R中的序数Logistic回归解释

Question

我想知道如何最好地解释序数逻辑回归的结果。

我有一个下面的MASS软件包中的调查数据示例。

# loading MASS package
library(MASS)

# regression to determine if how much the student smokes is dependent on how often the student exercises
survey %>% 
  mutate(Exer = factor(Exer, levels = c("None", "Some", "Freq")),
         Smoke = factor(Smoke, levels = c("Never", "Occas", "Regul", "Heavy"))) %>% 
  polr(Smoke ~ Exer, data = ., Hess = TRUE) %>% 
  summary()

Call:
polr(formula = Smoke ~ Exer, data = ., Hess = TRUE)

Coefficients:
           Value Std. Error t value
ExerSome -0.4471     0.5765 -0.7756
ExerFreq  0.1867     0.5436  0.3434

Intercepts:
            Value   Std. Error t value
Never|Occas  1.3272  0.4997     2.6558
Occas|Regul  1.9468  0.5137     3.7899
Regul|Heavy  2.9630  0.5657     5.2378

Residual Deviance: 333.3227 
AIC: 343.3227 
(1 observation deleted due to missingness)

我应该如何解释系数（无，有些和频率）和截距（从不重）？

在决策过程中还有其他方法可以合并p值（重要性/非重要性）吗？

谢谢。