TERPRETING
ROBABILITY MODELS
ogit, Probit, and
her Generalized
inear Models
“TIM FUTINGILIAO
Quantitative Applications
'n ‘ie Social Sciences
101
SACE UNIVERSITY PAPER“Quantitative Applications
/ in the Social Sciences
fmt ios Angle
INTERPRETING
PROBABILITY MODELS
Logit, Probit, and Other
Generalized Linear Models
TIM FUTING LIAO
University of inci, Urbana-Champaign
SAGE PUBLICATIONS
Internationa! Educational and Professional Publisher
‘Thousand Oaks Londen New Delhionsen 6198 Sage betes ne
egal Mpa hi eka Seater atria am ay 2
Is ean mei aheing ecoregion
‘Sager sen te pron wi ew Be ache
Sanaa
‘ett ents a rpg a he aye mel ee
(HUG. F350 lei Prt asl Loa, Pl nd Oe Oar
Seely pet Gace Aan
(2) tan 7. 188g rbot: Lag rand ter gene
‘Ete tei Get ope so ee Noo
‘CONTENTS
Series Réitors Introduction ¥
Acknowledgments
4. Turoducton 1
Why Probebiiy Models? 1
Why Iaterpetaton? 2
2. Generalized Linear Models and the Interpretation of
Parameters 3
‘Gonetlzed Lincar Models 3
Interpeeation of Parameter Esinates 6
3. Binary Logit and Probit Models 10
opt Models 11
Interpretation of Logit Models 12
Probit Models 21
Inzerpcctation of Probit Models 22
Logit or Probit Models? 24
4. Sequential Logit and Probit Models 25
The Model 25
Inzerpectation of Sequential Logit and Prebt Models 29
5. Ordinal Logit and Probit Models 37
"The Model 37
Iserpotaton of Osdnal Logit and robit Models 41
6. Multinomiat Logit Models 48
“The Model 48
Inverpcetation of Multinomisl Logk Models 50
7. Conditional Logit Modes 59
The Mode! 60
Incerpretation of Conditions! Logit Models 62