diff --git a/dev/index.html b/dev/index.html index 65240c3..6e6c2df 100644 --- a/dev/index.html +++ b/dev/index.html @@ -183,4 +183,4 @@ # calculate p-value polrtest(house_null, cont; test=:score)
0.0001648743597587817

Step 3: Now suppose we want to test significance of another predictor, z1. We just need to call polrtest with z1 and the same fiited null model. No model fitting is needed.

For demonstration purpose, we generate z1 randomly. The score test p-value of z1 is, not suprisingly, large.

z1 = randn(nobs(house_null))
 polrtest(house_null, z1)
0.1673512522966108

Step 4: We can also test a set of precitors or a factor.

z3 = randn(nobs(house_null), 3)
-polrtest(house_null, z3)
6.709335149358069e-10
+polrtest(house_null, z3)
6.709335149358069e-10
diff --git a/dev/search/index.html b/dev/search/index.html index 17b09d4..dd334ed 100644 --- a/dev/search/index.html +++ b/dev/search/index.html @@ -1,2 +1,2 @@ -Search · OrdinalMultinomialModels.jl

Loading search...

    +Search · OrdinalMultinomialModels.jl

    Loading search...