Model Building Notes.docx - Model Building We want estimates to be unbiased \u2013 to have the same statistics as parameters(the sample mean should equal

Model Building Notes.docx - Model Building We want...

This preview shows page 1 - 2 out of 2 pages.

Model BuildingWe want estimates to be unbiased– to have the same statistics as parameters (the sample mean should equal the population mean)oFor regression, this means that the estimated slope coefficient b1is an unbiased estimate of the true slope B1if the average of all the b1estimates equals B1. 4 Possible Outcomes when Building a Regression Model:A.Correctly Specified Model– regression equation contains all the relevant predictors, including any necessary transformations/interactions; yields unbiased statistics – best case scenarioB.Underspecified Model– model is missing one or more important predictor variables; worst case scenario, because it consistently yields biased statistics to over or underestimate the population parametersC.Regression Model Contains 1+ Extraneous Variables – variables that aren’t related to the response nor any of the other predictors; not terrible, because it yields unbiased regression
Background image
Image of page 2

You've reached the end of your free preview.

Want to read both pages?

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture

  • Left Quote Icon

    Student Picture