MSN Analysis Results

The most important section of the MSN report lies in the accuracy assessment and validation statistics which are listed below. To view the entire report, please visit the downloads section.

The following report contains accuracy assessment and validation statistics from the MSN imputation.

Number of reference observations                  =   2775

 Number of target observations                     =   2939

 **Number of ties broken with random numbers       =      0

   (**Random number seed =           55329.)

  3 variables selected for validation:

  model.9       model.8       model.10   

 Categorical variable group: model.     

                         Imputed

 Observed

              model.9       model.8       model.10       Total           %CC     

 model.9         291           271           116           678

      Row%     42.92         39.97         17.11         24.43

  Expected    170.05        440.27         67.68

    Chi-Sq     86.03         65.08         34.50

 model.8         285          1421            97          1803

      Row%     15.81         78.81          5.38         64.97

  Expected    452.21       1170.81        179.98

    Chi-Sq     61.83         53.46         38.25

 model.10        120           110            64           294

      Row%     40.82         37.41         21.77         10.59

  Expected     73.74        190.91         29.35

    Chi-Sq     29.02         34.29         40.92

 Total           696          1802           277          2775          1776

 Row%          25.08         64.94          9.98        100.00         64.00

 Total Chi-Sq:    443.4  DF:    4  Prob>Chi-Sq:  <0.0001

Kappa statistics: Khat:0.28886  Std Err:0.01537

 Interpretation of Khat: <=0 Poor, 0-.2 Slight, .2-.4 Fair, .4-.6 Moderate, .6-.8 Substantial, .8-1 Almost perfect.

Categorical variable group: model.

Correlations between Y and X-variables

-------------------------------------------------------------------------------

Randomized Validation Statistics

Randomizes the observed X-variables with respect to the observed Y-variables among the reference observations. It leaves intact the relationships within the Y’s and within the X’s while destroying the relationship between them. Note the low Chi-Sq test results in this randomized assessment vs. the actual assessment above.

Number of reference observations                  =   2775

 Number of target observations                     =   2939

 **Number of ties broken with random numbers       =      0

   (**Random number seed =           55329.)

  3 variables selected for validation:

  model.9       model.8       model.10   

 Categorical variable group: model.     

                         Imputed

 Observed

              model.9       model.8       model.10       Total           %CC     

 model.9         140           447            91           678

      Row%     20.65         65.93         13.42         24.43

  Expected    170.05        432.21         75.74

    Chi-Sq      5.31          0.51          3.07

 model.8         477          1142           184          1803

      Row%     26.46         63.34         10.21         64.97

  Expected    452.21       1149.37        201.42

    Chi-Sq      1.36          0.05          1.51

 model.10         79           180            35           294

      Row%     26.87         61.22         11.90         10.59

  Expected     73.74        187.42         32.84

    Chi-Sq      0.38          0.29          0.14

 Total           696          1769           310          2775          1317

 Row%          25.08         63.75         11.17        100.00         47.46

 Total Chi-Sq:     12.6  DF:    4  Prob>Chi-Sq: 0.0133

Kappa statistics: Khat:-.02479  Std Err:0.01437

 Interpretation of Khat: <=0 Poor, 0-.2 Slight, .2-.4 Fair, .4-.6 Moderate, .6-.8 Substantial, .8-1 Almost perfect.

Categorical variable group: model.    

Correlations between Y and X-variables

-------------------------------------------------------------------------------