Cross-national variation in the effects of school and national contexts on school principals’ time allocation for interaction with students

Moo Sung LEE

Research output: Contribution to conferencePaper


Principals’ attention to major realms of responsibility has been reported in educational leadership literature as one key organizational resource for school improvement (Hallinger & Heck, 1996; Robinson et al., 2008). Previous studies have documented this issue by examining how principals use their time through surveys of principals or daily logs (e.g., Goldring at al., 2008; Halland, 2008; Kieley, 2000; Kmetz & Willower, 1982; Martin & Willower, 1981; Peterson, 1981). While insightful, previous studies have focused on particular schools or districts within one country mostly with small sample groups of principals. Research examining principals’ time use on a larger scale is rarely found. This study focuses on principals’ time use for interaction with individual students. This study aims to reveal in what contexts school principals tend to spend their time for interacting with students. To this end, this study seeks to identify key school- and national-level contexts that influence principals’ time allocation for interaction with students. For the current research, the Progress in International Reading Literacy Study (PIRLS) 2006 was used. The current research includes 34 societies from 28 countries. To reconstruct the dataset for this study, relevant data files (i.e. school questionnaires completed by principals) from 34 societies were merged. As such, the final dataset included 5,297 principals from 34 societies. Because the nature of the data represented a unit of analysis (i.e. 5,297) nested within a larger unit (34 societies), a two-level hierarchical linear model (HLM) was utilized (Raudenbush & Bryk, 2002). Prior to the HLM analysis, due to the original variables having missing cases ranging from 5.1% (e.g. home-school involvement) to 14.2% (e.g. proportion of immigrant students), a multiple imputation technique equipped with expectation maximization (EM) estimation was first conducted. To prevent imputed values from falling outside the reasonable range of values (e.g. negative values of principal time allocation), a custom imputation model with constraints on the dependent variables was also used. As such, five datasets that are simulated versions of the sample were created. Together with a weighting variable of school, these five complete datasets (each of which contains imputed values for the missing values) were analyzed using HLM. Findings indicate that more students with free/reduced-price lunch in the school, more immigrant students in the school, and greater lack of school resources were associated with a decrease in the time allocation for interaction with students. In addition, principals in rural schools tended to show slightly less time allocation for interaction with students than their counterparts in urban schools. Unexpectedly, larger school size indicated a positive association with an increase in principals’ time allocation. With regard to the two level-2 (country or regional level) predictors of interest, power distance index (PDI) and curriculum standardization had significant effects on principals’ time allocation for interaction with students in an opposite way. Higher PDIs were negatively associated with principals’ time allocation for interaction with students whereas higher levels of curriculum standardization were positively associated with the same activity. Implications for policy and research are discussed in-depth.
Original languageEnglish
Publication statusPublished - 2010


linear model
rural school
model analysis


Lee, M. (2010, January). Cross-national variation in the effects of school and national contexts on school principals’ time allocation for interaction with students. Paper presented at the Asia Leadership Roundtable 2010, The Hong Kong Institute of Education, China.