Table 13. Multinomial regression showing the best fit model for predictors of practice level.

Area	Practice level	Model
Flooded area (R2 value = 0.3263)	Practice (High/Poor)	y = -231.871 int + 15.080 AGE—28.240 MA 1 + 219.474 MA3–36.475 MA4 + 36.416 DH2 + 33.341 K1–125.804 K2–59.361 A1–5.275 A2
Unflooded area (R2 value = 0.3577)	Practice (High/Poor)	y = -263.559 int + 19.275 AGE + 19.061 DH2

Multinomial regression JMP (Poor practice as reference category), int: Intercept; MA: Mosquito abundance; DH: Dengue History; K: Knowledge; A: Attitude; P-value is based on 2 log-likelihood regression using backward stepwise method, significant values are in bold, P<0.05.