Stata is a registered trademark of StataCorp LP , College Station, TX, USA, and the Stata logo is used with permission of StataCorp.
StataCorp社 国内正規販売代理店 インフォーマティック（株） [Back] [Home] [Bottom] お問い合わせ・注文
2009年6月26日、米国Stata社は、Stataの新しいバージョンStata 11を発表しました。
実に2年ぶりのバージョンアップで、新たな強化機能に加えて、自由プラットフォームの
サポート、PDFマニュアルの提供等、魅力的な機能を数多く提供しています。
出荷はすでに開始されています。
Stata 11 has many, many new features:
o Multiple imputation
Stata's new -mi- command provides a full suite of multiple-
imputation methods for the analysis of incomplete data, data for
which some values are missing.
An outstanding feature of Stata's multiple-imputation system
is
the data-management aspect. Handling multiple copies of the
data is the bane of multiple-imputation analysis, but Stata
solves this problem because -mi- automatically keeps all datasets
in sync. You can create or drop variables, append, merge, or
reshape datasets -- it does not matter, because Stata will carry
out these data-management tasks consistently on all datasets.
Anyone interested in multiple imputation will love Stata's
implementation.
o Factor (categorical) variables
Factor variables are one of those things that are difficult to
describe but that everyone is really going to like.
To give you just a hint as to what this is about, suppose that
you want to run a regression that contains an age-squared term
and a dummy variable for race being black.
In Stata 10, users did this:
. generate age2 = age^2
. generate black = (race==2)
. xtgee ln_w grade age age2 black
In Stata 11, users will do this:
. xtgee ln_w grade age c.age#c.age 2.race
(The "c" just indicates that age is a continuous
variable, and
the # specifies the interaction.)
For those working with factor (categorical) variables,
. regress y gender##agegroup
will create indicators for male and female, indicators for
each
category in agegroup, and indicators for all combinations of
categories in gender and agegroup.
In Stata 11, you do not have to create the extra variables, so
you will have fewer variables in your dataset.
Also, suppose that you want to compute the average marginal
effect of age on wages. If your model was specified using age
and age2 (the old way), Stata could not consider the contribution
of age from age2 because it had no way of knowing that these
variables are related. On the other hand, if you specify
c.age#c.age, then Stata deeply understands that this age-squared
term is related to age.
o Marginal means, adjusted predictions, and marginal effects
The new -margins- command makes the old -mfx- and -adjust-
commands obsolete, and does a lot more than they did. Among the
statistics calculated is predictive margins, which are very
important for survey data. Other statistics include what are
sometimes called estimated marginal means or least-squares means.
-margins- is a postestimation command that works after almost
every estimation command in Stata.
This is a great command!
o Additional graph features
Graphs now support multiple fonts and symbols.
Regular, bold, and italics formatting; sans serif, serif,
monospace, and symbol fonts; lowercase and uppercase Greek
alphabet; and over 70 mathematical symbols are now available.
These new features may be specified in command mode, through
dialog boxes, and through the interactive Graph Editor.
We have had requests for this feature, so I expect that users
will be happy to see it in the new release.
o GMM (generalized method of moments estimation)
This is a very nice implementation because it makes GMM
estimation as simple as nonlinear least-squares regression. We
expect this to be an important addition for social scientists
and, in particular, economists.
o Competing-risks regression
We have been asked frequently about competing risks. In fact,
several of our staff members have written Stata Journal and STB
articles on the subject. With this estimator, Stata now has a
built-in way of handling survival or duration models when
individuals face more than one kind of risk. Expect health
researchers, sociologists, and labor economists to be excited.
o State-space models
State-space models can be used to represent and estimate an
extremely wide range of multivariate time-series processes. One
nice thing about Stata's implementation is that, unlike most
packages that do state-space models, it does not require the
user to specify whether the model is stationary or nonstationary.
Instead, Stata's implementation automatically flips between the
two.
o Dynamic-factor models
This is a foray into the increasingly popular approach of
modeling sets of related variables as they relate to one or
more unobservable characteristics or processes. Dynamic factors
are a decidedly modern approach with a heritage in classic
factor analysis and structural-equation modeling.
o Multivariate GARCH
This is like the current -arch- command, but now users can
analyze the interactions of multiple time series all at once.
State-space models, dynamic-factor models, and multivariate
GARCH
will be popular with users working in finance, banking, portfolio
analysis, and advanced econometrics.
o Unit-root tests for panel data
This extends Stata's current ability to test for unit roots to
panel data.
o New Data Editor
There are a lot of improvements to the editor. The biggest
change is that the Data Editor is now a live view onto your
data. You can leave the Data Editor open while you perform
your analysis; if you change your data by running a Stata
command, you will see the results in the Data Editor instantly.
My favorite improvement is how easy it is to enter dates and
times. I believe that users are really going to like this.
Also, you can take a snapshot of your data so you can easily
undo a change. You can apply filters so you can view subsets
of your data.
o Variables Manager
I particularly like the Variables Manager. You click on the
Variables Manager icon, a window appears showing you the variable
name, label, storage type, format, value label, and any notes
for
the variables in your dataset currently in memory. Changing any
of these can easily be done by clicking with your mouse. Those
with many variables can type a word in the filter field to see
only the variables that contain that word.
o Do-file Editor
This is a big improvement. In Stata for Windows, changes
include syntax highlighting, code folding (the ability to hide
blocks of code while you focus on the rest), the ability to work
on multiple files simultaneously, and no limit on file size.
Users should be happy about this because these improvements
have been requested in every Users Group meeting for the last
few years.
o PDF manuals
These are nice! They are integrated with Stata's help system.
Every copy of Stata will come with PDF manuals. Of course, we
will still have printed copies of the manuals.
o And lots more!
You can find out more about Stata 11 by visiting
http://www.stata.com/stata11.
___ ____ ____ ____ ____
/__ / ____/ / ____/ Gretchen Farrar gfarrar@stata.com Email
___/ / /___/ / /___/ International Sales 1-800-782-8272 ext. 219
U.S.
StataCorp LP 1-979-696-4600 ext. 219 Intl
4905 Lakeway Drive 1-800-248-8272 ext. 219 Canada
College Station, TX 77845 1-979-696-4601 Fax
"VISIT OUR WEB SITE" http://www.stata.com
[Top]
[Home] [Top] [Back] お問い合わせ・注文