Skip to main content

Table 3 Percentage of detection using MELS models and MRMs and different influence measures. For scenarios with the first subject as the influential subject, “large(%)" represents the percentage of simulations in which this subject has the largest value of the respective influence measure among all subjects, and “small(%)" represents the percentage of simulations in which this subject has the smallest value of the respective influence measure among all subjects. For scenarios with the first three subjects as the influential subjects, “large(%)" represents the percentage of simulations in which all three of these subjects have the largest values of the respective influence measure among all subjects, and “small(%)" represents the percentage of simulations in which all these three subjects have the smallest values of the respective influence measure among all subjects. Results in the column named “MRMs" were obtained through analyses of using MRMs with no scale components. The simulated datasets used in the MELS model and MRM analyses are the same in each scenario

From: Detecting influential subjects in intensive longitudinal data using mixed-effects location scale models

 

MELS models

MRMs

 

Single

Multiple (3)

Single

Scenario 1

   

     \(C^{\tau }\) [large(%)]

75.9

16.7

-

     \(\text {COVRATIO}^{\tau }\)[large(%)/small(%)]

0/100

0/99.6

-

     \(\text {DFBETAS}^{\tau _0}\) [large(%)/small(%)]

99.4/0

25.4/0.4

-

     \(\text {DFBETAS}^{\tau _1}\) [large(%)/small(%)]

5.4/34.6

1.4/15.7

-

     \(\text {DFBETAS}^{\sigma ^2_{\omega }}\) [large(%)/small(%)]

100/0

99.6/0

-

     \(\text {C}^{\beta }\) [large(%)]

3.5

5.8

0

     \(\text {COVRATIO}^{\beta }\)[large(%)/small(%)]

15.1/0.8

2.0/0.8

41.6/0

     \(\text {DFBETAS}^{\beta _0}\) [large(%)/small(%)]

0.8/1.7

0.6/0.6

0/0

     \(\text {DFBETAS}^{\beta _1}\) [large(%)/small(%)]

21.0/44.4

3.6/22.7

6.8/13.9

     \(\text {DFBETAS}^{\sigma ^2_{\nu }}\) [large(%)/small(%)]

0.6/0

1.0/0

0/20.1

Scenario 2

   

     \(C^{\tau }\) [large(%)]

99.8

62.3

-

     \(\text {COVRATIO}^{\tau }\)[large(%)/small(%)]

0/100

0/99.8

-

     \(\text {DFBETAS}^{\tau _0}\) [large(%)/small(%)]

99.8/0

69.9/0

-

     \(\text {DFBETAS}^{\tau _1}\) [large(%)/small(%)]

11.3/16.0

0.4/0.8

-

     \(\text {DFBETAS}^{\sigma ^2_{\omega }}\) [large(%)/small(%)]

100/0

100/0

-

     \(\text {C}^{\beta }\) [large(%)]

21.9

0

1.2

     \(\text {COVRATIO}^{\beta }\) [large(%)/small(%)]

0/2.6

0/0

2.0/1.2

     \(\text {DFBETAS}^{\beta _0}\) [large(%)/small(%)]

1.4/2.0

0/0

1.4/1.6

     \(\text {DFBETAS}^{\beta _1}\) [large(%)/small(%)]

38.9/44.9

0/0.8

1.4/0.6

     \(\text {DFBETAS}^{\sigma ^2_{\nu }}\) [large(%)/small(%)]

1.4/6.3

0/0

1.2/2.0