Pengujian Model Prediktif Kinerja Pelatihan Diplomat: Analisis Jaringan dan Regresi Hierarkis Berdasarkan Teori Kinerja Campbell

Nina Vania Aristawati, Campbell’s performance theory, network analysis, predictive model, psychological potential, training performance

2026 | Tesis | S2 Psikologi

Penelitian ini menguji model prediktif kinerja Campbell (1990) pada pelatihan diplomat Sesdilu menggunakan data sekunder (N=140). Melalui Regresi Linear Berganda Hierarkis dan Analisis Jaringan, hasil menunjukkan tingkat pendidikan (H1), SKP (H2), dan empat klaster potensi psikologis (H3) memberikan kontribusi prediktif signifikan (seluruh hipotesis diterima). Validasi melalui regresi stepwise dan penyetaraan Z-score mengonfirmasi bahwa pendidikan dan kecerdasan emosional adalah prediktor paling stabil, sementara daya prediksi SKP gugur pada model akhir akibat Leniency Bias dan Restriction of Range birokrasi. Terkait potensi, dominasi Kecerdasan Emosional selaras dengan hasil jaringan yang terfragmentasi. Fragmentasi mengindikasikan ada bias metode antar-vendor. Hal ini terbukti dari gugurnya signifikansi potensi saat efek vendor dikontrol serta rendahnya stabilitas jaringan (CS=0). Namun, prosedur equating melalui Z-score terbukti mampu meningkatkan stabilitas jaringan menjadi 0.049 dan menghasilkan pemetaan talenta yang lebih kohesif. Disarankan penggunaan data pelatihan sebagai standar objektif, penerapan penyetaraan skor secara statistik, dan pengembangan alat ukur internal untuk menjamin objektivitas sistem manajemen talenta.

This study examined Campbell’s (1990) performance prediction model within the context of Sesdilu diplomatic training using secondary data (N=140). Utilizing Hierarchical Multiple Linear Regression and Network Analysis, the results demonstrate that educational level (H1), prior performance/SKP (H2), and four psychological potential clusters (H3) offer significant predictive contributions, leading to the acceptance of all hypotheses. Validation through stepwise regression and Z-score standardization confirmed that education and emotional intelligence are the most robust predictors. Conversely, the predictive power of SKP became non-significant in the final model, attributed to bureaucratic leniency bias and restriction of range. Regarding psychological potential, the dominance of emotional intelligence is consistent with the fragmented network structure, which indicates cross-vendor methodological bias (idiosyncratic rater effect). This was evidenced by the attenuation of potential's significance when vendor effects were controlled and the lack of network stability (CS=0). However, the Z-score equating procedure proved essential in increasing network stability to 0.049 and generating more cohesive talent mapping. This study recommends utilizing training data as an objective benchmark, implementing statistical score equating, and developing internal measurement tools to ensure the objectivity of the talent management system.

Kata Kunci : kinerja pelatihan, model prediktif, network analysis, potensi psikologis, teori kinerja Campbell