Bioinformatician here. I'm wondering what the main differences between both of these approaches is. I just want to clarify that a formal mathematical treatment would likely be too much for me.

From what I can tell, a GLM does not assume a normal distribution of response variables. Am I to infer that an LM does? What other important differences are there? If the distribution is normal, would an LM and GLM give the same results? I don't know enough to ask any more questions but would welcome any other details you think are important.

  • 2
    $\begingroup$ Generalized liner models involve link functions whereas linear models do not. $\endgroup$ Jan 31 '17 at 23:06