Validating Student-Parent Actions Model (SPA) of Consistency Using the Rasch Measurement Model

Abstract

The objective of this paper is to investigate the consistency of item selection and the existence of unidimensionality within each construct of the Student-Parent Actions Model (SPA) measurement scale using item response theory (Rasch measurement model). Based on Rasch model, which uses the WINSTEPS program, 373 students' responses were analysed. The sample group, which had a mean age of 16.41 years, was selected from five secondary schools through convenience sampling. The initial 62 items of SPA were analysed within the eight 'a priori' hypothesised dimensions. Rasch's misfit statistics were used to eliminate items that violated the infit and outfit MNSQ criteria of being between 0.70 and 1.30. In combination with the first principal component analysis of standardised residual (PCAR), Rasch's fit statistics were used to analyse unidimensionality within each individual construct. From the Rasch analysis results, 35 items were selected. In total, 27 items were similar to the SPA selected through structural equation modelling (SEM) analysis. All eight constructs exhibited unidimensionality, as PCAR analysis showed explained variances ranging from 47.10% to 73.6% for each individual construct. However, the eigenvalues of unexplained variances in the first contrast were all less than 3.0 indicating unidimensionality within each construct. Overall, SPA exhibited high consistency in the selection of items between the Rasch and SEM methods. When compared to the original SPA, the high percentage of similarity (approximately 80%) for the selected items indicated item selection stability for SPA across the two methods. Multidimensionality for the instrument was also detected.

Objektif kajian adalah untuk menyelidik konsistensi dalam pemilihan item dan kewujudan unidimensionaliti untuk konstruk-konstruk dalam Student-Parent Actions Model (SPA) menggunakan teori item respons (Model Pengukuran Rasch). Respons daripada 373 pelajar dianalisis berdasarkan model Rasch dengan menggunakan program WINSTEPS. Min umur sampel adalah 16.41 tahun dan dipilih menggunakan kaedah convenience sampling daripada lima buah sekolah menengah. Enam puluh dua item yang merangkumi lapan konstruk yang dihipotesiskan a priori dianalisa. Statistik Rasch digunakan untuk pemilihan item. Kriteria yang digunakan untuk memilih item ialah MNSQ 0.70–1.30. Statistik Rasch yang digabungkan dengan PCAR digunakan untuk menganalisis unidemensionaliti dalam setiap konstruk. Tiga puluh lima item dipilih berdasarkan kriteria Rasch sementara 27 item dipilih apabila menggunakan kaedah structural equation modelling (SEM). Kesemua konstruk menunjukkan unidimensionaliti apabila nilai PCAR berada dalam lingkungan 47.10% hingga 73.60%. Nilai eigen kurang daripada 3.0 turut memberikan indikasi bahawa konstruk adalah unidimensi. Secara keseluruhannya, SPA mempamerkan konsistensi yang tinggi walaupun menggunakan dua kaedah yang berbeza di antara Rasch dan SEM. Persamaan yang mencapai 80% di antara item terpilih memberikan implikasi bahawa pemilihan item adalah agak stabil walaupun menggunakan dua kaedah berbeza. Akhir sekali, multidimensionaliti instrumen turut dikesan.