Skip to main content

Table 4 Summary of observations from the three-treatment case

From: Investigating the relationship between predictability and imbalance in minimisation: a simulation study

 

Predictabilityat P= 1.0, %

Na

Recommended P-value

Predictability at this P-value, %

Reduction in predictability, %

P-value at which imbalance occurs

One variable

Two categories

33.0

100

0.8b

32.0

1.0

0.7

  

200 to 300

0.7 b

33.0

0

0.5

  

≥400

0.7 b

33.0

0

-

Two categories - unequal prevalence

 

100

1.0

33.0

0

1.0

  

200

0.8 b

33.0

0

0.7

  

300

0.7 b

33.0

0

0.5

  

≥400

0.7 b

33.0

0

-

Three categories

 

100

1.0

32.0

0

0.9

  

200

0.7 b

32.5

<1.0

0.6

  

300

0.7 b

33.0

<1.0

0.5

  

≥400

0.7 b

33.0

<1.0

-

Four categories

 

100

1.0

32.0

0

0.9

  

200

0.9 b

32.5

<1.0

0.8

  

300

0.7 b

32.7

<1.0

0.6

  

≥400

0.7 b

32.5

<1.0

0.5

Two variables

Both with 2 categories

56.0

100

0.9

54.0

2.0

0.8

  

200

0.7

48.0

8.0

0.6

  

300

0.7

48.0

8.0

0.5

  

≥400

0.7

48.0

8.0

-

Both with 2 categories - unequal prevalence

 

100

1.0

56.0

0

1.0

  

200

0.9

54.0

2.0

0.8

  

300

0.7

49.0

7.0

0.6

  

≥400

0.7

48.0

8.0

0.5

Three variables

All with 2 categories - equal prevalence

67.0

100

1.0

67.0

0

0.9

  

200 to 300

0.7

55.0

12.0

0.6

  

≥400

0.7

56.0

11.0

-

All with 2 categories - unequal prevalence

 

100

1.0

67.0

0

0.9

  

200 to 300

0.7

55.0

12.0

0.6

  

≥400

0.7

55.0

12.0

0.5

Four variables

All with 2 categories - equal prevalence

74.0

100

1.0

74.0

0

1.0

  

200

0.8

64.0

10.0

0.7

  

300

0.7

57.0

17.0

0.6

  

≥400

0.7

58.0

16.0

0.5

  1. aThe categories of N are dependent upon the point at which imbalance is observed (the value of probability of assignment P). bFor one prognostic variable the reduction in predictability is so small as the probability of assignment P is reduced that the recommended P- value is 1.0.