Saturday, September 10, 2011

What is specification error in research?

With some degree of regularity I find myself having to explain how model "specification error" can distort research findings, especially studies that use predictive (multiple regression, causal modeling). So, below is a definition with an analogy from a 2005 publication. This is one of the most series problem with many research studies that claim to have identified THE important predictors of some dependent variable (e.g, reading disability research)

Double click on images to enlarge

- iPost using BlogPress from Kevin McGrew's iPad

No comments: