Using Hierarchical Growth Modeling to Promote District Systematic Improvement in Ohio and Texas

Abstract

The purpose of this study is to apply hierarchical linear growth modelling to district demographic and effectiveness data in the U.S. states of Ohio and Texas to identify districts that are significantly outperforming or underperforming their demographic and resource characteristics from the entire population of districts in a state. The study focuses on administrative panel data for the years 2006-2012 for all districts in Ohio (n=608) and 2005-2011 for all districts in Texas (n=1028). Our findings indicate that: 1) Multiple district demographic variables and financial expenditures, such as enrollment, student ethnicity, district location, and general and school administration expenditures per pupil have a significant relationship to growth in district academic achievement; 2) we identified 32 significantly outperforming districts in Texas and 15 districts in Ohio; 3) we compare this identification system with state accountability indicators through a quadrant plot to provide a new framework for site selection in district effectiveness research, which has important implications for the policy and practice of re-envisioning school district effectiveness.

Examining a congruency-typology model of leadership for learning using two-level latent class analysis with TALIS 2018

“…Across the teacher-level part of the model, while the five domains of the leadership for learning model provide a useful framework to group the indicator variables, Figure 3 demonstrates that there are some discrepancies, as within the five domain item clusters, the lines for the three groups cross. In the previous research using this leadership for learning framework with the five domains and a two-level LCA (Bowers et al, 2017 [20]), the teacher Level 1 types had an ordinal set of responses, with high, medium, and low groups. Thus, each group in the LCA with the five domain leadership for learning framework behaved similarly within the five clusters of items.…”

“…The CALL validation dataset included n=3 367 teachers and 117 schools and their principals in the United States. In their study, Bowers et al (2017 [20]) proposed a "congruency-typology" model of school leadership, in which they proposed that there may be a multiple group typology of teachers as well as school leaders (typology theory), and that these two groups may agree or disagree in how they are aligned or misaligned across the core aspects of leadership for learning (congruency theory). Thus, Bowers et al…”

“…(2017 [20]) postulated that there may be four types of leadership for learning schools as measured through CALL, a low-low type in which teachers and principals agree that leadership for learning practices do not occur often (congruency), a high-low type in which teachers have high perceptions of leadership for learning but principals have low perceptions (misalignment), a low-high type in which teachers are low but principals high in their perceptions (misalignment), and a high-high type in which teachers and principal agree that the practices happen often (congruency). Interestingly, (Bowers et al, 2017[20]) found a three-group typology of teachers, and a three-group typology of principals and their schools, with evidence for the overall congruency-typology of only low-low, low-high, and high-high.…”

“…Multilevel LCA model estimation is a three-phase process(Henry and Muthén, 2010[92]), and here I follow the recommendations from the previous multilevel leadership for learning LCA from Bowers et al (2017[20]) which also nested teachers within schools. In phase 1, a Level 1 only teacher within-school LCA model is estimated.…”