ABSTRAKTujuan penelitian ini adalah untuk mengetahui indeks validitas prediktif dan sumbangan efektif predictor (Tes Pengetahuan Agama) dalam memprediksi kriterium (prestasi mahasiswa) yang direpresentasikan dari IP semester 1. Namun sebelumnya, penulis melakukan analisis kuantitatif terhadap instrumen TPA untuk melihat seberapa baik karakteristik item yang digunakan yang notabene dirakit dan disusun oleh panitia penerimaan mahasiswa STAIN dibantu oleh para dosen yang latar belakang kelimuannya sesuai dengan instrumen yang akan dibuat. Dalam melakukan analisis validitas prediktif penulis menggunakan regresi sederhana dengan software SPSS V.13 sementara untuk analisis instrumen penulis menggunakan teori tes klasik (Classical Test Theory) yaitu ITEMAN dan teori tes modern (Item Response Theory) yaitu BIGSTEPS.Hasil analisis instrumen terhadap 100 item soal dan 693 responden menunjukkan bahwa 5 item yang ditolak dan 26 item (termasuk 5 item yang ditolak), 94 responden yang tidak fit dengan model. Hasil analisis validitas prediktif terhadap 336 responden yang lulus sebagai mahasiswa STAIN, dari 12 hipotesa, 11 Ho ditolak dengan indeks validitas prediktif antara 0,450 -0,933 dan sumbangan efektif antara 0,308-0,871 dan 1 Ho diterima dengan indeks validitas prediktif 0,319 dan sumbangan efektif 0,102.

---

ABSTRACTThe aim of this research is to find out the index of predictive validity and the effective contribution of predictors (Religious Subject Test) in predicting the criterion (students' achievement) which is represented by the students' first semester Achievement Index (IP). Prior to the research, the researcher carried out the quantitative analysis toward the Academic Potentiality Test to find out how good were the item characteristics used which were developed by the Students Entrance Test Committee of STAIN (State Islamic Religion School) who were supported by the lecturers who have the background knowledge on the instrument made. In doing the analysis of predictive validity, the researcher used simple regression with the temporal SPSS V.13 software and to analyze the instrument the researcher used Classical Test Theory which is called as ITEMAN and the modern test Theory (Item Response Theory) which is called as BIGSTEPS.The result of instrument analysis toward 100 items of the test and 693 respondents shows that 5 items were rejected and 26 items (including 5 rejected items), 94 respondents were not suitable with the model. The result of the predictive validity toward 336 students who passed the test and accepted as the STAIN students, from 12 hypotheses, 11 Ho were rejected with the predictive validity index between 0.450-0.933 and the effective contributions were between 0.308-0.871 and 1 Ho accepted with the predictive validity index 0.319 and the effective e contribution is 0.102.