5 May
2003
5 May
'03
8:50 p.m.
If we are calculating confidence intervals by simulation, do we still need
to calculate panel corrected standard errors? (We have a pooled
cross-sectional time-series data set and a linear regression model.)
Thanks,
Anna
5 May
5 May
8:54 p.m.
simulation is done conditional on whatever estimate of the variance matrix
you have. PCSE's are supposed to better estimate the variance matrix
under certain conditions. if those conditions hold, then yes, you'd first
do those and then simulate.
Gary
On Mon, 5 May 2003, Anna Lorien Nelson wrote:
If we are calculating confidence intervals by
simulation, do we still need
to calculate panel corrected standard errors? (We have a pooled
cross-sectional time-series data set and a linear regression model.)
Thanks,
Anna
_______________________________________________
gov2001-l mailing list
gov2001-l(a)fas.harvard.edu
http://www.fas.harvard.edu/mailman/listinfo/gov2001-l
6 May
6 May
8:12 a.m.
It looks like we need to compute PCSE's. So we read Beck and Katz (1995),
but we're having trouble with the actual formula. What is the "Kronecker
product," for instance? Can anyone offer guidance on how to compute
PCSE's in R?
Many thanks,
Anna
On Mon, 5 May 2003, Gary King wrote:
simulation is done conditional on whatever estimate of
the variance matrix
you have. PCSE's are supposed to better estimate the variance matrix
under certain conditions. if those conditions hold, then yes, you'd first
do those and then simulate.
Gary
10:39 a.m.
kronecker(X,Y) will give you the kronecker product of X and Y. The entire
matrix, X, will be multiplied by each element of Y forming a submatrix of
the resulting giant matrix.
Kosuke
On Tue, 6 May 2003, Anna Lorien Nelson wrote:
It looks like we need to compute PCSE's. So we
read Beck and Katz (1995),
but we're having trouble with the actual formula. What is the "Kronecker
product," for instance? Can anyone offer guidance on how to compute
PCSE's in R?
Many thanks,
Anna
On Mon, 5 May 2003, Gary King wrote:
simulation is done conditional on whatever
estimate of the variance matrix
you have. PCSE's are supposed to better estimate the variance matrix
under certain conditions. if those conditions hold, then yes, you'd first
do those and then simulate.
Gary
_______________________________________________
gov2001-l mailing list
gov2001-l(a)fas.harvard.edu
http://www.fas.harvard.edu/mailman/listinfo/gov2001-l
5:41 p.m.
New subject: [gov2001-l] [gov2001-l]CLARIFY and Cox hazard
Dear all,
Does anyone know if CLARIFY can be used to simulate first differences for a Cos
proportionate hazards model? There is a hint right at the end of the
documentation (the section on weibull) but nothing definite. If there is a way
to do so, how would one create the survival object (in R or Stata you create a
'Surv' object combining the time and status of the case, e.g. 1963, 0 1964, 1)
I have used in the normal statistics packages, esp R?
Thanks
Ben
10:13 p.m.
New subject: [gov2001-l] [gov2001-l]CLARIFY and Cox hazard
you can't compute FD's from a cox model. there's no constant term.
have a look at my (unlectured from) lecture notes. you can compute a
relative rate.
Gary
On Tue, 6 May 2003, Benjamin William Ansell wrote:
Dear all,
Does anyone know if CLARIFY can be used to simulate first differences for a Cos
proportionate hazards model? There is a hint right at the end of the
documentation (the section on weibull) but nothing definite. If there is a way
to do so, how would one create the survival object (in R or Stata you create a
'Surv' object combining the time and status of the case, e.g. 1963, 0 1964,
1)
I have used in the normal statistics packages, esp R?
Thanks
Ben
_______________________________________________
gov2001-l mailing list
gov2001-l(a)fas.harvard.edu
http://www.fas.harvard.edu/mailman/listinfo/gov2001-l
7683
days inactive
7684
days old
gov2001@lists.gking.harvard.edu
5 comments
4 participants
participants (4)
-
alnelson＠fas.harvard.edu -
ansell＠fas.harvard.edu -
Gary King -
kimai＠fas.harvard.edu