a) Fisher test was completed to provide F value of 344.9. In this case, the Fisher test shows that there is the maintenance score difference between expert, inconsistent and novice cluster. The value of 344.9 proves that there is significant difference between the clusters. For the maintenance score, maintenance inconsistency, management score and management inconsistency F values prove that clusters differentiate.
b) Chi-square test was completed to provide χ2 results for adequate maintenance and adequate management results.
c) χ2 analysis allows proving that adequate maintenance and adequate management differ significantly between the clusters. In particular, χ2 = 269.7 (high values of χ2) state that there is significant difference between expert, inconsistent and novice clusters.
d) Yes, there are. All the results in Table 2 are significant at the p < 0.05 level. The column p of the Table 2 states that the results are significant at p < 0.0001. This means that the F and χ2values are significant at p < 0.0001. If we needed p = 0.00001, the results would be insignificant.
Table 3 questions
a) p = 0.005 shows that the odds ratio of 0.948 is significant at level 0.005. Duke Activity Status Index is significant at level 0.005 or confidence level of 99.995%. For this research, statistical significance was predetermined at p < 0.05, which means that the Duke Activity Status Index value of 0.948 is significant for the multinomial model predicting the self-care cluster membership.
b) Duke Activity Status Index parameter estimate is 0.998. Confidence interval of 0.971-1.026 is the interval estimate of a parameter; it is used to indicate the reliability of the parameter estimate. The 95% confidence interval means that 95% of the parameter values are within the interval 0.971 - 1.026, and only 5% lay beyond it.
c) Yes, there are. The value of the Duke Activity Status Index parameter for inconsistent over novice cluster membership is not significant at the p < 0.05 level. According to the table, it is significant at p = 0.895. This means, that it should not be used for the self-care cluster membership prediction for inconsistent over novice; only SCHFI confidence parameter should be used.
d) The odds ratio presents the association of the parameter with a model. The Duke Activity Status Index or/and SCHFI confidence odds ratios describe the association with multinomial model that predicts self-care cluster membership.
e) The odds ratios presented in Table 3 are used to predict the cluster membership. The higher values of odds ratios show higher contribution of the parameter to the result. Thus, for expert over novice section, the Duke Activity Status Index odds ratio contribution is 0.948, while SCHFI confidence odds ratio is 1.069. This results in higher contribution of self-care heart failure practices to expert cluster membership than the activity status. On the contrary, the models for novice over expert and inconsistent over expert supply higher values of Duke Activity Status Index ratios comparing to SCHFI confidence. Practically, this means that the patients with higher activity will be more probably graded as inconsistent or novice and those with high confidence in self-care will be graded as experts. To rate a patient as inconsistent over novice, only odds ratio of SCHFI confidence is used, activity does not contribute significantly.