Skip to main content

Table 4 Multiple regression analysis to explain the variability of the parameter "reported effect size". Shown are the top four models taking into account one or two variables with the effect size measured in absolute or in relative dimensions

From: The fading of reported effectiveness. A meta-analysis of randomised controlled trials

 

Ranking of model

Effect size measured in absolute terms

Effect size measured in relative terms

  

Variables in model

R2*

Variables in model

R2

Pravastatin

1

B, Y

0.8477

Y, B

0.2941

 

2

B, T

0.8077

Y, T

0.2916

 

3

B, n

0.8060

Y, n

0.2910

 

4

B

0.8037

Y

0.2873

Atorvastatin

1

B, T

0.7473

T, Y

0.2014

 

2

B, Y

0.6961

T, B

0.2001

 

3

B, n

0.6960

T, n

0.1957

 

4

B

0.6959

T

0.1882

Timolol

1

B, Y

0.5453

Y, B

0.2351

 

2

B, T

0.4948

Y, n

0.2206

 

3

B, n

0.4829

Y, T

0.1979

 

4

B

0.4829

Y

0.1958

Latanoprost

1

B, Y

0.7958

B, Y

0.5636

 

2

B, T

0.7798

B, T

0.5306

 

3

B, n

0.7649

B, n

0.5030

 

4

B

0.7647

B

0.5030

  1. Abbreviations: Y, year of publication; B, baseline of parameter of interest; T, treatment group; n, study size.
  2. * R2: Determination coefficient.