperament in infancy may be based largely on emotionality, but among older children it may relatemore to manageability and therefore be conceptually closer to behavior problems (Rothbart, Posner, &Hershey, in press).

    DIRECT ASSOCIATIONS BETWEEN CHILD TEMPERAMENT AND PARENTING

    The expectation that child temperament and parenting would be associated seems reasonable,but it has proved difficult both to predict on theoretical grounds what the nature of the associationsshould be, and to obtain fully persuasive empirical evidence of such links. Methodological issues areone important basis for this difficulty, and we now consider these issues.

    Parent report is the most frequent source of data on infant temperament. If parent report isalso used to assess parenting, there is a clear potential for non-independence of measures, becausecharacteristics of the parent may affect both their parenting practices and their report of their child'stemperament. Because the child's temperament is likely to be affected by prior parenting, anyassociation between concurrent parenting and child temperament may also be the result ofchildrearing history. In her review of the links between infant emotionality and parenting,Crockenberg (1986) noted that few studies attain independence between temperament and parentingvariables. In addition, the apparent effects of parenting on the child may be related to the geneticsimilarity of parent and child (Scarr, 1992).

    Given these caveats, it is not surprising that relatively few studies of parenting and

    temperament allow unambiguous interpretation of results. Because of these problems, we do notattempt a comprehensive review of studies in this area. More detailed reviews of the associationbetween infant negative emotionality and parenting can be found in Crockenberg (1986), and Bates(1987) has reviewed the literature on the influence of temperament on parent-child interaction. Weemphasize here studies whose results are more clearly interpretable. We also stress the importance ofthird variables such as age and gender in influencing relations between temperament and parenting,and finally suggest in a section on joint contributions that some of the most promising thinking andresearch involves using combinations of temperament and parenting variables to predict outcomes inchildren.

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    7/29

    Running Head: TEMPERAMENT AND PARENTING 7

    Empirical Studies of Parenting-Temperament LinkagesDirect associations. Some general positive relations between parenting and temperament

    have been expected: The adaptable, easy to soothe, or sociable child may elicit warm and responsiveparenting, whereas the irritable, demanding, or withdrawing child may elicit parental irritation andwithdrawal of contact or stimulation. Conversely, warm and responsive parenting may decrease the

    expression of negative emotionality in the child, and distant or inconsistent parenting may increase it.There is evidence in favor of these expectations. Most of the relevant studies have focused on distress-related temperament attributes (e.g., irritability, "difficultness," negative mood) which tend to covarywith poor parenting and general unresponsiveness (e.g., Buss, 1981; Campbell, 1979; Crockenberg &Acredolo, 1983; Hinde, 1989; Kelly, 1976; Linn & Horowitz, 1983). Others have noted associationsbetween the child's positive affect and self-regulation and parental responsiveness, social interaction,and use of rewards (e.g., Hinde, 1989; Kyrios & Prior, 1990; West, McLaughlin, Rieser, Brooks, &O'Connor, 1986). The direction of causation is, of course, not clear in these studies.

    However, it is also possible to argue for another association between parenting and childtemperament. If we assume most parents to be highly invested in their children, we might predict thatparents with more irritable or difficult children will exert more positive efforts with them than witheasier children. Crockenberg (1986) cited seven studies finding greater parent involvement withgreater infant irritability; we have been unable to find more. In none of these studies, however, is

    there evidence that "positive" child temperament qualities are related to "negative" parenting; instead,the links are between characteristics like child irritability and higher maternal contact and stimulation.For example, Fish and Crockenberg (1986) found that crying and time to calm at 1 and 3 months in asmall sample of infants were associated with more caregiving and social interaction with the mother at9 months. Similarly, Caron and Miller (1981) found that African mothers were more responsive tohighly irritable babies.

    Moderating variables. Evidence for the two contrasting relations between temperament andparenting discussed above combined with findings of mixed effects (e.g., Klein, 1984) leads to theconjecture that third factors may also be involved. There are also several published accounts of noassociation between temperament and parenting, and given the difficulty in getting such null resultspublished, these may be an underestimate (Daniels, Plomin & Greenhalgh, 1984; Rothbart, 1986;Vaughn, Taraldson, Crichton, & Egeland, 1981; Wachs & Gandour, 1983). It is notable that, althoughthe majority of the null findings are obtained from large-sample studies, most of the studies findingassociations have been obtained from relatively small samples--these findings may be spurious, or itmay be that the influence of competing third variables "cancels out" effects found with smaller andmore homogeneous samples.

    Age. There is indeed suggestive evidence of the role of third variables in moderating theassociation between parenting and child temperament. One, suggested by Crockenberg (1986), is thechild's age. Parents may begin by investing greater effort in their distress-prone child, but not be ableto sustain this effort over time. Consistent with this notion are findings of Peters-Martin and Wachs(1984): At 6 months, infant withdrawal as assessed by mothers' report was related to more maternalemotional and verbal responsiveness, and less restriction and punishment. By 12 months, intensity(another negative affect temperament dimension) was related to less maternal involvement and morerestriction and punishment. It should be noted, however, that these were the only significantcorrelations out of a large set. Similarly, in an observational study by Maccoby, Snow, and Jacklin

    (1984), mothers of boys who were "difficult" (fussy, intense, hard to soothe) at 12 months showed areduction in their teaching efforts in a joint teaching-learning task at 18 months. Among easy-goingboys, mothers' teaching efforts increased over this time. Greater teaching effort at 12 months alsopredicted a decline in boys' difficultness, suggesting bidirectional effects.

    Bates, in a series of reports of a longitudinal study following children from 6 months to 2 years(Bates, Olson, Pettit, & Bayles, 1982; Lee & Bates, 1985; Pettit & Bates, 1984), also found a reversal ofrelations. At both 6 and 13 months, babies with high ratings on a "fussy/difficult" factor from maternalreport and observation received more affectionate contact and object stimulation from their mothers.At 24 months, however, more difficult children resisted their mothers' efforts at control, and received

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    8/29

    Running Head: TEMPERAMENT AND PARENTING 8

    more negative control from their mothers. Although this may partially reflect changes in underlyingdefinitions of "difficult," these findings suggest that some mothers respond to their harder-to-parentinfants with greater efforts, but cannot--or do not--sustain this over time.

    Sex. A recent meta-analysis by Lytton and Romney (1991) found little difference in parentingof boys and girls overall, but the authors did not consider potential sex by temperament interactions.Differences in temperament and parenting associations for boys and girls have been documented.

    Gordon (1983) observed 2- to 4-year-old children interacting with their mothers. Children classified as"easy" on the basis of maternal report did not differ in behavior from "difficult" children, but mothersgave more commands to "easy" than to "difficult" boys, and fewer commands to "easy" than to "difficult"girls. Crockenberg (1986), reanalyzing data from Crockenberg and Smith (1982), found mothers to bemore responsive to the crying of irritable girls than boys.

    Klein (1984) found that children who typically showed intense reactions to stimulation differedin the types of maternal contact they received--highly intense boys received high levels of physicalcontact and intense girls more distal vocal stimulation. Simpson and Stevenson-Hinde (1985)documented better relationships with mothers for shy than for non-shy girls, but the opposite for boys.Two studies of fathers' parenting also found sex differences: Lamb and colleagues (Lamb, Frodi,Hwang, Forstromm, & Corry, 1982) found fathers to be more involved with difficult sons and easydaughters, and Rendina and Dickerscheid (1976) found fathers to be more involved in social activitieswith difficult boys and less with difficult girls. A subgroup of children in the ATP were followed closelyfrom 3 to 7 years (Sanson, Smart, Prior, & Oberklaid, 1993). For these subjects, early child inflexibilitywas related to later parental punishment for girls only, suggesting less parent acceptance of negativityin girls than in boys. Overall, there were more significant correlations between early parenting andlater temperament for girls, suggesting possibly greater responsivity in parenting practices todaughters.

    Differential beliefs about the acceptability and desirability of temperamental attributes forboys and girls might explain these patterns of parental responses, with the predominant patternemerging from these data being more positive responses to boys' difficultness, and lower acceptance ofdifficultness in girls, especially on the part of fathers. It is likely that failing to differentiate betweengirls and boys has resulted in some of the inconsistencies in findings that we have previously noted.

    Maternal characteristics. A third category of moderating variables involves mothers'psychological and social characteristics. Escalona (1968) noted that more anxious mothers tended to

    lose confidence when their usual soothing techniques failed to work for their infants, whereas theconfidence of other mothers was relatively unaffected. She suggested that a sense of maternalincompetence might have far-reaching consequences for mother and child. Her hypothesis wassupported in a study by Teti and Gelfand (1991), who concluded that maternal self-efficacy mediated alink between "fussy-difficult" ratings for infants and their mothers' lower competence (sensitivity,warmth, engagement). Similarly, Gowen, Johnson-Martin, Goldman, and Appelbaum (1989) found thatinfant irritability predicted both depression and a sense of parenting incompetence, and Cutrona andTroutman (1986) found that infant difficultness was strongly related to post-partum depression, bothdirectly and through self-efficacy. Because depressed and nondepressed mothers vary in theirparenting (see Field, in this handbook; Tronick, 1989), temperament difficultness might have anindirect impact on parenting. Genetic overlap between parent and child needs to be taken intoaccount, however, in considering these relationships (Scarr, 1992).

    Extreme children. Temperament and parenting may be more closely related for some childrenthan for others. Buss and Plomin (1984) argued that children with extreme temperamentcharacteristics should be less affected by their environments than those with less extremecharacteristics. A relative immunity to environmental influences would provide one explanation forthe greater stability found for children at extremes on temperamental attributes in the ATPlongitudinal study (Sanson et al., 1991a) and in other data sets (e.g., Kagan, Reznick, & Snidman,1986). Data from two studies suggest that parenting may affect children with some temperamentprofiles more than others. Crockenberg and McCluskey (1985) found a relation between motherresponsiveness at 3 months and babies' crying at 12 months only for babies who were low in irritabilityat 3 months. Feinman and Lewis (1983) noted that 10-month-old infants followed the example of their

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    9/29

    Running Head: TEMPERAMENT AND PARENTING 9

    mothers in being friendly or unfriendly to strangers. This social referencing effect was stronger for"easy" than "difficult" babies.

    In the above two studies, the implied direction of effects is from parenting to temperament, orat least to temperament expression. Maccoby et al. (1984) noted that potential parenting effects ontemperament have been under-researched, probably because of the belief that temperament is in itsnature relatively immutable. However, as indicated in our discussion of stability, temperament is not

    unchanging in its expression, and a case can be made that parenting would be an impetus to change.In many of the studies reviewed to this point, the direction is moot, and may as easily be parenting-to-temperament as the hypothesized temperament-to-parenting.

    Social and cultural factors. In a study of social support, Crockenberg and McCluskey (1985)found that, when mothers had low social support and their babies were more irritable as measured inneonatal testing, the mothers showed less sensitivity to their babies at 12 months. In studies of SES,Bates et al. (1982) and Bates, Maslin, and Frankel (1985) found no consistent SES interactions on theeffect of temperament on parent-child relationships. However, Prior, Sanson, Carroll, and Oberklaid(1989b) examined temperament and parenting practices among groups of 3- to 4-year-old childrendrawn from the upper and lower SES quartiles of a large sample. They found almost twice as manysignificant correlations between temperament factors and parenting dimensions in the high SES groupas in the low SES group, and interpreted this result as evidence of possible greater sensitivity to theindividuality of their children among high SES mothers.

    Numerous studies have found mean differences on temperament scales between children indifferent cultural contexts (e.g., Ahadi, Rothbart, & Ye, in press; Kohnstamm, 1989; Kyrios, Prior,Oberklaid, & Demetriou, 1989), and accounts of cultural differences in parenting practices also abound(see Whiting & Whiting, 1973). No studies appear to have addressed parenting-temperamentinteractions in different cultural groups explicitly, although indirect evidence of a link is often found.For example, DeVries and Sameroff (1984) assessed the temperament of infants from three East Africansocieties. Substantial differences were found between cultural groups in temperament scores, andthese were argued to be due to "child-rearing behavior encoded in social custom" (p. 93).

    Conclusions About Parenting-Temperament LinkagesThe studies reviewed here do not lend themselves to simple conclusions about parenting and

    temperament. Even when methodological problems are taken into account, variability in findings iscommon, and the same temperament characteristics (e.g., negative emotionality, irritability) havebeen shown to be related to both "good" and "poor" parenting. To an extent, Bates' (1987) conclusionthat effects are small, inconsistent, and inconclusive thus appears to remain valid. On the other hand,although the amount of research investigating the influence of third variables is limited, it providespersuasive evidence that non-temperamental characteristics of the child (age and sex), characteristicsof the caregiver (sex, psychological health), and of the caregiving environment (social support, lifestresses, social class and cultural affiliation), all affect links between temperament and parenting.

    These effects in turn suggest that parent attitudes and beliefs about parenting and children arelikely to be very important in temperament-parenting interaction. Super and Harkness (1981; Harkness& Super, this handbook) refer to this set of attitudes and beliefs as the caregiver's "ethnotheory." Thetrend in the data on age effects, for example, makes sense if parents believe the irritable infant "can'thelp it," whereas the more negative toddler is "naughty" or will be "spoiled" if parents give in to a

    negative disposition. Much of the data on sex effects is interpretable in terms of parental beliefs thatboys will be active, intense, and hard to manage, and need to "stand up for themselves," whereas girlswill be more docile and compliant, and need to "be cooperative."

    The ability of parents to adapt to their child's temperamental characteristics may also berelated their own psychological characteristics, such as a sense of competence in the parenting role,stresses they experience, and supports available to them. In her review of parenting associations withinfant irritability, Crockenberg (1986) noted that mothers in three of the studies where an associationwas found between irritability and low maternal responsiveness were in stressed circumstances.Variations are also likely in the extent to which parent behavior is governed by unconscious (automatic)

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    10/29

    Running Head: TEMPERAMENT AND PARENTING 10

    reactions, or by active seeking to understand the child and to adapt parenting accordingly (Papousek &Papousek, this handbook). Social class and cultural differences suggest that parents may differ in theircore beliefs about the nature of the child and whether child individuality "matters." In some culturalsettings (e.g., more collectivistic ones that may value individuality less), it may be seen as lessnecessary or appropriate to adapt parenting styles to a child's particular characteristics. Clearly, morework must be done to understand contextual influences on parenting-temperament relations. Thepromising effects of third variables, however, leads us to turn to more complex relations betweentemperament and parenting, i.e., joint influences of these variables on child outcomes.

    COMBINED EFFECTS OF TEMPERAMENT AND PARENTING

    Early formulations of the idea that similar parenting may have different consequences forchildren with different temperament characteristics have included Escalona's (1968) concept of"effective experience," further developed in Wachs and Gruen's (1982) notion of "organismicspecificity," and the term "goodness of fit," first used in the NYLS (Thomas, Chess, & Birch, 1968) andfurther developed by Lerner and Lerner (1983). In these conceptualizations, variation in an outcomemeasure such as psychosocial adjustment, cognitive development, or behavior problems is seen toresult from differential reactions of children with differing temperament to similar parenting.

    We believe that these more complex views of temperament and parenting may prove to be agood deal more useful than searching for simple direct associations such as those described above.Here we consider two general classes of outcome variable. One is security of attachment, viewed asan outcome of transactions between parent and infant; the second is behavioral adjustment. Thesignificance of temperament on its own for behavioral adjustment has already been outlined; here theemphasis is on interactions between temperament and parenting in relation to behavioral adjustment.Other potential outcome constructs, including cognitive development, school achievement and self-esteem, are not reviewed due to space limitations, and because the research base is generally thinnerin these areas. Bates (1989) has touched upon some of these areas in his review.

    AttachmentThere is an ongoing debate on the relationship of temperament to attachment. Attachment is

    an outcome of the parent-child relationship to which both parent and child contribute, however, and

    hence is relevant to the concerns of this chapter. Although it has been argued that parents' sensitivityto infant behavior is the crucial antecedent to security of attachment (e.g., Sroufe, 1985), there issubstantial empirical support for the notion that the child's temperament is related to how the childreacts during separation and reunion with the parent in the Strange Situation procedure, and thisaffects the child's attachment classification as securely or insecurely attached (see Goldsmith &Alansky, 1987). Some researchers have failed to find direct relations between temperament andattachment security (e.g., Bates et al., 1985; Sroufe, 1985). However, Calkins and Fox (1992) observedthat this most frequently occurs in studies using parent reports of temperament rather thanobservational measures.

    A variety of temperament attributes in infancy has been found to relate to later attachmentsecurity, including sociability to strangers and mother ratings as "easy" (Frodi, 1983), proneness todistress (Belsky & Rovine, 1987), neonatal distress reactivity such as crying at removal of pacifier(Calkins & Fox, 1992), and "object-orientation" versus "person-orientation" (Lewis & Feiring, 1989).

    Temperament characteristics of fear (Thompson, Connell, & Bridges, 1988) and "difficultness" (Weber,Levitt, & Clarke, 1986) have also been found to relate to the infant's negative reactions, such asresistance to the mother in the Strange Situation. One can conjecture about the processes by whichany of these temperamental attributes may be related to the mother-child relationship andattachment. However, none of these studies included measures of parenting.

    Some studies have included measures of both temperament and parenting as predictors ofattachment, allowing us to examine parenting-temperament interactions. Mangelsdorf, Gunnar,Kestenbaum, and Lang (1990) found no main-effect relations between proneness to distress at 9months and security of attachment classification at 13 months. However, they did find that

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    11/29

    Running Head: TEMPERAMENT AND PARENTING 11

    attachment could be predicted from the interaction between maternal personality and infanttemperament. An insecure attachment was more probable when distress-prone infants' mothers hadhigh constraint scores, indicating rigidity, traditionalism and low risk taking. Teti and Gelfand (1991)found secure attachment in preschool children to be related to sensitive, involved and flexibleparenting and the child's sociability towards the mother, with both measures taken from free-playobservations. Insecure attachment was related to the child's negative affectivity (irritability,avoidance, and resistance).

    A final study comes even closer to letting us examine mother-child interactions over time asthey affect attachment, and we consider it in more detail. Van den Boom (1989) used a neonatalbehavior scale administered at 10 and 15 days of age to select an extremely irritable group of infantsrepresenting the top 17% of the low-SES sample tested, and a group of non-irritable infants, drawnfrom the remaining 83% of the sample. She observed the selected infants with their mothers twice amonth to the age of 6 months, and measured mother sensitivity (looking, affective, stimulating, andsoothing behaviors) and infant behavior (positive and negative social signals). A rating scale ofmaternal sensitivity including general attitude, availability, and physical and social contact was alsoused. Both mothers and observers completed temperament scales at 6 and 12 months, and attachmentsecurity was assessed at 12 months.

    Mother and child behaviors differed in relation to newborn infant temperament. More of theirritable infants were classified as anxious/avoidant in attachments to their mothers at 12 months, andtheir mothers were rated as more unresponsive. This suggested bidirectional effects, with childirritability hindering maternal responsiveness, and ineffective maternal soothing behaviors failing toinhibit child irritability. Among the irritable group, van den Boom also found a gradual retreat fromcontact with the baby associated with maternal unresponsiveness. Irritability was associated withperceptions of the infant as "difficult" at 6 and 12 months, and "difficultness" was associated with morematernal noninvolvement with age. These results are both consistent with and extend the studiesreported above, demonstrating how individual differences in irritability and mothers' responsivenessmutually interact to affect the attachment processes.

    Perhaps the most persuasive part of van den Boom's study is its intervention component (vanden Boom, in press). Here 50 low-SES mothers of 6-month-old irritable infants assessed as newbornsreceived specific training in soothing and playing with their babies, and were compared to a matcheduntreated control group of irritable infants. Differences between the intervention and control groups

    were found on measures of quality of mother-child interaction, quality of infant exploratory behaviorat 9 months, and attachment status at 12 months. Intervention group mothers were also moreresponsive, stimulating, and controlling. Their babies were more sociable and exploratory and criedless, and were more cognitively sophisticated in their exploratory behavior. Secure attachment wassignificantly more common in the intervention group (68% versus 28% of the control group). Thuschanges in mothers' behavior clearly led to changes in mother-child interaction and to changes in childbehavior. This study highlights the importance of tracking both temperament and parenting in detailto unravel the bidirectional processes involved. An irritable infant is predisposed to insecureattachment, likely due at least in part to the mother coming to ignore the infant. Interventionprevents this maternal component from further leading the child to develop avoidant (independent)coping strategies.

    This set of studies provides persuasive evidence that temperament facilitates or impedes theattachment process in an ongoing transactional process with parenting. This transactional nature of

    attachment development was suggested by Rothbart and Derryberry (1981, p. 68) over a decade ago:"As important as the mother's sensitivity and flexibility may be, the role of the child's constitutionalcapacities and limitations in shaping her behavior should not be underestimated. Nor should thesensitivity and flexibility of the infant be neglected, for infants vary greatly in their capacity toaugment or reduce their own reactivity, and to bring distress or pleasure to their care-givers. It seemsessential that the mother-infant interaction and the resulting attachment process be viewed as afunction of two intricate and flexible interactional systems, which can achieve a 'balance' in a numberof ways."

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    12/29

    Running Head: TEMPERAMENT AND PARENTING 12

    Behavioral AdjustmentThe first set of studies to investigate the joint influence of temperament and parenting on

    behavioral adjustment was the NYLS (Thomas et al., 1968). They developed the notion of "goodness offit" as a means of analyzing the ontogenesis and evolution of behavior disorders, and a framework fortreatment and intervention. Goodness of fit was said to result "when the child's capacities, motivationsand temperament are adequate to master the demands, expectations and opportunities of the

    environment" (Chess & Thomas, 1989, p. 380). The specific nature of "fit" is undefined, but suggeststhat different styles of parenting will fit different children. The implication is that, because "difficult"babies are more demanding to parent, the usual parenting strategies may be ineffective with them.They may also often elicit poorer parenting; this provides a link between early temperament and laterbehavioral maladjustment.

    Within the NYLS, multiple examples have been reported of apparent mismatches between childand parent preceding poor behavioral outcomes (Chess & Thomas, 1984; Thomas & Chess, 1977). Oneexample is the case of Roy, a highly distractible child. As an infant, Roy's "easy" distractibility allowedparental soothing to be quick and effective. As an older child, however, Roy was often unreliable andforgetful. His mother engaged in extensive nagging to try to gain the child's cooperation. In time, Roycame to "tune out" the mother's messages, and Roy's mother increasingly judged her child in negativeterms. The child's behavior did not improve, and the mother was not willing to appreciate theconsistency of her child's temperament characteristics.

    Other studies can also be conceptualized within a "goodness of fit" framework. For example,Kochanska (1991) found in the development of conscience that anxious children are more affected byparenting practices such as power assertion than non-anxious children. Crockenberg (1987) found thatirritable infants who had angry, punitive mothers were more angry, non-compliant, and less confidentas 2-year-olds than less irritable infants with similar parenting.

    "Goodness of fit" has more recently been operationalized by the Lerners (e.g., Lerner & Lerner,1983) as the discrepancy between the child's actual temperament and others' (usually parent's orteacher's) concepts of the "ideal" temperament for the child. They argue that the same temperamentmay be associated with positive or negative parent-child interactions and outcomes, depending onparents' values and expectations about that temperament attribute. Studies have sought todemonstrate that the discrepancy between "real" and "ideal" temperament is more strongly related tooutcome measures than is the child's temperament on its own. Several studies have provided some

    support for this notion within the school setting, although the improvement in predictive powerprovided by the discrepancy score has usually been small (Keogh, 1986; Lerner, Nitz, Talwar, & Lerner,1989; Talwar, Nitz, & Lerner, 1990). There has been mixed support in other settings (e.g., Hagekull &Bohlin, 1990; Mangelsdorf et al., 1990; Wallander, Hubert, & Varni, 1988). These studies have notincorporated measures of parenting per se, however, so there is no direct evidence that the effect ofthe discrepancy between "real" and "ideal" is mediated by parenting variables.

    This operationalization of "goodness of fit" is also somewhat problematic. Notions of what is an"ideal" temperament tend to have little variability (Windle & Lerner, 1986), so they contribute little todiscrepancy scores. There has also been a tendency to look at a limited number of parental or teacherexpectations or ideals, ignoring other potentially more complex relations such as those between parentand child temperament, or between temperament and the physical environment. Finally, the term"goodness of fit" tends to suggest a symmetrical and unchanging relationship between the child andhis/her world, whereas, as Windle and Lerner (1986) note, it is a dynamic transactional process, and

    measures of it need to capture child-environment interactions over time.

    Another approach to conceptualizing differential effects of parenting on children of differingtemperaments has been Escalona's (1968) concept of "effective experience." She noted how an activechild will seek out toys, whereas an inactive one may require an adult to present them. Thus, thepresence of toys in the house implies different learning experiences for different children. Similarly,Gandour (1989) found children's activity level and the intensity of stimulation provided by the parentsto interact in predicting the exploratory competence of toddlers and the total amount of explorationthey display. The notion of "effective experience" has parallels to "active" effects in Scarr andMcCartney's (1983) model of genotype-environment interactions, wherein children seek out and create

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    13/29

    Running Head: TEMPERAMENT AND PARENTING 13

    environments compatible with their genotypes. Of course, passive effects (mediated throughgenetically based similarities in temperament between parent and child) and evocative effects (wherethe child's temperament "draws out" particular parenting characteristics) likely also apply.

    Others have conducted research framed in terms of "organismic specificity," which like"effective experience," suggests that the same event may have different effects on children who differtemperamentally (Wachs & Gruen, 1982). Wachs (1987), for example, found that for highly-active 12-

    month-olds, parents' naming of objects was related to less mastery behavior, whereas the oppositerelation held for low-active children. A high level of person traffic in the home was related to lowermastery behavior for "difficult" children, but not for "easy" ones.

    Concepts such as effective experience predict differences in outcome for children dependingon particular combinations of parenting and temperament characteristics (that is, multiplicative ornon-linear interaction effects). Bates (1989) comments that, in the interests of parsimony, theindependent contributions of temperament and parenting to outcome (additive effects) should beassessed before addressing any interactive effects. This is a fair comment, but researchers, havingfound additive effects, have often not gone on to investigate potential multiplicative ones. In fact,both additive and multiplicative effects have been found.

    Findings of additive effects are of course also relevant to this discussion, because theydemonstrate that parenting alone (or temperament alone) is not the sole predictor of the child

    outcome. Some theorists predict no additive effects. According to the model proposed by Reid andPatterson (1989) for antisocial behavior, no independent contribution of temperament would beexpected. They posit parenting practices as the intervening variables between temperament andbehavior. They suggest that temperament characteristics can disrupt parental discipline andmonitoring of the child's behavior, but that discipline and monitoring are the proximal causes ofantisocial behavior.

    Additive effects have nevertheless been found by several researchers. Bates and Bayles (1988)followed children from infancy to 6 years. By adding together 0 to 3 year temperament, maternalpositive involvement (affection, teaching) and 3-year-old behavior problems they were able to stronglypredict internalizing (anxious, fearful) and externalizing (acting out) behavior problems for both boysand girls at 6 years. Similarly, Cameron (1978) in analyzing NYLS data found that an index ofdifficultness and persistence at 1 year, along with poorer parenting at 3 years, predicted later behaviorproblems. Additive effects were also found by Fisher and Fagot (1992); here, toddler temperament

    and parental discipline practices were independently related to children's antisocial and coercivebehavior at 5-7 years.

    In a Canadian longitudinal study, Maziade (1989) found evidence for both additive andmultiplicative effects. Of those children who at 7 years had a "difficult" temperament and came fromdysfunctional families (characterized by a lack of rule clarity, consistency, and parental consensus),most had oppositional disorders at 12 years. In contrast, almost none of the children with "difficult"temperament but superior family functioning had behavioral disorders. Maziade labeled thisinteractive effect "synergy." In contrast, at 4 years only additive effects of temperament and parentingon disorder had been found. A study by Martin (1981) of children followed from 10 to 42 months alsofound both additive and multiplicative effects for compliance and coerciveness at 42 months. Forboys, the temperament-like characteristic of demandingness at 10 months and maternalresponsiveness both contributed to later child compliance, but a multiplicative interaction of the two

    also affected outcome. For girls, only infant demandingness predicted (non)compliance. Coercivenessin boys resulted from additive effects of infant demandingness and maternal non-involvement at 10months; these effects were not significant for girls. Thus, despite the somewhat scattered nature ofthe evidence, interactions of particular temperament characteristics with particular parenting do seemto affect behavioral outcome.

    Temperament and parenting have also been conceptualized as risk or protective factors forbehavioral outcome, and have been shown to operate cumulatively. For example, prediction frominfancy to 4-5 year externalizing behavior problems (hostile-aggressive and hyperactive-distractible)and internalizing behavior problems (anxious-fearful) was undertaken for over 1500 subjects from the

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    14/29

    Running Head: TEMPERAMENT AND PARENTING 14

    ATP (Sanson et al., 1991c). Temperament on its own had little impact on outcome: The most"difficult" children, as assessed by an infant "easy-difficult" scale including approach-adaptability,cooperation-manageability and irritability, had only a slightly raised incidence of problems on theoutcome measures compared to the remainder of the children. However, when difficult temperamentoccurred in a context reflecting poor mother-child relationship and presumably poor parenting style,the level of risk for behavioral problems increased substantially. Although other biological andenvironmental factors likely to affect parenting quality also contributed to cumulative risk, thecombination of difficult temperament and poor mother-infant relationship was the most reliable riskindicator. This combination was also particularly characteristic of children who were described ashostile-aggressive at 7 to 8 years (Sanson et al., in press a). These findings are highly reminiscent ofSameroff and Chandler's (1975) construct of the "continuum of caretaking casualty," where infantsidentified at biological risk tended to show negative outcomes chiefly when caretaking was alsodeficient.

    In other studies, temperament has been conceptualized as a mediator affecting the relationbetween aspects of parenting and outcome (Rutter, 1987). For example, temperament has been seenas a resilience factor when there is a high level of psychosocial stress and parenting is poor (e.g.,Werner, 1986; Werner & Smith 1982); in these situations, the sociable or adaptable child may be ableto elicit more care and concern from parents and from significant others, who can act as mentors toprotect the child from adverse outcomes. Temperament has been seen to be important in the divorce

    literature; Hetherington (Hetherington & Henderson, in this handbook; Hetherington, Stanley-Hagan, &Anderson, 1989) suggests that both temperament and parental warmth and control are important indetermining a child's adjustment to marital transitions. Puckering (1989) posits positive aspects oftemperament as a resilience factor for children with depressed mothers.

    These findings of additive effects for both parenting and temperament, and sometimesmultiplicative effects, suggest that temperament cannot be seen as operating only through its effectsof parenting. There seems to be strong evidence that combinations of temperament and parentingvariables affect behavioral outcome.

    For both outcome variables examined here, attachment and behavior problems, there is thussubstantial evidence for interactive processes posited under the concepts of goodness of fit, organismicspecificity, and effective experience. However, the number of studies which have explicitly examinedinteractive effects is quite small, and we are still far from specifying precisely the particular

    configurations of temperament and parenting that constitute a "good fit" and optimize outcome. Aftera discussion of practical implications, future research directions are suggested.

    IMPLICATIONS OF TEMPERAMENTAL VARIATION FOR PARENTING

    On the basis of the empirical evidence reviewed above, what can be said about implications of childtemperament for parenting? Although answers to this question are necessarily speculative because ofthe incompleteness of the research literature, some tentative conclusions can be drawn in three areas:Parental attention to and respect for the child's individuality, parental structuring of the child'senvironment, and applications of the "difficult child" construct.

    Attention to and Respect for IndividualityAn implication of taking children's individuality seriously is that it becomes more difficult to

    give any universal prescription for "good parenting," other than perhaps specifying the need forparental sensitivity and flexibility. Because children may differ in their responses to similar patterns ofparenting, parents need to be attentive to temperament characteristics of their children, and to beable to adapt their parenting behaviors to them. This requires attention to the signals of the childconcerning the child's state and needs. A goal of parenting may then be accomplished in one way forone child, and in a different way for another, depending on the child's temperament characteristics.

    Schaffer and Emerson (1964), for example, noted that babies who dislike being cuddled andresist mothers' attempts at close physical contact are often highly active. Tactile soothing is notcomforting to these children. If mothers recognize this and substitute other more active forms of

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    15/29

    Running Head: TEMPERAMENT AND PARENTING 15

    contact (e.g., play or distraction with toys) for cuddling, problems are less likely to result. Parentsneed to be able to "operationalize" their warmth, concern, and approaches to parenting goals indifferent ways, given the likely reaction of the child to the treatment. Sensitivity is also needed whenparents attempt to modify nonadaptive or unacceptable behavior of the child. Some children will behighly sensitive to punishment; others will be so strongly driven by potential reward that self-controlwill be a problem. Some children will thus need extra encouragement; others will need help withlimits and controls on behavior (Rothbart & Ahadi, in press).

    Another conclusion emerging from the literature is that some temperament characteristicspose more parenting challenges than others, at least in modern Western societies. Although infantcrying and irritability may elicit more maternal contact, this contact often does not seem to besustainable over time, and children's proneness to distress can contribute to the emergence of avoidantor negative, mutually coercive parent-child interactions. Van den Boom's (1989) study, however, showsthat these influences can be countered with extra support and training for mothers of distress-proneinfants. The importance of thoughtful socialization is thus enhanced rather than diminished when thechild's temperament is taken into account.

    Prescriptions for good parenting are also dependent on goals, values, and assumptions aboutoutcomes. Similar temperament characteristics (e.g., shyness, intensity, fussiness) may be reacted todifferently in boys and girls. We suggested that these differences are likely to reflect parents'"ethnotheories" about sex differences and gender roles. Sociocultural variations in ethnotheories arelikely, affecting characteristics seen as desirable and outcomes that are valued. In both of thesecases, however, individuality of temperament should be a consideration, avoiding a tendency to try tofit all children, all girls, or all boys into a single mold. If a "surgent," outward-oriented disposition maybe a more valued style in some cultures (e.g., the U.S.), this may not be the case in other cultures(e.g., China). Recognition of the legitimacy and value of multiple patterns of children's behavior isneeded.

    Structuring the Child's EnvironmentTemperament variation is also important when we take a broader perspective on parenting:

    not only in direct parent behavior toward the child but also in decisions made about daycare, timing ofschool entry, size and structure of school, kinds of extracurricular activities, and so forth. Althoughlittle research provides a direct guide to these decisions, Wachs's (1987; Wachs & Gandour, 1983) work

    suggests that crowded, noisy environments will pose greater problems for some children than forothers. We might also expect that a fearful, withdrawing child would benefit from slower entry to newsituations. Individual differences in attentional self-control should also guide decisions about the timeto start school. Evidence of possible slower maturation of these attributes for boys is relevant to issuesof school readiness.

    "The Difficult Child" and Packaged Parenting ProgramsManuals and courses on parenting abound, and community interventions have been directed

    toward improving parenting skills. How well can these efforts take individuality in child temperamentinto account? Any program giving prescriptions about "the right way to do it" will clearly be deficient ifit does not also direct parents' attention to individuality and to the need to be flexible in theirapproach to childrearing. Some books and programs specifically focus on temperament; examplesinclude Turecki and Tonner (1989) and Cameron, Hansen, and Rosen (1989). In these, there is a focus

    on "difficult" temperament. As noted above, some characteristics are often (but not invariably) asource of difficulty for parents in modern individualistic societies, and acknowledgement that somechildren are harder to parent is often helpful. Advice on how to handle particular "difficult"temperament characteristics can also be useful.

    Against these potential advantages however, there need to be weighted several disadvantages.All the previously-noted problems associated with concept of the "difficult child" apply here. As hasbeen stressed, whether a particular characteristic is "difficult" depends on its fit with the environment,whereas the notion of "difficult temperament" implies that the problem lies in the child. To label achild as "difficult" also has the danger of becoming a self-fulfilling prophecy. The stability of

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    16/29

    Running Head: TEMPERAMENT AND PARENTING 16

    temperament is quite low from infancy to later childhood, and moderate after infancy. If a childbecomes identified as "difficult," this may in itself serve to maintain that status. Indeed, such labelingas "difficult" or "easy" may be one basis for the finding (Kagan et al., 1986; Sanson et al., 1991a) thatextreme temperament characteristics are generally more stable than more moderate characteristics.Family systems theory stresses the importance of assigned roles within families; to be assigned the roleof the "difficult child" may both intensify and maintain the expression of "difficult" characteristics.

    The notion of "difficult temperament" may also lead to the expectation that the parent's majorefforts should be directed toward modifying the child's temperamental expressions, when there isinstead an initial asymmetry in parent and child contributions to interactions (Rothbart, 1989a). Younginfants react to their own internal states and to the immediate situation, including the caregiver'ssoothing and activating stimulation. Caregivers interpret the infant's emotional reactions as signals ofthe need for increasing, decreasing, or changing stimulation. Only at later ages can the child beexpected to play a more active and anticipatory role in the interaction. Thus the initial responsibilityfor adaptation lies strongly with the parent.

    It is clear that their children's temperament should be an important focus for adults whenconsidering their caregiving behavior. Although the research evidence to date does not allow for manyhighly specific recommendations, general principles of sensitivity to the individual characteristics oftheir child, flexibility in parenting behavior in response to these characteristics, and avoidance ofnegative labeling of the child are all important.

    FUTURE DIRECTIONS FOR RESEARCH

    Temperament is an important aspect of the child's contribution to the parent-childrelationship, but our knowledge of parenting-temperament interactions is as yet incomplete. We nowhighlight areas where further understanding may be especially useful. We have noted above thedifficulty of obtaining "clean" measures of both temperament and parenting before they start toinfluence each other. Parents have attitudes about parenting before the birth of their child, however,and although it is well known that the relationship between attitudes and behavior is far from perfect,these attitudes may give us hints about what parenting would be before it is affected by the child'sindividuality. Some researchers have shown that prenatally measured attitudes are related topostnatal temperament ratings (e.g., Heinecke, in this handbook; Vaughn, Bradley, Joffe-Lyle, &

    Seiffer, 1987). Others have indicated that values about parenthood remain relatively constant fromthe prenatal period to 5 months postnatally (Lamb et al., 1982).

    To address the question of parenting-temperament interactions, however, a central issue iswhether parents modify their parenting attitudes and behavior once the child is born, and whether thisis systematic for children with different temperament characteristics. We need to be asking questionslike: You thought X about parenting before your child was born; do you still think so? Is this what youdo? Does it work as you thought it would? There is little direct data on these questions. Observingseveral adults interacting with the same children, each differing in their temperament characteristics,may also inform our understanding of the "active" and "evocative" effects of temperament (Scarr &McCartney, 1983).

    Similarly, to investigate whether and how parenting affects expressions of temperament, weneed evidence of change in temperament when parenting varies, or of mutual parenting-temperament

    changes. Van den Boom (1989), in both the observational and intervention phases of her study,provided a good model of the detailed fine-grained analysis needed to address these issues. Furtherwork following her model of observation and experimental intervention, that might systematicallyaddress other aspects of temperament, other facets of parenting, and other child outcome variablesamong different age groups, would be most beneficial.

    The studies we have reviewed provide a strong case for the importance of third variables forthe temperament-parenting relationship. We suggested that the concept of parental "ethnotheories"may provide a parsimonious explanation of many of the findings. Parents' underlying beliefs, valuesand expectations are likely to significantly affect their responses to children's individual

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    17/29

    Running Head: TEMPERAMENT AND PARENTING 17

    characteristics, including whether parents respond to an irritable and inflexible child with efforts tofind effective ways of soothing and managing the child, or by labeling the child as "difficult" or "bad."Little research attention has been directed to this issue.

    Another area where additional study of parenting-temperament interactions could be highlyinformative is in relation to gender. We noted that interactions between temperament and parentingoften differed for boys and girls. Once again, parental ethnotheories may be involved, but another

    possibility is that actual differences in temperament between boys and girls may also be involved. Ininfancy, while gender differences in temperament are small (Prior, Smart, Sanson, & Oberklaid, 1993),boys have been found to have higher activity levels (Eaton & Enns, 1986). This, combined with thepossibility of earlier language development among girls (Maccoby & Jacklin, 1974), may make girls onaverage more susceptible to early caregiving socialization than boys (Rothbart, 1989a). Activity leveldifferences may also promote cycles of coercive interaction with some boys. There is opportunity forcareful longitudinal work from infancy to address these questions.

    Current research has been somewhat more successful in identifying combined effects oftemperament and parenting for behavioral and cognitive outcomes. Interactive effects occur, bothadditive and multiplicative. However, there is a need for more specificity here, moving from globalmeasures of temperament such as "difficult," which confound several facets of temperament, to morespecific dimensions (self-regulation, negative emotionality). Comparison with normative data is lackingin most studies, so that those classified as, for instance, "irritable" in one study are not necessarilycomparable to those so labeled in another. Observations of parenting in extreme temperament groupsmay be a useful starting point, and studies need to be specifically designed to be able to detect bothindependent and interactive contributions of temperament and parenting dimensions.

    If any of the interactive models (goodness-of-fit, organismic specificity, and so forth) are to betaken seriously, we also need to consider the implications of temperament for broader aspects ofparenting than have traditionally been the focus of research. As noted above, a somewhat neglectedissue has been optimal childcare requirements for children with differing temperament (Bradley &Sanson, 1992). Wachs's (1993) work suggests that optimal levels of stimulation (both physical andsocial) can be defined for children with different temperament characteristics. The implications ofsuch findings for the introduction of children to new settings, the size of social grouping, the age ofentry, and their relevance for a wider age range than has so far been investigated, are all in need ofresearch attention.

    The question of generalizability of results takes us back to "ethnotheories." It may be noaccident that the current interest in temperament and parenting has arisen in individualistic Westerncultures. In more collectivistic cultures, where individuals are defined by their relation to the group,temperamental variation among individuals may be of less relevance and salience than in moreindividualistic Western culture (Kitayama & Marcus, in press). The applicability of conclusions based onWestern samples to other cultural groups thus also needs investigation.

    Finally, a research need that applies to all of the above areas is for investigation oftemperament dimensions that are empirically and theoretically well-grounded. As indicated above, alimited number of temperament dimensions are emerging in the literature, and for some of these,biological underpinnings have been put forward. However, most of the research on temperament andparenting has used more diffuse and sometimes value-laden temperament dimensions such as"difficultness," limiting our ability to study the processes involved.

    For example, if there are differences among children in the strength of reward and punishmentsystems (Behavioral Activation and Behavioral Inhibition systems), and also in the maturation of theirattentional control systems, one could expect the same parenting behavior to have predictablydifferent effects on different children (Kochanska, 1991, 1993; Kochanska et al., 1993; Rothbart &Ahadi, in press; Rothbart et al., 1993). For some children, parents' low-level punishment attempts maynot work, so parents may escalate their punishment in their effort to gain child compliance, thussetting the scene for the development of coercive parent-child cycles of interaction. Researchfocusing upon basic levels of temperamental variation, especially along the lines developed by van den

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    18/29

    Running Head: TEMPERAMENT AND PARENTING 18

    Boom (1989, in press) and Kochanska (1991, 1993) holds great promise for increasing our understandingof temperament and parenting.

    GENERAL CONCLUSIONS

    A child's temperament is apparent from early infancy, and is an important influence on

    individual differences among older children. Variations in reactivity and self-regulation are related tocharacteristic patterns of positive and negative emotionality, sociability, and attentional persistencefor each child. These patterns are fairly stable over time but by no means immutable. They have animpact on a wide range of child outcomes in behavioral, cognitive and social domains. The task forparents in thinking about temperament is to take their child's particular characteristics intoconsideration in choosing strategies to soothe, control and stimulate their child, and in arranging theoverall childrearing environment.

    Guidance can be drawn from analyses of the effects on child outcome of particularcombinations of parent behavior and child temperament. Parental handling can lead to positiveoutcomes even for children extreme in temperament. For instance, van den Boom's (in press) study hasshown us how maternal training in dealing with irritable infants can lead to improved cognitive, socialand emotional outcomes for these infants. Similarly, studies by Wachs and his colleagues (e.g., Wachs,1987) reveal how different levels of social and environmental stimulation appear to be optimal forchildren with differing temperament profiles. More adaptable, sociable and persistent children mayalso cope with stressful life events such as parental divorce better than those with other profiles,presumably partly because they are better able to elicit support from adults around them. It may thusrequire special efforts to provide support to children with less positive profiles when they encounterstressful events.

    We have suggested that some of the consequences of taking temperament into account mightbe to adapt parenting behavior and the child's environment to provide as good a "fit" to the child'stemperament as possible, while at the same time encouraging the child's adaptations to situations; torecognize that, while a child's temperament is not immutable, changes over time are unlikely to bedramatic; and to avoid value judgments about these individual differences. Even though it may berecognized that in a given social and cultural context some children take more effort to parent, thereis nothing inherently inferior about these children, nor are temperament characteristics the result of

    "naughtiness."In sum, the concept of temperament directs our attention to important aspects of child

    individuality that must be considered in parenting. It has long been recognized that appropriateparenting depends on the age of the child; the child's temperament characteristics also determinewhat is appropriate. Even if this recognition complicates both the task of the parent and that of theresearcher, such complication is unavoidable. The task then for the parent and the practitioner is tofoster "respect for the individuality and integrity of each child, and flexibility in creating environmentsthat may lead to positive outcomes for them and for us" (Rothbart, 1989a, p. 236).

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    19/29

    Running Head: TEMPERAMENT AND PARENTING 19

    ACKNOWLEDGMENTS

    Preparation of this chapter was supported in part by National Institutes of Mental Health GrantMH43361 to the second author. The authors wish to thank S. Ahadi, M. Prior and M. Rothbart for theirgenerous help on a previous version of the chapter. Direct inquiries to Ann Sanson, Department ofPsychology, School of Behavioural Science, University of Melbourne, Parkville 3052, Australia, or Mary

    K. Rothbart, Department of Psychology, University of Oregon, Eugene, OR 97403.

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    20/29

    Running Head: TEMPERAMENT AND PARENTING 20

    REFERENCES

    Ahadi, S. A., & Rothbart, M. K. (in press). Temperament, development and the Big Five. In C. F.Halverson, G. A. Kohnstamm, & R. P. Martin (Eds.), The developing structure of temperamentand personality from infancy to adulthood. Hillsdale, NJ: Erlbaum.

    Ahadi, S. A., Rothbart, M. K., & Ye, R. M. (1993). Child temperament in the U.S. and China:Similarities and differences. European Journal of Personality.

    Barron, A. P., & Earls, F. (1984). The relation of temperament and social factors to behavior problemsin three-year-old children. Journal of Child Psychology and Psychiatry, 25MDRV, 23-33.

    Bates, J. E. (1987). Temperament in infancy. In J. D. Osofsky (Ed.), Handbook of infant development(pp. 1101-1149). New York: Wiley.

    Bates, J. E. (1989). Applications of temperament concepts. In G. A. Kohnstamm, J. E. Bates, & M. K.Rothbart (Eds.), Temperament in childhood (pp. 321-355). Chichester, England: Wiley.

    Bates, J. E. (1990). Conceptual and empirical linkages between temperament and behavior problems:A commentary on the Sanson, Prior, and Kyrios study. Merrill-Palmer Quarterly, 36, 193-197.

    Bates, J. E., & Bayles, K. (1988). The role of attachment in the development of behavior problems. InJ. Belsky & T. Nezworski (Eds.), Clinical implications of attachment (pp. 253-299). Hillsdale,NJ: Erlbaum.

    Bates, J. E., Maslin, C. A., & Frankel, K. A. (1985). Attachment security, mother-child interaction, andtemperament as predictors of behavior-problem ratings at age three years. In I. Bretherton &E. Waters (Eds.), Growing points in attachment theory and research. Society for ChildDevelopment Monographs (Vol. Serial No. 209, pp. 167-193).

    Bates, J. E., Olson, S. L., Pettit, G. S., & Bayles, K. (1982). Dimensions of individuality in the mother-

    infant relationship at 6 months of age. Child Development, 53, 446-461.

    Bell, R. Q. (1968). A reinterpretation of the direction of effects in studies of socialization.Psychological Review, 75, 81-95.

    Bell, R. Q. (1974). Contributions of human infants to caregiving and social interaction. In M. Lewis &L. A. Rosenblum (Eds.), The effect of the infant on its caregiver (pp. 1-20). New York: Wiley.

    Belsky, J., & Rovine, M. (1987). Temperament and attachment security in the strange situation: Anempirical rapprochement. Child Development, 58, 787-795.

    Bohlin, G., Hagekull, B., & Lindhagen, K. (1981). Dimensions of infant behavior. Infant Behavior andDevelopment, 4, 83-96.

    Bornstein, M. H., Gaughran, J., Homel, P. (1986). Infant temperament: Theory, tradition, critique,and new assessments. In C. E. Izard & P. B Read (Eds.), Measuring emotions in infants andchildren (Vol. 2, pp. 172-199). New York: Cambridge University Press.

    Bradley, B., & Sanson, A. (1992). Promoting quality in infant day care via research: Conflicting lessonsfrom the day care controversy? Australian Journal of Early Childhood, 17, 3-10.

    Buss, A. H., & Plomin, R. (1984). Temperament: Early developing personality traits. Hillsdale, NJ:Erlbaum.

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    21/29

    Running Head: TEMPERAMENT AND PARENTING 21

    Buss, D. M. (1981). Predicting parent-child interactions from children's activity level. DevelopmentalPsychology, 17, 598-605.

    Calkins, S. D., & Fox, N. A. (1992). The relations among infant temperament, security of attachment,and behavioral inhibition at twenty-four months. Child Development, 63, 1456-1472.

    Cameron, J. R. (1978). Parental treatment, children's temperament, and the risk of childhoodbehavioral problems: I. Relationships between parental characteristics and changes inchildren's temperament over time. Annual Progress in Child Psychiatry and Child Development,233-244.

    Cameron, J. R., Hansen, R., & Rosen, D. (1989). Preventing behavioral problems in infancy throughtemperament assessment and parental support programs. In W. B. Carey & S. C. McDevitt(Eds.), Clinical and educational applications of temperament research (pp. 155-165).Amsterdam, Netherlands: Swets & Zeitlinger.

    Campbell, S. (1979). Mother-infant interaction as a function of maternal ratings of temperament.Child Psychiatry and Human Development, 10, 67-76.

    Caron, J., & Miller, P. (1981). Effects of infant characteristics on caregiver responsiveness among theGusii. Paper presented at the meeting of the Society for Research in Child Development,Boston.

    Chess, S., & Thomas, A. (1984). Origins and evolution of behavior disorders. New York: Bruner/Mazel.(Reprinted in paperback in 1987, Harvard University Press.)

    Chess, S., & Thomas, A. (1989). Issues in the clinical application of temperament. In G. A.Kohnstamm, J. E. Bates, & M. K. Rothbart (Eds.), Temperament in childhood (pp. 378-386).Chichester: Wiley.

    Chess, S., Thomas, A., & Birch, H. G. (1965). Your child is a person. New York: Viking.

    Crockenberg, S. B. (1986). Are temperamental differences in babies associated with predictabledifferences in care giving? In J. V. Lerner & R. M. Lerner (Eds.), Temperament and socialinteraction during infancy and childhood. New Directions for Child Development, No. 31, 53-73. San Francisco: Jossey-Bass.

    Crockenberg, S. (1987). Predictors and correlates of anger toward and punitive control of toddlers byadolescent mothers. Child Development, 58, 964-975.

    Crockenberg, S., & Acredolo, C. (1983). Infant temperament ratings: A function of infants, ormothers, or both? Infant Behavior and Development, 6, 61-72.

    Crockenberg, S., & McCluskey, K. (1985). Predicting infant attachment from early and currentbehavior of mothers and infants. Paper presented at the meeting of the Society for Researchin Child Development, Toronto, Ontario.

    Crockenberg, S., & Smith, P. (1982). Antecedents of mother-infant interaction and infant irritability inthe first three months of life. Infant Behavior and Development, 5, 105-119.

    Cutrona, C. E., & Troutman, B. R. (1986). Social support, infant temperament, and parenting self-efficacy: A mediational model of postpartum depression. Child Development, 57, 1507-1518.

    Daniels, D., Plomin, R., & Greenhalgh, J. (1984). Correlates of difficult temperament in infancy. ChildDevelopment, 55, 1184-1194.

  • 7/27/2019 Child Temp and Parenting_Sanson-Rothbart

    22/29

    Running Head: TEMPERAMENT AND PARENTING 22

    DeVries, M. W., & Sameroff, A. J. (1984). Temperament and infant mortality among the Masai of EastAfrica. American Journal of Psychiatry, 141, 1189-1194.

    Diamond, S. (1974). The roots of psychology. New York: Basic Books.

    Digman, J. M. (1990). Personality structure: Emergence of the five-factor model. Annual Review of

    Psychology, 41, 417-440.

    Eaton, W. O., & Enns, R. (1986). Sex differences in human motor activity level. Psychological Bulletin,100, 19-28.

    Escalona, S. A. (1968). The roots of individuality: Normal patterns of development in infancy.Chicago: Aldine.

    Eysenck, H. J., & Eysenck, M. W. (1985). Personality and individual d