Stability Study

Coefficients

  

The coefficients table shows the coefficients (Coef) for each term in the final model and for the constant. The table also includes the standard error (SE Coef), the t-value, and the p-value for each coefficient.

If the VIF values are all close to 1, this indicates that the predictors are not correlated. VIF values greater than 5 suggest that the regression coefficients are poorly estimated.

Example Output

Coefficients

 

Term             Coef  SE Coef  T-Value  P-Value   VIF

Constant      100.085    0.143   701.82    0.000

Month        -0.13633  0.00769   -17.74    0.000  1.07

Batch

  1            -0.232    0.292    -0.80    0.432  3.85

  2             0.068    0.292     0.23    0.818  3.85

  3             0.394    0.275     1.43    0.162  3.41

  4            -0.317    0.292    -1.08    0.287  3.85

  5             0.088    0.275     0.32    0.752     *

Month*Batch

  1            0.0454   0.0164     2.76    0.010  4.52

  2           -0.0241   0.0164    -1.47    0.152  4.52

  3           -0.0267   0.0136    -1.96    0.060  3.65

  4            0.0014   0.0164     0.08    0.935  4.52

  5            0.0040   0.0136     0.30    0.769     *

Interpretation

The final model for the pill data includes three terms: Month, Batch, and the Month by Batch interaction (Month*Batch). Thus the relationship between the response (Drug%) and time for each batch is described by 4 coefficients.

The coefficients for Constant and Batch together give you the fitted value for each batch at time zero (also called the intercept). For example, the fitted value for Drug% for Batch 1 at time zero is equal to 100.085 - 0.232, or 99.853.

The coefficients for Month and the Month by Batch interaction together give you the slope for the linear relationship between Month and Drug%. For example, the slope for Batch 1 is equal to -0.13633 + 0.0454, or -0.09093. This value indicates that over a period of 1 month, Drug% decreases by 0.09093.

Because the final model includes the batch factor, the intercept and slope for each batch are also shown in the regression equation table.