Probit vs Logit

• both are types of generalized linear models
• both can be used for modeling the relationship between:
  • one or more numerical or categorical predictor variables
  • categorical outcome

uses something called the cumulative distribution function of the logistic distribution

uses the logistic function:

• 𝜃₀ + 𝜃₁𝑥₁ + … + 𝜃_𝑘𝑥_𝑘 = 𝒛 = 𝑙𝑛(𝐏/(1-𝐏))

uses something called the cumulative distribution function of the standard normal distribution

uses the inverse standard normal distribution function:

• 𝜃₀ + 𝜃₁𝑥₁ + … + 𝜃_𝑘𝑥_𝑘 = 𝒛 = 𝚽^-1(𝐏)