The first part of this report will have 3 immediate goals based on the priorities of expediency, relevant psychometric analyses and testing regression and ‘prediction’ models for our external criteria, like ‘Quality of Self Care’ and health status (and cost reduction).
In the second part, based upon the multiple regressing modeling, will be a stage 4: ‘Prediction’ of Diabetes Self Management Outcomes (page 17 ff).
0.1 STAGE 1: THE TECHNICAL PSYCHOMETRIC CHARACTERISTICS
In stage 1, we will examine the individual items for their psychometric characteristics, distributional and discrimination characteristics.
Then we will examine the internal consistency reliability (Cronbach’s coefficient Alpha) of the individual scales. For each item in a given scale the internal consistency reliability coefficient alpha will be calculated for the scale in the absence of that item. In that manner, each item’s contribution to the overall unidimensional construct is evaluated.
Factor analysis will be used to examine the structure of the scales and to form factors based on the scales within a construct group and across construct groups. In future reports, item (scale) level factor analyses will be conducted across the entire questionnaire.
0.2 STAGE 2: EXTERNAL VALIDITY
In the second stage, the external validity of the scales will be examined by determining the relationship between the scales and demographic, clinical, and communication characteristics. In future reports the inter-correlations of the scales will be tabulated. However it should be noted, that the factors formed are indicative of the pattern of scale inter-correlations.
0.3 STAGE 3: MULTIPLE REGRESSION MODELING
In the third stage, we will employ multiple regression modeling to determine the relationship between the health outcomes (e.g., blood sugar testing) and the scales, factors, communication styles, demographic and clinical characteristics.
Because of the large sample size, these models will allow the simultaneous testing of all these factors to determine those most significantly and independently related to health outcomes.
Future reports will include refitting the models with other distributions and other refinements of the models. Item level models will also be tested in the future.
0.4 STAGE 4: PREDICTION OF DIABETES SELF MANAGEMENT
In this stage, the relationship between Diabetes Health Outcomes and demographic, diabetes clinical,
I-D-E-A scales and the 4 ICS Factors are described.