• No results found

Chapter 3 Ecological systems theory in relation to the present thesis Introduction

3.2 The conceptualisation and specification of the model based on EST

Model conceptualisation

EST provided a general theoretically-guided framework that identified the underlying processes according to which the above discussed home- and school-related factors were hypothesised to influence the development of adolescent educational expectations. EST defined what a proximal process is as distinguished from actors’ subjective interpretations of a proximal process. Ecological system theory has provided theoretical support to my claim that LSYPE respondents’ reports, assessments and estimates captured the important subjective understandings of the home- and school-related processes in which mothers and young people participated. In short, EST

theoretically validated these responses as valid operationalisations of the underlying home- and school-related proximal processes and provided the theory that justified why they are expected to impact on each other. The thesis will examine these mesosystemic relations (Neal and Neal, 2013) between the home and school and explore their role in the development of adolescent educational expectations between ages 14 to 16.

But most importantly, EST provided theoretical support for the conceptualisation of maternal ethnicity as a distal moderator of proximal processes in the microsystem and the framework that specified the mechanisms via which this is done. The same could be said about social class. Ecological systems theory does not specify which of the two distal influences is the most potent

chapter 2 showed that social class was less likely to affect parental and adolescent educational expectations. Ethnicity on the other hand, was shown to provide a buffer from class-induced disadvantage and was associated with higher educational expectations than expected just on the basis of social class. This evidence suggests ethnicity as a more probable moderator of the influence of parental social class on the development of adolescent educational expectations rather than social class as a moderator of the effect of ethnicity on the same outcome. My analysis will provide a direct test of the first hypothesis.

Model specification

EST also guided the actual specification of the model in this thesis. First, EST provided the

theoretical grounds to explore the potential interrelations among the hypothesised home- and school- related factors. The theory presupposes that the contexts that sustain human interaction involve multiple proximal processes. The form, power, content and direction of these proximal processes producing development were argued to vary systematically as a joint function of personal

characteristics and surrounding context. Thus, EST provided the theoretical basis for testing simultaneously the effect of multiple proximal processes on the developmental outcome in this thesis.

EST provides theoretical support for specifying a quantitative model capable of investigating all possible causal routes that may involve the outcome. Such a model investigates systematically

‘the relation of development to variously combined characteristics of the person and the environment involving a set of processes through which the course and consequence of development are determined. These processes may involve […] feedback mechanisms, sequential stages and alternative paths of direct and indirect influence’ (Bronfenbrenner, 2005, p. 69).

Thus, EST provided the framework for systematic exploration of three types of potential mediation of background influences at pupils’ age 14 on pupils’ expectations at age 16 via multiple proximal processes represented by the above four factors at age 15. These background influences at pupils’ age 14 are first, the effect of parental social position; second, the prior effects of the four factors, and third the feedback of pupils’ prior expectations. The same framework of EST also supports a model specification for systematic exploration of potential moderation of the above three types of

mediation by maternal ethnicity. It is thus well-suited to test for moderated mediation (Muller, Judd and Yzerbyt, 2005). Thus specified, the model will explore whether the hypothesized proximal processes exert effects of different type, magnitude and direction on the developmental outcome;

whether these effects change as a function of an external factor representing surrounding distal context (i.e., across categories of different maternal ethnicity); whether these effects change as a function of time.

Modelling time explicitly is also theoretically supported. Thus, chronosystem designs (i.e., longitudinal contextual models) ‘permit one to identify the impact of prior life events and experiences, singly or sequentially, on subsequent development’ (Bronfenbrenner, 2005, p. 83). Following this principle and related methodological issues to be discussed in chapter 4, every home- and school-related proximal process as well as the developmental outcome was followed over ages 14-16. The model which will be formally presented in chapter 4, included measurements of all factors at baseline (age 14) as well as at least another measurement occasion at age 15, followed in most cases by another at age 16. Thus, each hypothesised proximal process at ages 15 and 16 explicitly controlled for its prior occasions under the theoretical premise of EST that each proximal process is a function of its own (prior) development.

Specifying my model as a Person-Process-Context-Time model

The Person-Process-Context-Time (PPCT) model is a dynamic ecological systems model representing Bronfenbrenner’s (Bronfenbrenner and Morris, 1998) later developments of his theoretical formulation. It incorporates all the insights gained from EST referred to above. It provides a blueprint for conceptually coherent and operational research designs (Bronfenbrenner, 1988) informed by EST. This section defines the conceptual specification of the present research design as a person-process-context-time model. It identifies which elements of the model to be presented in chapter 4 were specified as ‘person’, ‘process’, ‘context’ and ‘time’.

Process, person, context and time are the four basic elements in the PPCT model. Process refers to an enduring form of interaction that the person is exposed to and participates in, like a pupil’s engagement with homework. Person refers to personal characteristics. A process or a person model simply offer very limited information of the ecological process because they measure either the extent of interaction or categorize people according to their characteristics.

The person-process models are capable of testing the hypothesis that various combinations of contextual and personal characteristics can produce developmental effects that cannot be predicted

In quantitative models these effects on the outcome are typically tested as person by context

statistical interactions controlling for the main effects of each predictor. Class by ethnicity or gender by ethnicity are examples of such interactions. If shown to be significant, these statistical

interactions would be typically interpreted as showing moderation of the effect of the predictor on an outcome by ethnicity, social class or gender.

The person-process-context model permits testing of a much larger set of hypotheses involving multiple interactions between personal characteristics, multiple proximal processes and different contexts. These models are capable of testing ‘synergism, [i.e.,] the joint operation of two or more forces producing an effect that is greater than the sum of the individual effects’ (Bronfenbrenner, 2005, p. 117). Thus, it is capable of testing the hypothesis that different contexts are associated with different developmental processes and outcomes (Bronfenbrenner, 2005, p. 118). This typically leads to more expanded regression models with more interaction terms and their combinations.

When specified as structural equation models (SEM), person-process-context models are capable of exploring and analyse direct and indirect effects (Bronfenbrenner, 2005, p. 85). However a major limitation of person-process-context models is that they are cross-sectional. In other words, they do not model time specifically. Failing to model time specifically typically creates a number of

methodological and interpretive problems that will be discussed in chapter 4.

Person-process-context-time (PPCT) models are described by Bronfenbrenner as chronosystem models (Bronfenbrenner, 2005, p. 84). These models include time explicitly in their specification acknowledging the fact that time ‘alters the existing relation of a person and the environment […] creating a dynamic that may instigate developmental change’ (Bronfenbrenner, 2005, p. 119). They are longitudinal models that specify that the hypothesized predictors precede the hypothesized mediators which in turn precede the outcome. In case of multiple mediators, as is the case here, these models also permit analysis of cross-lagged effects among mediators and feedback effects from earlier occasions of the outcome on the mediators at later time points and via them, on subsequent occasions of the outcome.

My model, presented in chapter 4 and estimated in chapter 7, is a short-term chronosystem model (Bronfenbrenner, 2005, p. 120 ) as it will examine developmental changes in adolescent

expectations between ages 14 to 162. ‘Person’ is defined by mothers and young people. ‘Process’ is defined by the four factors representing home and school-related proximal processes; ‘context’ is defined by the white, Indian, Pakistani, Bangladeshi and Black African group memberships; ‘time’ is defined by the two-year time window between pupils’ ages 14 to 16. As explained in the

introduction, this is a short but critically important time window. Age 16 signals the end of compulsory education in England and the time when educational expectations begin to actualise. Year 11 (age 16) marks therefore an important transition in an adolescents’ life course. This ecologically important two-year period leading to this transition is captured by three consecutive yearly measurement occasions, at age 14, 15 and 16 based on LSYPE panel data.

In the chronosystem model of this thesis, factors are placed at different times depending on the role they are hypothesized to play in the longitudinal mediation model, following Cole and Maxwell’s (2003) recommendations, as will be explained in chapter 4. The predictor, parental social position and family material circumstances for example, is measured at age 14. The mediators, parent-child conflict, pupils’ engagement with homework, feelings about school and assessments about teachers’ effectiveness are measured at age 15. The outcome, adolescent expectations, is measured at age 16. However the prior occasions of all the mediators at age 14 and of the outcome at ages 14 and 15 are also specified in the model. As it will be explained in chapter 4, this model specification permits more reliable assessment of the three types of potential influences to adolescent expectations via the longitudinal mediation of the four mediators. This is because the model allows for time to operate between the following critical time lags. First, between the hypothesized predictor at age 14 and the mediators at age 15 as well as between the mediators at age 15 and the outcome at age 16. Second, between the prior measure of the mediators at age 14 and their repeated measure at age 15. Finally, between the prior measure of the outcome at age 14, the measure of the mediators at age 15 and the measure of the outcome at age 16. To analyse potential moderation by specific distal contexts, like, social class and ethnicity, Bronfenbrenner (1988; 1992) suggests that the same SEM should be estimated across different categories of the potential moderator. Elder (1999) used such a technique to test for moderator effects, and it is also the technique typically recommended for mediation SEM (Baron and Kenny, 1986; MacKinnon, 2008; Muller, Judd and Yzerbyt, 2005).

3.3 Conclusions

The purpose of this chapter was to justify why EST provides a theoretically consistent framework supporting the potential roles parent-child conflict, pupils’ engagement with homework, feelings about school and assessments of teachers’ effectiveness are hypothesised to have in this thesis. The chapter presented and discussed the principles of ecological systems theory relevant to the analysis that follows, described its topology and situated the factors included in my model in it.

It was shown how the conceptualisation and the specification of the quantitative model were informed by EST to become a mesosystemic ecological model (testing relations in a mesosystem). Specifying the model as a mesosystemic ecological model will enable testing explicit hypotheses about the potential influence of these proximal microsystems on the development of adolescent expectations. The model will permit systematic exploration of three types of background influences at pupils’ age 14 on pupils’ expectations at age 16 via the potential mediation of parent-child conflict, engagement with homework, feelings about school and assessment of teachers’ effectiveness at age 15. These background influences are first, the influence of parental social position; second, the influence the mediators at ages 14; and third, the feedback of pupils’ prior expectations at age 14. The model will also permit exploration of the potential moderation of all of these potential causal routes by maternal ethnicity. EST offers strong theoretical support for these analyses. Finally, it was shown that the model that captures these effects simultaneously is a chronosystem model in that it includes all the elements for the person-process-context-time model. The formal presentation of the model follows in chapter 4.

Chapter 4 Models and measurement methodology

Outline

Related documents