help whatsnew9to10
-------------------------------------------------------------------------------
What's new in release 10.0 (compared with release 9)
This file lists the changes corresponding to the creation of Stata
release 10.0:
+------------------------------------------------------------+
| help file contents years |
|------------------------------------------------------------|
| whatsnew Stata 11.0 2009 to present |
| whatsnew10to11 Stata 11.0 new release 2009 |
| whatsnew10 Stata 10.0 and 10.1 2007 to 2009 |
| this file Stata 10.0 new release 2007 |
| whatsnew9 Stata 9.0, 9.1, and 9.2 2005 to 2007 |
| whatsnew8to9 Stata 9.0 new release 2005 |
| whatsnew8 Stata 8.0, 8.1, and 8.2 2003 to 2005 |
| whatsnew7to8 Stata 8.0 new release 2003 |
| whatsnew7 Stata 7.0 2001 to 2002 |
| whatsnew6to7 Stata 7.0 new release 2000 |
| whatsnew6 Stata 6.0 1999 to 2000 |
+------------------------------------------------------------+
Most recent changes are listed first.
--- more recent updates -------------------------------------------------------
See whatsnew10.
--- Stata 10.0 release 25jun2007 ----------------------------------------------
Remarks
We will list all the changes, item by item, but first, here are the
highlights:
1. Stata 10 has an interactive, point-and-click editor for your
graphs. You do not need to type anything; you just right-click
within the Graph window and select Start Graph Editor. You can
do that any time, either when the graph is drawn or when you have
graph used it from disk. You can add text, lines, markers,
titles, and annotations, outside the plot region or inside; you
can move axes, titles, legends, etc.; you can change colors and
sizes, number of tick marks, etc.; and you can even change
scatters to lines or bars, or vice versa. See [G] graph editor.
2. You can now save estimation results to disk. After fitting a
model, whether with regress, logistic, ..., or even a
user-written command, you type estimates save filename to save
it. You type estimates use filename to reload it later. See [R]
estimates.
3. Stata now fits nested, hierarchical, and mixed models with binary
and count responses; i.e., you can fit logistic and Poisson
models with complex, nested error components. See [XT] xtmelogit
and [XT] xtmepoisson.
4. Stata now has exact logistic and exact Poisson regression. In
small samples, exact methods have better coverage than asymptotic
methods, and exact methods are the only way to obtain point
estimates, tests, and confidence intervals from covariates that
perfectly predict the observed outcome. See [R] exlogistic and
[R] expoisson.
5. Stata now supports LIML and GMM estimation in addition to 2SLS.
Tests of instrumental relevance and tests of overidentifying
restrictions are available. See [R] ivregress and [R] ivregress
postestimation.
6. Stata now has more estimators for dynamic panel-data models,
including the Arellano-Bover/Blundell-Bond system estimator.
This estimator is an extension of the Arellano-Bond GMM estimator
for dynamic panel models. It is more efficient and has smaller
bias when the AR process is too persistent. These new estimators
can also be used to fit models with serially correlated
idiosyncratic errors and where the structure of the predetermined
variables is complicated. These new estimators can compute the
Windmeijer biased-corrected two-step robust VCE in addition to
the standard one-step i.i.d., one-step robust, and two-step
i.i.d. VCEs. See [XT] xtabond, [XT] xtdpdsys, and [XT] xtdpd.
7. New estimation command nlsur fits a system of nonlinear equations
by feasible generalized least squares, allowing for covariances
among the equations. See [R] nlsur.
8. Stata has new estimation commands for fitting categorical and
ranked outcomes, often used for choice models. The new commands
allow separate equations for each outcome and support the
easy-to-use alternative-specific notation. Joining Stata's
alternative-specific multinomial probit are alternative-specific
conditional logit (McFadden's choice model) and
alternative-specific rank-ordered probit (for modeling ordered
outcomes). See [R] asclogit and [R] asroprobit.
9. Stata's svy: prefix now works with 48 estimators, 27 more than
previously. Most importantly, svy: now works with Cox
proportional-hazards regression models (stcox) and parametric
survival models (streg). See [SVY] svy estimation.
10. The new stpower command provides sample-size and power
calculations for survival studies that use Cox
proportional-hazards regressions, log-rank tests for two groups,
or differences in exponentially distributed hazards or log
hazards. Available are (1) required sample size (given power),
(2) power (given sample size), and (3) the minimal detectable
effect (given power and sample size). stpower allows automated
production of customizable tables and has options to assist with
creating graphs and power curves. See [ST] stpower.
11. Stata 10 provides several discriminant analysis techniques,
including linear discriminant analysis (LDA), quadratic
discriminant analysis (QDA), logistic discriminant analysis, and
kth-nearest-neighbor (KNN) discrimination. See [MV] discrim.
12. Stata now provides multiple correspondence analysis (MCA) and
joint correspondence analysis (JCA). See [MV] mca.
13. Stata now provides modern as well as classical multidimensional
scaling (MDS), including metric and nonmetric MDS. Available
loss functions include stress, normalized stress, squared stress,
normalized squared stress, and Sammon. Available transformations
include identity, power, and monotonic. See [MV] mds.
14. Stata 10 has time/date variables in addition to date variables,
so now you can work with data that say an event happened on
12may2007 14:03:22.234 or events that happen every day at
14:03:22.234. Note the millisecond resolution. Time variables
are available in two forms: adjusted for leap seconds and
unadjusted. See [D] dates and times.
15. New Mata function optimize() performs minimization and
maximization. You can code just the function, the function and
its first derivatives, or the function and its first and second
derivatives. Optimization techniques include Newton-Raphson,
Davidon-Fletcher-Powell, Broyden-Fletch-Goldfarb-Shanno,
Berndt-Hall-Hall-Hausman, and the simplex method Nelder-Mead.
See [M-5] optimize().
16. Stata's online help now provides saved results and examples that
you can run.
17. Stata for Windows now supports Automation, formerly known as OLE
Automation, which means that programmers can control Stata from
other applications and retrieve results. See [P] automation.
18. Stata for Unix now supports unixODBC [sic], making it easier to
connect to databases such as Oracle, MySQL, and PostgreSQL; see
[D] odbc.
Another change is the introduction of Stata/MP, but that really happened
during Stata 9. Stata/MP is the parallel version of Stata for
multiple-CPU and multicore computers; see flavors. Stata/MP runs faster.
In Stata 10, many more commands now exploit the multiple processors,
which means that they run faster, too. This includes both survey and
nonsurvey mean, total, ratio, and proportion estimators and the survey
linearized variance estimator, which is available with many Stata
estimation commands.
There is much more, and the important changes for you may not be what we
list as highlights. Below are the details.
What's new is presented under the following headings:
What's new in the GUI and command interface
What's new in data management
What's new in statistics (general)
What's new in statistics (longitudinal/panel data)
What's new in statistics (time series)
What's new in statistics (survey)
What's new in statistics (survival analysis)
What's new in statistics (multivariate)
What's new in graphics
What's new in programming
What's new in Mata
What's new in the GUI and command interface
1. The Review window has been redesigned. It now shows the return
code of each previous command and highlights errors. From the
window, you can now select multiple commands -- not just single
ones -- to save or execute.
2. The Variables window has been redesigned. It nows shows the
storage type and display format for each variable in addition to
the variable's name and label. You can change any of these from
the window, including the name; just right-click.
3. Stata's Viewer has been redesigned. In addition to an all-new
look, it has a Forward button and the search capability is now
built in rather than provided by a dialog box.
4. The Graph window has been redesigned. In addition to providing
an interactive editor, under Windows, it now allows tabs. You
can have either one window containing multiple graphs each on its
own tab, or you can have each graph in a separate window.
5. You can copy and paste from Stata's Results and Viewer windows in
AS-IS mode, meaning that you can include Stata output in
documents and slides looking exactly as it looked on your screen.
6. Multiple Do-file Editors can now be opened simultaneously; just
click on the files in the Open dialog. (Unix users: You too can
now open multiple Do-file Editors.)
7. Concerning dialogs,
a. Stata now uses child dialogs, making dialogs for Stata
commands easier to use.
b. Programmers can program child dialogs, too; see [P] dialog
programming.
c. Graph dialogs have an all-new look that makes specifying the
most important items and options easier, yet provides access
to even more of graph's capabilities.
d. Dialogs that need matrices now allow the user to create the
matrix via a matrix-input child dialog and to show that new
matrix in the original dialog box after it has been created.
e. Dialogs that need formats now allow the user to specify the
format via a format builder.
f. Data from ODBC sources can now be accessed via a dialog.
g. Dialogs now scale with Microsoft Windows' DPI settings.
h. Dialogs now load faster.
8. Stata for Unix has an all new, more modern GUI.
9. Stata now executes sysprofile.do, if it exists, in addition to
profile.do when Stata is launched. This allows system
administrators to provide global customization. See [GS]
Appendix A: More on starting and exiting Stata.
10. New command adoupdate automates the process of updating
user-written ado-files; see [R] adoupdate.
11. New command hsearch searches the help files for specified words
and presents a ranked, clickable list in the Viewer; see [R]
hsearch.
12. Stata's help files are now named *.sthlp rather than *.hlp,
meaning that user-written help files can be sent via email more
easily. Many email filters flag .hlp files as potential virus
carriers because Stata was not the only one to use the .hlp
suffix. You need not rename your old help files. See [R] help.
13. There are now console versions of Stata/SE and Stata/MP for Mac,
just as there are for Unix. They are included on the
installation CD and installed automatically.
14. Stata's in command modifier now accepts F and L as synonyms for f
and l, meaning first and last observations.
15. Multiple log files may be opened simultaneously; see [R] log.
16. Intercooled Stata has been renamed to Stata/IC.
What's new in data management
1. Stata 10 has new date/time variables, so you can now record
values like 14jun2007 09:42:41.106 in one variable. They are
called %tc and %tC variables. The first is unadjusted for leap
seconds; the second is adjusted.
What used to be called "daily variables" are now called %td
variables. This is just a jargon change; daily (%td) variables
continue to work as they did before -- 0 means 01jan1960, 1 means
02jan1960, and so on.
%tc and %tC variables work similarly: 0 means 01jan1960
00:00:00. Here, however, 1 means 01jan1960 00:00:00.001, 1000
means 01jan1960 00:00:01.000, and 02jan1960 08:00:00 is
115,200,000. The underlying values are big -- so it is important
you store them as doubles -- but the %tc and %tC formats make the
values readable, just as the %td format makes daily (%td) values
readable.
There are many new functions to go along with this new value
type. clock(), for instance, converts strings such as "02jan1960
08:00:00" (or even "8:00 a.m., 1/2/1960") to their numeric
equivalents. dofc() converts a %tc value (such as 115,200,000,
meaning 02jan1960 08:00:00) to its %td equivalent (namely, 1,
meaning 02jan1960). cofd() does the reverse (the result would be
86,400,000, meaning 02jan1960 00:00:00).
See [D] dates and times.
2. The previously existing date() function, which converts strings
to %td values, is now smarter. In addition to being able to
convert strings such as "21aug2005", "August 21, 2005", it can
convert "082105", "08212005", "210805", and "21082005". See [D]
dates and times.
3. New command datasignature allows you to sign datasets and later
use that signature to determine whether the data have changed.
An early version of the command was made available during the
Stata 9 release. That command is now called _datasignature and
was used as the building block for the new, improved
datasignature. See [D] datasignature and [D] _datasignature.
4. Existing command clear now clears data and value labels only.
Type clear all to clear everything. This change will bite you
the first few times you type clear expecting it to clear all.
The problem was that new users were surprised when clear by
itself cleared everything, whereas use filename, clear loaded new
data and value labels but left everything else in place. The new
users were right.
clear now has the following subcommands:
a. clear all clears everything from memory.
b. clear ado clears automatically loaded ado-file programs.
d. clear programs clears all programs, automatically loaded or
not.
c. clear results clears saved results.
d. clear mata clears Mata functions and objects from memory.
See [D] clear.
5. Stata for Unix now supports unixODBC [sic], making it easier to
connect to databases such as Oracle, MySQL, and PostgreSQL; see
[D] odbc.
6. Existing command describe now allows option varlist that was
previously allowed only by describe using. Existing command
describe using filename now allows option simple that was
previously allowed only by describe. Option varlist saves the
variable names in r(varlist), and option simple displays the
variable names in a compact format. See [D] describe.
7. Existing command collapse now supports four additional stats:
first, the first value; last, the last value; firstnm, the first
nonmissing value; and lastnm, the last nonmissing value. See [D]
collapse.
8. Existing command cf (compare files) now provides a detailed
listing of observations that differ when the verbose option is
specified. Setting version to less than 10.0 restores the
earlier behavior. See [D] cf.
9. Existing command codebook has new option compact that produces
more compact output. See [D] codebook.
10. Existing command insheet has new option case that preserves the
case of variable names when importing data; see [D] insheet.
11. Existing command outsheet has new option delimiter() that
specifies an alternative delimiter; see [D] outsheet.
12. Existing commands infile and infix can now read up to 524,275
characters per line; the previous limit was 32,765. See [D]
infile and [D] infix (fixed format).
13. Existing commands icd9 and icd9p have now been updated to use the
V24 codes; see [D] icd9.
14. New function itrim() returns the string with consecutive,
internal spaces collapsed to one space; see String functions in
[D] functions.
15. New functions lnnormal() and lnnormalden() provide the natural
logarithm of the cumulative standard normal distribution and of
the standard normal density; see Probability distributions and
density functions in [D] functions.
16. New functions for calculating cumulative densities are now
available:
binomial(n, k, p) lower tail of the Binomial distribution
ibetatail(a, b, x) reverse (upper tail) of the cumulative
Beta distribution
gammaptail(a, x) reverse (upper tail) of the cumulative
Gamma distribution
invgammaptail(a, p) inverse reverse of the cumulative Gamma
distribution
invibetatail(a, b, p) inverse reverse of the cumulative Beta
distribution
invbinomialtail(n, k, p) inverse of right cumulative binomial
See Probability distributions and density functions in [D]
functions.
17. Existing function Binomial(n, k, p) has been renamed
binomialtail(n, k, p), thus making its name consistent with the
naming convention for probability functions. The accuracy of the
function has also been improved for very large values of n. At
the other end of the number line, the function now returns the
appropriate 0 or 1 value when n = 0, rather than returning
missing. Binomial() continues to work as a synonym for
binomialtail().
18. The behavior and accuracy of the following probability functions
have been improved:
a. F(n_1, n_2, f) and Ftail(n_1, n_2, f) are more accurate for
small values of n_1 and large values of n_2. Also, F() is
more accurate for large f where n_1 and n_2 are less than 1.
b. gammap(a, x) is more accurate when a is large and x is near
a.
c. ibeta(a, b, x) now is more accurate when x is near a/(a+b)
and a or b is large.
d. invbinomial(n, k, p), invchi2(n, p), invchi2tail(n, p),
invF(n_1, n_2, p), and invgammap(a, p) are more accurate for
small values of p or for returned values close to zero.
e. invFtail(n_1, n_2, p) and invibeta(a, b, p) are more accurate
for small values of p or for returned values close to zero.
f. invttail(n, p) is more accurate for small values of p or for
returned values close to zero.
g. ttail(n, t) is more accurate for exceedingly large values of
n.
19. Existing function invbinomial(n, k, p) now returns the
probability of a success on one trial such that the probability
of observing k or fewer successes in n trials is p. The previous
behavior of invbinomial() is restored under version control.
20. New function fmtwidth() returns the display width of a %fmt
string; see Programming functions in [D] functions.
21. The maximum length of a %fmt has increased from 12 to 48
characters; see [D] format. (This change was necessitated by the
new date/time variables.)
22. Existing commands corr2data and drawnorm now allow singular
correlation (or covariance) structures. New option forcepsd
modifies a matrix to be positive semidefinite and thus to be a
proper covariance matrix. See [D] corr2data and [D] drawnorm.
23. Existing command hexdump, analyze now saves the number of \r\n
characters in r(Windows) rather than in r(DOS). r(DOS) is still
set when version is less than 10. See [D] hexdump.
What's new in statistics (general)
1. As mentioned above, you can now save estimation results to disk.
You type estimates save filename to save results and estimates
use filename to reload them. In fact, the entire estimates
command has been reworked. The new command estimates notes
allows you to add notes to estimation results just as you add
them to datasets. The new command estimates esample allows you
to restore e(sample) after reloading estimates, should that be
necessary (usually it is not). The maximum number of estimation
results that can be held in memory (as opposed to saved on disk)
is increased to 300 from 20. See [R] estimates.
2. Stata now has exact logistic and exact Poisson regression.
Rather than having their inference based on asymptotic normality,
exact estimators enumerate the conditional distribution of the
sufficient statistics and then base inference upon that
distribution. In small samples, exact methods have better
coverage than asymptotic methods, and exact methods are the only
way to obtain point estimates, tests, and confidence intervals
from covariates that perfectly predict the observed outcome.
Postestimation command estat se reports odds ratios and their
asymptotic standard errors. estat predict, available only after
exlogistic, computes predicted probabilities, asymptotic standard
errors, and exact confidence intervals for single cases.
See [R] exlogistic and [R] expoisson.
3. New estimation command asclogit performs alternative-specific
conditional logistic regression, which includes McFadden's choice
model. Postestimation command estat alternatives reports
alternative-specific summary statistics. estat mfx reports
marginal effects of regressors on probabilities of each
alternative. See [R] asclogit and [R] asclogit postestimation.
4. New estimation command asroprobit performs alternative-specific
rank-ordered probit regression. asroprobit is related to
rank-ordered logistic regression (rologit) but allows modeling
alternative-specific effects and modeling the covariance
structure of the alternatives. Postestimation command estat
alternatives provides summary statistics about the alternatives
in the estimation sample. estat covariance displays the
variance-covariance matrix of the alternatives. estat
correlation displays the correlation matrix of the alternatives.
estat mfx computes the marginal effects of regressors on the
probability of the alternatives. See [R] asroprobit and [R]
asroprobit postestimation.
5. New estimation command ivregress performs single-equation
instrumental-variables regression by two-stage least squares,
limited-information maximum likelihood, or generalized method of
moments. Robust and HAC covariance matrices may be requested.
Postestimation command estat firststage provides various
descriptive statistics and tests of instrument relevance. estat
overid tests overidentifying restrictions. ivregress replaces
the previous ivreg command. See [R] ivregress and [R] ivregress
postestimation.
6. New estimation command nlsur fits a system of nonlinear equations
by feasible generalized least squares, allowing for covariances
among the equations; see [R] nlsur.
7. Existing estimation command nlogit was rewritten and has new,
better syntax and runs faster when there are more than two
levels. Old syntax is available under version control. nlogit
now fits the random utilities maximization (RUM) model by default
as well as the nonnormalized model that was available previously.
The new nlogit now allows unbalanced groups and allows groups to
have different sets of alternatives. nlogit now excludes entire
choice sets (cases) if any alternative (observation) has a
missing value; use new option altwise to exclude just the
alternatives (observations) with missing values. Finally,
vce(robust) is allowed regardless of the number of nesting
levels. See [R] nlogit.
8. Existing estimation command asmprobit has the following
enhancements:
a. The new default parameterization estimates the covariance of
the alternatives differenced from the base alternative,
making the estimates invariant to the choice of base. New
option structural specifies that the previously structural
(nondifferenced) covariance parameterization be used.
b. asmprobit now permits estimation of the constant-only model.
c. asmprobit now excludes entire choice sets (cases) if any
alternative (observation) has a missing value; use new option
altwise to exclude just the alternatives (observations) with
missing values.
d. New postestimation command estat mfx computes marginal
effects after asmprobit.
See [R] asmprobit and [R] asmprobit postestimation.
9. Existing estimation command clogit now accepts pweights and may
be used with the svy: prefix.
Also, clogit used to be willing to produce cluster-robust VCEs
when the groups were not nested within the clusters. Sometimes,
this VCE was consistent, and other times it was not. You must
now specify the new nonest option to obtain a cluster-robust VCE
when the groups are not nested within panels.
predict after clogit now accepts options that calculate the
Delta(beta) influence statistic, the Delta(chi^2) lack-of-fit
statistic, the Hosmer and Lemeshow leverage, the Pearson
residuals, and the standardized Pearson residuals.
See [R] clogit and [R] clogit postestimation.
10. Existing estimation command cloglog now accepts pweights, may now
be used with the svy: prefix, and has new option eform that
requests that exponentiated coefficients be reported; see [R]
cloglog.
11. Existing estimation command cnreg now accepts pweights, may be
used with the svy: prefix, and is now noticeably faster (up to
five times faster) when used within loops, such as by statsby.
See [R] cnreg.
12. Existing estimation commands cnsreg and tobit now accept
pweights, may be used with the svy: prefix, and are now
noticeably faster (up to five times faster) when used within
loops, such as by statsby. Also, cnsreg now has new advanced
option mse1 that sets the mean squared error to 1. See [R]
cnsreg and [R] tobit.
13. Existing estimation command regress is now noticeably faster (up
to five times faster) when used with loops, such as by statsby.
Also,
a. Postestimation command estat hettest has new option iid that
specifies that an alternative version of the score test be
performed that does not require the normality assumption.
New option fstat specifies that an alternative F test be
performed that also does not require the normality
assumption.
b. Existing postestimation command estat vif has new option
uncentered that specifies that uncentered variance inflation
factors be computed.
See [R] regress postestimation.
14. Existing estimation commands logit, mlogit, ologit, oprobit, and
probit are now noticeably faster (up to five times faster) when
used within loops, such as by statsby.
15. For existing estimation command probit, predict now allows the
deviance option; see [R] probit postestimation.
16. Existing estimation command nl has the following enhancements:
a. Option vce(vcetype) is now allowed, with supported vcetypes
that include types derived from asymptotic theory, that are
robust to some kinds of misspecification, that allow for
intragroup correlation, and that use bootstrap or jackknife
methods. Also, three heteroskedastic- and
autocorrelation-consistent variance estimators are available.
b. nl no longer reports an overall model F test because the test
that all parameters other than the constant are jointly zero
may not be appropriate in arbitrary nonlinear models.
c. The coefficient table now reports each parameter as its own
equation, analogous to how ml reports single-parameter
equations.
d. predict after nl has new options that allow you to obtain the
probability that the dependent variable lies within a given
interval, the expected value of the dependent variable
conditional on its being censored, and the expected value of
the dependent variable conditional on its being truncated.
These predictions assume that the error term is normally
distributed.
e. mfx can be used after nl to obtain marginal effects.
f. lrtest can be used after nl to perform likelihood-ratio
tests.
See [R] nl and [R] nl postestimation.
17. Existing estimation command mprobit now allows pweights, may now
be used with the svy: prefix, and has new option probitparam that
specifies that the probit variance parameterization, which fixes
the variance of the differenced latent errors between the scale
and the base alternatives to one, be used. See [R] mprobit.
18. Existing estimation command rologit now allows vce(bootstrap) and
vce(jackknife). See [R] rologit.
19. Existing estimation command truncreg now allows pweights and now
works with the svy: prefix. See [SVY] svy estimation.
20. After existing estimation command ivprobit, postestimation
commands estat classification, lroc, and lsens are now available.
Also, in ivprobit, the order of the ancillary parameters in the
output has been changed to reflect the order in e(b). See [R]
ivprobit and [R] ivprobit postestimation.
21. All estimation commands that allowed options robust and cluster()
now allow option vce(vcetype). vce() specifies how the
variance-covariance matrix of the estimators (and hence standard
errors) are to be calculated. This syntax was introduced in
Stata 9, with options such as vce(bootstrap), vce(jackknife), and
vce(oim).
In Stata 10, option vce() is extended to encompass the robust
(and optionally clustered) variance calculation. Where you
previously typed
. estimation-command ..., robust
you are now to type
. estimation-command ..., vce(robust)
and where you previously typed
. estimation-command ..., robust cluster(clustervar)
with or without the robust, you are now to type
. estimation-command ..., vce(cluster clustervar)
You can still type the old syntax, but it is undocumented. The
new syntax emphasizes that the robust and cluster calculation
affects standard errors, not coefficients. See [R] vce_option.
Going along with this change, estimation commands now have a term
for their default variance calculation. Thus, you will see
things like vce(ols), and vce(gnr). Here is what they all mean:
a. vce(ols). The variance estimator for ordinary least squares;
an s^2(X'X)^{-1}-type calculation.
b. vce(oim). The observed information matrix based on the
likelihood function; a (-H)^{-1}-type calculation, where H is
the Hessian matrix.
c. vce(conventional). A generic term to identify the
conventional variance estimator associated with the model.
For instance, in the Heckman two-step estimator,
vce(conventional) means the Heckman-derived variance matrix
from an augmented regression. In two different contexts,
vce(conventional) does not necessarily mean the same
calculation.
d. vce(analytic). The variance estimator derived from first
principles of statistics for means, proportions, and totals.
e. vce(gnr). The variance matrix based on an auxiliary
regression, which is analogous to s^2(X'X)^{-1} generalized
to nonlinear regression. gnr stands for Gauss-Newton
regression.
f. vce(linearized). The variance matrix calculated by a
first-order Taylor approximation of the statistic, otherwise
known as the Taylor linearized variance estimator, the
sandwich estimator, and the White estimator. This is
identical to vce(robust) in other contexts.
The above are used for defaults. vce() may also be
g. vce(robust). The variance matrix calculated by the sandwich
estimator of variance, VDV-type calculation, where V is the
conventional variance matrix and D is the outer product of
the gradients, sum_i g_ig_i'.
h. vce(cluster varname). The cluster-based version of
vce(robust) where sums are performed within the groups formed
by varname, which is equivalent to assuming that the
independence is between groups only, not between
observations.
i. vce(hc2) and vce(hc3). Calculated similarly as vce(robust)
except that different scores are used in place of the
gradient vectors g_i.
j. vce(opg). The variance matrix calculated by the outer
product of the gradients; a (sum_i g_ig_i')^(-1)-type
calculation.
k. vce(jackknife). The variance matrix calculated by the
jackknife, including delete one, delete n, and the
cluster-based jackknife.
l. vce(bootstrap). The variance matrix calculated by bootstrap
resampling.
You do not need to memorize the above; the documentation for the
individual commands, and their corresponding dialog boxes, make
clear what is the default and what is available.
22. Estimation commands specified with option vce(bootstrap) or
vce(jackknife) now report a note when a variable is dropped
because of collinearity.
23. The new option collinear, which has been added to many estimation
commands, specifies that the estimation command not remove
collinear variables. Typically, you do not want to specify this
option. It is for use when you specify constraints on the
coefficients such that, even though the variables are collinear,
the model is fully identified. See [R] estimation options.
24. Estimation commands having a model Wald test composed of more
than just the first equation now save the number of equations in
the model Wald test in e(k_eq_model).
25. All estimation commands now save macro e(cmdline) containing the
command line as originally typed.
26. Concerning existing estimation command ml;
a. ml now saves the number of equations used to compute the
model Wald test in e(k_eq_model), even when option lf0() is
specified.
b. ml score has new option missing that specifies that
observations containing variables with missing values not be
eliminated from the estimation sample.
c. ml display has new option showeqns that requests that
equation names be displayed in the coefficient table.
See [R] ml.
27. New command lpoly performs a kernel-weighted local polynomial
regression and displays a graph of the smoothed values with
optional confidence bands; see [R] lpoly.
28. New prefix command nestreg: reports comparison tests of nested
models; see [R] nestreg.
29. Existing commands fracpoly, fracgen, and mfp have new features:
a. fracpoly and mfp now support cnreg, mlogit, nbreg, ologit,
and oprobit.
b. fracpoly and mfp have new option all that specifies that
out-of-sample observations be included in the generated
variables.
c. fracpoly, compare now reports a closed-test comparison
between fractional polynomial models by using deviance
differences rather than reporting the gain; see [R] fracpoly.
d. fracgen has new option restrict() that computes adjustments
and scaling on a specified subsample.
See [R] fracpoly and [R] mfp.
30. For existing postestimation command hausman, options sigmaless
and sigmamore may now be used after xtreg. These options improve
results when comparing fixed- and random-effects regressions
based on small to moderate samples because they ensure that the
differenced covariance matrix will be positive definite. See [R]
hausman.
31. Existing postestimation command testnl now allows expressions
that are bound in parentheses or brackets to have commas. For
example, testnl _b[x] = M[1,3] is now allowed. See [R] testnl.
32. Existing postestimation command nlcom has a new option noheader
that suppresses the output header; see [R] nlcom.
33. Existing command statsby now works with more commands, including
postestimation commands. statsby also has new option forcedrop
for use with commands that do not allow if or in. forcedrop
specifies that observations outside the by() group be temporarily
dropped before the command is called. See [D] statsby.
34. Existing command mkspline will now create restricted cubic
splines as well as linear splines. New option displayknots will
display the location of the knots. See [R] mkspline.
35. In existing command kdensity, kernel(kernelname) is now the
preferred way to specify the kernel, but the previous method of
simply specifying kernelname still works. See [R] kdensity.
36. Existing command ktau's computations are now faster; see [R]
spearman.
37. In existing command ladder, the names of the transformations in
the output have been renamed to match those used by gladder and
qladder. Also, the returned results r(raw) and r(P_raw) have
been renamed to r(ident) and r(P_ident), respectively. See [R]
ladder.
38. Existing command ranksum now allows the groupvar in option
by(groupvar) to be a string; see [R] ranksum.
39. Existing command tabulate, exact now allows exact computations on
larger tables. Also, new option nolog suppresses the enumeration
log. See [R] tabulate twoway.
40. Existing command tetrachoric's default algorithm for computing
tetrachoric correlations has been changed from the Edwards and
Edwards estimator to a maximum likelihood estimator. Also,
standard errors and two-sided significance tests are produced.
The Edwards and Edwards estimator is still available by
specifying the new edwards option. A new zeroadjust option
requests that frequencies be adjusted when one cell has a zero
count. See [R] tetrachoric.
41. Existing command areg now works like regress with indicator
variables when cluster() is specified. See [R] areg.
What's new in statistics (longitudinal/panel data)
1. New command xtset declares a dataset to be panel data and
designates the variable that identifies the panels. In previous
versions of Stata, you specified options i(groupvar) and
sometimes t(timevar) to identify the panels. You specified the
i() and t() options on the xt command you wanted to use. Now you
xtset groupvar or xtset groupvar timevar first. The values you
set will be remembered from one session to the next if you save
your dataset.
xtset also provides a new feature. xtset allows option delta()
to specify the frequency of the time-series data, something you
will need to do if you are using Stata's new date/time variables.
Finally, you can still specify old options i() and t(), but they
are no longer documented. Similarly, old commands iis and tis
continue to work but are no longer documented. See [XT] xtset.
2. New estimation commands xtmelogit and xtmepoisson fit nested,
hierarchical, and mixed models with binary and count responses;
i.e., you can fit logistic and Poisson models with complex,
nested error components. Syntax is the same as for Stata's
linear mixed-model estimator, xtmixed. To fit a model of
graduation with a fixed coefficient on x1 and random coefficient
on x2 at the school level, and with random intercepts at both the
school and class-within-school level, you type
xtmelogit graduate x1 x2 || school: x2 || class:
predict after xtmelogit and xtmepoisson will calculate predicted
random effects. See [XT] xtmelogit, [XT] xtmelogit
postestimation, [XT] xtmepoisson, and [XT] xtmepoisson
postestimation.
3. New estimation commands are available for fitting dynamic
panel-data models:
a. Existing estimation command xtabond fits dynamic panel-data
models by using the Arellano-Bond estimator but now reports
results in levels rather than differences. Also, xtabond
will now compute the Windmeijer biased-corrected two-step
robust VCE. See [XT] xtabond.
b. New estimation command xtdpdsys fits dynamic panel-data
models by using the Arellano-Bover/Blundell-Bond system
estimator. xtdpdsys is an extension of xtabond and produces
estimates with smaller bias when the AR process is too
persistent. xtpdsys is also more efficient than xtabond.
Whereas xtabond uses moment conditions based on the
differenced errors in producing results, xtpdsys uses moment
conditions based on differences and levels. See [XT]
xtdpdsys.
c. New estimation command xtdpd fits dynamic panel-data models
extending the Arellano-Bond or the
Arellano-Bover/Blundell-Bond system estimator and allows a
richer syntax for specifying models and so will fit a broader
class of models then either xtabond or xtdpdsys. xtdpd can
be used to fit models with serially correlated idiosyncratic
errors, whereas xtdpdsys and xtabond assume no serial
correlation. xtdpd can be used with models where the
structure of the predetermined variables is more complicated
than that assumed by xtdpdsys or xtabond. See [XT] xtdpd.
d. New postestimation command estat abond tests for serial
correlation in the first-differenced errors. See [XT]
xtabond postestimation, [XT] xtdpdsys postestimation, and
[XT] xtdpd postestimation.
e. New postestimation command estat sargan performs the Sargan
test of overidentifying restrictions. See [XT] xtabond
postestimation, [XT] xtdpdsys postestimation, and [XT] xtdpd
postestimation.
4. Existing estimation command xtreg, fe now accepts aweights,
fweights, and pweights. Also, new option dfadj specifies that
the cluster-robust VCE be adjusted for the within transform.
This was previously the default behavior. See [XT] xtreg.
5. Existing estimation commands xtreg, fe and xtreg, re used to be
willing to produce cluster-robust VCEs when the panels were not
nested within the clusters. Sometimes this VCE is consistent and
other times it is not. You must now specify the new nonest
option to obtain a cluster-robust VCE when the panels are not
nested within the clusters.
6. The numerical method used to evaluate distributions, known as
quadrature, has been improved. This method is used by the xt
random-effects estimation commands xtlogit, xtprobit, xtcloglog,
xtintreg, xttobit, and xtpoisson, re normal.
a. For the estimation commands, the default method is now
intmethod(mvaghermite). The old default was
intmethod(aghermite).
b. Option intpoints(#) for the commands now allows up to 195
quadrature points. The default is 12, and the old upper
limit was 30. (Models with large random effects often
require more quadrature points.)
c. The estimation commands may now be used with constraints
regardless of the quadrature method chosen.
d. Command quadchk, for use after estimation to verify that the
quadrature approximation was sufficiently accurate, now
produces a more informative comparison table. Before, four
fewer and four more quadrature points were used, and that was
reasonable if the number of quadrature points was, say, n_q =
12. Now you may specify significantly larger n_q and the +4
is not useful. Now quadchk uses n_q - int(n_q/3) and
n_q + int(n_q/3).
e. quadchk has new option nofrom that forces refitted models to
start from scratch rather than starting from the previous
estimation results. This is important if you use the old
intmethod(aghermite), which is sensitive to starting values,
but not important if you are using the new default
intmethod(mvaghermite).
See [XT] quadchk.
7. All xt estimation commands now accept option vce(vcetype). As
mentioned in the What's new in statistics (general), vce(robust)
and vce(cluster varname) are the right ways to specify the old
robust and cluster() options, and option vce() allows other VCE
calculations as well.
8. Existing estimation command xtcloglog has new option eform that
requests exponentiated coefficients be reported; see [XT]
xtcloglog.
9. Existing estimation command xthtaylor now allows users to specify
only endogenous time-invariant variables, only endogenous
time-varying variables, or both. Previously, both were required.
See [XT] xthtaylor.
10. Most xt estimation commands have new option collinear, which
specifies that collinear variables are not to be removed.
Typically, you do not want to specify this option. It is for use
when you specify constraints on the coefficients such that, even
though the variables are collinear, the model is fully
identified. See [XT] estimation options.
11. Existing command xtdes has been renamed to xtdescribe. xtdes
continues to work as a synonym for xtdescribe. See [XT]
xtdescribe.
12. The [XT] manual has an expanded glossary.
What's new in statistics (time series)
1. All time-series analysis commands now support data with
frequencies as high as 1 millisecond (ms), corresponding to
Stata's new date/time variables. Since your data are probably
not recorded at the millisecond level, existing command tsset has
new option delta() that allows you to specify the frequency of
your data. Previously, time was recorded as t_0, t_0 + 1, t_0 +
2, ..., and if time = t in some observation, then the
corresponding lagged observation was the observation for which
time = t-1. That is still the default. When you specify
delta(), time is assumed to be recorded as t_0, t_0 + delta, t_0
+ 2delta, and if time = t in some observation, then the
corresponding lagged observation is the observation for which
time = t - delta. Say that you are analyzing hourly data and
time is recorded using Stata's new %tc values. One hour
corresponds to 3,600,000 ms, and you would want to specify tsset
t, delta(3600000). Option delta() is smart; you can specify
tsset t, delta(1 hour). See [TS] tsset.
2. tsset now reports whether panels are balanced when an optional
panel variable is specified.
3. Many ts estimation commands now accept option vce(vcetype). As
mentioned in What's new in statistics (general), vce(robust) and
vce(cluster varname) are the right ways to specify the old robust
and cluster() options, and option vce() allows other VCE
calculations as well.
4. Options vce(hc2) and vce(hc3) are now the preferred way to
request alternative bias corrections for the robust variance
calculation for existing estimation command prais. See [TS]
prais.
5. Existing estimation commands arch and arima have new option
collinear that specifies that the estimation command not remove
collinear variables. Typically, you do not want to specify this
option. It is for use when you specify constraints on the
coefficients such that, even though the variables are collinear,
the model is fully identified. See [TS] estimation options.
6. Existing command irf now estimates and reports dynamic-multiplier
functions and cumulative dynamic-multiplier functions, as well as
their standard errors. See [TS] irf.
7. The [TS] manual has an expanded glossary.
What's new in statistics (survey)
1. Stata's svy: prefix now works with 48 estimators, 28 more than
previously. Most importantly, svy: now works with Cox regression
(stcox) and parametric survival models (streg). Other commands
with which svy: now works include
biprobit bivariate probit regression
clogit conditional (fixed effects) logistic regression
cloglog complementary log-log regression
cnreg censored-normal regression
cnsreg constrained linear regression
glm generalized linear models
hetprob heteroskedastic probit model
ivregress instrumental-variables regression
ivprobit probit model with endogenous regressors
ivtobit tobit model with endogenous regressors
mprobit multinomial probit regression
nl nonlinear least-squares estimation
scobit skewed logistic regression
slogit stereotype logistic regression
stcox Cox proportional hazards regression
streg parametric survival models (5 estimators)
tobit tobit regression
treatreg treatment-effects model
truncreg truncated regression
zinb zero-inflated negative binomial regression
zip zero-inflated Poisson regression
ztnb zero-truncated negative binomial regression
ztp zero-truncated Poisson regression
See [SVY] svy estimation.
2. svy: prefix now calculates the linearized variance estimator 2 to
100 times faster, the larger multiplier applying to large
datasets with many sampling units; see [SVY] svy.
3. svy: mean, svy: proportion, svy: ratio, and svy: total are
considerably faster when the over() option identifies many
subpopulations.
4. svy:, svy: mean, svy: proportion, svy: ratio, and svy: total now
take advantage of multiple processors in Stata/MP, making them
even faster in that case.
5. Concerning svyset,
a. New option singleunit(method) provides three methods for
handling strata with one sampling unit. If not specified,
the default in such cases is to report standard errors as
missing value.
b. New option fay(#) specifies that Fay's adjustment be made to
the BRR weights.
See [SVY] svyset.
6. estat has two new subcommands for use with svy estimation
results:
a. estat sd, used after svy: mean, reports subpopulation
standard deviations.
b. estat strata reports the number of singleton and certainty
strata within each sampling stage.
See [SVY] estat.
7. svy: tabulate now allows string variables. See [SVY] svy:
tabulate oneway and [SVY] svy: tabulate twoway.
8. Existing command svydes has been renamed svydescribe; svydes
continues to work. svydescribe now puts missing values in the
generate(newvar) variable for observations outside the specified
estimation sample. Previously, the variable would contain a zero
for observations outside the estimation sample. See [SVY]
svydescribe.
9. The [SVY] manual has been reorganized. Stata's survey estimation
commands are now documented in [SVY] svy estimation. All
model-specific information is now documented in the manual entry
for the corresponding estimation command.
What's new in statistics (survival analysis)
1. Existing estimation commands stcox and streg may now be used with
the svy: prefix and so can fit models for complex survey data;
see [ST] stcox and [ST] streg.
2. New command stpower provides sample-size and power calculations
for survival studies that use Cox proportional-hazards
regressions, log-rank tests for two groups, or differences in
exponentially distributed hazards or log hazards.
a. stpower cox estimates required sample size (given power) or
power (given sample size) or the minimal detectable
coefficient (given power and sample size) for models with
multiple covariates. The command provides options to account
for possible correlation between the covariate of interest
and other predictors and for withdrawal of subjects from the
study. See [ST] stpower cox.
b. stpower logrank estimates required sample size (given power)
or power (given sample size) or the minimal detectable hazard
ratio (given power and sample size) for studies comparing
survivor functions of two groups by using the log-rank test.
Both the Freedman (1982) and the Schoenfeld (1981) methods
are provided. The command allows for unequal allocation of
subjects between the groups and possible withdrawal of
subjects. Estimates can be adjusted for uniform accrual.
See [ST] stpower logrank.
c. stpower exponential estimates sample size (given power) or
power (given sample size) of tests of the difference between
hazards or log hazards of two groups under the assumption of
exponential survivor functions (also known as the exponential
test). Both the Lachin-Foulkes (1986) and
Rubinstein-Gail-Santner (1981) methods are provided. Unequal
group allocation, uniform or truncated exponential accrual,
and different exponential losses due to follow-up in each
group are allowed. See [ST] stpower exponential.
The stpower commands allow automated production of customizable
tables and have options to assist with creating graphs of power
curves. See [ST] stpower.
3. Concerning existing command sts graph,
a. New option risktable() places a subjects-at-risk table
underneath and aligned with the survivor or hazard plot.
b. New option ci replaces old options gwood, cna, and cihazard.
sts graph will choose the appropriate confidence interval on
the basis of the function being graphed.
c. Confidence intervals are now graphed using shaded areas and
new options plotopts() and ciopts() allow you to control how
plots and confidence intervals look.
d. Overlaid confidence intervals are now allowed and are
produced when new option ci is combined with existing option
by(varlist).
e. New option censopts() controls the appearance of ticks and
markers produced by existing option censored().
f. Boundary computations for smoothing hazards have been
improved. New option noboundary specifies that no boundary
correction be done.
g. The lower bound of the range to plot the hazard function now
extends to zero.
h. Option na has been renamed cumhaz. na may still be used.
{pmore2} See [ST] sts graph. Setting version to less than 10
restores previous behavior. {phang2} 4. For sts list,
option na has been renamed cumhaz. na may be used as a
synonym for cumhaz. See [ST] sts list. {phang2} 5.
Improvements to stcurve analogous to those of sts graph have
been made. {p 12 16 2} a. Boundary computations for
smoothing hazards have been improved. New option noboundary
specifies that no boundary correction be done. {p 12 16 2}
b. The lower bound of the range to plot the hazard function
now extends to zero. {pmore2} See [ST] stcurve. {phang2} 6.
All st estimation commands accept option vce(vcetype). As
mentioned in the What's new in statistics (general),
vce(robust) and vce(cluster varname) are the right ways to
specify the old robust and cluster() options, and option
vce() now allows other VCE calculations as well. {phang2} 7.
Existing command predict after stcox has a new option,
scores, that allows generating variables with the partial
efficient score residuals; see [ST] stcox postestimation.
{phang2} 8. Existing command ltable has new options
byopts(), plotopts(), plot#opts(), and ci#opts() that allow
for more customization of the graph. New option ci adds
confidence intervals to the graph. See [ST] ltable.
{phang2} 9. Existing command stphplot has a new option
plot#opts() that allows for more customization of the graph.
See [ST] stcox diagnostics. {p 7 12 2} 10. Existing command
stcoxkm has new options byopts(), obsopts(), obs#opts(),
predopts(), and pred#opts() that allow for more customization
of the graph. See [ST] stcox diagnostics. {p 7 12 2} 11.
Existing command cc has new option tarone that produces
Tarone's (1985) adjustment of the Breslow-Day test for
homogeneity of odds ratios. See [ST] epitab. {p 7 12 2} 12.
Existing command stdes has been renamed to stdescribe. stdes
continues to work. See [ST] stdescribe. {p 7 12 2} 13. The
[ST] manual has an expanded glossary. What's new in
statistics (multivariate) {phang2} 1. New estimation
commands discrim and candisc provide several discriminant
analysis techniques, including linear discriminant analysis
(LDA), quadratic discriminant analysis (QDA), logistic
discriminant analysis, and kth nearest neighbor
discrimination. See [MV] discrim, [MV] discrim estat, and
[MV] candisc. {phang2} 2. Existing estimation commands mds,
mdslong, and mdsmat now provide modern as well as classical
multidimensional scaling (MDS), including metric and
nonmetric MDS. Available loss functions include stress,
normalized stress, squared stress, normalized squared stress,
and Sammon. Available transformations include identity,
power, and monotonic. mdslong also now allows aweights and
fweights, and mdsmat has a weight() option. See [MV] mds,
[MV] mdslong, and [MV] mdsmat. {phang2} 3. New estimation
command mca provides multiple correspondence analysis (MCA)
and joint correspondence analysis (JCA); see [MV] mca and
[MV] mca postestimation. You can use existing command
screeplot afterward to graph principal inertias; see [MV]
screeplot. {phang2} 4. Concerning existing estimation
command ca (correspondence analysis), {p 12 16 2} a. ca now
allows crossed (stacked) variables. This provides a way to
automatically combine two or more categorical variables into
one crossed variable and perform correspondence analysis with
it. {p 12 16 2} b. ca's existing option normalize() now
allows normalize(standard) to provide normalization of the
coordinates by singular vectors divided by the square root of
the mass. {p 12 16 2} c. ca's new option length() allows
you to customize the length of labels with crossed variables
in output. {p 12 16 2} d. New postestimation command estat
loadings, used after ca and camat, displays correlations of
profiles and axes. {p 12 16 2} e. Existing postestimation
command cabiplot has new option origin that displays the
origin within the plot. cabiplot also now accepts
originlopts(line_options) to customize the appearance of the
origin on the graph. {p 12 16 2} f. Existing postestimation
commands cabiplot and caprojection now allow row and column
marker labels to be specified using the mlabel() suboption of
rowopts() and colopts(). {pmore2} See [MV] ca and [MV] ca
postestimation. {phang2} 5. Existing commands cluster,
matrix dissimilarity, and mds now allow the Gower measure for
a mix of binary and continuous data; see [MV] measure_option.
{phang2} 6. Existing command biplot has new options. dim()
specifies the dimensions to be displayed. negcol specifies
that negative column (variable) arrows be plotted.
negcolopts(col_options) provides graph options for the
negative column arrows. norow and nocolumn suppress the row
points or column arrows. See [MV] biplot. {phang2} 7. New
postestimation command estat rotate after canon performs
orthogonal varimax rotation of the raw coefficients, standard
coefficients, or canonical loadings. After estat rotate, new
postestimation command estat rotatecompare displays the
rotated and unrotated coefficients or loadings and the most
recently rotated coefficients or loadings. See [MV] canon
postestimation. {phang2} 8. Existing commands pcamat and
factormat now allow singular correlation or covariance
structures. New option forcepsd modifies a matrix to be
positive semidefinite and thus to be a proper covariance
matrix. See [MV] pca and [MV] factor. {phang2} 9. Existing
commands rotate and rotatemat now refer to "Kaiser
normalization" rather than "Horst normalization". A search
of the literature indicates that Kaiser normalization is the
preferred terminology. Previously option horst was a synonym
for normalize. Now option horst is not documented. See [MV]
rotate and [MV] rotatemat. {p 7 12 2} 10. Existing command
procrustes now saves the number of y variables in scalar
e(ny); see [MV] procrustes. What's new in graphics {phang2}
1. Stata 10 has an interactive, point-and-click editor for
your graphs. You do not need to type anything; you just
right-click within the Graph window and select Start Graph
Editor. You can do that any time, either when the graph is
drawn or when you have graph used it from disk. You can add
text, lines, markers, titles, and annotations, outside the
plot region or inside; you can move axes, titles, legends,
etc.; you can change colors and sizes, number of tick marks,
etc.; and you can even change scatters to lines or bars, or
vice versa. See [G] graph editor. {phang2} 2. New command
graph twoway lpoly plots a local polynomial smooth; see [G]
graph twoway lpoly. New command graph twoway lpolyci plots a
local polynomial smooth along with a confidence interval; see
> [G] graph twoway lpolyci. {phang2} 3. Concerning command
graph twoway, {p 12 16 2} a. graph twoway now allows more
than 100 variables to be plotted. {p 12 16 2} b. New
suboption custom of axis_label_options allows you to create
custom axis ticks and labels that have a different color,
size, tick length, etc., from the standard ticks and labels
on the axis. Such custom ticks can be used to emphasize
points in the scale, such as important dates, physical
constants, or other special values. See the custom suboption
in [G] axis_label_options. {p 12 16 2} c. New suboption
norescale of axis_label_options specifies that added ticks
and labels be placed directly on the graph without rescaling
the axis or associated plot region for the new values; see
[G] axis_label_options. {p 12 16 2} d. New advanced options
yoverhangs and xoverhangs adjust the graph region margins to
prevent long labels on the y or x axis from extending off the
edges of the graphs; see [G] advanced_options. {phang2} 4.
graph twoway pcarrow and graph twoway pcbarrow may now be
drawn on plot regions with log scales or reversed scales; see
> [G] graph twoway pcarrow. {phang2} 5. graph bar and
graph dot no longer require user-provided names when a
variable is repeated with more than one statistic; see [G]
graph bar and [G] graph dot. {phang2} 6. graph twoway lfit
and graph twoway qfit now use value labels to annotate the x
axis when using the existing suboption valuelabel of
xmlabel(); see [G] axis_label_options. {phang2} 7. graph
export has new options width(#) and height(#) that specify
the width and height of the graph when exporting to PNG or
TIFF, thus allowing the resolution to be greater than screen
resolution; see png_options and tif_options in [G] graph
export. {phang2} 8. graph twoway area and graph twoway
rarea now allow option cmissing() to control whether missing
values produce breaks in the areas or are ignored; see [G]
graph twoway area and [G] graph twoway rarea. {phang2} 9.
Typing help graph option now displays the help file for the
specified option of the graph command. See [R] help. What's
new in programming {phang2} 1. First, a warning for
time-series programmers: Stata's new date/time values, which
contain the number of milliseconds since 01jan1960 00:00:00,
result in large numbers. 21apr2007 12:14:07.123 corresponds
to 1,492,776,847,123. Date/time values must be stored as
doubles. Programmers should use scalars to store these
values whenever possible. If you must use a macro, exercise
caution that the value is not rounded. It would not do at
all for 1,492,776,847,123 to be recorded as "1.493e+12"
(which would be 24apr2007 02:13:20). If you must use macros,
our recommendations are {p 12 16 2} a. If a date/time value
is stored in one macro and you need it in another, code {p 16
18 2} local new `old' {p 12 16 2} b. If a date/time value is
the result of an expression, and you must store it as a
macro, code {p 16 18 2} local new = string(exp, "%21x") {p 12
16 2} or {p 16 18 2} local new : display %21x (exp) {pmore2}
Now we will continue with What's new. {phang2}2. Stata for
Windows now supports Automation, formerly known as OLE
Automation, which means that programmers can control Stata
from other applications and retrieve results. See [P]
automation. {phang2} 3. New command confirm
{numeric|string|date} format verifies that the format is of
the specified type; see [P] confirm. {phang2} 4. New
function fmtwidth(s) returns the display width of a %fmt
string, including date formats; see Programming functions in
[D] functions. {phang2} 5. Expression limits have been
increased in Stata/MP, Stata/SE, and Stata/IC. The limit on
the number of dyadic operators has increased from 500 to 800,
and the limit on the number of numeric literals has increased
from 150 to 300. See limits. {phang2} 6. Intercooled Stata
has been renamed to Stata/IC. c(flavor) now contains "IC"
rather than "Intercooled" if version > 10.
Backward-compatibility old global macro $S_FLAVOR continues
to contain "Intercooled". See [P] creturn and [P] macro.
{phang2} 7. c() now contains values associated with
Stata/MP: c(MP) (1 or 0 depending on whether you are running
Stata/MP), c(processors) (the number of processors Stata will
use), c(processors_mach) (the number of processors on the
computer), c(processors_lic) (the maximum number of
processors the license will allow you to use), and
c(processors_max) (the maximum number of processors that
could be used on this computer with this license). {phang2}
8. New command include is a variation on do and run.
include executes the commands stored in a file just as if
they were entered from the keyboard or the current do-file.
It is most often used in advanced Mata situations to create
the equivalent of #include files. See [P] include. {phang2}
9. New commands signestimationsample and
checkestimationsample are useful in writing
estimation/postestimation commands that need to identify the
estimation sample; see [P] signestimationsample. {p 7 12 2}
10. New command _datasignature is the building block for
Stata's datasignature command and the programming commands
signestimationsample and checkestimationsample. In advanced
situations, you may wish to call it directly. See [P]
_datasignature. {p 7 12 2} 11. New extended macro function
:copy copies one macro to another and is faster when the
macro being copied is long. That is, coding {p 12 16 2}
local new : copy local old {pmore2} is faster than {p 12 16
2} local new `old' {pmore2} See [P] macro. {p 7 12 2} 12.
New command timer times sections of code; see [P] timer. {p
7 12 2} 13. Existing command matrix accum is now faster when
some observations contain zeros; see [P] matrix accum. {p 7
12 2} 14. Existing command ml display has new option
showeqns that requests that equation names be displayed in
the coefficient table; see [R] ml. {p 7 12 2} 15. Existing
command mkmat has new options rownames(), roweq(),
rowprefix(), obs, and nchar() that specify the row names of
the created matrix; see [P] matrix mkmat. {p 7 12 2} 16.
Existing command _rmdcoll's nocollin option has been renamed
to normcoll. nocollin will continue to work as a synonym for
normcoll. See [P] _rmcoll. {p 7 12 2} 17. Existing command
describe's option simple no longer saves the names of the
variables in r(varlist); you must specify option varlist if
you want that. Also, existing command describe using
filename now allows options simple and varlist. See [D]
describe. {p 7 12 2} 18. New extended macro function
adosubdir returns the subdirectory in which Stata would
search for a file along the ado-path. Determining the
subdirectory in which Stata stores files is now a function of
the file's extension. adosubdir returns the subdirectory in
which to look. See [P] macro. {p 7 12 2} 19. Existing
command syntax {[optionname(real ...)]} now returns the
number specified in %18.0g format if version is set to 10.0
or higher. For version less than 10, the number is returned
in %9.0g format. See [P] syntax. {p 7 12 2} 20. New
functions _byn1() and _byn2(), available within a
byable(recall) program, return the beginning and ending
observation numbers of the by-group currently being executed;
see [P] byable. {p 7 12 2} 21. Existing command program
drop may now specify program drop _allado to drop programs
that were automatically loaded from ado-files; see [P]
program. {p 7 12 2} 22. Concerning SMCL, {p 12 16 2} a.
Existing directive {synoptset} has new optional argument
notes that specifies that some of the table entries will have
footnotes and results in a larger indentation of the first
column. {p 12 16 2} b. Existing directive {p} now has an
optional fourth argument specifying the paragraph's width.
{pmore2} See [P] smcl. {p 7 12 2} 23. Concerning classes,
you can now define an oncopy member program to perform
operations when a copy of an object is being created. See
[P] class. {p 7 12 2} 24. Concerning programmable menus,
the maximum number of menu items that can be added to Stata
has increased to 1,250 from 1,000; see [P] window
programming. {p 7 12 2} 25. Concerning programmable dialogs,
{p 12 16 2} a. Child dialogs can now be created. {p 12 16
2} b. New control TEXTBOX allows displaying multiline text.
{p 12 16 2} c. In the dialog programming language, (1) if
now allows else and (2) new command close closes the dialog
programmatically. {p 12 16 2} d. Messages can be passed to
dialogs when they are launched; see [R] db. {p 12 16 2} e.
Dialogs can now be designated as modal, meaning that this
dialog must be dealt with by the user before new dialogs
(other than children) can be launched. {p 12 16 2} f.
Several controls have new options and new member programs.
For instance, FILE and LISTBOX now have option multiselect,
which lets the user pick more than one item. {pmore2} See
[P] dialog programming. {p 7 12 2} 26. Stata's help files
are now named *.sthlp rather than *.hlp, meaning that
user-written help files can be sent via email more easily.
Many email filters flag .hlp files as potential virus
carriers because Stata was not the only one to use the .hlp
suffix. You need not rename your old help files. See [R]
help. {p 7 12 2} 27. Two new C functions have been exposed
from Stata for use by plugins: sstore() and sdata().
sstore() stores string data in the Stata dataset and sdata()
reads them. See http://www.stata.com/plugins/. What's new
in Mata {phang2} 1. New Stata command include is a variation
on do and run and is useful for implementing #include for
header files in advanced programming situations. See [P]
include and type viewsource optimize.mata for an example of
use. {phang2} 2. Mata now has structures, which will be of
special interest to those writing large systems. See [M-2]
struct and [M-5] liststruct(). {phang2} 3. Mata now engages
in more thorough type checking, and produces better code, for
those who explicitly declare arguments and variables.
{phang2} 4. Mata inherits all of Stata's formats and
functions for dealing with the new date/time variables and
values; see [M-5] date() and [M-5] fmtwidth(). {phang2} 5.
New functions inbase() and frombase() perform base
conversions; see [M-5] inbase(). {phang2} 6. New function
floatround() returns values rounded to float precision. This
is Mata's equivalent of Stata's float() function. See [M-5]
floatround(). {phang2} 7. New function nameexternal()
returns the name of an external; see [M-5] findexternal().
{phang2} 8. Concerning matrix manipulation, {p 12 16 2} a.
Matrix multiplication is now faster when one of the matrices
contains many zeros, as is function cross(). {p 12 16 2} b.
Appending rows or columns to a matrix using , and \ is now
faster. {p 12 16 2} c. New function _diag() replaces the
principal diagonal of a matrix with a specified vector or
scalar; see [M-5] _diag(). {p 12 16 2} d. New functions
select() and st_select() select rows or columns of a matrix
on the basis of a criterion; see [M-5] select(). {p 12 16 2}
e. Existing functions rowsum(), colsum(), sum(),
quadrowsum(), quadcolsum(), and quadsum() now allow an
optional second argument that determines how missing values
are handled; see [M-5] sum(). {p 12 16 2} f. New functions
runningsum(), quadrunningsum(), _runningsum(), and
_quadrunning-sum() return the running sum of a vector; see
[M-5] runningsum(). {p 12 16 2} g. New functions minindex()
and maxindex() return the indices of minimums and maximums
(including tied values) of a vector; see [M-5] minindex().
{phang2} 9. Concerning statistics, {p 12 16 2} a. New Mata
function optimize() performs minimization and maximization.
You can code just the function, the function and its first
derivatives, or the function and its first and second
derivatives. Optimization techniques include Newton-Raphson,
Davidon-Fletcher-Powell, Broyden-Fletcher-Goldfarb-Shanno,
Berndt-Hall-Hall-Hausman, and the simplex method Nelder-Mead.
See [M-5] optimize(). {p 12 16 2} b. New function
cvpermute() forms permutations; see [M-5] cvpermute(). {p 12
16 2} c. New function ghk() provides the
Geweke-Hajivassiliou-Keane multivariate normal simulator; see
[M-5] ghk(). New function ghkfast() is faster but a little
more difficult to use; see [M-5] ghkfast(). {p 12 16 2} d.
New functions halton(), _halton(), and ghalton() compute
Halton and Hammersley sets; see [M-5] halton(). {p 12 16 2}
e. The new density and probability functions available in
Stata are also available in Mata, including binomial(),
binomialtail(), gammaptail(), invgammaptail(),
invbinomialtail(), ibetatail(), invibetatail(), lnnormal(),
and lnnormalden(); see [M-5] normal(). Also, as in Stata,
convergence and accuracy of many of the cumulatives, reverse
cumulatives, and density functions have been improved. {p 12
16 2} f. Existing Mata functions mean(), variance(),
quadvariance(), meanvariance(), quadmeanvariance(),
correlation(), and quadcorrelation() now make the weight
argument optional. If not specified, unweighted estimates
are returned. See [M-5] mean(). {p 7 12 2} 10. Concerning
string processing, {p 12 16 2} a. New function stritrim()
replaces multiple, consecutive internal spaces with one
space; see [M-5] strtrim(). {p 12 16 2} b. New functions
strtoreal() and _strtoreal() convert strings to numeric
values; see [M-5] strtoreal(). {p 12 16 2} c. New function
_substr() substitutes a substring into an existing string;
see [M-5] _substr(). {p 12 16 2} d. New function
invtokens() is the inverse of the existing function tokens();
see [M-5] invtokens(). {p 12 16 2} e. New function
tokenget() performs advanced parsing; see [M-5] tokenget().
{p 7 12 2} 11. Concerning I/O, {p 12 16 2} a. New function
adosubdir() returns the subdirectory in which Stata would
search for a file; see [M-5] adosubdir(). New function
pathsearchlist(fn) returns a row vector of full
paths/filenames specifying all the locations, in order, where
Stata would look for the specified fn along the official
Stata ado-path; see [M-5] pathjoin(). {p 12 16 2} b. New
function byteorder() returns 1 if the computer is HILO and
returns 2 if the computer is LOHI; see [M-5] byteorder(). {p
12 16 2} c. New undocumented function st_fopen() makes
opening files easier; see [M-5] st_fopen(). {p 12 16 2} d.
New functions bufput() and bufget() copy elements into and
out of buffers; see [M-5] bufio(). {p 12 16 2} e. Existing
function cat() now allows optional second and third arguments
that specify the beginning and ending lines of the file to
read; see [M-5] cat(). {p 7 12 2} 12. New functions
ferrortext() and freturncode() obtain error messages and
return codes following an I/O error; see [M-5] ferrortext().
{p 7 12 2} 13. Concerning the Stata interface, {p 12 16 2}
a. New function stataversion() returns the version of Stata
that you are running, and new function statasetversion()
allows setting it. See [M-5] stataversion(). {p 12 16 2} b.
New function setmore() allows turning more on and off. New
function setmoreonexit() allows restoring more to its
original setting when execution ends. See [M-5] more(). {p
12 16 2} c. New function st_lchar() allows storing
exceedingly long strings (such as a varlist) in Stata dataset
characteristics. See [M-5] st_lchar(). What's more {pstd}
We have not listed all the changes, but we have listed the
important ones. {pstd} Stata is continually being updated,
and those updates are available for free over the Internet.
All you have to do is type . update query {pstd} and follow
the instructions. {pstd} To learn what has been added since
this manual was printed, select Help > What's new? or type .
help whatsnew {pstd} We hope that you enjoy Stata 10.
References {phang} Freedman, L. S. 1982. Tables of the
number of patients required in clinical trials using the
logrank test. Statistics in Medicine 1: 121-129. {phang}
Lachin, J. M., and M. A. Foulkes. 1986. Evaluation of sample
size and power for analyses of survival with allowance for
nonuniform patient entry, losses to follow-up, noncompliance,
and stratification. Biometrics 42: 507-519. {phang}
Rubinstein, L. V., M. H. Gail, and T. J. Santner. 1981.
Planning the duration of a comparative clinical trial with
loss to follow-up and a period of continued observation.
Journal of Chronic Diseases 34: 469-479. {phang}
Schoenfeld, D. A. 1981. The asymptotic properties of
nonparametric tests for comparing survival distributions.
Biometrika 68: 316-319. {phang} Tarone, R. E. 1985. On
heterogeneity tests based on efficient scores. Biometrika
72: 91-95. --- previous updates {hline} {pstd} See
whatsnew9. {hline} Also see {psee} Help: whatsnew {p_end}
