Hamden Project
Overview | General Aims | Procedures | Specific Aims | Significant Features | Constructs

In the following, I describe the central features of the large-scale project that our group conducted between Fall of 1999 and Fall of 2000 (three assessments separated by 6 month intervals). The overarching goal of the project was to examine self-regulatory aspects of adolescents' academic and social development. The project’s central focus is on the role of personal agency as a predictor of successful transitions and positive adaptations during adolescence and as a resilience factor in the prevention of adverse outcomes such as depression, substance abuse, and antisocial behaviors.

In our framework, personal agency refers to the intersection of central self-regulatory processes (e.g., motives, beliefs, and behaviors; see
Little, in press, and Little, Hawley, Henrich, & Marsland, in press, for detailed theoretical discussions of personal agency). The purpose of this overview is to describe our three-wave assessment of 1,505 U.S. adolescents (grades 6-9): October, 1999, April, 1999, and October, 2000. 
The project was supported by grants from the Smith Richardson Foundation ($35K), Yale College ($10K), and the Graustein Foundation ($5K). During this project we have enjoyed the enthusiastic support of the Hamden School District and the School's administrators. As a result, the sample of approximately 1,500 students reflects an active consent participation rate of about 75% from the district's 9 elementary schools (grade 6), 1 middle school (grades 7 & 8), and 1 high school (grade 9).

Hamden reflects a diverse urban setting with approximately 70% of the sample comprised of European-American participants, 25% African-American, and 5% of mixed ethnic origin. Although the SES in Hamden is variable, it is predominantly lower-middle class. Our initial analyses indicate that the sample is representative of the city’s population characteristics. Because of the involvement and support of the Hamden schools. Over the course of the study, we had low and non-selective attrition (Little, Lindenberger, & Maier, 2000).

The focus of this overview is on the longitudinal investigation of the U.S. sample (we have also collected cross-cultural data in Germany and Moscow on many of the constructs described herein). The cross-time follow-ups are extremely critical for this project because they allow us to examine the antecedents, consequents, and reciprocal influences of personal agency in development, particularly during the inherent transitions captured in our study design (e.g., the dramatic transition from 6th grade elementary school to 7th grade middle school). A significant advantage of the intensive (twice yearly as opposed to once yearly) longitudinal data is that we can examine and contrast changes that occur within the school year with those that occur over the summer recess. Moreover, we can more accurately track important (and rapidly changing) influences such as pubertal development and patterns of peer association during the course of the study. 

General Aims
The general aim of this project has been to clarify how adolescents' self-regulatory systems (personal agency) influence successful outcomes in various facets of their lives, spanning schoolwork, friendships, and peer relationships (cliques, crowds, & social networks) as well as substance use, mental health, and aggression-victimization. To do so, we designed a protocol that includes a broad set of self-related constructs and commenced with a longitudinal investigation in an effort to isolate potential mechanisms of influence.

A primary aim of the project is to identify and accurately model the mechanisms by which personal agency regulates behavior across both the academic and social domains. A significant feature of this study is that we bring together three theoretical perspectives on personal agency: self-determination theory (Deci & Ryan, in press), resource-control theory (Hawley, 1999c) and action-control theory (
Little, 1998). These three perspectives share a common meta-theoretical base that allows us to unite their central constructs into a common framework (see Little et al., in press). The core of our protocol contains measures of motivational orientation (intrinsic to extrinsic), resource control strategies (e.g., prosocial, coercive), and action-control beliefs (means-ends, agency, and control-expectancy beliefs) in both the academic and social domains. 

With our unified framework among these three views on the agentic self, we seek to understand how agency both directly and indirectly influences important outcomes such as academic achievement, friendship quality, peer integration, and personal well-being. A central feature of personal agency is the action-control belief profile that one can call upon during challenging situations. Action-control beliefs are self-perceptions about the means and competencies one has to reach one's goals. Our research program focuses on three kinds of beliefs that are related to goal achievement (for overviews see Little, 1998; Little et al., in press): (a) judgments about which specific means are most effective for reaching one's goals ('means-ends beliefs'), (b) beliefs about whether these means are personally available for use ('agency beliefs') and (c), perceptions of the degree to which a person feels that s/he can generally attain his/her goals ('control-expectancy beliefs'). 

Agentic beliefs about one's own potential are particularly important because they can energize one toward successful outcomes and help one to avoid potentially harmful influences. In other words, these beliefs are central influences on developmental outcomes because they guide one's actions and behaviors (Little et al., in press; Skinner, 1996). In general, people who feel that they have very little personal agency often do not even try to achieve their goals, have low personal standards to which they aspire, and are at-risk for various negative developmental outcomes (e.g., alienation, drug use, violence, ill-being, etc.).

If one lacks a sense of personal agency, goals are harder to reach and further failures can undermine trust in one's own capabilities which, in many cases, can lead to feelings of depression and helplessness as well as delinquent behavior (Bandura, 1997; Seligman, 1975; Skinner, 1995). On the other hand, those who have a sense of agency both try and persist in their goal pursuits. Highly agentic persons are able to reach their goals more easily, and, in turn, their successes strengthen their feelings of personal agency and well-being. 

The development of these beliefs is strongly influenced by contextualized experiences, especially in school settings. For example, in an earlier study (Little et al., 1995b), we found striking differences between U.S. and European samples in the link between achievement and beliefs in one’s ability and effort. Namely, U.S. children showed much weaker correspondence between the beliefs and performance (rs around .30) than did European children (rs as high as .75). However, the mean-levels of these beliefs in the U.S. sample were the highest among the various cultures we tested. These differences emerged even though the means-ends beliefs, which are a reflection of one’s naïve theory of how successful outcomes are attained, were remarkably similar across the various school contexts. Little et al. (1995b) identified a number of contextual factors such as classroom structure and type of feedback as possible mechanisms shaping the characteristic profiles of the U.S. sample (i.e., high levels of belief but low correspondence between beliefs and actual performance). Their conclusions, however, were only tentative because of the cross-sectional nature of the data. One important contribution of the proposed study will be a test of whether the mechanisms identified by Little et al. do, in fact, shape the link between beliefs and performance. 

To better understand the role of these self-regulatory processes in development, our design not only is longitudinal in nature but also it explicitly compares and contrasts the functioning of the regulatory processes across two life-goal domains: academics and social relations. Spanning both of these critical life domains allows us to explicitly test the generality of the links among motivations, behaviors, beliefs, and performance as well as test the cross-domain relationships that affect learning and adjustment during adolescence. In other words, we adopt a systems view of psychological development wherein the functioning of one system is assumed to influence the functioning of another (Gottleib, 1997; Magnusson & Cairns, 1996). One goal of this study, then, is to examine the interactions among sub-systems (e.g., across the social and academic domains) and to model how they contribute to successful developmental outcomes during the transitions of adolescence.

In addition, because we have incorporated other theoretically important constructs in our protocols, we can place our action-regulatory constructs within the context of the larger self system. For example, we measure social and academic self-concept (Harter, 1988), general personality (McCrae & John, 1992), and intellective skill ('QuickRaven'; Little, Wanner, & Mauch, 1999). Within the social domain, we have measures of peer status (Newcomb, Bukowski, & Pattee, 1993; Rubin, Bukowski, & Parker, 1997), aggression-victimization (Crick & Grotpeter, 1995; Dodge, 1989; Little, Jones, Henrich, & Hawley, 2000; Olweus, 1993; Perry, Kusel, & Perry, 1988), clique and crowd membership (Henrich, Kuperminc, Sack, Blatt, & Leadbeater, in press), and reciprocal best friendships (Krappmann, 1996).

Important questions that we will address include: to what degree are action-control processes unique predictors above and beyond general self-concept? To what degree do action-control processes function similarly across the two life domains? To what degree does personality and/or intellective skill moderate one's action-control profile? In other words, because we bring together a broad yet theoretically integrated set of constructs into a single encompassing study (see Table 1), we can examine and test numerous critical hypotheses about (a) the interrelations among the constructs, (b) the developmental changes that underlie them, and (c) their role and importance in successful development during the challenging transitions of the adolescent period (see specific aims, below). In addition, we will thoroughly examine both gender and ethnic moderation of the relationships among the various constructs. Given the longitudinal nature of the design and the breadth of the measured constructs, our study should provide rich and detailed information about the role of personal agency as a unique and powerful influence contributing to both positive adjustment and adaptive pathways of development during adolescence. 

An additional strength of the study is the explicit attention we have paid to the reliability and validity of all measured constructs. Through rigorous pilot testing and years of actual usage in diverse sociocultural settings, the measurement properties of the constructs have been finely tuned and their validity has been very well supported (Brendgen, Little, & Krappmann, 2000; Lopez & Little, 1996; Little et al., 1995b; Little, Oettingen, & Baltes, 1995a; Little, Brendgen, Wanner, & Krappmann, 1999; Little & Wanner, 1998; Stetsenko, Little, Oettingen, & Baltes, 1995; Stetsenko, Little, Gordeeva, Grasshof, & Oettingen, 2000). The precision and quality of our measures allow us make very aggressive tests of competing models of how the various self-regulatory processes function in development: for example, whether one dimension (motives, beliefs, or behaviors) mediates another, how a dimension may moderate a set of relationships, and the conditions under which a particular dimension has a direct or indirect effect on adjustment and well-being outcomes. Although such questions can be difficult to address, our team is quite experienced with the analytic techniques needed to perform such tests (Hawley & Little, in press, 1999; Little, 1997; Little, Lindenberger, & Nesselroade, 1999; Little, Lindenberger, & Maier, 2000; Little, Schnabel, & Baumert, 2000).

Aside from answers to basic questions about the role and functioning of personal agency in development, the study's results should assist educators, social workers, and policy makers to better understanding how such beliefs are shaped during adolescence (i.e., the mechanisms by which such beliefs naturally evolve), how they affect various performance and well-being outcomes (both academically and socially), and, ultimately, how they can be effectively improved. With such knowledge, interventions can be developed that optimally bolster the development of adolescents' beliefs about their own personal agency which, in turn, bolsters successes in negotiating life's challenges.

Before turning to specific aims and questions, we present the procedures and instruments.

Procedures and Instruments
The Hamden schools are organized around a middle-school model (grades 7 and 8). 9 elementary schools feed into 1 middle school which in turn feeds into 1 high school. The middle school is organized around 10 teams. Incoming students are randomly assigned to one of five teams with approximately 100 students per team. Staying with the same team for two years, the youth spend much of their academic and free time within teams. The study design focuses on the transition from elementary-school to the middle-school, adjustment in the middle-school, the transition from the middle-school to the high school, and adjustment in the high school. Participants were followed into the next grade level (grades 7-10 in year 2, 8-11 in year 3, and 9-12 in year 4) and assessed once during the Fall of the school year.

For each wave of data collection,we assembled our instruments into three packets which were administered three to five days apart. The packets included four comprehensive batteries: (a) the Multi-CAM: Social Domain, (b) the Multi-CAM: School Domain, (c) the About Me and My Life inventory, and (d) the Friendship/I Feel inventory.

In the teacher’s absence, one to two trained assistants proctor a survey during a standard 40-minute class period. Our preliminary analyses of the 1st wave data indicate that all constructs are valid and reliable in this sample.

Multi-CAM: Social Domain and Multi-CAM: School Domain. The two Multi-CAM surveys measure the three kinds of action-control beliefs (described above) with regard to (a) friendships and (b) learning in school. They also include questions on (a) the behavioral coping strategies used to resolve problems within both domains (
Lopez & Little, 1996; Little & Wanner, 1997; Little, Lopez, & Wanner, 2001), (b) the intrinsic and extrinsic motivations for goal pursuit in each domain (based on Deci & Ryan, 1985) and (c) the general resource-control strategies of the participants (Hawley, 1999a, 1999b, 1999c).

The About Me and My Life inventory. This inventory assesses numerous self-perceptions of general personal characteristics such as personality, social influence, leadership, aggression, self-concept, identity, and basic problem-solving ability ('QuickRaven'; Little, Wanner, & Mauch, 1999). We expect these general attributes to influence both the choice and the degree of engagement that the adolescents have in pursuing their academic and relationship goals. We also expect these attributes (or their absence) to influence the tendency toward ill-being (e.g., depressive feelings and feelings of loneliness) and externalizing behaviors. Moreover, because we assess many of these constructs in the Friendship inventory (see below) and with sociometric ratings, we can explicitly contrast the self-perceptions of one's social positioning with one’s friends’ assessments as well as the peer world's assessments. We expect the different rater's perspectives would reflect unique views that, together, offer a complete picture of adjustment. 

The Friendship/ I FEEL inventories. In the Friendship inventory, the adolescents are asked to name all the classmates (within the same grade-level) with whom they are friends and to rate these friends on general features (e.g., how close they are). With this information we are able to identify clique affiliations among the participants as well as their social networks. The second part of the inventory assesses the quality of the relationships that the children have with their three closest friends (see Brendgen et al., 2000; Little, Brendgen et al., 1999). The Friendship inventory also allows assessment of the quality of a friendship from both friends' perspectives (i.e., reciprocal friendship dyads; see Gonzalez & Griffin, 1997; Little, Brendgen et al., 1999).
Included in this battery is the Inventory of Felt Emotion and Energy in Life (I FEEL; Little, Ryan, & Wanner, 1997). The I FEEL measures the general positive and negative moods during the previous two-week period. It also assesses feelings of loneliness, isolation, connectedness, depression, and anxiety; all of which are central outcome variables in our study. 

Additional measured constructs. Also embedded in our protocol is an extensive set of sociometric measures (both traditional and newly developed) that allow us to assess the peer relationships of the adolescents. Participants write down the names of others who match a statement such as “Who do you like the most” "Who is hung out with" "Who is very shy?", "Who is a leader?" "Who often gets into fights?", "Who do you like to spend time with?", and so on. Up to 10 names per question can be nominated. Names are selected from lists of all persons in the same grade at the school (for grades 6 and high school) or lists of all persons within the same team.

We include a set of questions regarding drug, alcohol, and tobacco use. This self-report inventory is used to identify adolescents who engage in risk behaviors. One important strength of our study is that we can examine the extensive array of relationships between such risk behaviors and the motivational and self-regulatory processes contained within our larger protocol as well as the peer influences that can affect such behavior (e.g., the use patterns of one's friends and the various cliques with which one associates).

Given that our target sample is an adolescent group of significant ethnic diversity, we also measure pubertal development using multiple methods, including health records provided by the schools, ratings conducted by the school’s health provider, and self-report measures. Ethnic background and ethnic identity is assessed by school records, parent-report, and self-report.

School performance and behavior. As indicators of actual school performance, we use the grades in language, mathematics, and the other subjects that are assigned by the teacher for the preceding and current school year as well as each additional data wave. We will also collect the scores on the students' standardized achievement tests. To assess school behavior and adjustment, we collect all relevant information from the school records such as disciplinary actions, absenteeism, reports and referrals for conduct problems, and the like. 

Table 1:
Constructs in the data collection protocol: U.S. longitudinal sample
General Self-Rated Constructs

Friend-Rated Constructs

Ethnic Identity

(reciprocal best friends only)
Personality (Big-5) Friend-rated Coercive & Prosocial Influence
Social Desirability> Friend-rated Resource Control
>Problem-solving Skill> Friend-rated Overt & Relational Aggression
Drug and Alcohol Use Friend-rated Shyness and Hostility
Social Influence (Positive & Negative) Friend-rated Academic Motivation
Overt & Relational Aggression Friend-rated Friendship Quality
Resource-control strategies (inc. sub-dimensions)
>Self-reported victimization and bullying Class-rated Constructs
Personal beliefs about school and social performance (peer nominations by class, team, or grade-level)
Class-rated Coercive & Prosocial Influence
Self-Regulation: Parallel Social and Academic Class-rated Resource Control
Perceived self-concept Class-rated Overt & Relational Aggression
Perceived goal difficulty and importance Class-rated Shyness & Peer Status
Intrinsic & extrinsic motivation (inc. sub-dimensions) Class-rated Bullying & Victimization
Action-control beliefs (inc. sub-dimensions)
Coping behaviors Teacher and Parent-rated Constructs
Personality (Big-5)
Self-report outcomes Overt & Relational Aggression
Perceived Social and School Well-being Social Acceptance
Positive & Negative Affect School & Social Autonomy
Depression & Loneliness Trait Anxiety
Shyness and General Hostility
Frustration Tolerance Parents' & Teachers' Behaviors
Perceived Friendship Quality Autonomy Support
Externalizing behavior Involvement
School & Social Support
Relevant School Records
Grades, achievement scores, attendance, etc. Index Variables
Height, weight, and overt pubertal indices Family structure and Socio-economic status
Gender & Ethnicity Extra-curricular activities and involvement
Free & reduced lunches> Neighborhood safety and violence exposure

Teacher and Parent Questionnaires. In addition to the adolescents’ surveys, we ask the teachers and the parents to fill out parallel sets of questions regarding characteristics of the children in relation to their school performance and their friendships. We also ask the teachers about the relationships among the children in order to contrast the teachers' judgments and the peer world's judgments with the adolescents' own judgments. Teacher's are also asked to rate general classroom climate and how they behave in relation to each student (e.g., autonomy support, feedback). In the middle school, the students within a team are randomly assigned to one of the team's principal instructors (5 to 7 instructors per team). In the high school, students were randomly assigned to one of the six instructors of the required freshman English course. 

We ask parents to assess their child's skills in interacting with peers. We also ask parents about their sociodemographic statuses, attitudes toward parenting, and parenting practices (e.g., involvement, autonomy support, structure, etc.; see Table 1). 

Specific Aims across the Different Areas of Investigation
Given the multi-disciplinary nature of the data collection protocol, our research team will investigate a number of theoretically driven hypotheses both within and between sub-domains of inquiry. Three general sub-areas of specific aims and hypotheses are presented. 

The Role of Agency in Successful Development
We adopt an organismic approach to understanding the developing individual (Hawley, 1999c;
Little et al., in press; Ryan, Sheldon, Kasser, & Deci, 1996). From this perspective, the individual is viewed as an integrated organism that both influences and is influenced by the contexts in which s/he acts and develops. This perspective presumes that individuals are active agents in their own development. This premise is the basis of the three primary theories that we utilize for understanding the agentic self in context: self-determination theory (Ryan & Deci, in press), resource-control theory (Hawley, 1999a, 1999c), and action-control theory (Little, 1998). Self-determination theory posits central mechanisms of human agency that stem from the ontogenetic advantages of satisfying basic psychological needs for relatedness, competence, and autonomy. Resource-control theory posits central mechanisms of human agency that stem from an evolutionary perspective, emphasizing the advantages afforded individuals who are the beneficiaries of the finite resources of an environment (i.e., biologic needs). Action-control theory posits central mechanisms of human agency that stem from episodes of motivated activity.


As each adolescent continues to discover who s/he is and what s/he is capable of, the evolving competence system gives rise to a sense of personal agency. The resulting system of motivational orientation, control strategies, and action-control beliefs provides a personal foundation that is called upon to negotiate various developmental tasks and challenges throughout the life course. In facing these challenges, an agentic individual has, for example, high aspirations, perseveres in the face of obstacles, sees more and varied options, learns from failures, and has a greater sense of personal empowerment and well-being. On the other hand, a non-agentic individual has low aspirations, is hindered with problem-solving blinders, and often feels helpless and unempowered (Little et al., in press; Ryan & Deci, in press; Skinner, 1995; Weisz, 1990).

Much of the work on action-control regulation has focused on the competence system, particularly as it occurs in the academic domain (see e.g., Skinner, 1995). However, action-control regulation is a general process that applies to any domain of goal-directed activity (e.g., social relationships; Little, 1998; Lopez & Little, 1996). Moreover, the profiles of action-control regulation that emerge across different domains of functioning provide a window that can inform how the basic needs systems operate within a given individual. In our view, action-control theory is a tool with which one can gain further insight into the interplay among the needs for competence, relatedness, and autonomy. For example, in action-control theory, causality beliefs (i.e., means-ends beliefs about the effectiveness of a given means) can include intra-agent means, such as effort or ability, and extra-agent means, such as luck and powerful others (e.g., parents, teachers, and peers).

In order for children to develop an optimally agentic sense of self, they must understand that one can utilize not only intra-agentic resources, but also extra-agentic resources as means to meet their goals -- a proposition held also by resource-control theory (Hawley, 1999c). However, for individuals to believe that they personally can utilize others prosocially as means to achieve their goals, they must feel connected to them. Furthermore, for individuals to be intrinsically motivated to utilize these means, they must feel that they are engaging in the process autonomously. In this way, the needs for relatedness and autonomy shape the motivations as well as the nature and quality of the agency beliefs guiding individuals actions (e.g., beliefs in personal access to powerful others would be quite high in this case). These features of agency beliefs in turn influence the extent to which the need for competence is supported.

The causality beliefs, on the other hand, have distinct differential relationships. By the early elementary years, most children readily learn that to perform at a certain level, more effort implies less ability and vice versa (Nicholls, 1978). Moreover, the relationships between intra-agent and extra-agent means is generally independent (Little & Lopez, 1997). However, when examined as personal agency beliefs, effort, ability, and powerful others are all personal resources that one can call upon to surmount any obstacles. Thus, having (a) a sense that one possesses the ability to accomplish a goal, (b) a sense that one is capable of putting forth any needed expenditure of effort (should progress toward the goal become impeded), and (c) that one can effectively utilize others in the process, reflect a very powerful and adaptive profile for an individual. 

Action-control theory posits that the set of control-related perceptions (i.e., action-control beliefs), because they are the proximal self-related resources that one calls upon during goal pursuit, would mediate motivational orientation (i.e., intrinsically or extrinsically motivated). However, motivational orientation is, at the same time, a moderator of action-control perceptions. Here, for example, intrinsically motivated individuals should have greater beliefs in the intra-agentic means of effort and ability and these perceptions should mediate the well-documented effects of intrinsic motivation on performance and well-being (Ryan & Deci, in press). On the other hand, externally regulated individuals should have lower intra-agentic beliefs and perhaps greater beliefs in the extra-agentic means of luck and powerful others.

These self-regulatory beliefs and perceptions would function as the mediators of performance and well-being. In other words, the nature of a person's action-control beliefs will vary depending on the general motivational impetus for actions. Together, the motivational orientation and the action-control belief profiles reflect the overall quality of the agentic self.
Similarly, the action-control profile of individuals who employ different strategies of resource control would reflect the differential orientation of the strategies. For example, coercive strategies would be associated with a highly intra-agent profile and a strong belief that coercive means are effective for achieving one's goals. In contrast, prosocial strategies would be associated with a balanced profile of both intra- and extra-agent (prosocial) means viewed as most effective. In other words, the action-control belief system is a direct reflection of the regulatory orientation that compels individuals in the service of their biological and psychological needs.

The growing literature on self-perceptions points to the middle school years as an important time in the development of self-regulatory skills. Studies show that children's competency-related beliefs tend to decline during the middle school years, particularly following the transition from elementary school (Stipek & MacIver, 1989; Eccles & Midgely, 1989; Simmons & Blyth, 1987), perhaps because of the rather impersonal environment and increased emphasis on grades and performance-related feedback (Eccles & Midgely, 1989; Eccles et al., 1993; Eccles, Lord, & Roeser, 1996). Moreover, declines in perceptions of competence are associated with decreased school interest and engagement, lowered intrinsic motivation, poor academic performance, and poor emotional well-being (Harter, 1990; Roeser, Eccles, & Sameroff, 1998; Stipek, 1996). In turn, these socio-emotional factors have been associated with poor long-term educational outcomes. Such links, however, have not be examined in a combined form. In this study, we will explicitly test these multiple linkages -- that lowered self-perceptions influence motivation (i.e., are amotivating), causing disengagement from school, which, in turn, results in poor performance, ill-being, and externalizing behaviors such as alcohol and tobacco usage.

We also hypothesize that children's beliefs about the causes of school achievement and their personal access to those causes (i.e., action-control beliefs) will meditate the effects of the middle-school transition on their motivation and performance. For example, we predict that the combination of high means-ends and high agency beliefs for effort will promote academic engagement and performance both cross-sectionally and over time. In contrast, high means-ends but low agency beliefs for extra-agent causes (e.g., luck, powerful others, and unknown) are expected to undermine engagement and performance (Skinner, 1995). We predict that the combination of both high agency and high means-ends beliefs for both effort and ability will serve as a protective factor during spring-fall transitions, particularly over the transition from 6th to 7th grade. Children who endorse this combination of beliefs will show less of a decline in perceived competence, control expectancy, intrinsic motivation, and achievement following the transition than will their classmates. Conversely, children who endorse high means-ends beliefs for extra-agent causes, and low agency beliefs for these causes are expected to be the most negatively affected by the transitions.

By assessing adolescents' action-control motives, beliefs, and behaviors across both the social and academic domains, we can examine both the domain-general and the domain-specific characteristics of these critical self-regulatory processes. Our hypotheses here are that (a) the cross-domain linkages will be quite low because the experiences that shape the processes are mostly unique to a given domain of functioning. On the other hand, we expect (b) that the linkages among the action-control constructs within a domain will evince the same patterns across the two domains (invariant predictive relationships) and (c) that the role of these agentic processes in predicting adjustment and well-being will show strong and significant relationships.

Peer Relationships in Adolescence

Peer relationships are critical for adolescents' successful adjustment (Berndt & Perry, 1990; Buhrmester & Furman, 1987). Having close friendships is associated with enhanced well-being and interpersonal competence. Friendships also provide models for either positive adjustment or a host of behavior problems (Berndt & Keefe, 1995; Cairns & Cairns, 1994). Peer acceptance versus rejection is also associated with adolescents' social and academic adjustment (Wentzel, 1994). Furthermore, both close friendships and peer acceptance predict adult adjustment (Bagwell, Newcomb, & Bukowski, 1998). Friendship theory and research have traditionally focused on dyads and best friends and have assumed that peer relationships are generally egalitarian and symmetric in nature (i.e., equal give and take; Bukowski, Newcomb, & Hartup, 1996; Sullivan, 1953). Other approaches, however, have integrated broader notions of context and have looked at friendships beyond the dyad (Brown, 1989; Cairns & Cairns, 1994; Durbin, Darling, Steinberg, & Brown, 1993; Urberg, Degirmencioglu, Tolson, & Halliday-Sher, 1995). Furthermore, peer relationships have been found to be more asymmetrical in subtler ways than previously thought (Hawley & Little, 1999). 

The peer data in our study will allow us to investigate the dynamics of both best-friend dyads and larger networks of social groups (e.g., cliques & crowds) as well as patterns of peer acceptance and rejection. We will explore both symmetry and asymmetry in the various types of peer relationships from multiple perspectives. Most importantly, however, our general aim is to examine the mediating role of the agentic self-regulatory processes across three primary sub-areas of peer relationships: friendship groups, peer influences, and peer aggression-victimization.

Friendship groups. As mentioned, friendships bolster well-being and overall adaptation (Berndt & Perry, 1990; Buhrmester & Furman, 1987). Furthermore, having competent, achievement-oriented friends can lead to beneficial outcomes (Epstein, 1983) while deviant friends can promote antisocial behavior (Cairns & Cairns, 1994; Dishion, Eddy, Haas, & Li, 1997; Kandel, 1978). Adolescent relationships, however, extend well beyond intimate friendships. Adolescence is a period of unsurpassed growth in the number and diversity of peers to which one is exposed, and accordingly is associated with increases in the number of peers with whom one associates. These factors add complexity to the influences on a youths' life, for better or worse. When considering how to promote positive outcomes in adolescence, the sources of complexity must be examined in the broader social context in order to understand both the positive and deleterious influences on, for example, achievement, risk behavior, and competence.

We know that, in general, close friends tend to be similar in important ways. For example, their levels of academic achievement and behavior problems, and their engagement in behaviors such as substance use tend to be matched. Longitudinal research indicates that this similarity of friends is about equally due to the bi-directional forces of selection and influence (Aseltine, 1995; Bauman & Ennett, 1996; Epstein, 1983; Kandel, 1978). However, few have ventured into the more complex domain of peer groups as sources of influence. The work that has been done suggests that members of larger social groups, such as cliques, also tend to have similar characteristics (Brown, 1989; Cairns & Cairns, 1994; Henrich et al., in press; Kindermann, McCollam, & Gibson, 1996; Shrum & Cheek, 1987; Urberg et al., 1995; Urberg, Degirmencioglu, & Pilgrim, 1997; Cohen, 1977). For example, children who are achievement-oriented tend to affiliate with others who are also achievement-oriented.

Some evidence suggests that this effect, as with dyadic friendships, is also bi-directional: Children appear to select each other based on some similarity, and then grow more similar over time (Cohen, 1977; Kindermann et al., 1996). The dynamics of influence within the groups and how these dynamics differ from the characteristic of friendships are largely still a mystery. Qualitative research indicates that cliques consist of hierarchical social orders with a range of roles from central influential leaders to peripheral conforming “wanna-be’s” (Adler & Adler, 1995; Damico, 1975, 1976). To date, however, little is known about the mechanisms through which friendship groups influence their members or how friends influence one another. We will examine possible mechanisms of influence in both friendships (e.g., friendship closeness) and clique groups (e.g., group cohesion) and explicitly contrast these two sources of influence on individual development.

The tendency of adolescents to form cliques has important implications for their learning at school and their social adjustment. It is therefore important to explore and understand the magnitude of similarity and the mechanisms of influence (e.g., agentic processes) among friends and members of peer groups. We hypothesize that group membership may support different motivational needs than do reciprocal close friendships and that the patterns of influence in cliques are more unidirectional and asymmetric than in friendships. Specifically, our protocol allows us to identify both reciprocal close friendships and patterns of social groups in the sample. We will be able to assess and follow longitudinally the degree of centrality of one’s membership in these groups and the degree and direction of influence. We will be able to quantitatively assess the internal social dynamics of cliques and to compare the patterns of peer influence in the social group and the dyadic close friendships. 

If the motivational needs fulfilled by these two contexts of peer relationships are indeed different, then the mechanisms of influence may also differ. Our primary hypothesis is that agentic self-regulatory processes will influence the direction and nature of these spheres of influence. For example, we predict that socially central individuals will have high action-control beliefs for their social goals. Similarly, we predict that well developed action-control beliefs will (a) serve as a buffer against unwanted pressures for conformity or deviant behavior and (b) provide a mechanism by which quality friendship relationships are developed and maintained.

Peer influence. Peer relationships (in contrast to relationships with adults) traditionally have been assumed to be symmetrical and on a horizontal plane in terms of power assertion (e.g., egalitarian; Piaget, 1962). However, early ethological work demonstrated that peer groups are beset with inequity (McGrew, 1972; Sluckin & Smith, 1977; Strayer & Strayer, 1976). Groups of preschoolers, for example, develop structures (i.e., hierarchies) that arise from children’s differential abilities to aggressively dominate their peers. This focus on aggression, however, has obscured children’s tendencies to adopt different strategies to control their social and material environments. Additionally, although this view alleged that dominance is an important aspect of dyadic relationships (Strayer & Strayer, 1976), it did not explore qualities relevant to interpersonal relationships (e.g., familiarity) or individual-level characteristics that children bring to relationships or groups (cf. Hawley & Little, 1999). More recent approaches to social dominance shift the focus from aggression to the different strategies of control that children employ (i.e., a shift from a structural to a functional approach; Hawley, 1999b, 1999c). According to this perspective, the optimally functioning individual considers others’ goals while pursuing his or her own (a question of social competence; e.g., Vaughn & Bost, 1998) and may well use prosocial behavior to manipulate others. In this sense, cooperation is an effective means to compete (Charlesworth, 1988, 1996).

This functional perspective raises novel questions about the development of adolescent’s control strategies, the qualities of individuals that foster successful control, the qualities that move a one to adopt socially acceptable (i.e., prosocial ) versus unacceptable strategies (i.e., coercive or overtly aggressive), and about the self and social outcomes of individuals employing various strategies in the peer group. Successful resource control in young children, for example, is predicted by persistence, extraversion, and social competence (Hawley, 1999a; Hawley & Little, 1999). In older children (around age 9), controlling children are composed of those who are drawn in positive ways to peers (e.g., agreeable) and those who are not (e.g., hostile; Hawley, Pasupathi, & Little, 1999). This perspective leads us to hypothesize not only about how prosocially and coercively controlling adolescents are different, but also about how they are the same (generally motivated to control). Importantly, by identifying more subtle strategies of control, we expect to find that girls are as successful at controlling as boys.

Social dominance is associated with social centrality (e.g., dominant individuals should be looked at and sought after; Abramovitch & Strayer, 1978; Chance, 1967). This association in children, however, appears to break down at about age eight (Dodge, Coie, Pettit, & Price, 1990; Pettit, Bakshi, Dodge, & Coie, 1990; Wright, Zakriski, & Fisher, 1996). We assume that this breakdown is not because social dominance is unimportant to human social organization, but rather because social dominance has been misleadingly associated with aggression to the exclusion of alternate strategies. From our functional perspective, we expect resource controlling adolescents to be highly socially central (e.g., popular), provided that they are employing prosocial strategies of control. Accordingly, we also expect to find non-resource controlling adolescents to be socially invisible (e.g., neglected). Furthermore, we hypothesize that friendships are formed around the power structure; relationships with powerful others are desirable, but not always achievable. Therefore, friends must settle with those of near-equal power.

Our questions concern the adaptiveness of social influence. By collecting personality assessments from multiple sources, nominations of who leads, who bullies, who is popular, who controls resources, who engages in overt aggression, and so on, from the peer group, and well-being assessments from the adolescents themselves, we can test hypotheses about what personality characteristics predict various levels and strategies of control and their relationship to agency (e.g., resource controlling adolescents will have high levels of agency). In addition, we can examine the degree and type of social influence one’s friends possess, and thus the friendship structure of a group in relation to its power structure.

Peer aggression. Aggressive behavior has weighty implications for adolescents’ social and academic functioning. Hostility poisons an atmosphere conducive to learning (Cairns, Cairns, Neckerman, Gest, Gariepy, 1988) and aggressive individuals, or groups of individuals, influence others asymmetrically by their very nature (Boulton, 1996; Olweus, 1993). Aggressive conduct has also been linked to the quality of peer relationships and peer victimization (Cairns et al., 1988; Coie & Dodge, 1998; Crick, 1997; Crick & Bigbee, 1998; Crick & Grotpeter, 1995; Dishion, French & Patterson, 1995; Farrington, 1991; Parker, Rubin, Price & DeRosier, 1995). This research, however, has focused primarily on physical aggression (i.e., overt aggression) which appears to be more characteristic of boys than of girls (Bjoerkqvist & Niemelae, 1992; Crick & Dodge, 1994, Dishion, 1990; Maccoby, 1998; Maccoby & Jacklin, 1980). In fact, girls may not be less aggressive, but rather appear to emphasize more subtle forms of aggression such as damaging or manipulating relationships among peers (i.e., relational aggression; Crick, Bigbee & Howes, 1996; Crick, 1995; Crick & Grotpeter, 1995, Rys & Bear, 1997).

Despite this work’s valuable contribution to our understanding of the nature and functioning of peer relationships, except for a few cases (Grotpeter & Crick, 1996), studies incorporating these multiple measures of aggression have not investigated the unique patterns of relational and overt aggressive behavior in both dyadic friendships and larger friendship groups such as cliques. Additionally, there is very little work on the evolving nature of these patterns and relationships over the particularly tumultuous transitions that characterize schooling during adolescence (cf. Seidman, 1991). Furthermore, most research has focused on the structural dynamics of aggression in peer relationships rather than on its functional attributes. What, for example, is gained by the perpetrator when the relationship between other individuals is damaged?

Our functional approach to social dominance suggests important possibilities. For example, individuals may resort to relational aggression to access and defend important resources; namely, social relationships or social status. The work on relational aggression suggests an interesting avenue that individuals can take to be successful controllers. Namely, while overt aggression may be used more maladaptively by marginalized peer groups (Cairns et al, 1988), relational aggression may be used successfully as a means of social control by socially agentic and socially dominant adolescents. This latter prediction implicates a further role of social self-regulatory processes; namely, that depending on the type of aggression, persons may have high action-control beliefs (relationally aggressive) or low action-control beliefs (overtly aggressive).

Parenting style and involvement and adolescent motivation

Although peer relationships play an increasingly dominant role in the social lives of adolescents, parents continue to influence school and social adjustment in important ways. As posited by self-determination theory (Grolnick & Ryan, 1989; Grolnick, Deci, & Ryan, 1997; Ryan & Deci, in press), parenting styles that are autonomy supportive, such as authoritative parenting, are positively related to adolescents’ social adjustment and academic achievement (Dornbusch, Ritter, Leiderman, Roberts, & Fraleigh, 1987; Durbin et al., 1993; Fletcher, Darling, Steinberg, & Dornbusch, 1995; Grolnick & Slowiaczek, 1994; Maccoby & Martin, 1983; Steinberg, Lamborn, Dornbusch, & Darling 1994). Steinberg and colleagues have demonstrated that authoritative parenting is associated with greater psychosocial competence in adolescent children and even the children’s friends (Fletcher et al., 1995). Parenting style is further predictive of the types of friends adolescents affiliate with (Durbin et al., 1993). In terms of academic achievement, authoritative parenting predicts high achievement (Dornbusch et al., 1987; Steinberg, Lamborn, Darling, & Mounts, 1992).

Steinberg et al. (1992) also found that parental involvement at school mediates the relations between parenting style and academic achievement, and that non-authoritative parenting styles undermine the beneficial effects of parents’ school involvement on student achievement. These findings link the research on parenting style with the large body of evidence that parent involvement is associated with better school performance (Baker, 1996; Epstein, 1995; Henderson, 1987; Henderson & Berla, 1994).

Recent research approaching the effects of parent involvement from the framework of self-determination theory suggests that the relationship between parenting style, parent involvement, and student achievement is mediated by motivational processes (Grolnick, Ryan, & Deci, 1991; Grolnick & Slowiaczek, 1994). According to self-determination theory, parents promote intrinsic and internalized motivational orientations toward achievement in their children by supporting their inherent needs for autonomy, competence, and relatedness (Grolnick & Ryan, 1989; Grolnick et al., 1997; Ryan & Deci, in press). Grolnick and colleagues have found empirical support for the propositions that children’s perceived autonomy and competence mediate the effects of parent involvement (Grolnick et al., 1991; Grolnick & Slowiaczek, 1994).

Our parental protocol assess multidimensional profiles of parenting style and parent involvement. These profiles will be used to compare students’ intrinsic motivation, perceived competence, and their perceptions of parental expectations and support. Using the longitudinal data, we will explore the direct and indirect effects of the inter-relations between parenting style, parent involvement, child motivation, and child performance. Specifically, we hypothesize that parent involvement at school will be most effective when supported by autonomy-supportive parenting coupled with parental supervision and support of academics at home.

We will also test predictions for how parent involvement affects student motivation and academic achievement. Specifically, we hypothesize that the profile of effective parent involvement described above promotes students’ perceived competence and internalized/intrinsic motivation by supporting the needs for autonomy, competence, and relatedness. Through this motivational process, parent involvement is hypothesized to affect students’ feelings of personal agency and their achievement. In other words, we hypothesize that the action-control beliefs of the adolescents mediate the motivational influences on achievement. To the best of our knowledge, this connection between parenting factors, adolescents' general motivational orientations, and specific action-control processes has not been examined, either cross-sectionally or longitudinally.

Because we also assess parallel constructs of social motivation and have both peer and friendship variables as social outcomes, we will investigate the processes through which parents influence their children’s social adjustment. We will test this under-studied linkage and the generality of the proposed model by examining it in the social domain. Specifically, we will test whether the parent involvement and social motivational processes mediate social adjustment and whether the directions and magnitudes of these effects are similar to the processes influencing adolescent achievement. Moreover, the cross-domain influences have not been examined in previous research. We will examine, therefore, whether parental support and involvement in the social domain influences motivation and performance in the academic domain.

We hypothesize that those parents who provide supportive and involved parenting across these two life domains provide an optimal developmental context for their adolescent children and, thereby, contribute maximally to their development, not only in terms of their academic performance, but also their social adjustment, friendship quality, and general well-being. Furthermore, we will test these models longitudinally as students transition into middle school. Because this transition is a particularly salient developmental context and is associated with dramatic drops in both parental involvement and student motivation (Carnegie Corporation, 1996), examining the linkages among parenting, motivation, action-control beliefs, and adjustment offers a strong test of the generality of the processes. Particularly because action-control processes are most salient when a person is under stress, these transitions are critical to fully test the proposed model of the mediating role action-control beliefs play between parenting and successful adolescent outcomes.

Significant and Unique Features of the Overall Design
Theoretically based and integrative. A number of theoretical models have been integrated to provide a united perspective that drives the central questions of this study. These models include the action-theory model of perceived control (Skinner, Chapman, & Baltes, 1988), the self-determination model of motivation (Deci & Ryan, 1985), and the resource-control model of peer interactions (Hawley, 1999c). In addition, we have carefully selected features and constructs of other theoretical perspectives and integrated them into our guiding framework. These theoretical models include the 5-factor model of personality (Caspi, 1998), a multi-dimensional action-theory model of coping (
Lopez & Little, 1996; Little, Lopez, & Wanner, 2001), and a unification action-theory model of aggression (see Little, Jones et al., 2000). 

Comprehensive coverage of constructs. The data collection protocol was designed to be a thorough and integrated compilation of theoretically related constructs. Our goal is to fully understand the inter-relations among the self-related motives, beliefs, and behaviors that comprise the agentic self (see Little et al., in press; and see Table 1). Clearly, not all constructs related to the self have been included, however, the list reflects a comprehensive selection of theoretically interconnected constructs. In addition, the constructs in our protocol have been well validated and possess satisfactory to excellent psychometric properties (see discussion above).

A contextual approach. We have taken a deliberated contextual view on the development of the agentic self. Generally speaking, contexts reflect a specific constellation of features (i.e., where intra- inter- and extra-personal features converge). These features provide a quasi-experimental manipulation of many of the potential mechanisms that can shape and mold individual development. Because contextual designs vary the features in which development occurs, they can uncover fundamental mechanisms that influence development and dramatically increase our universe of generalization (see e.g., Little, 1998). 
In particular, the naturally occurring transitions that these adolescents face provide clear and powerful challenges to self development. Following these youth as they progress through these transitions offers tremendous opportunity to examine many mechanisms of self-regulation (because of the breadth of our protocol) and to evaluate which are most effective and under what circumstances.

Moreover, our inclusions of ethnicity, gender, and SES as index variables will allow us to examine the moderating effects of the various subgroups. For example, research indicates that low SES minority children are at greater risk for school failure (Luster & McAdoo, 1996; Graham, 1994). Furthermore, authoritative parenting and parental involvement appear to have less of a positive impact on the school performance of minority children (Dornbusch et al., 1987; Luster & McAdoo, 1996; Steinberg, 1996), and minority adolescents additionally tend to be isolated from school-based peer social groups (Urberg et al., 1995). Considerable work from other theoretical perspectives has demonstrated pronounced differences in motivational systems of different subgroups (e.g., Graham, 1995; Stetsenko, Little et al., in press). One goal of this study will be to broaden and elaborate on this growing literature.

Cross-links between sub-areas. During the course of the project will focus on quite specific hypotheses within a sub-area of functioning; however, a significant number of hypotheses focus on the relationships between the various sub-areas. A comprehensive and integrated data set such as this also lends itself well to more exploratory investigations. Because of the substantial sample size of the data, these explorations can be conducted with cross-validation procedures, thereby minimizing the chances of inference errors and false conclusions. Because we have assessment in both the social and academic domains, we can explore various interrelations between the domains over time and across contextual transitions.

Longitudinal design. Critical transitions during adolescence highlight the question of continuity (Little, Stetsenko, & Maier, 2000). We will trace the developmental trajectories of each participant across the various transitions that are inherent in our design (e.g., the transitions to middle school and to high school). As mentioned above, issues of cross-lagged influence will be explicitly tested. For example, we will examine how personal agency, within the context of mutual friendships, contributes to quality friendship relations and, at the same time, examine agency within the context of achievement and, together evaluate their influence on adjustment and personal well being. In other words, understanding both the antecedents, reciprocal influences, and consequences of agentic processes is critical for understanding successful developmental pathways and outcomes during the turbulent transitions of the adolescent era.

Benefits and Contributions
Given the unique features of our protocol, the intensive longitudinal design, and the large and regionally representative sample, the potential benefits and contributions of this project are significant. Our project lies at the cusp between basic and applied paradigms. From a basic research perspective, the study will allow quite strong tests of competing theoretical models and provides important integrative findings. From an applied research perspective, the mass of knowledge that can be gained will allow policy-oriented programs the needed basis for designing interventions that are maximally effective. Such interventions can thereby reflect the complexities and multiple pathways in the development of the agentic self. Policies that provide both educational and social supports can be implemented with the goal of bolstering the agentic self-regulatory processes that adolescents need to negotiate the challenges and potential pitfalls of this critical life phase. Furthermore, developing and bolstering a sense of personal agency is important not only for individuals who are potentially at risk, but also for those who are within generally normative contexts; that is, by optimizing their unfolding pathways.