diff --git a/dev/.documenter-siteinfo.json b/dev/.documenter-siteinfo.json index 2998d51..031bf6d 100644 --- a/dev/.documenter-siteinfo.json +++ b/dev/.documenter-siteinfo.json @@ -1 +1 @@ -{"documenter":{"julia_version":"1.6.7","generation_timestamp":"2024-11-21T02:32:20","documenter_version":"1.8.0"}} \ No newline at end of file +{"documenter":{"julia_version":"1.6.7","generation_timestamp":"2024-11-22T02:32:56","documenter_version":"1.8.0"}} \ No newline at end of file diff --git a/dev/index.html b/dev/index.html index 055fa91..83f27ac 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