p********2
发帖数: 9939 | 1 有点头晕了。
今天问了个问题，发现问得有问题。
例子：
比如10个学校招faculty，市场上20个candidate，每个学校从20个人中选1－20个来发
offer。
看是multinomial logit. 但是multinomial logit只有一个选项，也就是只发一个
offer。现在我allow for multiple offers。有没有SAS model 可以做这样的adaption
呢？ |
A*******s
发帖数: 3942 | 2 刚才有大侠帮你贴出来资料了吧，你看看里面的这段是不是你想要的？
Multivariate logistic model
Simultaneously models multiple responses, taking into account the
correlations among all response functions.
How to fit it: In PROC CATMOD, specify the RESPONSE LOGITS; statement
and multiple response variables in the MODEL statement to fit the model by
using weighted least squares estimation. For example, these statements
simultaneously model logits that are defined separately on three response
variables: response logits; model x1*x2*x3 = group; The bivariate probit
model can be fit in SAS/ETS PROC QLIM: model y1 y2 = x / discrete;
你的目的是啥呢？预测10个学校发offer的joint distribution?
adaption
【在 p********2 的大作中提到】
: 有点头晕了。
: 今天问了个问题，发现问得有问题。
: 例子：
: 比如10个学校招faculty，市场上20个candidate，每个学校从20个人中选1－20个来发
: offer。
: 看是multinomial logit. 但是multinomial logit只有一个选项，也就是只发一个
: offer。现在我allow for multiple offers。有没有SAS model 可以做这样的adaption
: 呢？
|
p********2
发帖数: 9939 | 3 我一开始以为是multivariate logit，后然想想不对啊。
multivariate logit是有好几个logit model，那也就是一个logit有一个coefficient
set estimated。我的例子就会有20个coefficient sets estimated for each
candidate。
我想estimate choice specific attributes对选折的影响。比如candidate的文章数目
。我只要一个coefficient estimate for 文章数。
【在 A*******s 的大作中提到】
: 刚才有大侠帮你贴出来资料了吧，你看看里面的这段是不是你想要的？
: Multivariate logistic model
: Simultaneously models multiple responses, taking into account the
: correlations among all response functions.
: How to fit it: In PROC CATMOD, specify the RESPONSE LOGITS; statement
: and multiple response variables in the MODEL statement to fit the model by
: using weighted least squares estimation. For example, these statements
: simultaneously model logits that are defined separately on three response
: variables: response logits; model x1*x2*x3 = group; The bivariate probit
: model can be fit in SAS/ETS PROC QLIM: model y1 y2 = x / discrete;
|
A*******s
发帖数: 3942 | 4 糊涂了... candidate最后只会选一个学校啊，那不就是multinomial logit么？
你要只搞一组estimate，可以把学校当成random effect来加个random intercept么，
这样你只有一组slope，但是有20 intercepts for schools。
coefficient
【在 p********2 的大作中提到】
: 我一开始以为是multivariate logit，后然想想不对啊。
: multivariate logit是有好几个logit model，那也就是一个logit有一个coefficient
: set estimated。我的例子就会有20个coefficient sets estimated for each
: candidate。
: 我想estimate choice specific attributes对选折的影响。比如candidate的文章数目
: 。我只要一个coefficient estimate for 文章数。
|
t**c
发帖数: 539 | 5 可不可以这么来看这个问题：
每个candidate可以拿到0-10个offer，所以用multinomial logistic regression.但是
因为是ordinal的data,这样子predictor的coefficient只有一个，而不是10个。会有10
个intercept。
coefficient
【在 p********2 的大作中提到】
: 我一开始以为是multivariate logit，后然想想不对啊。
: multivariate logit是有好几个logit model，那也就是一个logit有一个coefficient
: set estimated。我的例子就会有20个coefficient sets estimated for each
: candidate。
: 我想estimate choice specific attributes对选折的影响。比如candidate的文章数目
: 。我只要一个coefficient estimate for 文章数。
|
p********2
发帖数: 9939 | 6 我现在是model学校会给谁发offer，选折pool是job candidate。multinomial logit只
有一个选折，也就是一个学校只发一个offer，现在情况是1个学校可以发大于1 个
offer，可以选折多个candidate，所以传统的multinomial logit不管用
10
【在 t**c 的大作中提到】
: 可不可以这么来看这个问题：
: 每个candidate可以拿到0-10个offer，所以用multinomial logistic regression.但是
: 因为是ordinal的data,这样子predictor的coefficient只有一个，而不是10个。会有10
: 个intercept。
:
: coefficient
|
D******n
发帖数: 2836 | 7 你这个叫简单的logit，甚至linear regression也可以。
每个学校有自己的模型，然后给每个人打分，只要高于一个cutoff就给offer。
所以你这个是简单linear regression + 一个cutoff
包子please
【在 p********2 的大作中提到】
: 我现在是model学校会给谁发offer，选折pool是job candidate。multinomial logit只
: 有一个选折，也就是一个学校只发一个offer，现在情况是1个学校可以发大于1 个
: offer，可以选折多个candidate，所以传统的multinomial logit不管用
:
: 10
|
p********2
发帖数: 9939 | 8 One thing I am very sure is that this is not simple logit model.
【在 D******n 的大作中提到】
: 你这个叫简单的logit，甚至linear regression也可以。
: 每个学校有自己的模型，然后给每个人打分，只要高于一个cutoff就给offer。
: 所以你这个是简单linear regression + 一个cutoff
: 包子please
|
|
p********2
发帖数: 9939 | 9 sigh, I found it from Greene's book. It's called panel logit which forced
all the coefficients to be the same across all the logit models.
It's a special case of multivariate logit model with further restriction on
the coefficients.
Any help with the estimation of multivariate logit model with restriction of
coefficients? |
l*********s
发帖数: 5409 | 10 I don't think the panel logit model is appropriate here. Most hiring procresses have budgetary constraints and candidates are picked on a
relative basis. |
|
|
t**c
发帖数: 539 | 11 唔，好像确实应该是panel logit。
其实也不是force all the coefficients to be the same。如果我没理解错的话，这
就是logistic regression with random effect。所以估计的是the average
coefficient across all the logit models以及这个coefficient的variance。
可以用SAS中的PROC GLIMMIX。
on
of
【在 p********2 的大作中提到】
: sigh, I found it from Greene's book. It's called panel logit which forced
: all the coefficients to be the same across all the logit models.
: It's a special case of multivariate logit model with further restriction on
: the coefficients.
: Any help with the estimation of multivariate logit model with restriction of
: coefficients?
|
A*******s
发帖数: 3942 | 12 your original question about "model学校会给谁发offer" is way too general,
which leads to thousands of possible candidate models. you must ask a
specific one. For example:
do u wanna predict the joint distribution of offers from various school?
wanna test marginal association between offer and certain attributes?
wanna estimate school-specific or applicant-specific effects?
I kinda feel that you wanna estimate the marginal association between prob.
of offer and some applicants' attributes, in which case GEE-type or mixed
models can meet ur purpose. Again, it is just my guess, u r the one who
knows ur needs. |
p********2
发帖数: 9939 | 13 我想要estimate是candidate的attributes对offer的marginal impact。（我只是举个
例子）每年限制在20个candidate中选。这20个candidate每年也会变。看来还不是
exactly panel logit。只能算是multivariate logit with same coefficient
estimates across logits， allowing correlation between disturbances for the
same school applications.
proc qlim是不是就可以做这个estimate了？
.
【在 A*******s 的大作中提到】
: your original question about "model学校会给谁发offer" is way too general,
: which leads to thousands of possible candidate models. you must ask a
: specific one. For example:
: do u wanna predict the joint distribution of offers from various school?
: wanna test marginal association between offer and certain attributes?
: wanna estimate school-specific or applicant-specific effects?
: I kinda feel that you wanna estimate the marginal association between prob.
: of offer and some applicants' attributes, in which case GEE-type or mixed
: models can meet ur purpose. Again, it is just my guess, u r the one who
: knows ur needs.
|
A*******s
发帖数: 3942 | 14 my 2 cents is GEE/mixed models fit exactly this scenario--"这20个candidate每
年也会变", in which case marginal effect is of interest while subject-
specific effect is nuisance.
the
【在 p********2 的大作中提到】
: 我想要estimate是candidate的attributes对offer的marginal impact。（我只是举个
: 例子）每年限制在20个candidate中选。这20个candidate每年也会变。看来还不是
: exactly panel logit。只能算是multivariate logit with same coefficient
: estimates across logits， allowing correlation between disturbances for the
: same school applications.
: proc qlim是不是就可以做这个estimate了？
:
: .
|
t**c
发帖数: 539 | 15 请教：
在这种情况下，谁是random factor,谁是fixed factor?
看楼主开始的描述，是要把candidate作为fixed，学校是random。
现在这20个candidate每年也会变，那是说candidate也是random的？
而且还要加一个时间变量？
在这种情况下学校还是random吗？
【在 A*******s 的大作中提到】
: my 2 cents is GEE/mixed models fit exactly this scenario--"这20个candidate每
: 年也会变", in which case marginal effect is of interest while subject-
: specific effect is nuisance.
:
: the
|
A*******s
发帖数: 3942 | 16 Don't 请教, i am also a beginner. :)
you have a good catch. actually i have no experience to model such nested
cluster correlations-- within-school and within-candidate. Can we just use
two random statement with _residual_ options to model two R-side effects in
GLIMMIX? any thought?
【在 t**c 的大作中提到】
: 请教：
: 在这种情况下，谁是random factor,谁是fixed factor?
: 看楼主开始的描述，是要把candidate作为fixed，学校是random。
: 现在这20个candidate每年也会变，那是说candidate也是random的？
: 而且还要加一个时间变量？
: 在这种情况下学校还是random吗？
|
t**c
发帖数: 539 | 17 拜读过你的找工经验，非常非常佩服。是我的奋斗目标，:)
我没有试过在SAS里define two random factors，应该可以吧。
不过对于楼主这种情况，貌似他不关心系数随学校或者申请人的变化。刚google了一下，楼主所说的PROC QLIM 可能比较靠谱。
in
【在 A*******s 的大作中提到】
: Don't 请教, i am also a beginner. :)
: you have a good catch. actually i have no experience to model such nested
: cluster correlations-- within-school and within-candidate. Can we just use
: two random statement with _residual_ options to model two R-side effects in
: GLIMMIX? any thought?
|
