.,

12 Agency in development

Todd D. LittleUniversity of Kansas, Lawrence, USA

Models and metatheories of human developmenthaveevolvedsignificantlyover the previous two centuries. Most changes are clearly reflections ofprogress and integration. Early mechanistic models, for example, haveincorporated organismic features such as when behaviour is linked tochanges in biology (Geriepy, 1996) or when social learning theorists speakof agents and agency (Bandura, 1997). Some organismic models, on theother hand, have revealed mechanistic tendencies such as when behaviourinterventions are focused on the first few years of life under the belief theywill inoculate children from future harm (Lewis, 1997). Because thesechanges have blurred traditional boundaries among metatheories, I revisitthe origins of the organismic model in an effort to revitalise it by high-lighting the place and functions of agency in development. In addition, Iwill examine context's role in influencing the expression of agency through-out development and discuss both statistical advances and methodologicallimitations for modelling the agentic self in context.

My primary goal is to discuss the linkages among the various dimensionsthat comprise personal agency across the lifespan. Although volitional goal-directed activity (i.e., agency) has an evolutionary basis (Chapman, 1984;see Hawley & Little, 2002, for a detailed exposition), I will focus on thevarious levels of ontogenetic meanings as they relate to the individual'ssense of personal agency. Generally speaking, one's sense of personalagency stems from goal-driven actions. It functions as a personal resourcefor facing the challenges that emerge throughout development (Little,1998). This sense of personal agency is particularly relevant given today'sdevelopmental milieu of increased challenges and potential disruptions tothe successful development of individuals. Even casual observationreinforces the notion that most cultural, societal, social, and familial struc-tures have dramatically transformed during the latter part of the previouscentury. Because sociocultural features such as educational choices, familialconstellations, behavioural norms, career directions, post-retirement life-styles, and so on will probably continue to become more varied anddifferentiated, individuals will be faced with new and diverse challenges tosuccessful development at every phase of the lifespan. How individuals

224 Little

meet and overcome such life-course challenges is the operative realm ofpersonal agency.

This chapter is organised around three related themes that I feel need tobecome focal directions of study in order to fully articulate the role thatagency plays in development. The first focal direction is to understandbetter the interrelations among the varied features of the self system and thesynergistic connections among the layered self-regulatory processes ofdevelopment (e.g., motives, goals, volitional control mechanisms, mentalrepresentations, subjective interpretations, and so on). As I describe indetail later, an organismic perspective on the individual offers a usefulstarting point from which to unpack fully the relationships among thequalities and quantities of one's sense of personal agency (e.g., Hawley &Little, 2002; Little, Hawley, Henrich, & Marsland, 2002; Magnusson &Cairns, 1996; Ryan, 1993; Ryan, Sheldon, Kasser, & Deci, 1996).

The second focal direction is to examine the influence of context moreprecisely (e.g., Bronfenbrenner, 1995;Lerner, 1996).Although the resurgenceof cross-cultural designs, for example, indicates progress in this direction, Isuggest that clearly articulated theories about the nature of contexts need toemerge and that correspondingly compelling designs for contextual studiesneed to be implemented. Simply demonstrating that a change in context isassociated with a change in behaviour provides only rudimentaryinformation about what aspects of contexts are operating and does little toclarify the individual-level mechanisms by which context has its effects.

The final focal direction is to increase further the methodological sophis-tication by which the interplay of context and self is disentangled. The latterquarter of the previous century saw the introduction of an extremelypowerful and versatile family of statistical modelling tools (e.g., generalstructural equation modelling techniques, including specific instantiationssuch as multilevel, growth-curve, and multiple-group models; Little,Schnabel, & Baumert, 2000; Marcoulides & Schumacker, 1996). Thesemodelling tools are continuing to be refined and will soon offer even greaterflexibility for researchers to ask and answer the types of complex questionsthat are needed for understanding developmental processes across contexts.

This chapter finishes with a discussion of a number of issues and broaderthemes that emerge when one considers these factors in the study of agencyin development.

ORGANISMIC PERSPECTIVE REVISITED

Because many research accounts rely on implicit assumptions about thenature of human behaviour, observers have noted that more rapid andefficient progress in the social and behavioural sciences will come about if wemake explicit the larger models of development that drive our various micro-theories (Shanahan, Valsiner, & Gottlieb, 1997). Historically, the organismic

,"

/2. Agl'lIcy III dl'l'l'/O/JI/IC'II/ 225

model has provided a unifying basis 1'01' II nUlIlbol' of pl'Odu()tiw andinfluential research traditions (see Cairns. I<)!{;\.I<)l)!{),In lighl of curr'enltrends towards larger and more eneolllpussing sludios ol'<.kVl'lopnwnt, I fedthat the organismic model remains a viable: IInd usoful vanlage point f['()m

which to view human development, pal'ticulal'ly tlw dl'volopmonl or [110self,At the heart of an organismic pel'~pec[ive is Ilw pr'usulllption that indi-

viduals are active agents and contributol's 10 Ihuil' own development. Anorganismic perspective views volitional bohaviolll' liS goul.dil'ccted {/c/ions,where actions are both purposive and self.iniliated (Bl'llndlsWdtel', 1984,1998; Chapman & Skinner, 1985), By eXlonsion, individuals III'Cinhel'entlyactive and self-regulating-that is, agentic, An ol'gunismk uppl'oaeh tounderstanding the developing individual also involves an explicit focus onthe interface between the self and contexl (Li(lk~el 111.,W02; Ryan et aI.,1996). Contexts reflect complex constdlulions of fuulul'cs, spanning themolar and micro levels, that both con~tl'llin und ul'l'ol'dbehavioul' (Bl'Onfl'cn-brenner, 1995; Gottlieb, 1997). Taken (ogollwl',such foa[lII'O~cl'eate IIpsychological carrying capacity within which Illl' individual opul'ates.

As an agent in his/her own developmunt. [Iw individuill functions as anintegrated organism (e.g., assimilative and acconlll1odativc processes areinvoked to maintain balance and equilibrium), Although an individual bothinfluences and is influenced by the contexts in which s/he acts and develops,the agentic self must negotiate the alTol'duncl;1sund constl'aints of eachcontext. Developmentally speaking, an individual continually progressesalong a predominantly self-guided path, giving fOl'mand meaning to his orher actions along the way. Such actions I'usult fl'om selcctive choices thatemanate primarily from the individual. The ol'ganismic perspective pre-sumes, then, that individuals are active agl.:l1tswho plot and navigate achosen course through the uncertainties and challenges of the social andecological environments; however, like trade winds and currents, environ-ments sometimes hinder, sometimes bol~tcl', IInd other times change thecourse of a developmental route (Little et al.. 2(02),

As part of their integrated functioning, individuals engage in a self-evaluative feedback process, continuously intel'pl'eting and evaluatingactions and their consequences. Across the episodes of activity in the varyingconstellations of context, the individual continually discovers and refineswho one is and what one is capable of and, perhaps most importantly, underwhat conditions one's actions can or cannot have II dcsired effect. Underoptimal circumstances, this continually evolving and actively monitored self-system gives rise to a strong and effective sense of personal agency. Differentindividuals with differing experiences and differing predispositions, how-ever, will yield varying profiles of pel'sonal agency because the sense ofagency is a multifaceted and striated systcmof motives,beliefs,and beha-viours (Hawley & Little, 2002; Little et aI., 2002; Little & Wanner, 1997).

This revitalised organismic perspective on agency necessarily entails asystems approach wherein the functioning of one system, such as the peer

226 Little

system, influences the functioning of another, such as the self (Gottlieb,1997). Because systems and subsystems are yoked, significant advances inunderstanding the breadth of human behaviour will require that they beexamined together rather than in isolation (Magnusson & Cairns, 1996).Such a view necessitates greater investment in planning a data collectionprotocol and greater investment in procuring appropriate samples ofparticipants (e.g., samples derived on the basis of theoretical considerationsrather than convenience). In my view, such larger-scale integrated studies ofdevelopment can succeed only if they share a common and unifying meta-theoretical perspective. When the pieces of a research programme comefrom different models and metaphors, they are unlikely to fit together into acoherent picture. The puzzle pieces would rarely match up and the resultingpicture would look more cubist than portraiture.

Not only does an organismic perspective encourage broad-based studiesof the agentic self, but it also supports a concerted focus on resilience andsuccessful developmental outcomes. As each child begins to discover whoslhe is and what slhe is capable of, the evolving competence system contri-butes to an integrated sense of personal agency-an agentic self. The result-ing systems of action-control motives, goals, beliefs, and behaviours providea developmental foundation that, as mentioned, is called upon to negotiatevarious developmental tasks and challenges throughout the life course. Infacing these challenges, an agentic individual is primarily the origin of his orher actions, has high aspirations, perseveres in the face of obstacles, seesmore and varied options, learns from failures, has a greater sense of well-being, and so on. A nonagentic individual, on the other hand, is primarilythe pawn of unknown extra-personal influences, has low aspirations, ishindered with problem-solving blinders, and often feels helpless and unem-powered (Little, 1998; Little et aI., 2002; Ryan et aI., 1996; Skinner, 1995;Weisz, 1990). Such undesirable characteristics can stimulate additionalnegative feedback from others, such as teasing and victimisation (Graham &Juvonen, 1998).

From a revitalised organismic perspective, one's sense of personal agencyis not simply a reinvention of constructs such as self-esteem, self-efficacy,self-concept, and so on. Rather, it is a layered concept starting with thepremise of volitional goal-directed actions. Such actions are motivated byboth biological and psychological needs (Hawley, 1999; Hawley & Little,2002; Little et aI., 2002). Such actions are propelled by specific action-controlbeliefs about the links among agents, means, and ends (Chapman, 1984;Little, 1998; Skinner, Chapman & Baltes, 1988). Such actions are guided bygeneral action-control behaviours that entail self-chosen forms andfunctions, means and ends (Lopez & Little, 1996; Little, Jones, Henrich, &Hawley, 2001; Little, Lopez, & Wanner, 2001).

Regarding action-control beliefs, this perspective explicitly differentiatesthe possible belief relations among the primary constituents of intentionalaction (Chapman, Skinner, & Baltes, 1990; Little, Oettingen, Stetsenko, &

Self as agent

/2. Agency ill development 227

Others as agent

Causalitybeliefs

Figure 12.1 Relations among the three constitucnts of human action. both whenthe self is the agent and when othel's an: Ihe agent. Means can varyfrom personal attributes such as effort IInd ability to external aidssuch as friends and teachers.

Baltes, 1995; Skinner & Chapman, 1987; Skinner et al.. 1988; Stetsenko,Little, Gordeeva, Grasshof, & Oetingen. 2(00). The three constituents of anaction sequence are the actor, the goal, and the various means by which thegoal (or end) can be obtained. Agentic action relleets an agent's generalawareness of goals and the means to the goals, taking personal respon-sibility in pursuing a chosen goal, and the ability to select and utilisepotential means (Chapman & Skinner, 1985). Given the three constituentsof volitional activity, a number of belief types about the relations amongthese constituents are possible.

Figure 12.1 displays six of these belief types (see also Skinner, 1996).Control expectancy reflects general beliefs about the link between the self asagent and the goal (e.g., "when I want to do -, I can."). Collectiveefficacy, on the other hand, reflects the belief link between one's focalcultural group and the goal. Capacity heliefi' rellcct the links between theself and the various means that are relevant for attaining a chosen end("I'm able to put forth enough effort to do this", "I possess the necessaryskills to do this"). The counterpart linkage, when others are viewed as theagent of behaviour, can be termed collective ('{//}(/('ityhelief~. Similarly, thelink between the various means and ends can have two levelsof meaning.When the self is the agent, the resulting system of beliefs would reflectstrategy beliefs (i.e., what means work for me). Causality beliefs, on theother hand, reflect general views of the utility 01' usefulness of a givenmeans such effort, luck, or ability for attaining a particular goal (Chapmanet aI., 1990; Little et aI., 1995; Oettingen et aI., 1994; Skinner et aI., 1988).

Ample research has supported these distinctions (see Little, 1998;Skinner, 1996, for overviews). A denning feature of the action-control viewthat separates it from other similar models of control perceptions is the

228 Little

Figure 12.2 An organismic classification of coping behaviours in terms of theiraction-control aspects (based on Little, Lopez, & Wanner, 2001).

explicit means-ends analysis of actions. For example, because ability isconsidered separately from other means such as effort and access topowerful others, Stetsenko et al. (2000) were able to identify a means-specific gender bias in the action-control beliefs of school-aged children.Girls whose actual school performance was on a par with boys, reportedlower beliefs in their own ability to accomplish getting good grades than didboys. All other beliefs about their personal agency, such as effort, luck, andaccess to teachers, were on an equal footing, as would be expected fromequally performing groups of children. Only the girls' beliefs in their accessto ability revealed the bias. In other words, by focusing on the uniquemeans that are utilised to achieve a goal, Stetsenko and co-workers identi-fied a specific source of bias in girls' personal agency profile that mayhinder them as they progress along the educational ladder.

A means-ends analytic framework not only applies to beliefs aboutbehaviour, but also to actual behaviours that are enacted when goal pursuitis challenging. For example, the various strategies of coping behavioursthat have permeated the literature can be categorised in terms of the natureof the means that are utilised (e.g., prosocial versus asocial) and the specificends that are pursued (negotiating the stressor directly, or indirectly allevi-ating the consequences of the stressor; Lopez & Little, 1996). Figure 12.2displays the conceptual quadrants into which many extant coping strategiescan be placed. This multi-axial representation was tested in a short-termlongitudinal study and found to be fully adequate in capturing the vari-ability in coping behaviours both within and across measurement occasions(Little, Lopez, & Wanner, 2001). The model suggests that coping beha-viours can be classified in terms of their form, such as utilizing others or

12. Agency in development 229

Instrumental Reactive

Figure 12.3 Structural relations and decompositions among the forms andfunctions of aggressive behaviour (from Little et aI., 2001).

relying on self-resources, and the function, such as taking direct actiontowards the stressor or indirect action to remediate the impact of thestressor .

Such a means-ends analysis can also be applied to other domains ofbehaviour. For example, the research on aggressive-agonistic behaviour hasrevealed muddied mixtures of means (e.g., physical, verbal, relations) andends (instrumental, reactive). Using an organismic meta theoretical perspec-tive, my colleagues and I developed a measurement and analysis system thatdifferentiates among the different forms and functions of aggressivebehaviour (Little et ai., 2001). Figure 12.3 displays this decomposition ofaggressive behaviour in terms of its form (means) and function (ends). Withthis measurement and analysis system, we have been able to show new andnovel relationships among the various aggression dimensions. For example,in the literature, proactive/instrumental measures of aggression typicallycorrelate in the .6 to .8 region with reactive measures of aggression (Dodge& Coie, 1987; Poulin & Boivin, 1999; Price & Dodge, 1989). Our work hasshown that, when one accounts for the form of the aggression (e.g., overt,relational), the instrumental and reactive functions of aggression are, infact, independent (uncorrelated) processes.

A number of commonalities, which emanate from the overarchingorganismic metatheoretical perspective, unite these different aspects ofagentic action. First, because actions are coordinated in specific domains ofgoal-directed activity, they are domain specific. Second, because a specificaction is chosen from among a number of possible actions, it is means

230 Little

specific. Third, because of the domain-specific and means-specific nature ofactions in service of self-chosen goals, they can be classified and categorisedin terms of a means-ends, form-function analytic framework.

Understanding the complexities of organismic activity clearly requires anunderstanding of how aspects of individuals such as their beliefs andbehaviours coordinate together in the pursuit of life-course goals. As I seeit, the linkages among the layered aspects of the agentic self are bestexamined when the operational nature of each layer stems from a commonmetatheoretical perspective, in this case, the organismic perspective.

1

THE ROLE OF CONTEXT

A contextual perspective on agency explicitly acknowledges that everydevelopmental epoch represents a unique constellation of influences(Bronfenbrenner, 1995; Lerner, 1995, 1996). Contexts provide both afford-ances and constraints on the range of individual actions and behaviours thatcan be expressed. Although some mechanisms of development are clearlyuniversal (i.e., transcend context), many of the mechanisms that willultimately account for the diversity of human behaviour across develop-mental contexts and the life course will probably be context-specific orcontext-moderated mechanisms (i.e., mechanisms that function in interac-tion with contextual features). When viewed in their entirety, both the pre-and post-millennial changes in the various social structures amount tounprecedented variability among the contexts in which development unfolds.As a result, a primary challenge to understanding the role of agency indevelopment is to examine the organismic processes of change, growth, andtransformation as they unfold within and between these varied contexts.

As suggested earlier, a given context reflects a specific constellation offeatures (i.e., where intra-, inter-, and extra-personal features converge).These multilayered features can provide, if carefully contrasted, a quasi-experimental manipulation of many potential mechanisms that can shapeand mould individual development. Given that the features in which devel-opmental processes occur can be quasi-systematically varied, contextualdesigns can intimate the working mechanisms that influence developmentand they can dramatically increase our universe of generalisation. In otherwords, comparing individual development across varying contexts canhighlight which facets of the self-system are independent of the contextualvariations and which are not (e.g., Little, 1998). Similarly, comparing thesame individual across numerous contexts would reveal which processesof the agentic self function similarly (transcend context) and which func-tion uniquely as a result of context-specific constraints or affordances(Mischel & Shoda, 1995). Well-conceived contextual studies can system-atically vary and thereby quasi-experimentally isolate specific features ofboth contexts and individuals in order to derive the nature and generality

of their respectiveLittle, 1999).

The dyad, for example, reflects a unique and informative context forstudy. Individuals carry with them orientations, response patterns, abilities,motivations, and so on. As soon as social behaviour begins, the charac-teristics of one individual collide with the characteristics of other indi-viduals. Much can be learned about the development of the agentic selfsimply by exploring in detail these individual-individual interactions. Witha general bias toward "large-n studies", developmentalists sometimes missopportunities for rich "small-n" data designs such as small-group, round-robin designs. Small groups allow for a microanalysis of the rich contexts ofinteraction among the different characteristic features of individuals.Research has shown that children will vary their behaviour according tothe past interactions and behaviour patterns of their partners. Across thevarious dyadic interactions, a child can exhibit agency with one child anddeference to another according to their history of wins and losses (seeHawley & Little, 1999).

Although context-related effects can be readily documented, advances inunderstanding how context influences individuals have been slow to emergebecause of the reliance on context-level comparisons of groups of indi-viduals. Rather than focus on documenting group-level differences indevelopment across various contexts, a challenge of contextual approacheswill be to attempt to reduce context to a set of context-level features andindividual-differences variables that can effectively explain away the group-level differences of the individuals in a given context. The basic action planinvolves first defining the dimensions of context and second developingmeasures of those dimensions at both the level of the context and the levelof the individual. By explicitly measuring the defining features of variouscontexts, greater precision in isolating the working mechanisms of change,growth, and transformation would be gained. For example, instead ofcharacterising two sociocultural contexts as individualistic versus collecti-vistic, a researcher could measure the degree to which each individual ineach context endorses individualistic and collectivistic values and ideals(i.e., one's personal balance of the cultural values). In this way, one couldthen examine the relative contribution of individual-level vs. context-levelinfluences on behaviour. The multilevel decomposition of variance fromrandom coefficient models (Kreft & De Leeuw, 1998; also known asHierarchical Linear Models [HLM] analyses; see Bryk & Raudenbush,1992) will be particularly useful for modelling the agentic self across contextand time.

The contextual perspective also encourages a typological approach tounderstanding individual development. Given that (a) few developmentalprocesses are likely to operate in the same way for all individuals and that(b) the self is composed of synergistically intertwined layers, one couldreasonably expect differing constellations of beliefs and behaviours that

12. Agency in development 231

influences on developmental outcomes (Hawley &

232 Little

characterise subsets of individuals. These subgroups or typologies wouldalso be characterised by a range of individual differences within them. Inthis sense, a typological profile would be considered a specific develop-mental context, with unique affordances and constraints on behaviour. Forexample, the profile of an individual who is both high on individualisticvalues and collectivistic values suggests constraints on behaviour thatwould not be the same for an individual who is low on both dimensions orhigh on one but medium or low on the other. Especially when typologiesare more theory-driven rather than data-driven, a typological approach canbe quite useful for understanding developmental processes (see, e.g.,Hawley, Little, & Pasupahti, in press; Smith & Baltes, 1999).

MODELLING THE AGENTIC SELF IN CONTEXT

The contextual and organismic perspectives on agency necessitate the use ofsophisticated statistical and methodological tools in order to disentangle(model) the various sources of influence. A primary goal of a modellingperspective is to make strong tests of underlying theoretical models (i.e.,bring implicit assumptions into the explicit realm of model specification).The goal is not to partition the person into discrete sources of variance, butrather to determine how the pieces interlock and what their boundaryconditions are and then to assemble the units into a coherent portrait of theindividual.

A modelling perspective is not limited to statistical procedures such asstructural modelling techniques, it is equally at home within the frameworkof classical, regression-based techniques (see, for example, Widaman, 2000).The modelling perspective views statistical procedures not as mechanisticends in themselves (i.e., fixed and rule-bound), but rather as flexible toolsthat can be adjusted and adapted into an appropriate means for testinga given substantive theory. From a modelling vantage point, statisticalmethods are used to fit models that are tested against observed datapatterns and are implemented such that issues of effect size, parsimony, andtheoretical integration are emphasised (cf. null hypothesis testing).

Numerous recent techniques would fit within this modelling framework.For example, dyadic, triadic, and social relations paradigms are extremelyuseful to model the interactions among individuals (see, e.g., Hawley &Little, 1999; Kenny, 1990). These designs allow decomposition of thesystematic sources of variance that contribute to individual behaviour; theyalso allow decomposition of the degree to which individuals contribute tothe behaviour ()f the larger unit (i.e., dyad, triad, family, and so on; see, forexample, Kreppner, 1992). Multiple-group and multilevel designs are usefulto compare and contrast the influences of different contexts as well as topartition the degree to which variability in behaviour is associated withcontextual features versus individual differences (Little, 1997; Little et aI.,

12. Agency in development 233

2000). These techniques will be particularly useful to examine differencesamong groups and subgroups (i.e., typologies) of individuals.

Because studies of the agentic self in context will often necessitate large-scale approaches, the demands on participants can be great. However,recent advances in data imputation techniques and incomplete designestimation have tremendous potential for reshaping how we organise ourdata collection protocols. Various forms of intentionally missing designscan be conceived that would allow researchers to collect data in a moreefficient and less costly manner and still be able to draw valid general-isations based on the whole sample of individuals and variables. Forexample, one could design a study of some large number of individuals, butwhere anyone individual responds to only a subset of the items of interestor where only a randomly selected sub-sample of participants receivesadditional intensive measurements. With proper care, the missing subset ofitems or individuals can be imputed veridically (or the missing relationshipsestimated) such that conclusions and generalisations are drawn on the fullset of individuals and variables (Graham & Hofer, 2000; McArdle, 1994).Clearly, at this stage, the boundaries and limitations of such designs stillneed careful testing, but the potential of such intentionally missing designsis certainly an appealing possibility for large-scale contextual-organismicresearch programmes.

A second future development that will be useful in this regard is a robusttechnique to evaluate individual fit of the models that we specify. Forexample, the Mx program calculates a psuedo-chi-squared value for eachindividual included in an analysis (see Neale, 2000). This information isuseful because it can identify persons for whom the model does not hold.Further advances in this direction will offer information about the source ofthe misfit and, ideally, provide an algorithm for clustering individuals basedon their concordance with the fit of the specified model. Current multi-variate analytic techniques generally only focus on two dimensions of thedata cube at anyone time, with one dimension being the focal dimensionand the second being the variance-producing dimension (e.g., variables bypersons, persons by variables, variables by occasions). Making multipledimensions be the simultaneous focus of investigation will yield greaterunderstanding of the complexities of the agentic self.

Finally, recent advances in general structural equation modelling pro-cedures are beginning to overcome one of the more frustrating limitationsof this class of modelling technique: Namely, the restriction to linearrelationships among continuous variables. Now, the procedures offerresearchers the ability to specify latent interactions and nonlinear relation-ships as well as to model the relations among a mixture of different types ofvariables and constructs (see, e.g., Little et aI., 2000; Marcoulides &Schumacker, 1996). Such advances will undoubtedly allow much greaterprecision in the manner by which theoretical propositions are specified,estimated, and tested.

234 Little

SPECIFIC ISSUES

A number of issues about the nature of development arise when one con-siders the organismic model as the meta theory. These issues highlight theprobabilistic complexity of development that will continue to challengeour efforts to model the expressions of personal agency throughoutdevelopment.

Equipotentiality versus omnipotentiality

Cicchetti (e.g., Cicchetti & Rogosch, 1999) has championed this criticaldistinction in the field of developmental psychobiology. Equipotentialityrefers to the idea that many paths or antecedent conditions can lead tothe same outcome or consequent condition. Omnipotentiality refers to theidea that common pathways and antecedent conditions can give rise to aplethora of outcomes or consequent conditions. Agentic choices and inputsthroughout development are features likely to yield greater predictability inthe varied directions and outcomes of developmental changes along the lifecourse.

Heterotypic versus homotypic constructs

The continuity of the nature of constructs across development posessignificant challenges to understanding the role of agency in development.Such challenges to measurement occur when a construct changes formsacross development or when it takes on various forms across individuals,but still serves the same function. The nature and function of the layers ofpersonal agency reflect a set of constructs that evolve in form throughoutdevelopment. Similar to the methodological challenge of cross-culturalcomparisons (Little, 1997), the challenge to developmentalists will be todevelop measurement instruments and statistical tools to ensure the metriccomparability of the dimensions of personal agency as they continue totransform across the lifespan.

A related and relevant distinction is that between the surface-structureand the deep-structure meaning of behaviours. For example, resourcecontrol theory (Hawley, 1999) posits that the deep-structure goals of anevolutionarily adapting organism do not change across development-thatis, the goal is optimal resource control. However, the surface-structuremanifestations of that driving force change dramatically with age. Intoddlerhood, coercive strategies of resource control are most common andeffective. By early childhood and throughout development, more differ-entiated structures of resource control emerge such as pro social strategies(Hawley, 1999).

12. Agency in development 235

Horizontal versus vertical cross-disciplinary integration

Many researchers have cried out for greater interdisciplinary cooperation.At least two dimensions of cross-disciplinary integration can be brought tobear in understanding agency in development. The vertical dimensionfocuses on the up-down linkages from genes, through a specific class ortype of behaviour, to higher levels of sociocultural constraints or influences.From this perspective, a behaviour such as aggression would be examinedat the level of hormones, the forms and functions of the behaviour, and thesocio-cultural context that promotes or inhibits the degree of aggression.The horizontal dimension focuses on the lateral linkages spanning aspectsof affect, behaviour, and cognition. From this perspective, a behaviour suchas school performance would be examined by integrating various sources ofinfluence such as beliefs, motivational orientation, teacher-skills, parentalsupport, and so on. In order to understand agency in development, researchprogrammes will need to be highly collaborative and designs will need to beboth more encompassing of the various micro-divisions of the field, that is,more multidisciplinary in nature.

Regarding collaborative work, a major goal of the next era should be topool the intellectual and talent capital of numerous researchers in order toensure that the large-scale research programmes are conducted well. Forexample, a specialist on control and motivation can join forces with aspecialist in personality and a specialist on emotion regulation to designa study that does more than just explore the inter-relations among theseparate categories of study, but rather, generates and tests meaningfulhypotheses regarding the interface among the areas. Such approacheswould advance our science by allowing us to contrast, reconcile, andintegrate the plethora of micro theories that abound in developmentalpsychology.

Cross-cultural approaches are just one manifestation of collaborativework. Another type are the interdisciplinary network studies that integratevarious disciplines in the social and behavioural sciences and cross over intothe biological and neurobiological realms (e.g., the Berlin Aging Study;Baltes & Mayer, 1999). The model of the network will certainly be oneinstantiation of the successful research programme in the years to come.Such a model will probably impact the culture of psychological enquiry.For example, it is likely that fewer studies will emerge with single-authoredcredits. Instead, the results of research teams and networks will continueto propagate our journals (e.g., Grob, Little, Wanner, Wearing, andEURONET,1996).

In my view, however, cross-disciplinary projects will suffer the caco-phony of asynchronous instruments if there is no consensus or agreementabout the nature of development. With a unified view such as the agentic-organismic model outlined here, the many parts would be able to work inconcert to yield a clear symphony of human development.

236 Little

Direct versus indirect and reciprocal effects

The distinction between direct and indirect influences on development hasbecome more relevant with the advent of structural equation modelling(SEM) procedures. The traditional approach to identifying a mediator(Baron & Kenny, 1986) has given way to the SEM approach, which entailsestablishing a set of simultaneously estimated linkages (i.e., A to B to C).The magnitude and significance of the parameters are directly interpreted.Mediation is tested by adding an additional path from A to C that bypassesB. If this path is nonsignificant and does not change the magnitude of the Ato Band B to C parameter estimates, mediation is concluded. Moreover,the significance of the indirect relationships between A and C is directlycalculated as part of the procedures. However, full mediation is an idealisedsituation. In many applications, both indirect and direct effects can occurand partial mediation would be concluded. Proximal versus distal nature ofthe constructs is used to justify the placement or ordering of constructs.With time-order measurements, direct and indirect effects can be calculatedto examine reciprocal effects among a set of constructs.

GENERAL IMPLICATIONS OF THE REVITALISEDORGANISMIC PERSPECTIVE

A determined focus on the role of personal agency in development can yielda better understanding of the complexities of the relationships betweencontext and self and how they influence successful development along thelife course. The type of research that I have outlined here lies at the cuspbetween basic and applied paradigms. Therefore, the results of studies thatfocus on the various themes I have presented have real potential to impactthe field. From a basic research perspective, they allow strong tests ofcompeting theoretical models and provide important integrative implica-tions. From an applied perspective, the mass of coordinated knowledge thatpotentially can be gained would allow policy-oriented programmes theneeded basis for designing interventions that are maximally effective. Suchinterventions would thereby reflect the complexities and multiple-pathwaysin the development of the agentic self.

A larger concern is clearly the degree to which we can make statementsabout causality in the context of quasi-experimental designs. As with anyscientific method, threats to validity are numerous and, therefore, greatcaution is required. On the other hand, solid designs, good theory, andappropriate use of the analytic machinery that is available has the potentialto advance our knowledge base at an incredible rate. Multiple levels ofanalysis techniques and the judicious use of experimentation procedurescan alleviate many of the concerns. In order to reap fully the benefits ofcontextual-organismic studies, however, training programmes will need

12. Agency in development 237

to focus greater emphasis on both the basics and advances in quasi-experimental analytic techniques and expand instruction on correspond-ingly sophisticated designs.

By explicitly examining the role of agency in development, we canidentify those features of both individuals and contexts that maximallycontribute to the agentic self and successful development along the lifecourse. Understanding the developmental avenues that lead to the emerg-ence of agentic and efficacious individuals can reap great rewards in termsof the productivity and well-being of individuals across the lifespan.

ACKNOWLEDGEMENT

Parts of this work were supported by a grant from Yale College, YaleUniversity.

REFERENCES

Baltes, P.B., & Mayer, K.U. (Eds.). (1999). The Berlin Aging Study: Aging from 70to 100. New York: Cambridge University Press.

Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman.Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction

in social psychological research: Conceptual, strategic, and statistical considera-tions. Journal of Personality and Social Psychology, 51, 1173-1182.

Brandstiidter, J. (1984). Personal and social control over development: Someimplications of an action perspective in lifespan developmental psychology. InP.B. Baltes & a.G. Brim, Jf. (Eds.), Lifespan development and behavior (pp. 1-32). New York: Academic Press.

Bronfenbrenner, U. (1995). Developmental ecology through space and time: Afuture perspective. In P. Moen & G.H. Elder, Jr. (Eds.), Examining lives incontext: Perspectives on the ecology of human development (pp. 619-647).Washington, DC: American Psychological Association.

Bryk, A.S., & Raudenbush, S.W. (1992). Hierarchical linear models. Newbury Park,CA: Sage.

Cairns, R.B. (1983). The emergence of developmental psychology. In P.H. Mussen(Series Ed.) & W. Kessen (Vol. Ed.), Handbook of child psychology: Vol. 1,History, theory, and methods (4th ed., pp. 41-102). New York: John Wiley.

Cairns, R.B. (1998). The making of developmental psychology. In W. Damon(Series Ed.) & R.M. Lerner (Vol. Ed.), Handbook of child psychology: Vol. 1.Theoretical models of human development (5th ed." pp. 25-105). New York: JohnWiley.

Chapman, M. (1984). Intentional action as a paradigm for developmentalpsychology: A symposium. Human Development, 27, 113-114.

Chapman, M., & Skinner, E.A. (1985). Actions in development/development inaction. In M. Frese & J. Sabini (Eds.), Goal-direct behavior: The concept of actionin psychology (pp. 199-213). Hillsdale, NJ: Lawrence Erlbaum Associates Inc.

238 Little

Chapman, M., Skinner, E.A., & Baltes, P.B. (1990). Interpreting correlationsbetween children's perceived control and cogitive performance: Control, agency,or means-ends beliefs? Developmental Psychology, 26, 246-253.

Cicchetti, D., & Rogosch, F.A. (1999). Conceptual and methodological issues indevelopmental psychopathology research. In P.C. Kendall & J.N. Butcher (Eds.),Handbook of research methods in clinical psychology (2nd ed., pp. 433-465). NewYork: John Wiley.

Dodge, K.A., & Coie, J.D. (1987). Social-information-processing factors in reactiveand proactive aggression in children's peer groups. Journal of Personality andSocial Psychology, 53, 1146-1158.

Gariepy, J-L. (1996). The question of continuity and change in development. InR.B. Cairns, G.H. Elder, Jr., & E.J. Costello (Eds.), Developmental science (pp.78-96). New York: Cambridge University Press.

Gottlieb, G. (1997). Commentary: A systems view of psychobiological development.In D. Magnusson (Ed.), The lifespan development of individuals: Behavioral,neurobiological, and psychosocial perspectives: A synthesis (pp. 76-103). NewYork: Cambridge University Press.

Graham, J.W., & Hofer, S.M. (2000). Multiple imputation in multivariate research.In T.D. Little, K.u. Schnabel, & J. Baumert (Eds.), Modeling longitudinal andmultilevel data: Practical issues, applied approaches, and specific examples (pp.201-218). Mahwah, NJ: Lawrence Erlbaum Associates Inc.

Graham, S., & Juvonen, J. (1998). Self-blame and peer victimization in middleschool: An attributiona1 analysis. Developmental Psychology, 34, 587-538

Grob, A., Little, T.D., Wanner, B., Wearing, A.J., & EURONET. (1996). Ado-lescents' well-being and perceived control across fourteen sociocultural contexts.Journal of Personality and Social Psychology, 71, 785-795.

Hawley, P.H. (1999). The ontogenesis of social dominance: A strategy-basedevolutionary perspective. Developmental Review, 19, 91-132.

Hawley, P.H., & Little, T.D. (1999). On winning some and losing some: Socialdominance in toddlers. Merrill Palmer Quarterly, 45, 185-214.

Hawley, P.H., & Little, T.D. (2002). Evolutionary and developmental perspectiveson the agentic self. In D. Cervone & W. Mischel (Eds.), Advances in personalityscience (pp. 177-195). New York: Guilford Press.

Hawley, P.H., Little, T.D., & Pasupahti, M. (in press). Winning friends andinfluencing peers: Strategies of peer influence in late childhood. InternationalJournal of Behavioral Development.

Kenny, D.A. (1990). Design issues in dyadic research. In C. Hendrick & M.S. Clark(Eds.), Review of personality and social psychology: Research methods in person-ality and social psychology, Vol. 11 (pp. 164-184). Newbury Park, CA: Sage.

Kreft, I., & De Leeuw, J. (1998). Introducing multilevel modeling. Thousand Oaks,CA: Sage.

Kreppner, K. (1992). Development in a developing context: Rethinking the family'srole for children's development. In L.T. Winegar & 1. Valsiner (Eds.), Children'sdevelopment within social context: Research and methodology, Vol. 2 (pp. 161-182). Hillsdale, NJ: Lawrence Erlbaum Associates Inc.

Lerner, R.M. (1995). Developing individuals within changing contexts: Implicationsof developmental contextualism for human development research, policy, andprograms. In T.A. Kindermann & J. Valsiner (Eds.), Development of person-context relations (pp. 13-37). Hillsdale, NJ: Lawrence Erlbaum Associates Inc.

12. Agency in development 239

Lerner, R.M. (1996). Relative plasticity, integration, temporality, and diversity inhuman development: A developmental contextual perspective about theory,process, and method. Developmental Psychology, 32, 781-786.

Lewis, M.D. (1997). Personality self-organization: Cascading constraints oncognition-emotion interaction. In A. Fogel, M.C.D.P. Lyra, & J. Valsiner(Eds.), Dynamics and indeterminism in developmental and social processes (pp.193-216). Mahwah, NJ: Lawrence Erlbaum Associates Inc.

Little, T.D. (1997). Mean and covariance structures (MACS) analyses of cross-cultural data: Practical and theoretical issues. Multivariate Behavioral Research,

32,53-76.Little, T.D. (1998). Sociocultural influences on the development of children's action-

control beliefs. In J. Heckhausen & C.S. Dweck (Eds.), Motivation and self-regulation across the life span (pp. 281-315). New York: Cambridge UniversityPress.

Little, T.D., Hawley, P.H., Henrich, C.C., & Marsland, K. (2002). Three views ofthe agentic self: A developmental synthesis. In R.M. Ryan & E.L. Deci (Eds.),Handbook. of self-determination research (pp. 389-404). Rochester, NY:University of Rochester Press.

Little, T.D., Jones, S.M., Henrich, c.c., & Hawley, P.H. (2001).Towarda unifiedmodelof aggression:Differentiating form from function. Manuscript submittedforpublication.

Little, T.D., Lopez, D.F., & Wanner, B. (2001). Children's action-control behaviors(coping): A longitudinal validation of the behavioral inventory of strategiccon.trol. Anxiety, Stress, and Coping, 14, 315-336.

Little, T.D., Oettingen, G., Stetsenko, A., & Baltes, P.B. (1995). Children's action-control beliefs and school performance: How do American children compare withGerman and Russian children?Journalof Personalityand Social Psychology,69,686-700.

Little, T.D., Schnabel,K.U., & Baumert,J. (Eds.). (2000).Modelinglongitudinalandmultilevel data: Practical issues, applied approaches,and specific examples.Mahwah, NJ: LawrenceErlbaum AssociatesInc.

Little, T.D., & Wanner, B. (1997).The Multi-CAM: A multidimensionalinstrumentto assesschildren'saction-controlmotives,beliefs,and behaviors(Materialen ausder Bildungsforchung, No. 59, ISBN 3-87985-064-x). Berlin: Max Planck Institutefor Human Development.

Lopez, D.F., & Little, T.D. (1996). Children's action-control beliefs and emotionalregulation in the social domain. Developmental Psychology, 32, 299-312.

Magnusson, D., & Cairns, R.B. (1996)Developmentalscience:Toward a unifiedframework. In R.B. Cairns, G.H. Elder, Jr., & J.E. Costello (Eds.), Developmentalscience (pp. 7-30). New York: Cambridge University Press.

Marcoulides, G.A., & Schumacker, R.E. (Eds.). (1996). Advanced structural equationmodeling: Issues and techniques. Mahwah, NJ: Lawrence Erlbaum Associates Inc.

McArdle, J.J. (1994). Structural factor analysis experiments with incomplete data.Multivariate Behavioral Research, 29, 409-454.

Mischel, W., & Shoda, Y. (1995). A cognitive-affective system theory of personality:Reconceptualizing situations, dispositions, dynamics, and invariance inpersonality structure. Psychological Review, 102, 246-268.

Neale, M.C. (2000). Individual fit, heterogeneity, and missing data in multi groupstructural equation modeling. In T.D. Little, K.U. Schnabel, & J. Baumert (Eds.),

240 Little

Modeling longitudinal and multilevel data: Practical issues, applied approaches, andspecific examples (pp. 249-267). Mahwah, NJ: Lwarence Erlbaum Associates Inc.

Oettingen, G., Little, T.D., Lindenberger, U., & Baltes, P.B. (1994). Causality,agency, and control beliefs in East versus West Berlin children: A naturalexperiment on the role of context. Journal of Personality and Social Psychology,66, 579-595.

Poulin, F., & Boivin, M. (1999). Proactive and reactive aggression and boys'friendship quality in mainstream classrooms. Journal of Emotional and BehavioralDisorders, 7, 168-177.

Price, J.M., & Dodge, K.A. (1989). Reactive and proactive aggression in childhood:Relations to peer status and social context dimensions. Journal of Abnormal ChildPsychology, 17, 455-471.

Ryan, R.M. (1993) Agency and organization: Intrinsic motivation, autonomy, andthe self in psychological development. In J.E. Jacobs (Ed.), Nebraska symposiumon motivation, 1992: Developmental perspectives on motivation (Vol. 40, pp. I-56).Lincoln, NE: University of Nebraska Press.

Ryan, R.M., Sheldon, K.M., Kasser, T., & Deci, E.L. (1996). All goals are not createdequal: An organismic perspective on the nature of goals and their regulation. InP.M. Gollwitzer & J.A. Bargh (Eds.), The psychology of action: Linking cognitionand motivation to behavior (pp. 7-26). New York: Guilford Press.

Shanahan, M., Valsiner, J., & Gottlieb, G. (1997). Developmental concepts acrossdisciplines. In J. Tudge, M. Shanahan, & J. Valsiner (Eds.), Comparisons inhuman development: Understanding time and context (pp. 13-7I). New York:Cambridge University Press.

Skinner, E.A. (1995). Perceived control, motivation, and coping. Thousand Oaks,CA: Sage.

Skinner, E.A. (1996). A guide to constructs of control. Journal of Personality andSocial Psychology, 71, 549-570.

Skinner, B.A., & Chapman, M. (1987). Resolution of a developmental paradox:How can perceived internality increase, decrease, and remain the same acrossmiddle childhood? Developmental Psychology, 23, 44-48.

Skinner, E.A., Chapman, M., & Baltes, P.B. (1988). Children's beliefs about control,means-ends, and agency: Developmental differences during middle childhood.International Journal of Behavioral Development, 11, 369-388.

Smith, J.S., & Baltes, P.B. (1999). Trends and profiles of psychological functioningin very old age. In P.B. Baltes & K.U. Maier (Eds.), The Berlin Aging Study:Aging from 70 to 100 (pp. 197-226). New York: Cambridge University Press.

Stetsenko, A., Little, T.D., Gordeeva, T.O., Grasshof, M., & Oettingen, G. (2000).Gender effects in children's beliefs about school performance: A cross-culturalstudy. Child Development, 71,273-287.

Widaman, K.F. (2000). Testing cross-group and cross-time constraints onparameters using the general linear model. In T.D. Little, K.U. Schnabel, & J.Baumert (Eds.), Modeling longitudinal and multilevel data: Practical issues, appliedapproaches, and specific examples (pp. 163-185). Mahwah, NJ: LawrenceErlbaum Associates Inc.

Weisz, J.R. (1990). Development of control-related beliefs, goals, and styles inchildhood and adolescence: A clinical perspective. In J. Rodin, C. Schooler, & W.Schaie (Eds.), Self-directedness: Cause and effects throughout the life course (pp.19-49). Hillsdale, NJ: Lwarence Erlbaum Associates Inc.

Growing Points inDevelopmental ScienceAn Introduction

Editedby

Willard W. HartupInstitute of Child Development, University of Minnesota,Minneapolis, USAand

Rainer K. SilbereisenDepartment of Developmental Psychology, Friedrich-Schiller-University, lena, Germany

\}' ~;IYr~~~~i?G~XP PressHOVE AND NEW YORK

First published2002 by PsychologyPress27 Church Road, ((ove, East Sussex,BN3 2FA

Simultaneouslypublished in the USA and Canadaby PsychologyPress29 West 35th Slreel, New YOI'k,NY 10001

l'.I'ycllO/II!(Y Pl'e,v.I' 1,1'11111'/Ii/' /lle 'IIIY/Ol' & 1;)'(11/1'1.1'(1/'111//1

iD 2002 ISSBD

TYPlJslJ1 in Til11lJs by Mnyh"w TyplJsollinjJ, Rhayader,PowysPl'inled nnd bound In <11'0111lll'ililin by '1'.1 IntlJl'I1alional Ltd,Padslow, COl'l1w1I1ICover dosi~n by .11111Wilide

All I'ijJhls rusurved, No 1'111'101' this book nmy blJ ropl'inled orJ'(:pl'Odueed 01' nlilisod In IIny IllI'Iu 01' by nny eleetl'Onic, mechanical,01' olhel' Inellns, now known 01' herulII'IIJJ' invlJnllJd, includingphotocopying lInd l'eeOl'dln~. ()I' in any inl'OI'l1\ation stornge orl'ell'ilJVlI1syslel11, withoul pOl'lulssion in writing 1'1'0111the publishers.

1l1'1/1,I'III./III'III'Y ('(I/O/II!(I//I/I/. III 1'1/11//1'(1//111//)(1/(1A IJlIllllo!lulJ 1'lJeol'd Ihr lhls book Is IIvailllblc 1'1'0111lhe British Library

UIII'(lI'Y I~/,('IIII}I.I'I'.V.I'('(1/%g/III/. /11I'IIMI('(I/frlll /)(1/(1Gl'Owing poinls in dlJVlJlopl1lcnllll sclcnelJ : lIn introduction I edited by William W. Hartup& RainlJl' 1<, SlIbor(jlson,

1', elll,Includes biblio!l/'llphicalrcl'crenees and index.ISBN I.X416<).JII.1

I, DlJvelopl11enlul psychology, I. Hartup, William W. II. Silbereisen, R. K. (Rainer K.)1<)44

1111713.5,(17652002155 d021

2002072796

ISBN I.X4169.31'.1