A multi-row deletion diagnostic for influential observations in small-sample regressions

详细信息    查看全文

• 作者：Daniel T. Kaffine^a ; ^{daniel.kaffine@colorado.edu} ; Graham A. Davis^b ; ^{gdavis@mines.edu}
• 关键词：Regression diagnostics ; Multi-row deletion ; Inference ; Influence points ; Outliers ; Growth regressions ; Institutions ; Resource curse
• 刊名：Computational Statistics & Data Analysis
• 出版年：2017
• 出版时间：April 2017
• 年：2017
• 卷：108
• 期：Complete
• 页码：133-145
• 全文大小：705 K
• 卷排序：108

文摘

The economic inferences from growth regressions are often sensitive to influence points in the sample. We introduce a multi-row deletion method for identifying influence points. The method is simple to use and is intuitively appealing. We illustrate our method using simulated and real data. Our method complements DFBETAS and robust regression.