Table 3 Results of the binary-logistic regression analysis
From: Counsellors contact dementia caregivers - predictors of utilisation in a longitudinal study
Regression coefficient β | Standard error | Wald coefficient | df | Sig. | |
|---|---|---|---|---|---|
Factor 1: Objective and subjective burden of care | .731 | .258 | 8.004 | 1 | .005 |
Factor 2: Spouses | 1.004 | .293 | 11.765 | 1 | .001 |
Factor 3: Comorbidities | .037 | .246 | .023 | 1 | .879 |
Factor 4: Attendant symptoms | -.236 | .257 | .846 | 1 | .358 |
Factor 5: Gender ratio | -.244 | .251 | .948 | 1 | .330 |
- How to predict utilization of counselling utilization for a new case:
- The probability for using counselling can be computed by
-
- e = Euler's constant
- while z = Factor1 × βFactor1 + Factor2 × βFactor2 + Factor3 × βFactor3 + Factor4 × βFactor4 + Factor5 × βFactor5 + constant (-.761)
- If p results in a value smaller than .5, it is assumed that the patient will not use counselling. If p is higher than .5 it is assumed that counselling will be used.
- The values for the different factors can be computed by using Table 2. The individual value for example in the burden scale for family carers has to be multiplied with .838 for Factor 1. But this is only part of Factor 1. In order to compute Factor 1 completely all individual values for the variables named in Table 2 have to be multiplied with the coefficient corresponding to the respective factor and than subsumed. In this way the individual Factors 2, 3, 4 and 5 can also be computed.
