青少年偏差行為發生與否及發生頻率之模型建構

Abstract

由於青少年偏差行為具有計數之資料型態，且常出現大量零之非常態分布，因此本研究提出零膨脹負二項迴歸模型，試圖突破過去青少年偏差行為研究受到資料異質性以及過度離散的限制，同時考量青少年依附關係、生活緊張及接觸偏差同儕分別對偏差行為發生與否，以及發生頻率兩大面向的影響，最後輔以根方圖作為模型適配度檢視的方法。本研究有效樣本為臺南市國中生947人。研究結果發現：一、生活緊張與偏差同儕是誘發青少年偏差行為的影響因素；二、接觸偏差同儕催化青少年偏差行為發生的頻率，家庭依附關係則降低偏差行為之發生頻率；三、對於出現大量零以及過度離散的計數資料，零膨脹負二項迴歸分析方法所建構之模型相對適配。研究最後並提出有關青少年學校輔導及後續研究之相關建議。

The purpose of this study was to explore the impact of attachment, life strains and association with deviant peers on frequency of juvenile delinquency by using the zero-inflated negative binomial (ZINB) regression model with R. The graphical tool “rootogram” was utilized for the goodness-of-fit visualization of the ZINB model. Research participants were 947 students in Tainan, Taiwan. The results were as follows: (1) life strains and association with deviant peers were significantly associated with the occurrence of delinquency; (2) attachment and association with deviant peers were significant predictors for the magnitude of delinquency; (3) ZINB outperformed negative binomial model for count data analysis with excessive number of responses of zero, and with over-dispersed distribution.