Skip to main content

Table 3 Summary of Methods and Results: Type I Error and Power

From: Tests of trend between disease outcomes and ordinal covariates discretized from underlying continuous variables: simulation studies and applications to NHANES 2007–2008

Method

1

2

3

4

5

6

7

8

9

10

Weighted?

No

No

No

Yes

Yes

Yes

Yes

Yes

Yes

Yes

Weights

N/A

N/A

N/A

(θj2σ2j)−1

(θjσj)−1

(σ2j)− 1

(σj)− 1

n j

n j

n j

Intercept?

No

Yes

Yes

Yes

Yes

Yes

Yes

No

Yes

Yes

Reference Group included?

Yes

Yes

No

Yes

Yes

Yes

Yes

Yes

No

Yes

Type I Error Controlled?

No

Yes

Yes

Yes

Yes

Yes

Yes

No

Yes

Yes

Power: Linear

N/A

M

L

VL

VL

H

M

N/A

H

M

Power: Weak Quadratic

N/A

H

M

L

L

M

M

N/A

M

H

Power: Strong Quadratic

N/A

M

L

VL

VL

H

M

N/A

H

H

  1. VL Very Low Power, L Low Power, M Moderate Power, H High Power
  2. Relative amounts of power refer to the relative ranks of the methods. Methods classified as with high power had the most power for at least one scenario in the corresponding type of simulation. Methods classified as with medium or low power were ranked next. Methods classified with very low power displayed the lowest power in every simulation of that type