the influence of temperament and mothering on attachment ...people.uncw.edu/hungerforda/graduate...

The Influence of Temperament and Motheringon Attachment and Exploration: AnExperimental Manipulation of SensitiveResponsiveness among Lower-Class Motherswith Irritable Infants

Dymphna C. van den BoomUniversity of Leiden

VAN DEN BOOM, DYMPHNA C. The Influence of Temperament and Mothering on Attachment andExploration: An Experimental Manipulation of Sensitive Responsiveness among Lower-ClassMothers with Irritable Infants. CHILD DEVELOPMENT, 1994, 65, 1457-1477. 6-month-old infantsselected on irritability shortly after birth and their mothers were randomly assigned to 2 interven-tion and 2 control groups to test the hypothesis that enhancing maternal sensitive responsivenesswill improve quality of mother-infant interaction, infant exploration, and attachment. The inter-vention lasted 3 months and ended when the child was 9 months of age. When infants were 9months of age, intervention group mothers were significantly more responsive, stimulating, visu-ally attentive, and controlling of their infant's behavior than control group mothers. Interventioninfants had higher scores than control infants on sociability, self-soothing, and exploration, andthey cried less. Quality of exploration also improved, with intervention infants engaged in cogni-tively sophisticated kinds of exploration more than control infants. At 12 months of age, signifi-cantly more intervention group dyads were securely attached than control group dyads.

Over the last 15 years, researchers have attachments have been identified. The pres-demonstiated that insecurely attached in- ent study is an intervention study conceptu-fants function less well than their securely ally linked to attachment theory,attached peers on a variety of tasks duringtheir toddler, preschool, and beginning According to attachment theory, differ-school years. Insecurely attached youngsters ences in attachment as assessed in theappear less effective in interpersonal rela- Strange Situation are related experientiallytions (Londerville & Main, 1981; Maslin & to differences in the sensitivity of caregiver'sBates, 1982; Sroufe, 1983; Sroufe, Fox, & responses to infant's signals and communi-Pancake, 1983; Thompson & Lamb, 1983) cations during the infant's first year of lifeand less successful in their efforts to master (Ainsworth, Blehar, Waters, 6f Wall, 1978).challenging tasks (Erickson & Farber, 1983; Empirical support for the central role of sen-Main, 1973; Matas, Arend, & Sroufe, 1978) sitivity has been provided by several studiesthan their securely attached peers, and they in which the 1-year-old infants of mothersshow a high incidence of behavioral prob- rated as highly sensitive were significantlylerQs (Erickson, Sroufe, & Egeland, 1985; more likely to be securely attached thanLewis, Feiring, McGuffog, & Jaskir, 1984). those of mothers rated as less sensitiveSuch evidence suggests that attachment the- (Bates, Maslin, & Frankel, 1985; Belsky,ory and research are of clinical significance. Rovine, & Tayior, 1984; Egeland & Farber,Nonetheless, attachment theory has had sur- 1982,1984; Crossmann & Grossmann, 1982;prisingly littie direct impact on the design Isabella & Belsky, 1991; Isabella, Belsky, &and the evaluation of clinical child care, de- von Eye, 1989; Smith & Pederson, 1988). Al-spite the fact that groups of parents and ba- though attachment classification describesbies who are at risk for developing anxious the quality ofthe infant's relationship with

This research was supported by a Netherlands Organization for the Advancement of PureResearch grant (56-212) to the author. Thanks are due to Noortje Gerritsen, Gera Kofnian, YvonneMathijsen, Danielle Mol, Ank Westerink, and Gerlach Zeegers for their help in data collectionand scoring. The author would also like to thank Jay Belsky, Mary Rothbart, and the anonymousreviewers for their thoughtful suggestions. Reprint requests should be addressed to DymphnaG. van den Boom, Department of Developmental and Educational Psychology, University ofLeiden, Wassenaarseweg 52, 2333 AK Leiden, The Netherlands.

[Child Development, 1994,65,1457-1477. © 1994 by the Society for Kesearch in Child Development, Inc.All rights reserved. 0009-3920/94/6505-0006$01.00]

1458 Child Development

a parent, situational factors as well as thecharacteristics of both the parent and the in-fant also contribute to the quality of this rela-tionship (Belsky & Isabella, 1988).

Recently, it has been hypothesized thattemperamental differences in infants affectthe development of infant-mother interac-tions and, thus, the mother-infant attach-ment relationship. In general, temperamentis regarded as a constitutionally based pre-disposition that is stable across time andgeneralizable across situations (Buss &Plomin, 1984; Rothbart & Derryberry, 1981).The temperamental dimension of irritabil-ity, or a low threshold to expression of nega-tive affect, has been shown to infiuence thequality of motherrinfant interaction. Associa-tions between infant irritability and mater-nal behavior have been investigated usingdifferent kinds of measures of irritability.Some studies used measures of infant irrita-bility that are independent of caregiving be-havior (Crockenberg & Acredolo, 1983;Crockenberg & Smith, 1982; Fish & Crock-enberg, 1981; Linn & Horowitz, 1983; Osof-sky, 1976; Osofsky & Danzger, 1974; vanden Boom, 1989), while others used parentreport measures (Bates, Olson, Pettit, &Bayles, 1982; Campbell, 1979; Caron &Miller, 1981; Crockenberg, 1986; Kelly,1976; Klein, 1984; Lee & Bates, 1985; MiUi-ones, 1978; Peters-Martin & Wachs, 1984;Pettit & Bates, 1984; Vaughn et al., 1992).The results ofthe studies using parent reportmeasures are confiicting. Some studies re-port more responsive mothering with irrita-ble babies while others report less involvedcontact. The reason for this inconsistencymay be that parent report measures do notavoid confounding irritability with maternalcaregiving behavior (Crockenberg, 1986).The majority of studies using observationalmeasures of infant irritability seem to sug-gest that a heightened level of irritability isassociated with less maternal involvement.This in tum may negatively influence thequality of attachment. In light of such find-ings, this intervention study was designedaround children selected for being espe-cially irritable as newborns as indexed bytwo separate Brazelton exams.

Current research also underscores theneed to look beyond characteristics of theparent and infant in order to understandwhat transpires between them (Belsky &Isabella, 1988). Thus, the question we nowaddress is whether characteristics of the so-cial context influence mothering behaviorand the quality of attachment. Several stud-

ies reported results indicating that a lowlevel of maternal responsiveness is preva-lent in lower-class families (Mey, 1992), re-sulting in an overrepresentation of anxiousattachments (Koot, 1988). Because wewanted to be able to test the effect of tem-perament and of intervention, we prese-lected a sample that would be at risk forinsecurity not only on a (presumed) temper-amental basis but also on the quality of carethat might be available to the infants.Croups were selected so that the incidenceof stressml life events would be similar andthus an improvement in the interventiongroup Could be attributed to the effective-ness of the intervention rather than tochanges in external circumstances. Hence,in addition to selecting irritable infants, aliinfants came from lower-class families.

In this study, the intervention strategywas directed at promoting secure attachmentby improving the mother's ability to monitorinfant signals attentively, perceive infantsignals accurately, and respond appropri-ately and contingently. More specifically,the concept of felt security (Ainsworth,1973; Waters, 1981) was used as a point ofdeparture for the intervention. Felt securityrefers to the emotional core of a relationshipwith a specific caregiver which is character-ized by feeling safe and protected in thepresence of that person and by feelings oflonging and desire to restore proximity andcontact when that person is absent. Recur-rent experiences of feeling unprotected dueto maternal insensitivity have a profound in-fiuence on the process of internalizing therole of the attachment figure as protector.When the working model of attachment in-corporates salient features of maternal un-protectiveness, the child's ability for self-protection is in turn jeopardized. Distortionsin self-protecting mechanisms may then beobserved in many areas ofthe child's behav-ior, particularly those that involve the needto negotiate a balance between the attach-ment system and the child's involvement inexploration (Lieberman & Pawl, 1990).Thus, the intervention is geared towardameliorating disorders of attachment by fo-cusing on their roots in the mother-child in-teraction. Since the population under studywas a basically well-functioning, lowersocio-economic status sample with normalbut irritable babies, our intervention useda skills-training format emphasizing specificbehavioral goals within an overall frame-work of using the power of contingencies toinfluence interpersonal behavior. The inter-

Dymphna C van den Boom 1459

vention was implemented during naturallyoccurring mother-child interactions in thehome, potentially the most direct way to af-fect behavior.

This report is unique in that it concernsa study designed to influence the critical ma-ternal processes postulated to affect securityto see not only whether the intervention in-fluences sensitive responsiveness but alsowhether intervention and control subjectsvary in rates of security. We hypothesizedthat mothers in the intervention groupwould be more sensitively responsive at theend of a 3-month intervention period (infantage 9 months) than mothers in the controlgroup and that intervention infants would bemore likely to be secure 3 months later (in-fant age 12 months) in the Strange Situationthan control infants. And, given that attach-ment and exploration are supposed to becomplementary behavioral systems^the at-tachment system aimed at promoting prox-imity to/contact with the mother and theexploratory system aimed at gaining infor-mation at)out the environment—we ex-pected that intervention infants wouldengage in more cognitively sophisticated ex-ploration after the intervention period thancontrol infants.

MethodSUBJECTS

The data are based on 100 mother-childpairs observed at infant ages 6, 9, and 12months for whom complete data were avail-able on all measures used. At the time ofthe child's birth, 97 of the 100 families wereintact; three mothers were single. Two fa-thers were unemployed. In all instances themother was the primary caretaker. All in-fants were Caucasian, firstborn, and fromloW-SES families. Social class was assignedaccording to the social group division of theNetherlands Central Bureau of Statistics(1971). All infants were carried to term (be-tween 38 and 42 weeks of gestation) andweighed more than 2,500 grams at birth.Pregnancies were uncomplicated, and deliv-eries did not involve mid- or high-forceps,caesarean section, or breech birth. Mothersdid not receive more than routine medica-tion during delivery. Apgar scores were atleast 7 at 1 min and 8 at 5 min.

The families were located through the

birth register of the Department of Obstet-rics and Gynecology of the Leiden Univer-sity Hospital and through midwives in theLeiden area. The mothers ranged in agefrom 19 to 33 years. The infants were se-lected on irritability on the tenth and thefifteenth day after birth with the NeonatalBehavioral Assessment Scale (NBAS;Brazelton, 1973). The author administeredthe Brazelton exams during the course of thestudy. To assess degree of irritability, scoreson the peak of excitement, rapidity ofbuildup, and irritability items were com-bined to form the neonatal irritability mea-sure identified by Kaye (1978). These threeitems have been used by other researchersinterested in infant irritability (Crocken-berg, 1981). Infants whose mean score fromboth exams on these three items was 6 orhigher were considered irritable; thosewhose mean scores were below 6 were con-sidered nonirritable. To find the predeter-mined number of 100 irritable infants, theNBAS had to be administered to 588 infants.Thus, 17% of infants from low-SES familiesproved to be irritable according to the crite-rion used in the present study. The finalsample consisted of 47 girls and 53 boys.

OVERALL DESIGN

Subjects were randomly assigned to theintervention and control groups, resulting in50 dyads who received the intervention and50 dyads who did not. In addition, all 100dyads received two immediate posttreat-ment assessments (mother-infant interactionand infant'exploration) and a delayed post-treatment assessment (attachment security).

A posttreatment-only control group de-sign poses several problems. On the onehand, it remains unclear whether the groupswere truly equivalent on critical variablesbefore the intervention, which suggests thatpretreatment evaluation is necessary. On theother hand, pretests may have an interactioneffect with the actual intervention. To over-come these problems, half of the interven-tion and half of the control subjects receiveda pretreatment assessment and half did not,creating a factorial design: two treatmentconditions (control and intervention) andtwo pretest conditions (pretest and no pre-test). This four-group design (Solomon &Lessac, 1968) enables assessment of pre-treatment differences between the interven-

' The author was trained to administer the scale at the Child Development Unit of theChildren's Hospital Medical Center in Boston by Dr. Brazelton and his group.


tion and control groups. It also enables as-sessment of the interaction of pretreatmenttesting with the intervention. In addition,the pretested control group was observed asoften as the intervention groups were visitedin order to control for the fact that gettingattention alone would affect maternal be-havior.

Thus, the sample consisted of 25mother-infant pairs per group. The four re-search groups do not differ in demographiccharacteristics (see Table 1).

Numerous tests were performed to as-sess the effect of pretreatment data collec-tion on the effect of the intervention. Virtu-ally no influence was exerted, at least nonethat substantially changed the overall effectsof the intervention found when pretreatmentstatus was not taken into consideration. Be-cause the reporting of the results when ef-fects of intervention are tested in concertwith pretreatment status is unwieldy andadds little of substance, all findings will bereported without regard to pretreatment sta-tus, except for the issue of pretreatmentgroup differences.^

PROCEDURES AND DATA REDUCTION

The reporting of procedures is dividedinto three sections, one dealing with pre-treatment measurement, another with post-treatment evaluation, and the last with theintervention.

Pretreatment AssessmentTwo sets of measurements were taken,

one pertaining to the quality of mother-infant interaction, the other to the quality ofinfant exploration. In addition, mothersfilled out an inventory of life events. Fifty ofthe 100 dyads participated in this phase ofthe study.

Mother-Infant InteractionThe 6-month assessment consisted of

two home visits. During each visit, 40 min ofmother-infant interaction in routine familycontexts (feeding, bath, play, etc.) was ob-served. One visit was scheduled in themorning, the other in the afternoon. Obser-vations were of interactions between motherand infant. When, at times, the mother de-cided not to interact, observations weremade out of the interactional context, with

focus on the child. The interactions were notvideotaped.

Coding of interaction.—All four observ-ers were trained intensively over a periodlasting at least 3 months and were blind togroup assignment of the dyads. While themothers were going about their daily rou-tines, conversation with the observer wasnot encouraged, and the observers weretrained to minimize the potential intru-siveness of their presence. Mother and childbehaviors were recorded at 6-sec intervalsidentified by a beep emanating from a timerand communicated to the observer throughan inconspicuous earplug. During thecourse of the study, formal reliability assess-ments were conducted during in-home ob-servations by having two observers observethe same interaction and code it indepen-dently. Interobserver agreement was com-puted using Cohen's kappa on 25% of thevisits per month for all combinations of ob-server pairs. Reliability coefficients for thecomposite variable sets used in the analysesare included in Table 2.

Coding categories for maternal interac-tive behavior.—Observational categories forthe mother consisted of behavior-based mea-sures of those dimensions of maternal be-havior that have been found to be related tolater quality of attachment, that is, affective,attentive, supportive, soothing, and stimula-tion behaviors. In addition, the coded obser-vational data were used to define a moreconceptually complex measure of respon-siveness: operational measures of a mother'sresponsiveness were computed based on ananalysis of contingencies between observedinfant and maternal behaviors. The infantbehaviors included in the observational sys-tem might be expected to elicit maternal re-sponses (e.g., crying, vocalizing). For eachof these infant behaviors, a set of maternalbehaviors was selected which would indi-cate an "appropriate" response to the spe-cific infant behavior. For example, for infantvocalizes to mother, appropriate responsesfrom the mother were: comes to infant, looksat infant, or verbalizes to infant. Each re-corded occurrence of each of the infant be-haviors was considered, its consequentswithin the same or the next 6-sec periodwere examined, and each materrial behaviorwas given a responsiveness code. Respon-sive behaviors were circled on the codingsheet; unresponsive behaviors were not.

^ More details on analyses examining the effect of pretreatment data collection in interactionwith the intervention are available from the author upon request.

1462 Child Developnient

TABLE 2

OBSERVATION CODE CATEGORY CLUSTERSWITH KAPPAS FOR EACH CLUSTER

Code Cluster Kappa

Matemal interactive behavior:Visual contact 75Physical contact 74Stimulation , 76Effective stimulation 74Control 80Soothing 79Noninvolvement 81Instruction 77Responsiveness to crying 79Responsiveness to positive signals 78Ignoring of crying 76

Infant interactive behavior;Distant contact 64Positive social behavior 72Negative emotionality 76Exploration 72Mobility 79Sucking 76Environmental interest 75

Infant exploratory behavior:Mouthing 67Undifferentiated manipulation 71Vigorous exploration 70Relational exploration 77Speciflc exploration 80Simple exploration 75

Two scores of maternal responsiveness fant signals. The responsiveness scores re-were calculated based upon the proportion fleet the inherent interdependence of inter-of infant behaviors to which the mother re- actors by combining the behaviors ofthe twosponded: responsiveness to the infant's dis- participants into a dyadic score that reflectstress and responsiveness to the infant's so- the functioning ofthe pair.cial signals. It was deemed important to \^ i. ^ u u r j i U i .

, , ^ . . . ^ Attachment researchers have found thatcalculate two separate responsiveness scores . • • e ^ • -n i j J- • ju i i . u i i i r i . 1 . 1 , 1. Ignoring lniant crying will lead to increasedbecause the absolute value ot the base rate . ° , . /n ii * A- _I.U in-roN

r . r . j . . 1 1 • . 1.- 1 1 , - L crying at later ages (Bell & Ainsworth, 1972).01 mtant distress behavior IS relatively high T, .r .i- \ • • l r

. . , , . r . ml.- i • t J To verify this conclusion in our sample ofin irritable infants. This puts special de- . .. i i . r . . j r

J II. iU J.U J irritable miants, a score was computed formands on responses trom the mother and . . c • r . • u iU TI, •1 i i . ^ ,.• r i- Ignoring of infant crying by mothers. This

also means that perception of contingency * ^ ., / cL, j-£D ll r ^1- • r 1 nxT^ score was the percentage of time as a propor-becomes difBcult for the infant (Watson, ^ r • c ,. a • j • tu t^ -u-inncw u 4.1, il, • 1-1 1 4- tion of iniant fussing and crying that these1979) because the mother is likely to re- . , It I ' dspond to infant distress with behaviors that t> y B •she is likely to use in response to other in- To compute the proportional measuresfant signals as well. For instance, if the of responsiveness and ignoring of infant cry-mother's base rate of vocalizing is quite high ing, the number of 6-sec periods in whichand she also uses vocalizing in response to each behavior unit or specified combinationher infant's distress, it will be difficult for of behavior units occurred was tallied, a be-the infant to detect the contingency between havior recorded more than once in a 6-seccrying and vocalizing. If, on the other hand, period being counted only once. Calcula-the mother resorts to effective soothing strat- tions were made for fussing/crying, positiveegies in response to distress, the infant's per- behaviors, and the matemal responsivenessception of contingency will be facilitated, codes. These frequencies were then con-since it is unlikely that the mother will use verted to proportions in order to providesoothing behaviors in response to other in- comparable scores for subjects whose

amount of fussing/crying or positive signal-ing behavior differed in length. Data reduc-tion was performed on the two proportionalscores of responsiveness, the proportionalscore of ignoring of infant crying, and thebehavior-based measures of matemal sensi-tive responsiveness from the observationalcode.

Data reduction of maternal interactivecategories.—The wealth of observationaldsita gathered was reduced to improve thecase-to-variable ratio. Composite variableswere created on empirical grounds (i.e., in-tercorrelations between behavioral catego-ries). Variables that, on inspection of meansand standard deviations, did not occur withsufficient frequency at 6 months weresummed on the basis of their intercorrela-tions at 9 months. Because the infants werestudied during a period of rapid develop-mental change, some variables increasedconsiderably over the age range investigated(e.g., instmction). Since these variableswere observed with high frequency at 9months, it was not deemed appropriate toeliminate these behaviors from subsequentanalyses on the basis of their frequency at 6months. Matemal verbal and physical in-struction were summed because of their lowindividual frequency of occurrence at 6months. For the same reason, the frequen-cies ofthe soothing behaviors were summedtogether. Correlations among the discretevariables included in the composites wereall high (s:.5O). Matemal neutral and posi-tive expression were significantly relatedand were less likely to occur when themother was attending to her child's needs.By summing the frequencies of these threecategories, visual contact was created. Highrates of physical stimulation were associatedwith high rates of bodily contact. The sumof their frequencies constitutes physicalcontact. Mothers who offered their infantsobjects to play with were also high in vo-calizing, and the frequencies of these twobehaviors were summed to obtain stimula-tion. Positive vocalization, stimulate to play,effective physical stimulation, and affection-ate contact were all significantly related.These frequencies were summed to make ef-fective stimulation. High rates of verbal dis-approval were associated with high rates ofphysical restraint. Control was created bysumming the frequencies of these two be-haviors. Matemal noninvolvement, sooth-ing, and instruction were not significantlyrelated to any ofthe other behavioral catego-ries and were retained separately. Together

Dymphna C. van den Boom 1463

with the two responsiveness scores and ig-noring of infant crying, these eight maternalcomposite variable sets were entered intofurther analyses (see Table 2).

Coding categories for infant interactivebehavior.—The observational system for theinfant was focused entirely on attachmentbehaviors that develop during the first yearof life (Ainsworth, 1973) and included ori-enting (look, look at), signaling (positive vo-calization, other vocalizations, smile, fuss,cry, suck), and proximity- and contact-promoting behaviors (positive and negativephysical expression, locomote). Exploratorybehavior (object manipulation) was also in-cluded because of our interest in the attach-ment-exploration balance.

Data reduction of infant interactivecategories.—Data reduction of infant behav-iors proceeded according to the same proce-dure as for matemal categories. Because oftheir Iow frequency of occurrence at 6months, dropping toys and manipulation,positive vocalization and contact mainte-nance, and crying and contact resistancewere summed. For the most part, correla-tions between specific infant behaviors in-cluded in the composites were clear-cut(>.5O). Looking and vocalizing were sig-nificantly related, and their summed fre-quencies constituted distant contact. Posi-tive social behavior was created bysumming the frequencies of positive vocal-ization and smiling, which were signifi-cantly related. High rates of fussing were as-sociated with high rates of crying, and bysumming together the frequencies of theseindividual behaviors negative emotionalitywas created. Picking up toys and manipula-tion were significantly related, and the sumof their frequencies captured exploration.Locomoting, sucking, and looking at per-sons, objects, or events had much lower cor-relations with the other behavioral catego-ries and were, therefore, retained separatelyas mobility, sucking, and environmental in-terest. This analysis resulted in seven com-posite variable sets for infant interactive be-havior (see Table 2).

Infant ExplorationQuality of exploration was measured in

a free play situation. The validity of such anapproach is suggested by previous researchshowing that infant free play behavior thatdraws upon both cognitive and motivationalskills is sensitive to ontogenetic change(Belsky, Garduque, & Hmcir, 1984; Sorce &Emde, 1981) and that the quality of such ac-

1464 Child Developnient

tivity can be influenced by the infant's earlyenvironment in a way predicted by contem-porary attachment theory (Belsky, Goode, &Most, 1980). Therefore, a free play task wasadministered during the first home visitprior to observation of mother-infant interac-tion. During administration, the infant sat onhis or her mother's lap at a feeding table.The mother was instructed to refrain frominteracting with or encouraging the infantwhile the infant played with the toys. At thestart of each session, a warm-up toy (a smallplastic booklet with attractive pictures) waspresented to the infant for 1 min in orderto attract the infant's attention to the task.Following the warm-up period, the infantwas presented with three sets of toys for 5min each. The toys were presented in a fixedorder.

Coding of infant exploration.—All ob-servers were extensively trained to reliabil-ity and blind to group assignment ofthe dy-ads. Every 3 sec, an exploratory behaviorand the toy with which it was performedwere registered. Exploratory behaviors werecoded in vivo. Interobserver agreement wascomputed as for the interactive categoriesusing Cohen's kappa on 75% ofthe visits forall combinations of observer pairs. Agree-ment coefficients are included in Table 2.

Coding categories for infant explora-tion.—The observational system consistedof looking, undifferentiated exploratory be-haviors (e.g., mouthing, simple manipula-tion, rotation), and functional behaviors, thatis, behaviors that fit the specific features ofthe toys. In addition to the categories in-cluded in the observational system, we em-ployed summary and derived measures thatrepresented indices of tempo, attentionspan, and level of exploration.

Several investigators have noted vast in-dividual differences in tempo of explora-tion—the extent to which the infarit movesfrom one toy to the next as opposed to main-taining sustained contact with a single object(Kagan, 1971; Maccoby & Feldman, 1972;McCall, 1974). Wenckstern, Weizmann, andLeenaars (1984), who studied the relationbetween temperament and tempo of play,found that rapid-tempo infants are more in-tense, more negative in mood, and less sen-sitive to environmental stimulation. Thus, tocontrol for the possibility that the irritableinfants, who are intense and negative inmood, might be less sensitive to free playthan nonirritable infants, we included ameasure on behavioral tempo. Perseverance

time (the index of tempo) was computed thefollowing way. First, we computed the totalamount of time spent in physical contactwith a toy (total contact time) and the totalnumber of times an infant made contact witha different toy (act changes). Multiple con-tacts with the same toy were counted as actchanges when the contacts were separatedby intervals of 3 sec or more. Finally, perse-verance times, derived separately for eachinfant, were computed by dividing the con-tact time scores by the number of actchanges.

Although Wenckstern et al. (1984) didnot find a relation between tempo and theability to sustain attention, they suspect thatsuch a relation might be revealed if atten-tional indices that are more subtle than therelatively gross measures of overall attentioncaptured by temperament questionnaireswere used. In exploration sessions whereseveral toys are offered at the same time.Power, Chapieski, and McGrath (1985)found that infants become distracted by onetoy while they are playing with another andthat they spend time sorting through the ob-jects before finding one that eventuallycatches their attention. Following Poweret al. (1985), two measures of attention spanwere computed: total amount of time spentexploring toys during the entire session andmean length in seconds of the two longestperiods of uninterrupted object involve-ment. Attention span was defined as the ten-dency to focus attention on an object or ob-jects for prolonged periods, especially oncesuitable object(s) for extended explorationhad been found. The second measure of at-tention span was chosen instead ofthe aver-age length of object involvement calculatedover all instances because the latter wouldinclude, inappropriately, the brief periods ofobject involvement attendant on the infant'ssorting behavior.

Level of exploration, that is, the highestlevel of exploration observed, was identifiedon a 12-step sequence-of-play scale devel-oped by Belsky and Most (1981). Belsky,Garduque, and Hmcir (1984) showed thatthe highest level of play spontaneously gen-erated during free play is related to the qual-ity of attachment and, possibly, to matemalsensitive responsiveness. It might be ex-pected, therefore, that the interventionwould affect the highest level of play. Thus,the four summary and derived measures in-cluded in the analyses, in addition to the sixcomposite variable sets derived from the ob-servational system (see below), were: perse-

verance time, two measures of attentionspan (total amount of time exploring toys [I];mean length of the two longest periods of'uninterrupted object involvement [II]), andhighest level of exploration.

Data reduction of infant exploratorycategories.—FOI infant exploration, the fol-lowing six composite variable sets were cre-ated (rs > .50). Other mouthing was nega-tively related to both active mouthing andalternate. Their summed frequencies consti-tute mouthing. A significant correlation wasobtained between detailed manipulationand rotating. The frequencies of these be-haviors were summed together to createundifferentiated manipulation. Vigorousmanipulation and throwing with visualguidance were significantly related. Thesummed frequencies ofthese behaviors con-stitute vigorous exploration. High rates ofsimple relational behavior were associatedwith high rates of grouping. Their summedfrequencies make up relational exploration.Specific exploration consisted of thesummed frequencies of functional and func-tional-relational exploration, which weresignificantly related. Since simple manipu-lation was not highly correlated with any ofthe other exploratory categories, it was re-tained separately. Together with persever-ance time, the two measures of attentionspan, and level of exploration, these six ex-ploratory composites constituted the basisfor further analyses (see Table 2).

Factor AnalysesBecause several of the composite vari-

able sets had skewed distributions, datawere logarithmically transformed to promotemultivariate normality. Complete data wereavailable for all subjects. To obtain the factorscores to test for differences between the in-te:rvention and control groups, principalcomponents analyses (VARIMAX rotation)had to be performed on both the total sampleand subsets ofthe data, that is, on data fromthe two pretested groups (N = 50) and thetotal sample (N = 100). The solutions ob-tained were very similar. In each solution,four components were extracted with thesaime variables loading on the components.Only slight differences emerged in the orderin which variables loaded on the factors. Toavoid repetitious presentation of results,only the solution obtained for the two pre-tested groups is presented here.

The 11 maternal interactive compositevariable sets were subjected to a principalcomponents analysis. Four components ac-


counted for 80% of the variance: respon-siveness, stimulation, visual attentiveness,and control. The seven infant interactivecomposite variable sets yielded three princi-pal components accounting for 71% of thevariance: social behavior, self-soothing, andexploration. A principal components analy-sis of the 10 infant exploratory compositevariable sets extracted four components ac-counting for 80% of the variance: sophisti-cated exploration, tempo of exploration, non-specific manipulation, and mouthing. Table3 presents an overview ofthe three principalcomponents solutions.

QuestionnairesDuring the home visits, an inventory of

life events (23 items) adapted from thelist reported by Egeland, Breitenbucher,Dodds, Pastor, and Rosenberg (1979) wasadministered to the mothers to assess stress-ful circumstances. Items covered areas suchas financial problems and family discord.Each item checked was assigned a weightbetween 1 and 5, depending on the degreeof disruption involved and readjustment re-quired. The authors report a mean score of4.0 for a normative sample and a mean scoreof 8.0 for a high-risk sample. A score of 11or higher places the family in the upper 20%for incidence of stressful events. Stressfullife events experienced by subjects in thepresent sample included financial problems,emotional problems, marital discord, andserious illness. The four groups of subjectsdo not differ in reported life stress (seeTable 1).

Posttreatment Assessment

Immediate PosttreatmentThe 9-month assessment again con-

sisted of two home visits. As during the pre-treatment assessment, mother-infant interac-tion was observed for 80 min, quality ofexploration during free play was evaluated,and the inventory of life events was adminis-tered. Pre- and posttreatment observationsfor the same family were performed by dif-ferent observers who were blind to treat-ment condition.

Delayed Posttreatment: AttachmentEach dyad was videotaped in the Ains-

worth and Wittig (1969) Strange Situationwhen infants were 12 months of age. Classi-fications were assigned according to thesecure (B), insecure avoidant (A), insecureresistant (C) (Ainsworth et al., 1978),and insecure disorganized/disoriented (D)groups (Main & Solomon, 1990).


TABLE 3

RESULTS OF THBEE PWNCIPAL COMPONENTS ANALYSES: (1) MATERNAL INTERACTIVE,(2) INFANT INTERACTIVE, AND (3) INFANT EXPLORATORY COMPOSITES

COMPONENTS

I II III IV

.91

.70

.56

.92- .67

Matemal interactive variables:I. Responsiveness (46%):

Responsiveness crying 92Ignoring crying - .92Responsiveness positive 83Effective stimulation 78Soothing 74

II. Stimulation (13%):StimulationPhysical contactInstruction

III. Visual attentiveness (11%):Visual contactNoninvolvement

IV. Control (10%):Control . . . . . . .92

Infant interactive variables:I. Social behavior (32%):

Negative emotionality —.77 . . . . . . . . .Distant contact 72Mobility 64Positive social behavior 63 . . . . . . . . .

II. Self-soothing (21%):Environmental interest —.85 . . . . . .Sucking .67 . . . . . .

III. Exploration (18%):Exploratory behavior . . . .94 . . .

Infant exploratory variables:I. Sophisticated exploration (27%):

Relational exploration 85 . . . . . . . . .Level of exploration 84 . . . . . . . . .Specific exploration 73 . . . . . . . . .

II. Tempo of exploration (26%):Perseverance time .92 . . . . . .Attention span (II) .91 . . . . . .

III. Nonspecific manipulation (15%):Simple manipulation . . . .91 . . .Attention span (I) . . . .64 . . .Vigorous exploration . . . —.62 . . .

IV. Mouthing (12%):Mouthing . . . . . . .79Undifferentiated manipulation . . . . . . - .67

Videotapes were independently coded all tapes; Cohen's kappa of interrater agree-by two observers trained by the author.'' ment was .94.Training was done on pilot data from 40mother-infant pairs. Coders had to achieve Interventionat least 90% reliability on these tapes. Cod- The intervention began 3 weeks afterers were blind to all other data on the fami- pretreatment assessments. Mothers werelies, including irritability or intervention sta- asked if they wished to participate in thetus. Interrater reliability was conducted on firstbom program being offered for first-time

' The author was trained to reliability at the Institute of Child Development in Minneapolisby Dr. Sroufe and his group.


mothers. The program was described as anopportunity for mothers to talk with staffabout their concems regarding child rearing.Mothers were not told about their infant'sindtability classification, since it is not feasi-ble to tell a mother of a healthy, full-termfirstborn that her infant is considered to beat risk for subsequent emotional difficulty.Such labeling might even set in motion aself-fulfilling prophecy; some evidence ex-ists that children labeled as atypical may betreated differently by caretakers (Aber ScBaker, 1990). Three intervention sessionswere scheduled for the dyads of the inter-vention groups while the infants were be-tween 6 and 9 months of age (i.e., one ses-sion every 3 weeks). During the entire3-month intervention period, the author car-ried out the interventions. Thus each dyadhad one and the same intervenor. All visitstook place in the home and lasted about 2hours. The intervention was implementedduring everyday interactions.

The intervention was focused on en-hancing matemal sensitive responsiveness.Toward this end, we assisted mothers in ad-justing their behaviors to the infant's uniquecues, and in doing so we elaborated on whatwe had leamed from the results of an obser-vational study. In the observation study, irri-table infants were found to be less attentiveand less responsive to their mothers andthey seemed to "enjoy" their interactionswith their mothers less than nonirritable in-fants: their smiles and contented vocaliza-tions were less frequent and their fusses andcries were more frequent than those of nor-mal infants. Mothers responded with in-creasing inactivity or passivity to their irrita-ble infants. Hence, the infant's crying seemsto focus events in an unhealthy way. As soonas the infant stops crying, interaction is re-duced to a minimum to prevent the infantfrom starting to cry again (van den Boom,1988). Although infants were selected on ir-ritability shortly afler birth, the interventionstEirted when infants were approximately 6months of age. The results of the observa-tional study showed that having an irritableinfant resulted in a negative cycle of interac-tion becoming apparent from about the fifthmonth of life, despite the fact that the actualamount of fiissing/crying decreased with age(vsm den Boom, 1991). Intervention wasguided by the four stages of the responseprocess: (a) perceiving an infant signal, (fo)interpreting the signal correctly, (c) select-ing an appropriate response, and (d) imple-menting the response effectively. Insensitiv-

ity Could be the result of a deficiency in anyof these stages of the reception-response se-quence.

To infiuence accurate signal perceptionand interpretation by mothers, we began byaffecting matemal attentive behavior. Weencouraged mothers to imitate their infant'sbehaviors, to repeat their own verbal expres-sions, and to maintain silence while their in-fant's gaze was averted. These matemal be-haviors have been shown to be effective inone-time laboratory manipulations (Field,1982). Attentiveness to infant signals is criti-cal for a sensitive maternal response to oc-cur. The accompanying infant gaze behaviorindicates whether the infant is being atten-tive or inattentive, and gaze aversion by theinfant is an often-reported manifestation ofdisturbed interactions (Stem, 1971). Thus,infant gaze behavior provides feedback tothe mother on whether she is accurately in-terpreting her infant's signals. In addition,interventions focused on matemal atten-tiveness have been shown to have the effectof simplifying and slowing down matemalinteractive behavior, presumed to occur be-cause of the greater attentiveness of themother to alterations in her infant's behav-ior. This was also deemed important be-cause it has been shown that irritable infantsare very active, that is, their signeJing behav-ior is high paced (van den Boom, 1991).

Interventions to infiuence the last twostages of the response sequence were in-tended to introduce variety and complexityin the mother's behaviors toward both nega-tive and positive infant signals. Instances offussing or crying served as targets for behav-ior change in sensitive responsiveness tonegative infant signals. If the infant startedto cry, some mothers spontaneously soothedthem, and feedback was provided on the ef-fectiveness of the soothing technique se-lected. If soothing did not occur, motherswere asked to soothe their infants, and theimportance of soothing a crying infant washighlighted. Individualization was very im-portant since some babies like close physicalcontact while others do not. Therefore, themother was offered a number of alternativeways for dealing with the situation to arriveat an optimal meshing of behavior ofthe twointeractive partners.

For mothers with irritable babies, it isalso important to show that interactions canbe positive and rewarding since these moth-ers are inclined to focus their interactivebehavior on negative signals and to ignore


positive ones (van den Boom, 1991). Tech-niques used here included asking themother to play with toys and play infantgames, which typically served to increasethe activity levels of the mothers duringplayful interactions. Here, too, attention wasdevoted to getting an optimal attunement ofmatemal behavior with infant signals.

Careful consideration in tailoring indi-vidual interaction intervention techniquesproved to be very important. While the ma-jority of mothers of irritable babies were ex-tremely active once they initiated interac-tion with their infants, there were somemothers in which activity was typically low.Clearly, interventions for simplifying andslowing down the activity of these motherswould have been inappropriate and ineffec-tive. Hence, the intervention aimed at recip-rocal reading and responding to cues, whichcan have long-range favorable results be-cause mothers will have gained skill in ob-serving their infants, in understanding theimportance of responding to both positiveand negative infant signals, and in leamingto anticipate changes in infant behavior andneeds as a part of development. Our inter-vention was not a psychotherapeutic inter-vention, but a mothering/skill-based one.

After each intervention session, mater-nal and infant behaviors were observed withthe observational system used for pretreat-ment observations. These observational datawere used to evaluate the attainment (ornonattainment) of behavioral goals, to deter-mine whether changes occurred, and toidentify the focus of the next interventionsession.

ResultsThe reporting of results, along with pre-

liminary findings pertaining to the pretreat-ment comparison of the intervention andcontrol groups, is divided into three sec-tions: the first deals with preliminary analy-ses to establish the initial equality of meansof the groups, the second with immediateposttreatment effects, and the last with de-layed posttreatment effects.

The identified patterns (factor scores)were submitted to (repeated-measures)MANOVAs to test for differences betweenthe intervention and control groups on ma-ternal and infant interactive behavior and oninfantile exploration behavior.

PRELIMINARY ANALYSES

To establish the initial equality ofmeans of the groups, preliminary ANOVAs

were applied to the pretreatment observa-tions of matemal interactive behavior. Nosignificant differences were found on any ofthe variable sets, with all F values (1, 48) <3.90, ps > .05, except for effective stimula-tion on which the intervention mothers hadlower rates (M = 1.46, SD = 1.05) than thecontrol mothers (M = 3.82, SD = .96), F(l,48) = 81.49, p < .001.

Preliminary ANOVAs applied to thepretreatment infant interaction rates did notdiffer significantly on five of the seven vari-able sets among the intervention and controlconditions, with all Fs (1, 48) < 3.00, ps >.05. Control group infants scored higher onpositive social behavior (M = 2.65, SD =.73), F(l, 48) = 57.89, p < .001, and mobility(M = 3.34, SD = 1.47), F(l, 48) = 50.76, p< .001, than intervention infants (M = 1.28and 1.29, SD = .83 and 1.08, respectively).

At the time ofthe pretreatment observa-tions, infant exploratory behavior did not dif-fer significantly on eight of the 10 variablesets among intervention and control condi-tions, with all F values < 3.25, ps > .10. In-tervention infants showed more simple ma-nipulation (M = 5.23, SD = .16), F(l, 48) =12.94, p < .001, and less vigorous explora-tion (M = 2.20, SD = 1.08), F(l, 48) =14.19, p < .01, when compared to the controlinfants (M = 4.91 and 2.80, SD = .41 and1.07, respectively).

No sex differences were obtained inseparate analyses centered on the pretreat-ment variable. All data, therefore, will be re-ported for the sexes combined.

IMMEDIATE POSTTREATMENT EFFECTS

At 9 months of age, effects of the inter-vention were estimated on the quality of in-teractive behavior. Maternal and infant in-teractive behavior were analyzed separatelyto document changes in each partner's be-havior. Effects ofthe intervention were alsoexamined for infant exploratory behavior.Finally, changes in reported life stress wereanalyzed over the two points of measure-ment (6 months and 9 months infant age).

Effect of Intervention on MatemalInteractive Behavior

To assess the effectiveness ofthe inter-vention a 2 (pretest: yes vs. no) X 2 (inter-vention vs. control) MANOVA was carriedout On the four matemal components: re-sponsiveness, stimulation, visual atten-tiveness, and control. With the use of Wilks'scriterion, the combined dependent variables

were significantly affected by interventionF(4, 93) = 74.55, p < .01, and by the interac-tion between pretest and intervention, F(4,93) = 2.64, p < .05. There was no significantmain effect of pretest. Univariate tests indi-cated that intervention mothers were sig-nificantly more responsive, F(l, 96) =176.75, p < .001, more stimulating, F(l, 96)= 35.56, p < .001, more visually attentive,F(l, 96) = 68.74, p < .001, and more control-ling of their infant's behavior, F(l, 96) =26.23, p < .001, compared to the controlmothers. The mean factor scores for the in-teivention mothers were 4.32 (SD = 1.71),1.57 (SD = 1.18), 1.04 (SD = .79), and 1.24(SD = .87); control group mothers, on theother hand, had mean factor scores of —1.84(SD = 2.48), - .74 (SD = 2.45), - .52 (SD= 1.04), and .15 (SD = 1.22). The interac-tion effect for intervention x pretest wassig;nificant only for visual attentiveness, F(l,96) = 4.77, p < .05. Post hoc comparisonsindicated that the pretested control grouphad significantly lower mean factor scores onvisual attentiveness (M = - .96, SD = .78)than did the control group that was not pre-tested (M = - .09, SD = 1.31). No signifi-cant differences were found between thetwo intervention groups.

A MANOVA for repeated measures wasperformed on mothers' behaviors in the in-tervention and control group that received apretest at 6 months of age. The within-subject factor was time (6 vs. 9 months). Thisanalysis showed multivariate main effects ofintervention, F(4, 45) = 9.09, p < .001, andtime, F(4, 45) = 66.24, p < .001, which werequalified by an interaction of intervention xtime, F(4,45) = 43.73, p < .001. The univar-iate tests showed that the interaction effectswere significant for all matemal compo-nents, that is, for responsiveness, F(l, 48) =122.19, p < .001, stimulation, F(l, 48) =46.49, p < .001, visual attentiveness, F(l, 48)= 38.90, p < .01, and control, F(l, 48) =41.46, p < .001. In the intervention group,rates of'maternal responsiveness (Men,onths =-3,.53, SDe^o jhs = 1-75; M^^,^^ = 4.18,SDQ nionths = 2.39), Stimulation (Mg months =-1.19, SDe ^««a,, = 1-6; Mg ^^^^, = 1.98,SDg months = 1.38), and visual attentiveness(^Smonths = ~ - 3 5 , SDgmonths = 1-12; Mgmonths= 1.01, SDg months = -84) increase over thetwo time points, these behaviors decreasedslightly in mothers from the control group(A eimonths = - -53 , -10 , and - .47, SDgmonths= 2.27, 1.66, and .93; Mg months = -1-62,- .64, and - .96, SDgmonths = 2.83, 2.36, and.78, respectively). Maternal control in-


creased in both groups between 6 and 9months of age, but the increase was largerin mothers from the intervention group(Me months = - 1 . 7 2 , SDe months = -85; Mg months= 1.49, SDg months = -76) compared to thecontrol group mothers (Mg months = - - 8 4 ,SD6 months = -78; Mg months = -17, SDg months— 1.13).

In sum, mothers who participated in theintervention did differ in meaningful waysfrom the control group mothers, and the dif-ferences were rooted in their responsive andstimulating orientation toward their infants.

Effect of Intervention on InfantInteractive Behavior

A 2 (pretest: yes vs. no) x 2 (interven-tion vs. control) MANOVA was used to as-sess differences on the infant components byintervention and pretest. Significant main ef-fects of intervention and pretest were foundusing Wilks's lambda test of significance,F(3, 94) = 56.55, p < .001, F(3, 94) = 10.60,p < .001, respectively. There was no sig-nificant intervention x pretest interaction.Univariate tests indicated that infants in theintervention groups were significantly moresociable (M = 1.35, SD = .92), F(l, 96) =31.73, p < .001, more self-soothing (Af = .82,SD = .71), F(l, 96) = 29.54, p < .001, andexplored more (M = 3.43, SD = .91), F(l,96) = 94.34, p < .001, when compared tothe control group infants (M = .22, - .04,and - .17, SD = 1.07, .86, and 1.01, respec-tively). The effect for pretest was significantonly for social behavior, F(l, 96) = 13.45, p< .001, and self-soothing behavior, F(l, 96)= 10.85, p < .001, with pretested interven-tion infants being more sociable (M = 1.88,SD = .87) and more self-soothing (M = 1.13,SD = .73) compared to the pretested controlgroup infants (M = .43 and .17, SD = .96and .91, respectively).

The entire population of the presentstudy consisted of irritable infants. Irritableinfants have been shown to fuss and crymore and to exhibit fewer positive social be-haviors than nonirritable infants (van denBoom, 1988). Interest in these specific be-haviors led to examination of changes frompretreatment to posttreatment for the vari-able sets positive social behavior and nega-tive emotionality. ANOVAs yielded signifi-cant differences for intervention and controlinfants on each ofthese variable sets. Thesedifferences indicate that intervention infantsdisplayed more positive social behavior di-rected to the mother (M = 4.72, SD = .36),F(l, 96) = 139.78, p < .001, and showed less


negative emotionality during interactions (M= 2.10, SD = 1.20), F(l, 96) = 44.11, p <.001, compared to control infants (M = 2.98and 3.39, SD = 1.01 and 1.14, respectively).For each of these variable sets, we found amain effect for pretest condition, with an Fvalue for positive social behavior of F(l, 96)= 5.72, p < .025, and for negative emotional-ity of F(l, 96) = 9.07, p < .01. The interac-tion between intervention x pretest wasalso significant for positive social behavior,F(l, 96) = 8.79, p < .01, and negative emo-tionality, F(l, 96) = 5.19, p = .025. Contraryto our expectations, post hoc comparisons re-vealed that the control group that did notreceive a pretest displayed higher mean fac-tor scores on positive social behaviors (M =3.33, SD = 1.04) than the pretested controlgroup (M = 2.62, SD = .85). No differencesemerged between the pretested and no-pretest intervention groups. On negativeemotionality, infants in the interventiongroup that did not receive a pretest scoredhigher (M = 2.68, SD = 1.09) than infantsin the intervention group that did receive apretest (M = 1.47, SD = 1.01). In this in-stance, no differences emerged between in-fants in the pretested and no-pretest controlgroup.

A repeated-measures MANOVA on theinfant components ofthe pretested interven-tion and control group yielded a significantmultivariate main effect of intervention, F(3,46) = 9.22, p < .001, and time, F(3, 46) =134.31, p < .001. The multivariate effect ofthe intervention x time interaction, F(3, 46)= 62.73, p < .001, indicated the followingdifferences between the intervention andcontrol group. Univariate tests showed thatthe interaction effect was significant for so-cial behavior, F(l, 48) = 67.06, p < .001,self-soothing, F(l, 48) = 51.49, p > .001, andexploration, F(l, 48) = 36.53, p < .001. In-fant social behavior and exploration in-creased in both groups between the twopoints of measurement (6 and 9 montibs ofage). In the intervention group, both behav-ioral components increased at a higher rate(Ms months = -2.81 and -1.67, SDg months =1.56 and 1.31; Mg months = 1-88 and 1.98,SDg months ~ .87 and 1.04, respectively) thanin the control group (Mg months "= — -35 and-1.41, SDg months = 1-02 and 1.05; Mg months= .43 and - .09, SDg months = -96 and .87,respectively). For self-soothing, the oppositetrend could be observed; that is, this behav-ior increased in the intervention group(M6 months = -1.77, SDg months = 1-32;Mg months = 1-13, SDg months = -73), but it de-

creased slightly in the control group over thetwo time points (Me months = -23, SDg months= .72; Mg months = -17, SDg months = -91).

Differences were predicted that favoredintervention groups over the control groupson the following aspects of infant interactivebehavior: more positive social behavior, lessfussing and crying, and more exploration.The most consistent finding was that onposttreatment observations intervention in-fants were more sociable, more able tosoothe themselves, and explored more com-pared to control group infants.

Effect of Intervention on InfantExploratory Behavior

A major purpose of observing the in-fant's exploratory behavior was to examinechanges in the quality of this behavior afterintervention. To assess the effectiveness ofthe intervention, a 2 (pretest: yes vs. no) x2 (intervention vs. control) MANOVA wasconducted on the exploratory composites.This analysis revealed significant multivari-ate intervention and pretest effects, F(4, 93)= 5.26, p < .001, F(4, 93) = 3.21, p < .025,respectively. No significant intervention Xpretest interaction was obtained. UnivariateANOVAs indicated that intervention infantsshowed significantly more sophisticated ex-ploration (M = .67, SD = 1.61), F(l, 96) =12.64, p = .001, and less mouthing (M =- . 3 1 , SD = 1.42), F(l, 96) = 4.61, p < .05,when compared to control group infants (M= - .67 and .31, SD = 2.08 and 1.44, respec-tively). No significant differences werefound for either tempo of exploration or non-specific manipulation. The effect for pretestwas significant for sophisticated exploration,F(l, 96) = 7.66, p < .01, and nonspecificmanipulation, F(l, 96) = 5.04, p < .05. Posthoc comparisons revealed that infants whowere pretested received higher mean factorscores on sophisticated manipulation (M =1.45, SD = 1.69) than did infants who didnot receive a pretest (M = — .12, SD = 1.53).The opposite was found for nonspecific ma-nipulation. In this case, infants who did notreceive a pretest had higher scores (M = .33,SD = 1.21) than pretested infants (M = .13,SD = 1.47).

A repeated-measures MANOVA on thepretested groups' sophisticated exploration,tempo of exploration, nonspecific manipula-tion, and mouthing revealed a significantmultivariate main effect of intervention, F(4,45) = 5.80, p < .001, and a multivariate in-teraction effect of intervention x time, F(4,45) = 6.10, p £ .001. There was no signifi-


TABLE 4

INTERCORRELATIONS BETWEEN INTERACTIVE AND FREE-PLAYMEASURES OF EXPLORATION

Free Play

Sophisticated exploration .Tempo of explorationNonspecific manipulationMouthing

*p< .05.

Explorationduring Interaction

.28*- .01

.09-.24*

cant main effect of time. Univariate tests in-dicated that the interaction effect was sig-nificant for sophisticated exploration, F(l,48) = 16.25, p < .001, with an increase intliis kind of exploratory behavior in infantsin the intervention group over time (Mg months=-- - .77, SDg months = 2.44; Mg months = 1-45,SDg months = 1-69), while in control groupinfants rates of sophisticated exploration de-creased overtime (Mgmonths = -55, SDgmonths== 2.57; Mg months = --42, SDg months = 2.41).Nonspecific manipulation was also signifi-cant, F(l, 48) = 5.49, p < .025, indicating adecrease over time in rates of this behaviorin intervention infants (Mg months ~ 1-02,SDg months = 1-08; Mg months = -13, SDg months=• 1.47), and an increase over time in thecontrol infants (Mg months = -1.43, SDg months= 2.55; Mgmonths = - .84, SDgmonths = 1-77).Finally, mouthing proved to be significant,F(l, 48) = 7.57, p < .01, that is, over time itdecreased in infants from the interventiongroup (Mg months = -40, SDg months = 1-55;Mgmonths = --25, SDgmonths = 1-57) andincreased in infants from the control group(j'- emonths = -.50, SDgmonths = 1-60; Mgmonths= .26, SDgmonths = 1-45).

Pearson product-moment correlationswere used to determine whether explorationobserved during mother-infant interaction'was related to quality of exploration as-seissed during free play. Quantity of explora-tion observed during mother-child interac-tion and sophisticated exploration in freeplay are significantly related, r = .28, p <.01 (see Table 4). In addition, high rates ofexploration during interaction are related tolow rates of mouthing during free play, r =— .24, p < .05. Although the correlations aresignificant, they are not high. It should bekept in mind, however, that the exploratorydata collected during mother-child interac-tion provided information about the quantityof exploration, while the measure of freeplay was a qualitative measure. That quality

of exploration requires a certain amount ofquantity of exploration is clearly refiected inthe significant correlations obtained.

Results from the analysis of the- datagathered during the free play assessment ofexploratory competence provide additionalsupport for the conclusion that the interven-tion was effective in enhancing the qualityof infant exploratory behavior. While thecontrol infants were significantly more likelyto engage in the least cognitively sophisti-cated kinds of play, the intervention infantsengaged in functional play significantlymore frequently.

Changes in Reported Life StressAn ANOVA with repeated measures for

life stress between infant age 6 and 9 monthsshows no significant intervention effect, asignificant effect for time, F(l, 48) = 28.6, p< .001, and no interaction. Reported lifestress was high at entry (M = 11.04) and de-clined similarly for the pretested interven-tion and control groups to an overall samplemean of 7.26.

DELAYED POSTTREATMENT EFFECTS

The follow-up measure assessed was se-curity of attachment at 12 months.

Effect of Intervention on Qualityof Attachment

Log-linear analyses were conducted toexamine the relation between infant-motherattachment and treatinent group. Becauseof the small number of subjects in the inse-cure resistant and insecure disorganized/disoriented attachment categories, all threeinsecure categories were collapsed for sub-sequent data analysis into one insecure cate-gory and together were compared to thegroup of infants who were classified as se-curely attached. Separate two-way log-linearanalyses ofthe interrelation ofthe pretestedand no-pretest groups and attachment secu-


TABLE 5

ATTACHMENT CLASSIFICATIONS AT 12 MONTHSFOR INFANTS IN EACH TREATMENT GROUP

SECURE INSECURE

B A C D

Pretest:Intervention group 14 8 1 2Control group 6 13 4 2

No pretest:Intervention group 17 4 2 2Control group 5 13 3 4

Total 42 38 10 10

NOTE.—L^ for pretested groups = 5.45, p < .025; Lfor groups without pretest = 12.23, p < .001; L^ total =16.96, p < .001. Attachment classification: B = secure, A= insecure avoidant, C = insecure resistant, D = inse-cure disorganized/disoriented.

rity (secure vs. insecure) generated signifi-cant associations between group and attach-ment security, L\l) = 5.45 and 12.23, p <.025 and p < .001, respectively. No pretreat-ment effects were found. Therefore, datawere collapsed within the intervention andcontrol groups. The result of the two-waylog-linear analysis on this pooled data setwas clear: An association was found be-tween intervention group and attachmentclassification, L\l) = 16.96, p < .001. In-spection of Table 5 indicates that 78% ofcontrol infants were classified in the inse-cure categories compared to 38% in the in-tervention groups. Only 28% of all controlinfants were assigned the traditional secureclassification, compared to 62% of the inter-vention infants. Infants in the control groupswere most likely to be classified as insecureavoidant (52%)—not as insecure resistant, aswould be expected from previous researchwith irritable infants.

DiscussionTo date, two aspects in the study ofthe

development of human attachment havebeen largely neglected despite increasingscientific endeavor in the field. Few investi-gators have directly addressed the problemof the infiuence of infant negative emotion-ality on the developing attachment relation-ship; the studies by Belsky, Fish, and Isa-bella (1991) and Crockenberg (1981) arenotable exceptions. Second, few if any in-vestigators have experimentally manipu-lated the aspect of behavior called maternalsensitivity, which in attachment theory isdeemed crucial for the development of a se-

cure attachment bond. The present study isunique in that infants were preselected onan aspect of temperament related to attach-ment security: negative emotionality. In ad-dition, matemal sensitive responsivenesswas experimentally manipulated in dyadswith irritable infants, which are at risk fordeveloping an anxious attachment rela-tionship.

The findings indicate that the interven-tion was effective in enhancing maternal re-sponsiveness, stimulation, visual atten-tiveness, and control. The fact that only onegroup difference in maternal behavior wasevident before the treatment phase of theexperiment, and that this difference was un-favorable for the intervention group, servesto buttress this conclusion. Infant behavioralso changed after intervention. Interven-tion infants were more sociable, more ableto soothe themselves, and engaged more fre-quently in cognitively sophisticated kinds ofexploration than control infants. Hence, theintervention promoted substantial improve-ment in the quality of mother-infant interac-tion and the quality of infant exploration, pa-rameters indicative of the quality of anattachment relationship.

A follow-up 3 months later revealed thatsignificantly more intervention infants thancontrol infants were securely attached, indi-cating that matemal sensitive respon-siveness is causally related to infant attach-ment, at least in firstborn, irritable infantsfrom lower-class families. In that they showsignificant intervention effects on attach-ment-related measures, our results are con-sistent with findings in high-risk samples byLyons-Ruth, Zoll, Connell, and Odom (1987)and Lieberman, Weston, and Pawl (1991).Our findings differ from the results of Bar-nett, Blignault, Holmes, Payne, and Parker(1987) and Booth, Mitchell, Barnard,Spieker, and Magyary (1987), which did notshow significant intervention effects on at-tachment classification. Like ours, the for-mer studies involved populations with an in-creased risk of insecure infant attachmentand focused on intervention on matemalsensitivity. In the latter studies, it is notclear whether the populations under studywere at risk for developing an insecure at-tachment relationship and the interventionwas not specifically geared toward enhanc-ing sensitive matemal responsiveness. Bothexperimental differences may have contrib-uted to the differences in effect.

The intervention model we employed

in our research emphasizes matemalstrengths rather than focusing on matemalinadequacies. We believe its strongest pointis that it enhances the mother's feelings ofeffectiveness. Because mothers are in-structed while they are interacting with theirown infants, they themselves perform the ac-tions that elicit the desirable infant behav-iors. At the same time, they are convincedtliat the directions given are effective fortlieir infant, something parents are not sureabout when they are just given advice, eitherorally or in writing. In addition, by makingmiothers self-aware of their own growth-promoting parenting practices, we not onlyincreased the incidence of such matemal be-haviors but also infiuenced infant behavior.

Although we demonstrated that enhanc-ing matemal sensitive responsiveness isbeneficial for the development of a secureattachment relationship between motherand infant during the first year of life, a wordof caution is in order. Sensitive respon-siveness seems to refer to the current voguefor intervention in early infancy in which a"the more the better" philosophy may pre-vail. As part of the model for the idealmother-infant relationship, a 100% level ofmaternal responsiveness could thus easilybe adopted, with relationships falling belowthat level being regarded as suboptimal andintervention being geared to raising respon-siveness to its maximum. However, such agoal would only be attainable in laboratorysettings, where the two interactants are com-pletely sheltered from all distractions and itis, therefore, very easy for the mother to at-tend to the infant virtually continuously.Mother-infant interactions in real-life set-tings such as the home present a rather dif-ferent picture. And might there not even besome benefits in a certain amount of ignor-ing of the infant's signals—benefits such asself-reliance and learning to share themother with others? Our conclusion, then,is that a fairly high level of responsivenessmay well be beneficial, indeed essential, tothe infant's development; it does not follow,however, that "the more the better" shouldbe adopted as a guideline and that all inter-vention efforts should be directed towardmaximum responsiveness. Such a conclu-sion seems to be in accord with results fromotlier researchers (Belsky, Rovine, & Taylor,1984) whose findings show that secure in-fants experience neither especially high norespecially low, but actually intermediate,levels of reciprocal interaction.

The most significant theoretical finding


of this work is its support for a relation be-tween negative emotionality (assessed neo-natally), quality of mother-infant interaction,and attachment classification in a group ofirritable infants. Attachment theorists (Ains-worth, 1973; Waters, 1981) state that securityis the functional goal of infant attachmentbehavior. An affective condition of securityrecognizes that the behavior required ofmothers to foster feelings of security in theirinfants in most situations varies across in-fants (Thompson, 1986). The high rate of in-security in the control groups may have beencaused by the dual risk status of our sample.In addition to being preselected on irritabil-ity, the infants were raised in lower-classfamilies where relatively high rates of insen-sitivity predominated (Mey, 1992). These as-pects of maternal and infant behavior seemto infiuence the infant's understanding andinterpretation of and adaptation to socialevents. The results of the experimental ma-nipulation of matemal sensitive respon-siveness in these dyads with irritable infantslends further support to this notion. Enhanc-ing maternal responsiveness seems to fosterthe infant's sense of security, since it leadsto an increase in exploratory behavior whichis also of a higher quality compared to irrita-ble infants receiving a lower level of mater-nal responsiveness. Moreover, our findingthat the majority of control dyads are avoid-antly attached directly challenges previousfindings with irritable infants (Miyake,Chen, & Campos, 1985) and the assumptionof Chess and Thomas (1982) and Kagan(1984) and the conclusion drawn by Cold-smith and Alansky (1987) that such infantswould be resistantly attached. The currentdata suggest a more complex process than adirect path between neonatal irritability and(Strange Situation) resistance. Neonatal irri-tability seems to be a general dispositionthat may predispose infants who are at riskfor insecure relationships for other, contex-tual reasons toward insecurity but is not di-rectly predictive of type of insecurity. Bel-sky and Isabella (1988) also found that asrisks increased so did rates of insecurity.

Infant irritability seems to be an influ-ential condition in the relationship betweenmother and infant in the first year of life. Itmakes it difficult for mothers to adjust theiractions to the child's disposition. That moth-ers can be taught to do so is suggested by theincreases in maternal responsiveness and inthe amount and quality of infant exploratorybehavior after an intervention geared to thecharacteristics of individual mother-infant


dyads. However, one must be wary of a newtrend or relation, however intriguing, dis-covered in only one study. So far, very fewstudies with irritable infants have been con-ducted, and those that have been done donot present concordant results (Fish &Crockenberg, 1981). Replications should ad-dress the question of how easily these tech-niques might be taught to others, since inthis study all the intervention sessions wereperformed by the same person. In addition,the caveats associated with the implementa-tion of only one kind of intervention mustbe reiterated in relation to our findings.Without the use of a comparison interven-tion group, the possibility cannot be ruledout that, in some way, the intervention dyadsshowed an improvement in their behaviorsimply because they received a great dealmore attention than the control dyads. In or-der to control for such nonspecific placebofactors, future studies should include at leastone other comparison intervention group sothat the critical aspects of intervention maybe identified more precisely than was possi-ble from the present study. Further studiesare needed to investigate the questionsraised above and to elucidate the long-termeffects of intervention through sustainedfollow-up of research populations who re-ceived an early intervention program. To un-cover such possible long-term effects of ourintervention, we are still collecting data onthe sample we reported on in this article.

References

Aber, J. L., & Baker, A. J. L. (1990). Security ofattachment in toddlerhood. In M. T. Green-berg, D. Cicchetti, & E. M. Cummings (Eds.),Attachment in the preschool years: Theory,research, and intervention (pp. 427-460).Chicago: University of Chicago Press.

Ainsworth, M. D. S. (1973). The development ofinfant-mother attachment. In B. M. Caldwell& H. N. Ricciutti (Eds.), Review of child de-velopment research (Vol. 3, pp. 1-94). Chi-cago: University of Chicago Press.

Ainsworth, M. D. S., Blehar, M. C , Waters, E.,& Wall, S. (1978). Pattems of attachment: Apsychological study ofthe Strange Situation.Hillsdale, NJ: Erlbaum.

Ainsworth, M. D. S., & Wittig, B. A. (1969). Attach-ment and exploratory behavior of one-year-olds in a strange situation. In B. M. Foss (Ed.),Determinants of infant hehavior (Vol. 4, pp.113-136). London: Methuen.

Barnett, B., Blignault, I., Holmes, S., Payne, A., &Parker, C. (1987). Quality of attachment in asample of 1-year-old Australian children.

Journal of the American Academy of Childand Adolescent Psychiatry, 26, 303-307.

Bates, J. E., Maslin, C. A., & Frankel, K. (1985).Attachment security, mother-child interac-tion, and temperament as predictors of behav-ior-problem ratings at age three years. In I.Bretherton & E. Waters (Eds.), Growingpoints of attachment theory and research (pp.167—195). Monographs ofthe Society for Re-search in Child Development, 50(1—2, SerialNo. 209).

Bates, J. E., Olson, S., Pettit, C , & Bayles, K.(1982). Dimensions of individuality in themother-infant relationship at six months ofage. Child Development, 53, 446-461.

Bell, S. M., & Ainsworth, M. D. S. (1972). Infantcrying and maternal responsiveness. ChildDevelopment, 43, 1171-1190.

Belsky, J., Fish, M., & Isabella, R. (1991). Conti-nuity and discontinuity in infant negative andpositive emotionality: Family antecedentsand attachment consequences. Develop-mental Psychology, 27, 421-431.

Belsky, J., Garduque, L., & Hrncir, E. (1984). As-sessing performance, competence, and execu-tive capacity in infant play: Relations to homeenvironment and security of attachment. De-velopmental Psychology, 20, 406-417.

Belsky, J., Goode, M. K., & Most, R. K. (1980).Matemal stimulation, and infant exploratorycompetence: Cross-sectional, correlational,and experimental analyses. Child Develop-ment, 51, 1163-1178.

Belsky, J., & Isabella, R. (1988). Maternal, infant,and social-contextual determinants of attach-ment security. In J. Belsky & T. Nezworski(Eds.), Clinical implications of attachment(pp. 41-94). Hillsdale, NJ: Erlbaum.

Belsky, J., & Most, R. K. (1981). From explorationto play: A cross-sectional study of infant free-play behavior. Developmental Psychology,17, 630-639.

Belsky, J., Rovine, M., & Taylor, D. C. (1984). ThePennsylvania infant and family developmentproject: HI. The origins of individual differ-ences in infant-mother attachment: Maternalarid infant contributions. Child Developm,ent,55, 718-728.

Booth, C. L., Mitchell, S. K., Barnard, K. E.,Spieker, S. J., & Magyary, D. (1987, April).Evaluating intervention with multirisk m,oth-ers: Links between social skills improve-ments and child outcomes. Paper presentedat the biennial meeting of the Society for Re-search in Child Development, Baltimore.

Brazelton, T. B. (1973). Neonatal behavioral as-sessment scale. Philadelphia: Lippincott.

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


Campbell, S. (1979). Mother-infant interaction asa function of matemal ratings of tempera-ment. Child Psychiatry and Human Develop-ment, 10, 67-76.

Caron, J., & Miller, P. (1981, April). Effects of in-fant characteristics on caretaker respon-siveness among the Gusii. Paper presented atthe biennial meeting of the Society for Re-search in Child Development, Boston.

Chess, S., & Thomas, A. (1982). Infant bonding:Mystique and reality. American Journal ofOrthopsychiatry, 52, 213-222.

Crockenberg, S. B. (1981). Infant irritability,mother responsiveness, and social support in-fluences on the security of infant-mother at-tachment. Child Development, 52, 857-865.

Crockenberg, S. B. (1986). Are temperamental dif-ferences in babies associated with predict-able differences in caregiving? In J. V. Lemer& R. M. Lemer (Eds.), New directions forchild development: Temperament and socialinteraction during infancy and childhood(pp. 53-73). San Francisco: Jossey-Bass.

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

Grockenberg, S. B., & Smith, P. (1982). Anteced-ents of mother-infant interaction and infant ir-ritability in the first three months of life. In-fant Behavior and Development, 5,105—119.

Egeland, B., Breitenbucher, M., Dodds, M., Pas-tor, D., & Rosenberg, D. (1979). Life stressand scoring manual. Unpublished manu-script. University of Minnesota, Minneapolis.

Egeland, B., & Farber, E. A. (1982). Antecedentsof infant-mother attachment relationships ineconomically disadvantaged families. Un-published manuscript. University of Minne-sota, Minneapolis.

Egeland, B., & Farber, E. A. (1984). Infant-motherattachment: Factors related to its develop-ment and changes over time. Child Develop-ment, 55, 753-771.

Erickson, M. F., & Farber, E. A. (1983, April). In-fancy to preschool: Continuity of adaptationin high-risk children. Paper presented at thebiennial meeting of the Society for Researchin Child Development, Detroit.

Erickson, M. F., Sroufe, L. A., & Egeland, B.(1985). The relationship between quality ofattachment and behavior problems in pre-school in a high-risk sample. In I. Bretherton& E. Waters (Eds.), Growing points of attach-ment theory and research (pp. 147-166).Monographs of the Society for Research inChild Development, 50(1-2, Serial No. 209).

Field, T. (1982). Interaction coaching for high-riskinfants and their parents. In H. A. Moss, R.Hess, & G. Swift (Eds.), Early intervention

programs for infants (pp. 5-24). New York:Haworth.

Fish, M., & Grockenberg, S. (1981). Gorrelatesand antecedents of nine-month infant behav-ior. Infant Behavior and Development, 4,69-81.

Goldsmith, H. H., & Alansky, J. A. (1987). Mater-nal and infant temperamental predictors of at-tachment: A meta-analytic review. Journal ofConsulting and Clinical Psychology, 55,805-816.

Grossmann, K., & Grossmann, K. E. (1982, April).Maternal sensitivity to infants' signals dur-ing the first year as related to the year old'sbehavior in Ainsworth's Strange Situation ina sample of Northern German families. Paperpresented at the International Gonference onInfant Studies, Austin, TX.

Isabella, R. A., & Belsky, J. (1991). Interactionalsynchrony and the origins of infant-mother at-tachment: A replication study. Child Devel-opment, 62, 373-384.

Isabella, R. A., Belsky, J., & von Eye, A. (1989).The origins of infant-mother attachment: Anexamination of interactional synchrony dur-ing the infant's first year. Developmental Psy-chology, 25, 12-21.

Kagan, J. (1971). Change and continuity in in-fancy. New York: Wiley.

Kagan, J. (1984). The nature of the child. NewYork: Basic Books.

Kaye, K. (1978). Discriminating among nonnal in-fants by multivariate analysis of Brazeltonscores: Lumping and smoothing. In A. J. Sam-eroff (Ed.), Organization and stability of new-bom behavior: A commentary on theBrazelton Neonatal Behavior AssessmentScale (pp. 60-80). Monographs ofthe Societyfor Research in Child Development, 43(5—6,Serial No. 177).

Kelly, P. (1976). The relation of infant's tempera-ment and mother's psychopathology to inter-actions in early infancy. In K. F. Riegel & J.A. Meacham (Eds.), The developing individ-ual in a changing world (Vol. 2, pp. 23—34).Chicago: Aldine.

Klein, P. (1984). The relation of Israeli motherstoward infants in relation to infants' per-ceived temperament. Child Development, 55,1212-1218.

Koot, J. M.'(1988). Relaties tussen sensitiviteit vanmoederS, de competentie van het kind engehechtheid: Observatie aan een instructie-taak [Relationships between matemal sensi-tivity, child competence, and attachment: Ob-servation in an instructional situation]. In J.R. M. Gerris & J. van Acker (Eds.), Gezinsrel-aties onderzocht (pp. 53-63). Amsterdam:Swets & Zeitlinger.

Lee, C , & Bates, J. E. (1985). Mother-child inter-


action at age two years and perceived difficulttemperament. Child Developm,ent, 56,1314-1325.

Lewis, M., Feiring, C , McGuffog, G., & Jaskir, J.(1984). Predicting psychopathology in six-year-olds from early social relations. ChildDevelopment, 55, 123—136.

Lieberman, A. F., & Pawl, J. H. (1990). Disordersof attachment and secure base behavior in thesecond year of life. In M. T. Greenberg, D.Gicchetti, & E. M. Gummings (Eds.), Attach-ment in the preschool years: Theory, re-search, and intervention (pp. 375-397). Chi-cago: University of Chicago Press.

Lieberman, A. F., Weston, D. R., & Pawl, J. H.(1991). Preventive intervention and outcomewith anxiously attached dyads. Child Devel-opment, 62, 199-209.

Linn, P., & Horowitz, F. (1983). The relationshipbetween infant individual differences andmother-infant interaction during the neonatalperiod. Infant Behavior and Development, 6,415-427.

Londerville, S., & Main, M. (1981). Security ofattachment, compliance, and matemal train-ing methods in the second year of life. Devel-opmental Psychology, 17, 289-299.

Lyons-Ruth, K., Zoll, D., Connell, D. B., & Odom,R. (1987, April). Maternal depression as medi-ator ofthe effects of home-based interventionservices. Paper presented at the biennialmeeting of the Society for Research in ChildDevelopment, Baltimore.

Maccoby, E. E., & Feldman, S. S. (1972). Mother-attachment and stranger-reactions in the thirdyear of life. Monographs of the Society forResearch in Child Development, 37(1, SerialNo. 146).

Main, M. (1973). Play, exploration and compe-tence as related to child-adult attachment.Unpublished doctoral dissertation, JohnsHopkins University, Baltimore.

Main, M., & Solomon, J. (1990). Procedures foridentifying infants as disorganized/disori-ented during the Ainsworth Strange Situa-tion. In M. T. Greenberg, D. Cicchetti, &E. M. Cummings (Eds.), Attachment in thepreschool years: Theory, research, and inter-vention (pp. 121—160). Chicago: University ofChicago Press.

Maslin, C. A., & Bates, J. E. (1982, May). Anxiousattachment as a predictor of disharmony inthe mother-toddler relationship. Paper pre-sented at the Intemational Conference on In-fant Studies, Austin, TX.

Matas, L., Arend, R. A., & Sroufe, L. A. (1978).Continuity of adaptation in the second year:The relationship between quality of attach-ment and later competence. Child Develop-ment, 49, 547-556.

McCall, R. B. (1974). Exploratory manipulationand play in the human infant. Monographs ofthe Society for Research in Child Develop-ment, 39(2, Serial No. 155).

Mey, J. Th. H. (1992). Sociale ondersteuning,gehechtheidskwaliteit en vroegkinderlijkecompetentie-ontwikkeling [Social support,quality of attachment, and the early develop-ment of competence]. Unpublished doctoraldissertation. University of Nijmegen.

Milliones, J. (1978). Relationship between per-ceived child temperament and matemal be-havior. Child Development, 49, 1255-1257.

Miyake, K., Chen, S., & Campos, J. J. (1985). In-fant temperament, mother's mode of interac-tion, and attachment in Japan: An interim re-port. In I. Bretherton & E. Waters (Eds.),Crowing points of attachment theory and re-search (pp. 276-297). Monographs ofthe So-ciety for Research in Child Development,50(1-2, Serial No. 209).

Netherlands Central Bureau of Statistics. (1971).Systematische classificaties: Deel 2. Beroe-penclassificaties [Systematic classifications:Vol. 2, Classification of professions]. Voor-burg: Netherlands Central Bureau of Sta-tistics.

Osofsky, J. D. (1976). Neonatal characteristics andmother-infant interaction in two observa-tional situations. Child Development, 47,1138-1147.

Osofsky, J. D., & Danzger, B. (1974). Relation-ships between neonatal characteristics andmother-infant interaction. DevelopmentalPsychology, 10, 124-130.

Peters-Martin, P., & Wachs, T. D. (1984). A longi-tudinal study of temperament and its corre-lates in the first 12 months. Infant Behaviorand Development, 7, 285-298.

Pettit, G., & Bates, J. (1984). Gontinuity of individ-ual differences in the mother-infant relation-ship from 6 to 13 months. Child Development,55, 729-739.

Power, T. G., Ghapieski, M. L., & McGrath, M. P.(1985). Assessment of individual differencesin infant exploration and play. DevelopmentalPsychology, 21, 974-981.

Rothhart, M. K., & Derryherry, D. (1981). Devel-opment of individual differences in tempera-ment. In M. E. Lamh & A. L. Brown (Eds.),Advances in developmental psychology (Vol.1, pp. 37-86). Hillsdale, NJ: Erlbaum.

Smith, P. B., & Pederson, D. R. (1988). Matemalsensitivity and pattems of infant-mother at-tachment. Child Development, 59, 1097-1101.

Solomon, R. L., & Lessac, M. S. (1968). A controlgroup design for experimental studies of de-velopmental processes. Psychological Bulle-tin, 70, 145-150.


Sorce, J. F., & Emde, R. N. (1981). Mother's pres-ence is not enough: Effects of emotionalavailability on infant exploration. Develop-mental Psychology, 17, 737-745.

Sroufe, L. A. (1983). Infant-caregiver attachmentand adaptation in the preschool: The roots ofcompetence and maladaptation. In M. Perl-mutter (Ed.), Minnesota sym.posium in childdevelopment (Vol. 16, pp. 41-81). Hillsdale,NJ: Erlbaum.

Sroufe, L. A., Fox, N. E., & Pancake, V. R. (1983).Attachment and dependency in develop-mental perspective. Child Development, 54,1335-1354.

Stem, D. (1971). A micro-analysis of mother-infantinteraction: Behavior regulating social con-tact between a mother and her 3V2-month-oldtwins. Journal of the American Academy ofChild Psychiatry, 10, 501-517.

Thompson, R. A. (1986). Temperament, emotion-ality, and infant cognition. In J. V. Lerner &R. M. Lemer (Eds.), New directions for childdevelopment: Temperament and social inter-action during infancy and childhood (pp. 35-52). San Francisco: Jossey-Bass.

Thompson, R. A., & Lamb, M. E. (1983). Securityof attachment and stranger sociability ininfancy. Developmental Psychology, 19,184-191.

van den Boom, D. C. (1988). Neonatal irritabilityand the development of attachment: Obser-vation and intervention. Unpublished doc-toral dissertation. University of Leiden.

van den Boom, D. C. (1989). Neonatal irritabilityand the development of attachment. In G. A.Kohnstamm, J. E. Bates, & M. K. Rothbart(Eds.), Temperament in childhood (pp. 299—318). New York: Wiley.

van den Boom, D. C. (1991). The influence of in-fant irritability on the development of themother-infant relationship in the flrst sixmonths of life. In J. K. Nugent, B. M. Lester,& T. B. Brazelton (Eds.), The cultural contextof infancy (Vol. 2, pp. 63-89). Norwood, NJ:Ablex.

Vaughn, B. E., Stevenson-Hinde, J., Waters, E.,Kotsaftis, A., Lefever, G. B., Shouldice, A.,Trudel, M., & Belsky, J. (1992). Attachmentsecurity and temperament in infancy and earlychildhood: Some conceptual clarifications. De-velopmental Psychology, 28, 463-473.

Waters, E. (1981). Traits, behavioral systems, andrelationships: Three models of infant-adult at-tachment. In K. Immelman, G. W. Barlow, L.Petrinovich, & M. Main (Eds.), Behavioraldevelopm,ent (pp. 621-650). Cambridge:Cambridge University Press.

Watson, J. S. (1979). Perception of contingency asa determinant of social responsiveness. In E.B. Thoman (Ed.), The origins of the infant'ssocial responsiveness (pp. 33-64). Hillsdale,NJ: Erlbaum.

Wenckstem, S., Weizmann, F., & Leenaars, A. A.(1984). Temperament and tempo of play ineight-month-old infants. Child Development,55, 1195-1199.

the influence of temperament and mothering on attachment ...people.uncw.edu/hungerforda/graduate...

Documents