Dffits

Fit Analyses

Dffits

The Dffits statistic is a scaled measure of the change in the predicted value for the ith observation. For linear models,

$F_{i} = \frac{\hat{ \mu_{i}}- \hat{ \mu}_{(i)}}{s_{(i)} \sqrt{ h_{i}} }$

where ${ \hat{ \mu}_{(i)}}$ is the ith value predicted without using the ith observation.

Large absolute values of F_i indicate influential observations. A general cutoff to consider is 2; a recommended size-adjusted cutoff is ${2\sqrt{p / n}}$ .

For generalized linear models,

$F_{i} = \frac{\hat{ \mu_{i}} - \hat{ \mu}_{(i)}}{\sqrt{ \hat{ \phi}_{(i)} h_{i}} }$

The Dffits statistics are stored in variables named F_yname for each response variable, where yname is the response variable name.

Top of Page