whatsnew8to9.hlp

是一个经济学管理应用软件 很难找的 但是经济学学生又必须用到

{p 5 9 2}
8.  New options {cmd:stdize()} and {cmd:stdweight()} on commands
    {cmd:svy: mean}, {cmd:svy: ratio}, {cmd:svy: proportion},
    {cmd:svy: tabulate oneway}, and {cmd:svy: tabulate twoway} allow direct
    standardization of means, ratios, proportions, and tabulations using any
    of the three survey variance estimators.

{p 5 9 2}
9.  Programmers of estimation commands can get full support
    for estimation with survey and correlated data almost automatically.  This
    support includes correct treatment of multistage designs, weighting,
    stratification, poststratification, and finite-population corrections, as
    well as access to all three variance estimators.
    See {help program properties:{bf:[P] program properties}}.

{p 4 9 2}
10.  The [SVY] manual now has a glossary that defines commonly used terms in
     survey analysis and explains how these terms are used in the
     manual; see {bf:[SVY]} glossary.


{marker panel}{...}
{title:What's new:  Longitudinal/panel data}

{p 5 9 2} 
1.  The big news is new command {cmd:xtmixed} -- Stata now fits linear
    mixed models, also known as hierarchical models or multilevel models.

{p 9 9 2} 
    Mixed models include what social scientists call random-effects
    models, including one-way, two-way, multi-way, and hierarchical models,
    and it includes random-coefficient models.

{p 9 9 2} 
    Estimates are obtained using maximum likelihood (ML), restricted maximum
    likelihood (REML), or expectation maximization (EM).  Covariances among
    random effects are estimated and may be independent (no covariance),
    exchangeable (common covariance), or unstructured (unique covariance for
    each pair of effects).  

{p 9 9 2} 
    {cmd:xtmixed} estimates standard errors and confidence intervals for
    the fixed parameters, and it estimates the standard deviations (variances)
    and correlations (covariances) of the random effects and the full VCE
    matrix among them.

{p 9 9 2} 
    For details, see {manhelp xtmixed XT}.

{p 9 9 2} 
    After estimation with {cmd:xtmixed}, 

{p 9 13 2}
    a.  {cmd:estat} {cmd:recovariance} reports the estimated
        variance-covariance matrix of the random effects for each level.

{p 9 13 2}
    b.  {cmd:estat} {cmd:group} summarizes the composition of the nested
        groups, providing minimum, average, and maximum group size for each
	level in the model.

{p 9 9 2} 
    {cmd:predict} after {cmd:xtmixed} can compute best linear unbiased
    predictions (BLUPs) for each random effect.  It can also compute the
    linear predictor, the standard error of the linear predictor, the fitted
    values (linear predictor plus contributions of random effects), the
    residuals, and the standardized residuals.

{p 5 9 2}
2.  New features have been added to the maximum-likelihood estimators
    that do not have closed-form solutions and require numeric evaluation of
    the likelihood.  These estimators include {helpb xtlogit}, 
    {helpb xtprobit}, {helpb xtpoisson}, {helpb xtcloglog}, 
    {helpb xtintreg}, and {helpb xttobit}.

{p 9 13 2}
    a.  The likelihood may now be approximated using adaptive Gauss-Hermite
        quadrature (the new default) or nonadaptive quadrature (the previous
        default).
        Adaptive quadrature
        substantially increases the accuracy of the approximation,
        particularly on difficult problems such as data with large panel sizes
        or data with a large variance for the random effects. 

{p 9 13 2}
    b.  Linear constraints may now be imposed using the new option 
        {cmd:constraints()}.  Constraints are specified the standard 
        way; see {manhelp constraint R}.

{p 9 13 2}
    c.  New option {cmd:intpoints()} replaces old option {cmd:quad()}, 
        although {cmd:quad()} continues to work.
        The new name is more meaningful, especially when used with estimators
        that integrate likelihoods using methods other than quadrature.

{p 5 9 2}
3.  Existing command {cmd:xtreg} now allows options {cmd:robust} and
    {cmd:cluster()} when estimating fixed-effects (FE) and random-effects (RE)
    models; see {manhelp xtreg XT}.

{p 5 9 2}
4.  Most {cmd:[XT]} commands that previously did not allow time-series
    operators now support them.  These commands include
    {helpb xtgls}, {helpb xtreg}, {helpb xtsum}, 
    {helpb xtcloglog}, {helpb xtintreg}, {helpb xtlogit}, {helpb xtpoisson},
    {helpb xtprobit}, {helpb xttobit}, and {helpb xtgee}.

{p 5 9 2}
5.  New command {cmd:xtrc} is old command {cmd:xtrchh}, renamed, and with 
    new features.
    New option {cmd:beta} reports the best linear predictors (BLUPs) for the
    group-specific coefficients, along with their standard errors and
    confidence intervals.  For details,
    see {manhelp xtrc XT}.

{p 5 9 2}
6.  {cmd:predict} after {cmd:xtrc} has the new option {cmd:group()} to compute
    the BLUPs of the dependent variable using the BLUPs of the coefficients.

{p 5 9 2}
7.  New command {cmd:xtline} plots panel data and
    allows either overlaid or separate graphs for each panel; 
    see {manhelp xtline XT}

{p 5 9 2}
8.  New section {bf:[XT]} {bf:glossary} defines commonly used
    terms and how they are used by us.


{marker timeseries}{...}
{title:What's new:  Time-series statistics}

{p 5 9 2}
1.  Existing command {cmd:arima} can now estimate multiplicative seasonal
    ARIMA (SARIMA) models; see new options {cmd:sarima()}, {cmd:mar()}, and
    {cmd:mma()} in {helpb arima:[TS] arima}.

{p 5 9 2}
2.  New command {cmd:rolling} performs rolling-window or recursive estimations,
    including regressions, and collects statistics from the estimation on each
    window;
    see {manhelp rolling TS}.

{p 5 9 2}
3.  The {bf:[TS]} manual now has a glossary that defines commonly used terms
    in time-series analysis and explains how we use them in the manual; see
    the glossary of {bf:[TS]}.

{p 5 9 2}
4.  Many existing commands that previously did not allow time-series
    operators now do.  These
    commands include {cmd:areg}, {cmd:binreg}, {cmd:biprobit}, 
    {cmd:boxcox}, {cmd:cloglog}, {cmd:cnsreg}, {cmd:glm}, {cmd:heckman},
    {cmd:heckprob}, {cmd:hetprob}, {cmd:impute}, {cmd:intreg}, 
    {cmd:logistic}, {cmd:logit}, {cmd:lowess}, {cmd:mvreg}, {cmd:nbreg},
    {cmd:orthog}, {cmd:pcorr}, {cmd:poisson}, {cmd:probit}, {cmd:pwcorr},
    {cmd:rreg}, {cmd:testparm}, {cmd:treatreg}, {cmd:truncreg}, 
    {cmd:xtcloglog}, {cmd:xtgls}, {cmd:xtintreg}, {cmd:xtlogit}, 
    {cmd:xtpoisson}, {cmd:xtprobit}, {cmd:xtgee}, {cmd:xtreg}, 
    {cmd:xtsum}, and {cmd:xttobit}.  

{p 5 9 2}
5.  Many commands requiring time-series data will now work on a single panel
    from a panel dataset when that panel is selected using an {cmd:if}
    expression or an {cmd:in} qualifier.  Those commands include {cmd:ac},
    {cmd:corrgram}, {cmd:cumsp}, {cmd:dfgls}, {cmd:dfuller}, {cmd:pac},
    {cmd:pergram}, {cmd:pperron}, {cmd:wntestb}, {cmd:wntestq}, and
    {cmd:xcorr}.  New commands {cmd:estat} {cmd:archlm}, {cmd:estat}
    {cmd:bgodfrey}, {cmd:estat} {cmd:dwatson}, and {cmd:estat}
    {cmd:durbinalt}, which replace commands {cmd:archlm}, {cmd:bgodfrey},
    {cmd:dwstat}, and {cmd:durbina}, also work on a single panel from a panel
    dataset.

{p 5 9 2}
6.  The dialogs for analyzing IRF results are much improved.  The dialogs
    now populate lists of models and variables from the current IRF
    results that may be chosen for producing tables and graphs.  The 
    improved dialogs include {bf:{stata db irf cgraph}}, 
    {bf:{stata db irf ctable}},
    {bf:{stata db irf graph}}, 
    {bf:{stata db irf ograph}}, and 
    {bf:{stata db irf table}}.

{p 5 9 2}
7.  Existing command 
    {cmd:dfuller} has new option {cmd:drift} for testing the null hypothesis
    of a random walk with drift.  The algorithm for calculating MacKinnon's
    approximate p-values is also now more accurate in cases where the p-value
    is relatively large; see {helpb dfuller:[TS] dfuller}.

{p 5 9 2}
8.  Existing commands
    {cmd:corrgram} and {cmd:pac} have new option {cmd:yw} that
    computes partial autocorrelations using the Yule-Walker equations instead
    of the default regression-based method; see {manhelp corrgram TS}.

{p 5 9 2}
9.  Time-series operators are now better displayed in estimation
    and other result tables.

{p 4 9 2}
10.  New command {cmd:estat} -- used after {cmd:regress} -- brings together
     what was previously done by commands {cmd:dwstat}, {cmd:durbina},
     {cmd:bgodfrey}, and {cmd:archlm}.
     The new commands are 
     {cmd:estat dwatson}, 
     {cmd:estat durbina}, 
     {cmd:estat bgodfrey}, and
     {cmd:estat archlm}.
     See {helpb regress postestimationts:[R] regress postestimation time series}.

{p 4 9 2}
11.  The ability of {cmd:arima} and {cmd:arch} to estimate standard errors
     using either the observed information matrix (OIM) or the outer product
     of gradients (OPG) has been consolidated under the new {cmd:vce()}
     option.

{pstd}
(What follows was first released in Stata 8.2.)

{p 4 9 2}
12.  New command {cmd:vec} fits cointegrated vector error-correction
     models (VECMs) using Johansen's method; see {manhelp vec TS}.

{p 4 9 2}
13.  New command {cmd:vecrank} produces statistics used to determine the
    number of cointegrating vectors in a VECM, including Johansen's trace and
    maximum-eigenvalue tests for cointegration; see 
    {manhelp vecrank TS}.

{p 4 9 2}
14.  New command {cmd:fcast} -- which replaces old command
    {cmd:varfcast} -- produces and graphs dynamic forecasts of the dependent
    variables after fitting a VAR, SVAR, or VECM; see 
    {manhelp fcast TS}.

{p 4 9 2}
15.  New command {cmd:irf} -- which replaces the old command {cmd:varirf} --
    does everything the old command did and more.  {cmd:irf} estimates the
    impulse-response functions, cumulative impulse-response functions,
    orthogonalized impulse-response functions, structural impulse-response
    functions, and forecast error-variance decompositions after fitting a VAR,
    SVAR, or VECM.  {cmd:irf} can also make graphs and tables of the results.
    See {helpb irf:[TS] irf}.

{p 9 9 2}
    {cmd:varirf} continues to work but is no longer documented.  {cmd:irf}
    accepts {cmd:.vrf} result files created by {cmd:varirf}.

{p 4 9 2}
16.  Existing command {cmd:varsoc} can now be used to obtain lag-order
     selection statistics for VECMs, as well as VARs;
     see {helpb varsoc:[TS] varsoc}.

{p 4 9 2}
17.  New command {cmd:veclmar} computes Lagrange-multiplier statistics for
    autocorrelation after fitting a VECM; see 
     {helpb veclmar:[TS] veclmar}.

{p 4 9 2}
18.  New command {cmd:vecnorm} tests whether the disturbances in a VECM
    are normally distributed.  For each equation and for all equations
    jointly, three statistics are computed: a skewness statistic, a kurtosis
    statistic, and the Jarque-Bera statistic.  See 
    {helpb vecnorm:[TS] vecnorm}.

{p 4 9 2}
19.  New command {cmd:vecstable} checks the eigenvalue stability condition
    after fitting a VECM; see {helpb vecstable:[TS] vecstable}.

{p 4 9 2}
20.  New command {cmd:vecstable} and the existing command {cmd:varstable}
    have a new graph option for presenting
    the stability results.  See 
    {helpb vecstable:[TS] vecstable} and 
    {helpb varstable:[TS] varstable}.

{p 4 9 2}
21.  The output of the following commands has been standardized to improve
    formatting:  {cmd:var}, {cmd:svar}, {cmd:vargranger},
    {cmd:varlmar}, {cmd:varnorm}, {cmd:varsoc}, {cmd:varstable}, and
    {cmd:varwle}.

{p 4 9 2}
22.  New command {cmd:haver} makes it easy to load and analyze 
     economic and financial databases available from Haver Analytics;
     see {helpb haver:[TS] haver}.



{marker multivariate}{...}
{title:What's new:  Multivariate statistics}

{pstd}
Stata has four all-new methods for analyzing multivariate data and many
more extensions to existing methods.  In addition, most methods now support 
direct analysis of matrices as well as raw data.

{pstd}
Be sure you check the postestimation documentation for the multivariate
estimators you use; many important new features are documented there.  In
particular, all the multivariate commands make extensive use of new
command {cmd:estat} for providing additional statistics and results after
estimation.

{p 5 9 2}
1.  New commands {cmd:mds}, {cmd:mdslong}, and {cmd:mdsmat} perform classic 
    metric multidimensional scaling:  {cmd:mds} performs the scaling with
    respect to the distances (dissimilarities) between observations,
    {cmd:mdslong} performs the scaling on a long dataset where each
    observation represents the distance between two points or objects, and
    {cmd:mdsmat} performs the scaling on a matrix of distances.  
    See {help mds:{bf:[MV] mds}}, {help mdslong:{bf:[MV] mdslong}}, and 
    {help mdsmat:{bf:[MV] mdsmat}}.

{p 9 9 2}
    {cmd:mds} supports all 33 similarity/dissimilarity measures
    available in Stata; see 
    {help measure_option:{bf:[MV]} {it:measure_option}}.

{p 9 9 2}
    The following new {cmd:estat} commands work after {cmd:mds},
    {cmd:mdslong}, or {cmd:mdsmat} and provide additional statistics and
    results:

{p 9 13 2}
    a.  {cmd:estat} {cmd:config} reports the coordinates of the
        approximating configuration.

{p 9 13 2}
    b.  {cmd:estat} {cmd:correlations} reports the Pearson and