Chapter 9 172 contrast has been demonstrated between participants motivated by boredom and science; participants motivated by boredom are more likely to drop out than participants motivated by science [16]. Such a sample bias could affect the overall results, and thus, the study’s validity. This study mainly involved people who were motivated by science. It is possible that if people with more different motivations had participated, the results would have shown more variation. More variation in the results will not change the conclusion as far as the RAdMAT-NL and the prediction model are concerned, but it is possible that adherence may still be somewhat less constant over time than has been shown. The percentage of missing values across all 83 variables throughout the main study varied between 13.7% and 22.9%. In total 151-169 out of 196 patients had a complete data set. The main reason for missing data was that patients indicated that they did not have to perform homework exercises. Patients who indicated this, skipped the questions related to the Theory of Planned Behavior. A number of patients did require exercises from the physiotherapist (this was indicated by the physiotherapist at baseline), but they did not understand that the exercises were actually ‘homework’ exercises, and indeed a number of patients did not have to perform home-based exercises. Following the in-dept considerations of the patterns of missing data, the data were assumed to be missing at random, and so missing data were handled using multiple imputation [17]. By doing so, bias was minimized (patients with complete datasets may differ from patients with missing data), and power was kept as optimal as possible (exclusion of patients may cause loss of precision and power), so that valid results were presented and appropriate conclusions could be drawn [18]. It has been taken into account that problems can occur in multiple imputation analyses; omitting the outcome variable from the imputation procedure; dealing with non-normally distributed variables; plausibility of missing at random assumption [19]. So, the RAdMAT-NL score was included, the distribution of the variables was checked, and the variables predictive of the missing values were included in the imputation model [19]. Since participation in the study required time from the physiotherapists, they were offered the option of a financial compensation of 20 euros per participating patient. The physiotherapists would receive this compensation after completing the study, so the financial compensation would not affect the recruitment of physiotherapists. Also, the financial compensation was too low (as indicated by the physiotherapists) to influence physiotherapist retention and therefore did not serve as an incentive to prevent a physiotherapist from dropping out of the study. However, at the end of the study, the majority of the physiotherapists did not want to use the financial compensation. About half of the physiotherapists participated in the study at the invitation of their patient. The fact that the patient felt participation in the study was important, made it important for the physiotherapist to participate as well. This observation may be valuable for future research. To whom do the results of the research matter, to whom are they of real use? The answers to these questions
RkJQdWJsaXNoZXIy MjY0ODMw