在迴歸分析中，相較於傳統線性迴歸（LinearRegression），分位數迴歸（Quantile Regression）可用於估計應變數在不同條件分位數下的變化趨勢。為進一步探討自變數與應變數之間可能存在的分段關係，本研究使用分段分位數迴歸（Piecewise Quantile Regression），並旨在估算分段分位數迴歸分析下所需之樣本數。由於目標是希望估計出之分位數迴歸線足夠準確，因此本研究以既有分位數迴歸下基於精確度的樣本數估算方法為基礎，將其延伸至單一自變數且具有單一分段點之分段分位數迴歸模型，進一步建構樣本數公式。為評估所提出方法之表現，本研究透過統計模擬檢視在不同參數設定下，依估算樣本數所得之涵蓋率是否能達到預設信心機率。此外，亦進一步探討迴歸模型假設錯誤與自變數分佈假設錯誤時，樣本數估算結果之穩健性。最後，本研究將所提出之方法應用於一筆人工模擬生成的運動員資料，探討肌肉量指標與肌耐力之間的關係，並於既有樣本數下，評估使用分段分位數迴歸之精確度。

This study develops a sample size estimation method for piecewise quantile regression. The method is based on a precision-based sample size approach and is extended to a model with one explanatory variable and one change point. Statistical simulations are used to evaluate whether the estimated sample size can achieve the pre-specified confidence probability. The results show that the proposed method performs well when the model assumptions are correct. However, when the regression model or the distribution of the independent variable is misspecified, the estimated sample size may not be sufficient. The method is also applied to athlete data to study the relationship between a muscle mass indicator and muscular endurance. Under the existing sample size, the precision of using piecewise quantile regression is also evaluated.

[1] C. Gutenbrunner and J. Jurečková. Regression rank scores and regression quantiles. The Annals of Statistics, 20(1):305–330, 1992.

[2] Christine Jennen-Steinmetz. Sample size determination for studies designed to estimate covariate-dependent reference quantile curves. Statistics in Medicine, 33(8):1336–1348, 2014.

[3] Roger Koenker and Gilbert Bassett. Regression quantiles. Econometrica, 46(1):33–50, 1978.

[4] Daniël Lakens. Sample size justification. Collabra: Psychology, 8(1), 2022.

[5] Chenxi Li, Ying Wei, Rick Chappell, and Xuming He. Bent line quantile regression with application to an allometric study of land mammals’ speed and mass. Biometrics, 67(1):242–249, 2011.

[6] Mingxiang Li. Moving beyond the linear regression model: Advantages of the quantile regression model. Journal of Management, 41(1):71–98, 2015.

[7] Stefan Wellek, Karl J. Lackner, Christine Jennen-Steinmetz, Iris Reinhard, Isabell Hoffmann, and Maria Blettner. Determination of reference limits: statistical concepts and tools for sample size calculation. Clinical Chemistry and Laboratory Medicine (CCLM), 52(12):1685–1694, 2014.

[8] Wei Zhong, Chuang Wan, and Wenyang Zhang. Estimation and inference for multi-kink quantile regression. Journal of Business & Economic Statistics, 40(3):1123–1139, 2022.

[9] Kenneth Q. Zhou and Stephen L. Portnoy. Statistical inference on heteroscedastic models based on regression quantiles. Journal of Nonparametric Statistics, 9(3):239–260, 1998.