本研究主要目的在利用美國長程鋪面績效(LTPP)資料庫對接縫式混凝土鋪面高差現有預測模式進行評估，並嘗試改善該預測模式。首先，在對鋪面高差進行資料探索分析時發現，高差資料之隨機誤差未必符合一般傳統迴歸要求常態分配的假設。因此，在不假設應變數為任何誤差分佈情形下，本研究採用概似估計法與泊松分配，配合廣義線性模式(GLM)與廣義相加模式(GAM)的迴歸分析方式，並配合Box-Cox次冪轉換法、視覺圖的技術、與系統化之統計與工程分析流程來構建預測模式。本研究最後並藉由統計檢定與相關參數的敏感度分析，以進一步檢查模式的適合度。建議未來亦可更進一步的分析與改進。 The main objective of this study is to investigate the goodness of the fit and strive to develop improved faulting prediction models for jointed concrete pavements using the Long-Term Pavement Performance (LTPP) database. Exploratory data analysis (EDA) of the response variables indicated that the normality assumption with random errors and constant variance using conventional regression techniques might not be appropriate for prediction modeling. Therefore, without assuming the error distribution of the response variable, generalized linear model (GLM) and general additive model (GAM) along with quasi-likelihood estimation method and Poisson distribution were adopted in the subsequent analysis. Box-Cox power transformation and visual graphical techniques were frequently adopted during the prediction modeling process. By keeping only those parameters with significant effects and reasonable physical interpretations in the model, various tentative performance prediction models were developed. The goodness of the fit was further examined through various sensitivity analyses of pertinent explanatory parameters. The tentatively proposed models appeared to reasonably agree with the performance data, although further enhancements are possible and recommended.