neighbayes.diagnostics.lmtests.bayesian_glm_lm_wx_test

neighbayes.diagnostics.lmtests.bayesian_glm_lm_wx_test(model)[source]

LM test for omitted \(WX\) (H₀: θ = 0).

At the GLM null the score is \((WX)^\top e\); at a reduced-form SAR null, \(WX\) enters inside the multiplier, \(\eta = S(\rho)(X\beta + WX\theta)\), and the score is \((S(\rho)WX)^\top e\). Either is the Neyman-orthogonal score against \(\beta\), its information partialled likewise (see the module, also on \(\rho\)), with the Fisher weights at the posterior mean.

Returns:

Per-draw LM samples, summary statistics, df = k_wx.

Return type:

BayesianLMTestResult