Conceptual and methodological issues in studying school leadership effects as a reciprocal process

Abstract

Over the past 3 decades, a substantial body of scholarship has examined the effects of school leadership on student learning. Most of this research has framed leadership as an independent variable, or driver for change, in relation to school effectiveness and school improvement. Yet, scholars have observed that leadership is also influenced by features of the organizational setting in which it is enacted. This leads us to conclude that predominant approaches to studying school leadership effects provide an incomplete picture of the processes and paths by which leadership contributes to school learning. This paper examines the potential offered by conceptualizations of leadership as a reciprocal, or mutual-influence, process to the study of leadership for learning. We explore a variety of conceptual and related methodological issues that confront researchers who wish to employ this potentially rich but challenging approach to understanding how school leadership contributes to student learning.

Keywords:

• leadership
• school improvement
• reciprocal
• change

Acknowledgements

The authors would like to acknowledge the useful feedback offered on this manuscript by Edwin Bridges, George Marcoulides, and anonymous reviewers. The authors wish to acknowledge the funding support of the Research Grant Council (RGC) of Hong Kong for its support through the General Research Fund (GRF 840509).

Notes

1. Given space constraints and the purpose of this paper, we do not describe the data set from which the example analyses were obtained in great detail. The survey subscales (consisting of 8–10 items each) defining the collaborative leadership and school improvement capacity constructs (with alpha coefficients above 0.80) were developed through confirmatory factor analysis and have been shown in previous studies over a 10-year period to explain levels of school achievement and school growth. Achievement data were obtained from state achievement tests at the individual student level as a series of repeated measures. The tests have been vertically equated to permit examining growth over the 3-year period for this student cohort. For a detailed description of the data set, instruments, and related psychometric procedures, we refer the reader to Heck and Hallinger (2009).

2. We used the Comparative Fit Index (CFI) and the Standardized Root Mean Square Residual (SRMR) to evaluate the models. Values of the CFI near 0.95 and SRMR of 0.08 are often considered evidence of adequate model fit (Raykov & Marcoulides, 2006).

3. The basic measurement model can be defined as

where Λ is a matrix of factor loadings measuring each latent variable 𝛈 _i , and 𝛜_i are errors associated with items defining each construct which are contained in Θ. Factor variances and covariances are contained in Ψ.

4. We used the multiple-group capacity of SEM to test the fit of the subscales to the factors across the three measurement occasions. At a minimum, the same factor structure, invariant loadings of items, and invariant item intercepts on factors should be observed. This analysis is conducted to establish the consistency (i.e., reliability) and validity of our conceptualization of collaborative leadership and school improvement capacity over three measurement occasions. Adequacy of the consistency in measuring these processes simultaneously over time is determined by examining the model fit indices. Once measurement invariance is established, it is possible to examine whether perceptions changed over time. The successive factor means can be simultaneously tested (i.e., with t tests) against the initial factor mean ( = 0.00, SD = 1), which has the advantage of equating the multiple sets of scores to a common metric. The results suggested that, on average, schools increased their improvement capacity over time (i.e.,  = 0.05;  = 0.21). Although the factor score metric does not reveal the magnitude of the change, the difference between T₁ and T₃ was statistically significant (t = 3.04, p < .01). We also examined changes in the collaborative leadership factor (which is comprised of one observed scale consisting of eight items). The estimated factor means suggested leadership perceptions were the same between T₁ and T₂ but were not the same between T₁ and T₃ (t =−2.34, p < .05).

5. Variance in improvement capacity accounted for at Times 2–4 was .68, .58, and .71, respectively. Variance in leadership accounted for at Times 2–4 was .38, .47, and .66, respectively. Variance in math accounted for at Times 1–3 was .05, .78, and .46.