The study reports results for doctors/dentists/pharmacists with n=34. Given the wide confidence intervals (e.g., WTP: 3,477–15,442 Baht) and the use of regression models with multiple attributes, how did you ensure statistical validity and avoid overfitting? What sensitivity or power analysis was conducted to justify this sample size for policy-reliable DCE estimates?
In Table 3, for nurses/interdisciplinary teams, income increases of 20% and 40% show p-values of 1.000, which is statistically implausible. Is this a reporting error? If not, how do you explain these p-values, and what does this imply about the robustness of income as a key motivator in your analysis?
The WTP values show nurses require ~17,000 Baht to work outside their hometown—almost double the doctors’ requirement (~9,100 Baht)—without clear theoretical or empirical justification. What specific formula and assumptions were used to derive WTP, and how do you explain such a large discrepancy between professional groups?
You cite Herzberg’s theory but use it to emphasize financial incentives, even though Herzberg classifies salary as a hygiene factor (prevents dissatisfaction) not a motivator. How do you reconcile this theoretical contradiction, and why did you not engage more deeply with Public Service Motivation (PSM) in your DCE attribute selection?
The study was conducted in only four remote districts of Northern Thailand. How can these findings be generalized to other remote regions in Thailand with different cultural, economic, and health-system contexts? Furthermore, what fiscal or implementation analysis supports the feasibility of your recommendations (e.g., large subsidies, private housing) at the national level?
If nurses indeed require a 40% salary premium to work outside their hometown while doctors only need 15%, this points to a fundamental difference in mobility constraints or attachment motivations between the professions. Yet your discussion treats both groups’ “hometown preference” as similarly important without explaining this large discrepancy. This weakens targeted policy design.
In Table 3, for doctors/dentists/pharmacists, the odds ratio for “Management pays attention to complaints” is 0.922 and statistically insignificant (p=0.571). But in your qualitative findings, you noted that handling complaints and good administrative relations were priorities for this group.
Why is there such a strong qualitative-quantitative mismatch? Could it be that your DCE design failed to capture the nuance of “management support,” or did the small sample size obscure a real effect that qualitative methods picked up?