Influential observation


In statistics, an influential observation is an observation for a statistical calculation whose deletion from the dataset would noticeably change the result of the calculation. In particular, in regression analysis an influential point is one whose deletion has a large effect on the parameter estimates.

Assessment

Various methods have been proposed for measuring influence. Assume an estimated regression, where is an n×1 column vector for the response variable, is the n×k design matrix of explanatory variables, is the n×1 residual vector, and is a k×1 vector of estimates of some population parameter. Also define, the projection matrix of. Then we have the following measures of influence:
  1. , where denotes the coefficients estimated with the i-th row of deleted, denotes the i-th row of. Thus DFBETA measures the difference in each parameter estimate with and without the influential point. There is a DFBETA for each point and each observation. Table shows DFBETAs for the third dataset from Anscombe's quartet :
xyinterceptslope
10.07.46-0.005-0.044
8.06.77-0.0370.019
13.012.74-357.910525.268
9.07.11-0.0330
11.07.810.049-0.117
14.08.840.490-0.667
6.06.080.027-0.021
4.05.390.241-0.209
12.08.150.137-0.231
7.06.42-0.0200.013
5.05.730.105-0.087

Outliers, leverage and influence

An outlier may be defined as a data point that differs significantly from other observations.
A high-leverage point are observations made at extreme values of the independent variables.
Both types of atypical observations will force the regression line to be close to the point.
In Anscombe's quartet, the bottom right image has a point with high leverage and the bottom left image has an outlying point.