[下载][推荐][讨论]Stata Questions and Answers

是 否 +2 论坛币

k人 参与回答

经管之家送您一份

应届毕业生专属福利!

求职就业群

赵安豆老师微信：zhaoandou666

经管之家联合CDA

送您一个全额奖学金名额~ !

立即领取

感谢您参与论坛问题回答

经管之家送您两个论坛币！

+2 论坛币

Dear list -members,

Please ignore it if you have read it before.

I'm trying to use logistic regression models to analyze the association
between SES and physical activity levels. My dependent variable is physical
activity level, which are grouped into three levels: a. regularly (ref), b.
occasionally, and infrequently. My questions are:
1. Can I conduct two separated logistic regression models to examine the
impact of SES on physical activity levels
a. DV (regularly vs occasionally)
b. DV (regularly vs infrequently)
or I have to run logistic regression with physical activity levels as
ordinal response variables?
2. if using two separated models is acceptable, is there anything missing in
comparing to using logistic regression with ordinal response variables?

3. what is the stata codes for this type analysis?

Thanks in advance for your help!

Jason Liu

扫码加我 拉你入群

请注明：姓名-公司-职位

以便审核进群资格，未注明则拒绝

分享0 收藏2 回帖

关键词：questions question answers Answer Stata 下载 讨论 Stata answers questions

You most likely want the Stata routines -ologit- or -oprobit-. Other
routines that might be appropriate include mlogit (which makes no
ordinality assumption), slogit, oglm, gologit2 or goprobit. To get a
better rundown on your alternatives, see the excellent Long and Freese book:

http://www.stata.com/bookstore/regmodcdvs.html

Dear Maarten and Nick,

The translog described in my email does not estimate Beta21, Beta31 and
Beta32. The reason is because when one estimates the translog,, LnX1.LnX2 is
the same as Ln.X2.LnX1, similarly for LnX2.LnX3=LnX3.LnX2 etc.. These cross
products are not viewed as different from each other and hence only one of
the pair is estimated.

But the homogeneity test requires equality between Beta(12)=Beta(21) etc in
addition to the others..

I am new with this test for homogeneity and am hoping that I have got it
correct.

Regards,

Gauri

以下是引用Trevor在2006-4-14 22:22:00的发言：
Dear Maarten and Nick,

The translog described in my email does not estimate Beta21, Beta31 and
Beta32. The reason is because when one estimates the translog,, LnX1.LnX2 is
the same as Ln.X2.LnX1, similarly for LnX2.LnX3=LnX3.LnX2 etc.. These cross
products are not viewed as different from each other and hence only one of
the pair is estimated.

But the homogeneity test requires equality between Beta(12)=Beta(21) etc in
addition to the others..

I am new with this test for homogeneity and am hoping that I have got it
correct.

Regards,

Gauri

If I understand you correctly, you have already imposed \beta_{ij} =\beta_{ji} -- that is, symmetry -- in your estimation. Hence, you cannot test your "fourth" restriction, as it holds by assumption. Hope that helps.

I would like to run an F test using multiple restrictions on a translog
production function with three inputs. This forms a test for homegeneity
using the famous F test.

The restrictions that must be jointly imposed are:

1. Beta11+Beta12+Beta13=0
2. Beta21+Beta22+beta23=0
3.
Beta31+Beta32+Beta33=0http://by112fd.bay112.hotmail.msn.com/cgi-bin/compose?&curmbox=00000000%2d0000%2d0000%2d0000%2d000000000001&a=8e97f2d6d33d18578baf68a0a3ed4538f9bf87ce76a05b31a9e3bf145a925032#
Send
4. Beta12=Beta21; Beta13=Beta31, Beta23=Beta32

Any one have a clue on how to perform this?

Thank you,

Gauri Khanna
Graduate Institute of International Studies
University of Geneva

以下是引用Trevor在2006-4-14 22:35:00的发言：
I would like to run an F test using multiple restrictions on a translog
production function with three inputs. This forms a test for homegeneity
using the famous F test.

The restrictions that must be jointly imposed are:

1. Beta11+Beta12+Beta13=0
2. Beta21+Beta22+beta23=0
3.
Beta31+Beta32+Beta33=0http://by112fd.bay112.hotmail.msn.com/cgi-bin/compose?&curmbox=00000000%2d0000%2d0000%2d0000%2d000000000001&a=8e97f2d6d33d18578baf68a0a3ed4538f9bf87ce76a05b31a9e3bf145a925032#
Send
4. Beta12=Beta21; Beta13=Beta31, Beta23=Beta32

Any one have a clue on how to perform this?

Thank you,

Gauri Khanna
Graduate Institute of International Studies
University of Geneva

Gauri Khanna:
If all parameters are estimated with one command than, this looks like job for the -test- command with the accum option.
HTH,
Maarten

-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

Thanks for the explanation. So, if I understand this,
the parameters are, or are forced to be, one and the same.

In that case, I can't see that you have a testable hypothesis,
or even a hypothesis. What you have ensured as a consequence
of your estimation is not an economic conjecture that needs to
be checked.

I'm reminded of a conversation I had in which I mentioned
that students from my university play ball games with
students from other universities on Wednesday. "Oh",
said the other person, from another university, "what
a coincidence, so too do students from my place". But
it wasn't a coincidence at all: it is all deliberate.
If Durham play Glasgow on Wednesdays, it follows ineluctably
that Glasgow play Durham on Wednesdays.

Nick
n.j.cox@durham.ac.uk

Gauri Khanna

Dear Maarten and Nick,

Thank you again for your replies. Yes, I think I will drop it because there
is no conjecture involved and it does hold by definition. It's just strange
to see it in the test..

Also, ball games and homogeneity now go together!!

Regards,

Gauri

Doesn't that equality hold by definition? In that case there is no need to
test
it.

-----------------------------------------
Maarten L. Buis
Department of Social Research Methodology
Vrije Universiteit Amsterdam
Boelelaan 1081
1081 HV Amsterdam
The Netherlands

Dear Stata-listers

I have a minor issue with the way STATA is rounding, for example a
variable to one decimal place.

I used the command

gen x=round(y, 0.1)

Comparing the two variables, I noticed the that when:

x= y=

8.750 8.7
5.752 5.8
23.256 23.3

Is there a way I can modify the function to round "x"=8.750 to be 8.8

Or should I put together my own code to get the rounding I want.

Many thanks in advance...


Amani

Amani:

Stata 8.2 doesn't show this behaviour:

. input x

x
1. 8.750
2. 5.752
3. 23.256
4. end

. gen y = round(x, 0.1)

. list

+---------------+
| x y |
|---------------|
1. | 8.75 8.8 |
2. | 5.752 5.8 |
3. | 23.256 23.3 |
+---------------+


Maybe 8.750 in your data is actually 8.7499?

HTH,
Maarten