关于ordered logit regression的问题

我在看一篇文章应用到了ordered logit regression模型，可是我看不懂其中的结论是怎么的出来的，图片如下

表中回归的是各项因素对个人感觉自己能力的影响，也就是说，假设个人能力从低到高是1-5，上过大学和上过研究生的对自我能力的感觉的差别
其中文章结论说男性比女性自我感觉高0.39，研究生比大学高0.06，这是怎么的出来的？
还有，在ordered logit regression中，回归系数到底是什么意思，比如图上Male是0.729，这是什么意思？？
谢谢大家了

It seems that the logit coefficient (回归系数) for male is 0.729. Logit coefficient is the log(odds). In other words, it is the log(p/(1-p)). You can get odds if you exponentiate the logit coefficient, 0.729. Here, you can get odds for males at 2.073, which means that males are more likely to feel competent at higher levels than females.

woodhaven 发表于 2012-2-19 05:06
It seems that the logit coefficient (回归系数) for male is 0.729. Logit coefficient is the log(odds) ...

谢谢。我还是有点不懂，因为之前没学过ordered logit regression，可是后天要做一个presentation. 我像请问这里的2.073是一个概率吗？是男人比女人自我感觉好高的概率？原文说Comparing a college-educated female investor, with average income and investments, to a male investor with the same demographics, the gender difference accounts for an increase of 0.39 in predicted investor competence, from 3.64 to 4.01
这0.39是怎么的出来的？
谢谢大虾

2.073 is not 一个概率. 2.073 is the odds ratio here. It can be interpreted as the odds of feeling competent for males are 2.073 times as large as the odds for females. Odds=p(success)/p(failure).

Please see the following website for detailed explanations and examples.

http://www.ats.ucla.edu/stat/stata/faq/oratio.htm


"Comparing a college-educated female investor, with average income and investments, to a male investor with the same demographics, the gender difference accounts for an increase of 0.39 in predicted investor competence, from 3.64 to 4.01."

Regarding your second questions, if I understand it correctly, the above value 0.39 was computed from the estimated logit equation which includes the above predictors (for example, college-educated, income, investment, and gender). Since 4.01-3.64 = 0.37, so 0.39 might be a typo (0.37 might be the right answer).

补充一下大神的解释，Logit coefficency is the exp(odds). 即odds=e^0.729=2.073.
再者，概率值只可能取[0,1]这是常识啊，这里面odds只是一个比率，即p/(1-p)=2.073, 所以p=0.675.

问一下你这个使用什么软件做的？