whatsnew.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

    {it:{help ml##eform_option:eform_option}}.

{p 5 9 2}
{* fix}
4.  {cmd:predict} with the {opt pr} option, when used after {helpb probit}
    with the {helpb svy} prefix or the {cmd:vce(bootstrap)} or
    {cmd:vce(jackknife)} options, did not produce the predicted probability
    values.  This has been fixed.


{hline 8} {hi:update 17jun2005} {hline}

    {title:Ado-files}

{p 5 9 2}
{* fix}
1.  {helpb areg} failed when attempting to fit a constant-only model.  This
    has been fixed.

{p 5 9 2}
{* enhancement}
2.  {helpb asmprobit} uses either {helpb clogit} or {helpb mlogit} for initial
    estimates.  These estimates are now scaled by the standard deviation of
    the extreme value distribution, {cmd:c(pi)/sqrt(6)}.

{p 5 9 2}
{* enhancement}
3.  {helpb asmprobit} does not attempt to refine the standard
    deviation/correlation estimates if the first call to {cmd:ml} (using the
    Cholesky factored variance/covariance parameterization) does not converge.
    In this case, a second call to {cmd:ml} is made to reparameterize the
    estimates, but {cmd:iterate(0)} is used to prevent another nonconvergent
    search.

{p 5 9 2}
{* fix}
4.  {helpb asmprobit}, {helpb mprobit}, and {helpb slogit} used inappropriate
    error codes.  This has been fixed.

{p 5 9 2}
{* enhancement}
5.  {helpb biplot} now allows plotting of any two dimensions, not just the
    first two, with the new {cmd:dim()} option.  {cmd:biplot} also now allows
    you to plot negative column (variable) arrows with the new {cmd:negcol}
    option.  Graph options for the negative column arrows are specified with
    the new {cmd:negcolopts()} option.  New options {cmd:norow} and
    {cmd:nocolumn} suppress the row points or column arrows.

{p 5 9 2}
{* enhancement}
6.  {helpb nlogit} now allows groups to have different sets of alternatives,
    so unbalanced groups are now allowed.  {cmd:nlogit} now saves
    {cmd:e(alt_min)}, {cmd:e(alt_max)}, and {cmd:e(alt_avg)} providing the
    minimum, maximum, and average number of alternatives per group.

{p 5 9 2}
{* fix}
7.  {helpb cluster measures} complained if a {it:measure} was not provided.
    It now defaults to using the {cmd:L2} {it:measure}, as documented.

{p 5 9 2}
{* fix}
8.  {helpb dotplot}, when supplied with an {opt if} or {opt in} restriction,
    produced graphs with an incorrect range for the frequency axis.  This has
    been fixed.

{p 5 9 2}
{* fix}
9.  {helpb _get_gropts}, a programmer's command, now recognizes the
    {opt altshrink} option as a {cmd:graph} {cmd:combine} option when the
    {opt getcombine} option is specified.

{p 4 9 2}
{* fix}
10.  {helpb newey} failed if a model contained variables with time-series
     operators and one of those variables was dropped due to
     multicollinearity.  This has been fixed.

{p 4 9 2}
{* fix}
11.  {helpb newey}, like {helpb regress}, showed "(dropped)" in the table of
     coefficients for a dropped variable.  Now, like most other commands, it
     displays "note: {it:variable} dropped due to collinearity" at the top of
     the output.

{p 4 9 2}
{* fix}
12.  {cmd:predict} with the {opt pr} option, when used after {helpb logistic},
     {helpb logit}, or {helpb probit} with the {helpb svy} prefix or the
     {cmd:vce(bootstrap)} or {cmd:vce(jackknife)} options, produced values of
     the linear prediction instead of the predicted probability.  This has
     been fixed.

{p 4 9 2}
{* fix}
13.  {helpb xtmixed_postestimation##predict:predict}
     {it:varlist}{cmd:, reffects} after {helpb xtmixed} now produces an
     appropriate error message and exits with an error code if any of the
     variables in {it:varlist} already exist.

{p 4 9 2}
{* fix}
14.  {helpb svy} estimation commands, when string variables were used to
     identify strata or sampling units, incorrectly counted the corresponding
     number of strata or sampling units resulting in incorrect variance
     estimates.  This has been fixed.

{p 4 9 2}
{* fix}
15.  {helpb svyopts} did not work as advertised in the Stata Press book,
     {it:{browse "http://www.stata.com/bookstore/mle.html":Maximum Likelihood Estimation with Stata, 2nd edition}}
     (Gould, Pitblado, and Sribney 2003).  This has been fixed.

{p 9 9 2}
    Programmers interested in getting their estimation commands to work with
    the new {helpb svy} prefix command should see the {hi:Remarks} section of
    {help program properties}.

{p 4 9 2}
{* fix}
16.  {helpb "svy: regress"} and {helpb "svy: ivreg"} exited with an
     inappropriate error message when the sampling weights summed to a number
     less than 1.  Although this has been fixed, we note that the sum of the
     sampling weights should sum to the population size (or at least estimate
     the population size) for the DEFF design effect to be valid when using an
     FPC in the first stage.  DEFT is invariant to the scale of sampling
     weights.


{hline 8} {hi:update 25may2005} {hline}

    {title:Ado-files}

{p 5 9 2}
{* enhancement}
1.  {helpb asmprobit} help now documents that the options
    {cmd:basealternative()} and {cmd:scalealternative()} may contain a number,
    a label value, or a string.

{p 5 9 2}
{* fix}
2.  {helpb biplot} did not label row markers properly when the {cmd:mlabel()}
    option was specified and a subsample of the data was used.  This has been
    fixed.

{p 5 9 2}
{* enhancement}
3.  {helpb ca_postestimation##cabiplot:cabiplot} and
    {helpb ca_postestimation##caprojection:caprojection} now allow row and
    column marker labels to be specified.  This is accomplished with the
    {cmd:mlabel()} suboption of {cmd:rowopts()} and {cmd:colopts()}.

{p 5 9 2}
{* fix}
4.  {helpb estimates:estimates replay} failed when called under version 8.2
    version control.  This has been fixed.

{p 5 9 2}
{* fix}
5.  {helpb glm}, {helpb mdslong}, {helpb merge},
    {helpb mds_postestimation##predict:predict} after {helpb mds}, and
    {helpb rolling} failed when the temporary file directory path contained
    spaces.  This has been fixed.

{p 5 9 2}
{* fix}
6.  {helpb graph bar} and {helpb graph dot}, after the 04may2005 update,
    displayed an extra set of bars when the {cmd:sort} suboption was specified
    with the {cmd:over()} option.  This has been fixed.

{p 5 9 2}
{* fix}
7.  {helpb graph twoway} error messages have been improved when too many or
    too few variables are specified for a plot.

{p 5 9 2}
{* fix}
8.  {helpb icd9:icd9[p] check} with the {cmd:generate()} option did not
    generate a variable if all codes were valid.  This has been fixed.

{p 5 9 2}
{* enhancement}
9.  {helpb mprobit} and {helpb slogit} help now document that the option
    {cmd:baseoutcome()} may contain a number or a label value.

{p 4 9 2}
{* fix}
10.  {helpb nl} when used with the pre-version-9 syntax accepted but ignored
     the {cmd:robust}, {cmd:cluster()}, {cmd:hc2}, and {cmd:hc3} options.
     Now specifying one of those options with the old syntax will result in
     an error message.

{p 4 9 2}
{* fix}
11.  {helpb asmprobit_postestimation##predict:predict} failed after
     {helpb asmprobit} with a string {cmd:alternatives()} variable.  This has
     been fixed.

{p 4 9 2}
{* fix}
12.  {helpb xtreg_postestimation##predict:predict} after
     {helpb xtreg:xtreg, fe} produced the linear prediction when options such
     as {cmd:residuals} or {cmd:cooksd} were specified.  An error message is
     now issued instead.

{p 4 9 2}
{* fix}
13.  {helpb stepwise} incorrectly exited with an error for backward stepwise
     regression (when using options {opt pr()} and {opt pe()}).  This has been
     fixed.

{p 4 9 2}
{* fix}
14.  {helpb xtmixed} sometimes failed when given an {cmd:if} condition that
     used value labels, e.g., {cmd:if foreign == "Domestic":origin}.  This has
     been fixed.

{p 4 9 2}
{* fix}
15.  {helpb xttobit} erroneously treated observations for which the dependent
     variable was missing as being right-censored instead of ignoring them.
     This has been fixed.


{hline 8} {hi:update 04may2005} {hline}

    {title:Ado-files}

{p 5 9 2}
{* fix}
1.  {helpb asmprobit} models with unstructured covariances produced an error
    message during the reparameterization phase of the {helpb ml} optimization
    if the model definition for the variance/correlation parameters exceeded
    80 characters.  This has been fixed.

{p 5 9 2}
{* enhancement}
2.  {cmd:bootstrap} has a {opt group()} option, which was not documented in the
    manual.  See {helpb bootstrap} for the updated online documentation.

{p 5 9 2}
{* fix}
3.  {helpb clogit} performed an extra, but unnecessary, calculation, which
    caused it to run slowly on large datasets.  This has been fixed.

{p 5 9 2}
{* fix}
4.  {helpb clogit} uses {helpb logit} to generate starting values when none
    are specified.  For some models, these starting values resulted in
    nonsensical {cmd:clogit} log-likelihoods.  {cmd:clogit} now checks that
    starting values make sense before using them.

{p 5 9 2}
{* enhancement}
5.  {helpb graph bar} and {helpb graph dot} no longer require user-provided
    names when a variable is repeated with more than one statistic.  The
    dialogs for these commands have been redesigned to take advantage of the
    new, simplified syntax.

{p 5 9 2}
6.  When managing multiple graphs or renaming graphs when used from disk,
    Stata sometimes displayed the wrong graph in a Graph window.  This has
    been fixed.

{p 5 9 2}
{* fix}
7.  {helpb pause} executed subsequent commands under version 6 control.  It
    now executes them under the control of the caller's version.

{p 5 9 2}
{* fix}
8.  {helpb roctab} produced a different standard error for the area under the
    curve when a monotonic transformation was applied to the rating.  This has
    been fixed.

{p 5 9 2}
{* enhancement}
9.  {helpb screeplot} has been enhanced to produce scree plots automatically
    after {helpb manova}, {helpb canon}, {helpb ca}, and {helpb camat}.

{p 4 9 2}
{* fix}
10.  {helpb stsum} sometimes reported percentiles as the next ordered survival
     time when the desired fraction exactly equaled the empirically observed
     fraction.  This would be similar to saying that the median of (1,2,5,7,9)
     is 7 rather than 5.  {cmd:stsum} rarely made this mistake.  This problem
     has been fixed.

{p 4 9 2}
{* fix}
11.  {helpb svy} dropped observations from the estimation sample that had zero
     as the value of the sampling weight.  This has been fixed.

{p 4 9 2}
{* fix}
12.  {helpb xtreg:xtreg, be} and {cmd:xtreg, fe} reported a missing value for
     the model F statistic when the {cmd:vce(bootstrap)} option was specified.
     They now report the correct model Wald chi-squared statistic.

{p 4 9 2}
{* fix}
13.  {helpb xtreg:xtreg, re} reported a missing value for the model
     chi-squared statistic when the {cmd:vce(jackknife)} option was specified.
     {cmd:xtreg, re} now reports the correct model F statistic.


{hline 8} {hi:previous updates} {hline}

{p 9 9 2}
See {help whatsnew8to9}.

{hline}


{title:Also see}

    {c TLC}{hline 67}{c TRC}
    {c |} help file       contents                   dates                  {c |}
    {c LT}{hline 67}{c RT}
    {c |} {bf:this file}       Stata 9.0                  22apr2005 to present   {c |}
    {c |} {help whatsnew8to9}    Stata 9.0 new release      22apr2005              {c |}
    {c |} {help whatsnew8_2}     Stata 8.2                  28oct2003 to 24feb2005 {c |}
    {c |} {help whatsnew8_1}     Stata 8.1                  01jul2003 to 30sep2003 {c |}
    {c |} {help whatsnew8_0}     Stata 8.0                  14jan2003 to 25jun2003 {c |}
    {c |} {help whatsnew7to8}    Stata 8.0 new release      02jan2003              {c |}
    {c |} {help whatsnew7}       Stata 7.0                  08jan2001 to 05dec2002 {c |}
    {c |} {help whatsnew6to7}    Stata 7.0 new release      15dec2000              {c |}
    {c |} {help whatsnew6}       Stata 6.0                  13jan1999 to 20nov2000 {c |}
    {c BLC}{hline 67}{c BRC}

{psee}
Manual:  {bf:[U] 1.3 What's new},{break}
{bf:[U] 3 Resources for learning and using Stata},{break}
{bf:[U] 3.2 The http://www.stata.com web site},{break}
{bf:[U] 3.3 The http://www.stata-press.com web site},{break}
{bf:[U] 3.4 The Stata listserver},{break}
{bf:[U] 3.5 The Stata Journal and the Stata Technical Bulletin},{break}
{bf:[U] 3.6 Updating and adding features from the web},{break}
{bf:[U] 17.8 How do I install official updates?},{break}
{bf:[U] 28 Using the Internet to keep up to date}

{psee}
Online:  {helpb updates};
{helpb net},
{help sj}, {help stb},
{helpb update}, {help upunix},
and table above
{p_end}