ESTIMATING LINEAR PROBABILITY FUNCTIONS: A COMPARISON OF APPROACHES

摘要: A linear probability function permits the estimation of occurrence or non-occurrence a discrete event. Nerlove and Press (p. 3–9) outline several statistical problems that arise if such is estimated via OLS. In particular, heteroskedasticity inherent in regression model leads to inefficient estimates parameters (Amemiya 1973, Horn Horn). Moreover, without restrictions on conventional OLS model, lying outside unit (0–1) interval are possible (Nerlove Press). Goldberger Kmenta suggest two approaches for alleviating model. Logit analysis will also alleviate ensure probabilities lie within 1974, Hauck Donner, Hill Kau, Horn, Duncan, Theil 1970).