245 followers
1. Logistics regression, not linear, should be used to fit 0/1 data…? https://t.co/HUtF8yk91c
1. Logistics regression, not linear, should be used to fit 0/1 data…? https://t.co/HUtF8yk91c
私の博論審査、経済(国際)・経済(国内)・国際幼児(質)・障害(質)という顔ぶれなので、①HH Fixed-effects vs OLS、②LPMが使えない結構例外的なデータ特性がある(下記論文)、③ロジット交差項をMEAかAMEのどっちに戻すか問題、をどうディフェンスするか頭痛い(続) https://t.co/olYUy7MxPf
@fgilardi This might be helpful. Not sure though if it is exactly what you need. https://t.co/ZN0Uv8flyh
@cabralestweet On that (eternal) issue, I would suggest Hellevik (2009), which provides a rather simple but useful contribution to this, splitting the usage of the two models (linear vs. non-linear) on the purpose of the anlysis: prediction vs. causal anal