1

I don't understand these terms. They both have "likelihood". How are they different? Can someone provide an intuitive explanation of them?

gung - Reinstate Monica
  • 132,789
  • 81
  • 357
  • 650
Peter
  • 71
  • 1
  • 3
  • 1
    The partial likelihood is not an estimation method. You can have maximum likelihood based on partial likelihood such as in the Cox model so it is not very clear what you are asking. – JohnK Jan 09 '16 at 13:36
  • 3
    @JohnK, my guess is he doesn't understand the terms to ask a better question (that's why he's asking). You can explain that they aren't really contrasting categories; that would probably be very helpful for the OP. – gung - Reinstate Monica Jan 09 '16 at 13:50
  • As far as I understand both maximum likelihood and partial (or pooled) likelihood are likelihood functions that can be used to estimate the parameters of an econometric model. But I don't understand the exact difference of both functions. – Peter Jan 09 '16 at 16:06
  • Peter, I believe your question was already answered here: http://stats.stackexchange.com/questions/622/what-is-the-difference-between-a-partial-likelihood-profile-likelihood-and-marg Does this help to answer your question? – Matt Brems Jan 09 '16 at 15:21
  • Yes somehow. Are L1(β|data) L2(θ|data) partial likelihood functions? So if I use a partial likelihood function I just ignore a part of the model? – Peter Jan 09 '16 at 16:11
  • Partial likelihood is where you write the likelihood function L(β,θ|data) as L(β|data)•L(θ|data), then only maximize one of the likelihood functions. In this case, if you seek to find an estimate for β, then you can maximize L(β|data) and not worry about maximizing L(θ|data). It's called partial likelihood because you're maximizing only part of the likelihood function rather than attempting to find the maximum of the entire likelihood function. – Matt Brems Jan 09 '16 at 17:35

0 Answers0