Testing

Initial Exploratory Factor Analysis

Principal Component Analysis with Varimax Rotation yielded 3 main factors with Eigenvalues greater than 1.0

     Stability (System Maintanance and Coherence)      6 items

     Growth (System Change and Individuation)          6 items

     External Support                                               3 items

Internal Consistency - Reliability of the factors

                          .60 to .80

     Total Scale    .84

Details, see Friedemann, 1991

In all subsequent samples, external support did not factor as an independent dimension but was folded into the other factors. This was congruent with the theoretical framework. The factor structure of multiple test runs was indicative of all four dimensions. In each testing, a few items were added, some were taken off, leading to an ASF-E with 20 items. In all test runs, system maintenance and coherence factored together. The additional items were added to the Growth dimension for better reliability; reliability for Growth ranged from .50 to .62 in multiple test runs. 

Test-Retest Reliability (with 43 substance abusing families [sample 3] was satisfactory:

       Stability at 1 month     r = .93

                             2 months   r = .71

      Growth at   1 month    r = .80

                             2 months  r = .56  (Growth was expected to improve in these families)

     Total Family Health

                 at  1 month    r = .94

                      2 months  r = .84

Concurrent Validity (same sample)

ASF-E was tested against FES (Moos)

Values were unsatisfactory. The internal consistency of the FES with this sample was very low (.008 to .34).

The FES was not found suitable for sample with minority substance abuse families. 

Testing for Type III Error:

Testing the congruence between individual items and the actual phenomenon. 

    1) Convergence of intended meaning and meaning perceived by respondents

    2) Convergence of the sub-scores and total score and the qualitative description of family strategies in the                 same dimension. 

Results were very satisfactory. 

Details, see Friedemann and Smith, 1997.