On detecting outlier in simple linear functional relationship model using COVRATIO statistic

Outlier detection is a concern in data analysis,as anomalous observation can have consequence pertaining to the inference made from the analysis. numerous diagnostic tools have developed especially for regression model to overcome the presence of contamination in data. in this paper,we derived a test statistic using covratio to detect outliers in linear functional relationship model. the cut-off point for the statistic is determined using simulation result. we found that the power of performance for detecting the outlier increases as the standard deviation of error variance decreases and it is independent of the sample size. finally,this method is illustrated using simulated and real data set. © 2014 pakistan journal of statistics.