|
Example 3
Quality of Life Data: Missing Categorical Covariates Method
(Model: Multiple Linear Regression)
The data from a ‘Quality of Life’ study were analyzed using Missing Categorical Covariates method introduced in LogXact 7. The results are as shown below.
A comparison of analyses with ‘complete cases’ and with all cases
(including cases having missing covariates):
|
Using Complete cases |
Using all cases
(by missing covariates method) |
Parameter |
Beta |
P-value |
Beta |
P-Value |
%Const |
5.95 |
0 |
6.3813 |
0 |
age |
0.009339 |
0.2227 |
0.0035 |
0.6561 |
trt_e |
-0.04467 |
0.7378 |
-0.2823 |
0.0311 |
trt_f |
-0.03466 |
0.7960 |
-0.2882 |
0.0309 |
trt_g |
-0.09303 |
0.4979 |
-0.1409 |
0.3116 |
lang |
-0.1327 |
0.2224 |
-0.2516 |
0.0221 |
PHY1 |
0.2063 |
0.2888 |
0.2531 |
0.2438 |
PHY2 |
-0.1861 |
0.3422 |
-0.2645 |
0.2444 |
Tau |
|
|
1.1447 |
0 |
Notice that p-values for the covariates in the analysis with ‘complete cases’ are all large values > 0.2, wheras in the analysis using all cases, there are three covariates with significant p-values (< 0.05).
|
