Estimating Perturbation and Long-Term Memory Processes in Daily Attendance in Six Small High Schools

• In Event: 25.068 - Roundtable Session 8
  In Roundtable Session: 25.068-11 - Educational Statistical Research: Accuracy in Analytics and in the Context of Quasi-Experimental Designs

Abstract

This presentation discusses the daily attendance rates in six small high schools over a ten-year period and evaluates how stable they are. ‘Stability’ is approached from two vantage points: pulse models are fitted to estimate the impact of sudden perturbations and their reverberation through the series; Autoregressive Fractionally Integrated Moving Average (ARFIMA) techniques are used to detect dependencies over the long range of the series. The analyses are meant to exemplify the utility of the time series approach in educational research. This approach questions the assumption of stability that underlies the reporting of weekly, monthly and yearly averages, and it reveals dynamical processes (perturbation, meta-stability) that would otherwise remain hidden.

Author

• Matthijs Koopmans, Mercy College