Stata 15 help for whatsnew11to12

What's new in release 12.0 (compared with release 11)

This file lists the changes corresponding to the creation of Stata release 12.0:

+---------------------------------------------------------------+ | help file contents years | |---------------------------------------------------------------| | whatsnew Stata 15.0 and 15.1 2017 to present | | whatsnew14to15 Stata 15.0 new release 2017 | | whatsnew14 Stata 14.0, 14.1, and 14.2 2015 to 2017 | | whatsnew13to14 Stata 14.0 new release 2015 | | whatsnew13 Stata 13.0 and 13.1 2013 to 2015 | | whatsnew12to13 Stata 13.0 new release 2013 | | whatsnew12 Stata 12.0 and 12.1 2011 to 2013 | | this file Stata 12.0 new release 2011 | | whatsnew11 Stata 11.0, 11.1, and 11.2 2009 to 2011 | | whatsnew10to11 Stata 11.0 new release 2009 | | whatsnew10 Stata 10.0 and 10.1 2007 to 2009 | | whatsnew9to10 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 whatsnew12.

--- Stata 12.0 release 25jul2011 ----------------------------------------------

Remarks

We will list all the changes, item by item, but first, here are the highlights.

What's new (highlights)

Here are the highlights. There is more, and do not assume that because we mention a category, we have mentioned everything new in the category. Detailed sections follow the highlights.

1. Automatic memory management, which means that you no longer have to set memory and never again will you be told that there is no room because you set too little! Stata automatically adjusts its memory usage up and down according to current requirements.

The memory manager is tunable. We recommend the default settings. See [D] memory if you are interested.

Old do-files can still set memory. Stata merely responds, "set memory ignored".

2. Structural equation modeling (SEM), via the new sem command, is itself the subject of the new Stata Structural Equation Modeling Reference Manual. SEM fits multivariate linear models that can include observed and latent variables. These models include confirmatory factor analysis, linear models, instrumental variables, 2SLS, 3SLS, multivariate regression, seemingly unrelated least squares, recursive systems, simultaneous systems, path analysis, latent variables, MIMIC, modeling of direct and indirect effects, and more. All the above can be estimated by maximum likelihood with or without missing values, GLS, or ADF (asymptotic distribution free, also known as GMM). Missing values are handled using FIML. Raw and standardized coefficients and effects are reported. All models may be fit across groups and include tests for group invariance. Modification indices and score tests are provided.

Models may be specified and reported using commands or interactive path diagrams. See [SEM] sem.

3. MI, multiple imputation,

a. Chained equations, which is to say, fully conditional specifications for imputing missing values given arbitrary patterns for continuous, binary, ordinal, cardinal, or count variables. See [MI] mi impute chained.

b. Four new imputation methods. You can impute

1) truncated data, 2) interval-censored data, 3) count data, and 4) overdispersed count data.

See [MI] mi impute truncreg, [MI] mi impute intreg, [MI] mi impute poisson, and [MI] mi impute nbreg.

c. Conditional imputation is now supported by all univariate imputation methods, which is to say, you can impute values for variables with restrictions, such as the number of pregnancies being imputed only for females, even if female itself is imputed. See Conditional imputation in [MI] mi impute and new option conditional() in the univariate imputation entries such as [MI] mi impute regress.

d. Imputation by groups, which is to say, imputations can be made separately for different groups of the data. See new option by() in [MI] mi impute.

e. Imputation by drawing posterior estimates from bootstrapped samples. See new option bootstrap in the univariate imputation entries such as [MI] mi impute regress.

f. Panel-data and multilevel models are now supported by mi estimate. Included are xtcloglog, xtgee, xtlogit, xtmelogit, xtmepoisson, xtmixed, xtnbreg, xtpoisson, xtprobit, xtrc, and xtreg. See [MI] estimation.

g. Linear and nonlinear predictions after MI estimation using new commands mi predict and mi predictnl. See [MI] mi predict.

h. Monte Carlo jackknife error estimates obtained by omitting one imputation at a time and reapplying the combination rules. See new option mcerror in [MI] mi estimate.

4. Longitudinal/panel data,

a. Survey feature support for xtmixed including multilevel sampling weights and robust variance estimators. See [XT] xtmixed.

b. Documentation for xtmixed, xtmelogit, and xtmepoisson has been modified to adopt the standard "level" terminology from the literature on hierarchical models. See the Introduction section of Remarks in both [XT] xtmixed and [XT] xtmelogit.

c. xtmixed now uses maximum likelihood (ML) as the default method of estimation. See [XT] xtmixed.

5. Contour plots. Filled and outlined plots are available. See [G-2] graph twoway contour and [G-2] graph twoway contourline.

6. Contrasts, which is to say, tests of linear hypotheses involving factor variables and their interactions from the most recently fit model, and that model can be virtually any model that Stata can fit. Tests include ANOVA-style tests of main effects, simple effects, interactions, and nested effects. Effects can be decomposed into comparisons with reference categories, comparisons of adjacent levels, comparisons with the grand mean, and more. See [R] contrast and [R] margins, contrast.

7. Pairwise comparisons of means, estimated cell means, estimated marginal means, predictive margins of linear and nonlinear responses, intercepts, and slopes. In addition to ANOVA-style comparisons, comparisons can be made of population averages. See [R] pwmean, [R] pwcompare, and [R] margins, pwcompare.

8. Graphs of margins, marginal effects, contrasts, and pairwise comparisons. Margins and effects can be obtained from linear or nonlinear (for example, probability) responses. See [R] marginsplot.

9. Time series,

a. MGARCH, which is to say, multivariate GARCH, which is to say, estimation of multivariate generalized autoregressive conditional heteroskedasticity models of volatility, and this includes constant, dynamic, and varying conditional correlations, also known as the CCC, DCC, and VCC models. Innovations in these models may follow multivariate normal or Student's t distributions. See [TS] mgarch.

b. UCM, which is to say, unobserved-components models, also known as structural time-series models that decompose a series into trend, seasonal, and cyclical components, and which were popularized by Harvey (1989). See [TS] ucm.

c. ARFIMA, which is to say, autoregressive fractionally integrated moving-average models, useful for long-memory processes. See [TS] arfima.

d. Filters for extracting business and seasonal cycles. Four popular time-series filters are provided: the Baxter-King and the Christiano-Fitzgerald band-pass filters, and the Butterworth and the Hodrick-Prescott high-pass filters. See [TS] tsfilter.

10. Business dates allow you to define your own calendars so that they display correctly and lags and leads work as they should. You could create file lse.stbcal that recorded the days the London Stock Exchange is open (or closed) and then Stata would understand format %tblse just as it understands the usual date format %td. Once you define a calendar, Stata deeply understands it. You can, for instance, easily convert between %tblse and %td values. See [D] datetime business calendars.

11. Improved documentation for date and time variables. Anyone who has ever been puzzled by Stata's date and time variables, which is to say, anyone who uses them, should see [D] datetime, [D] datetime translation, and [D] datetime display formats.

12. ROC adjusted for covariates, which is to say, you can model the ROC curve and obtain coefficients, standard errors, and graphs. Nonparametric and parametric estimation is supported. See [R] rocreg and [R] rocregplot.

13. Survey SDR weights, which is to say, successive difference replicate weights, which are supplied with many datasets from the U.S. Census Bureau. See [SVY] svy sdr.

14. Bootstrap standard errors for survey data using user-supplied bootstrap replicate weights. See [SVY] svy bootstrap.

15. Importing and exporting,

a. Excel files. And the new import preview tool lets you see the data before you import them. See [D] import excel.

b. EBCDIC files, importing. And you can convert between EBCDIC and ASCII formats; see [D] infile (fixed format) and [D] filefilter.

c. ODBC connection strings. See [D] odbc.

d. PDF export for graphs and logs lets you directly create PDFs from your Stata results. See [G-2] graph export and [R] translate.

16. Renaming groups of variables is now easy using rename's new syntax that is 100% compatible with its old syntax. You can change names, swap names, renumber indices within variable names, and more. See [D] rename group.

17. Stata interface,

a. New layout is wider and fits most screens better.

b. New Properties window lets you manage the properties of your variables including names, labels, value labels, notes, display formats, and storage types. And you can manage the properties of your dataset.

c. Filtering of Review and Variables windows lets you type text and see only the matches.

d. Searching in the Results window lets you find results.

e. Expression Builder now accesses parameter estimates, returned results, macros, and more, so you can build expressions for nlcom and testnl. It is worth a test drive.

f. Unified interface for Mac means no more lost windows; all the Stata windows are tied together.

g. Gesture support for Mac makes changing font sizes and moving forward and backward easy.

h. Tabbed graphs for Mac.

i. File drag-and-drop for Windows -- Stata for Mac already had it -- now Stata for Windows does, too.

18. Data Editor,

a. New tool for managing variables lets you hide/show variables (and includes filtering!), sort variables, and reorder variables via drag and drop. And it includes Stata's new Properties tool, so you can manage your data more easily from the Data Editor. Try it. Click on the Variables tool in the toolbar.

b. New Clipboard Preview Tool lets you see the data before you paste them into Stata and lets you control how the data will be pasted.

c. Clipboard preserves variable properties such as display formats and types when you copy-and-paste data within Stata.

19. All-new Viewer,

a. Quick access to dialogs, sections, and "also see" references via three pulldown menus at the top of the Viewer for quick navigation inside help files.

b. Tabbed Viewer lets you open multiple help files and documents and switch between them.

20. Do-file Editor,

a. Tabbed for Mac and Unix -- Stata for Windows already had it -- now Stata for Mac and Unix do, too.

b. Syntax highlighting and bookmarks for Mac -- Stata for Windows already had it -- now Stata for Mac does, too.

21. Estimation output improved,

a. Baseline odds now shown, which is to say, the exponentiated intercept is displayed by logistic and by logit with option or. In fact, all estimation commands show exponentiated intercepts when option eform() or its equivalent is specified. For example, poisson shows the baseline incidence rate when option irr is specified.

b. Implied zero coefficients now shown. When a coefficient is omitted, it is now shown as being zero and the reason it was omitted -- collinearity, base, empty -- is shown in the standard-error column. (The word "omitted" is shown if the coefficient was omitted because of collinearity.)

c. You can set displayed precision for all values in coefficient tables using set cformat, set pformat, and set sformat. Or you may use options cformat(), pformat(), and sformat() now allowed on all estimation commands. See [R] set cformat and [R] estimation options.

d. Estimation commands now respect the width of the Results window. This feature may be turned off by new display option nolstretch. See [R] estimation options.

e. You can now set whether base levels, empty cells, and omitted are shown using set showbaselevels, set showemptycells, and set showomitted. See [R] set showbaselevels.

22. More MP speed ups, meaning faster execution for those running Stata/MP.

a. Improved MP support for factor variables used in estimation. Execution is much faster when there are lots of levels.

b. Faster maximum likelihood execution with large numbers of covariates. Processors being assigned on the basis of variables rather than observations when there are 300 or more covariates results in improved performance.

c. Improved performance on 16 or more cores due to better tuning.

d. Up to 64 cores now supported, up from 32.

23. Installation Qualification is now provided by a new tool which you download for free. The tool produces a report for submission to regulatory agencies such as the FDA to establish that Stata is installed correctly. Visit http://www.stata.com/support/installation-qualification/.

What's new in the GUI and command interface

1. Highlights, a. New layout. b. New Properties window. c. Filtering of Review and Variable windows. d. Searching in the Results window. e. Expression Builder can access parameter estimates, .... f. Unified interface for Mac. g. Gesture support for Mac. h. Tabbed Graphs for Mac. i. File drag-and-drop for Windows. j. Data Editor, new tool for managing variables. k. Data Editor, new Clipboard Preview Tool. l. Data Editor, Clipboard preserves variable properties. m. New Viewer, quick access to dialogs, sections, .... n. New Viewer, tabbed. o. Do-file Editor, tabbed for Mac and Unix. p. Do-file Editor, syntax highlighting and bookmarks for Mac. See What's new (highlights).

2. Aero Snap functionality for Viewer in Windows 7.

3. Stata for Unix dialog boxes now have full varlist controls.

What's new in data management

1. Highlights, a. Automatic memory management. See [D] memory.

b. Excel files, importing and exporting. See [D] import excel.

c. EBCDIC files, importing. See [D] infile (fixed format) and [D] filefilter.

d. ODBC connection strings, importing and exporting. See [D] odbc.

e. PDF files, exporting of graphs and logs. See [R] translate.

f. Business dates. See [D] datetime business calendars.

g. Improved documentation for date and time variables. See [D] datetime, [D] datetime translation, and [D] datetime display formats.

h. Renaming groups of variables. See [D] rename group.

See What's new (highlights).

2. New functions,

a. Tukey's Studentized range, cumulative and inverse, tukeyprob() and invtukeyprob().

b. Dunnett's multiple range, cumulative and inverse, dunnettprob() and invdunnettprob().

c. New date conversion functions dofb() and bofd() convert between business dates and standard calendar dates. See [D] datetime business calendars.

See [FN] Functions by category.

3. ODBC support for Oracle Solaris. See [D] odbc.

4. New Stata commands getmata and putmata make it easy to transfer your data into Mata, manipulate them, and then transfer them back to Stata. getmata and putmata are especially designed for interactive use. See [D] putmata.

5. New Stata commands import sasxport, export sasxport, and import sasxport, describe replace existing commands fdause, fdasave, and fdadescribe. fdause, fdasave, and fdadescribe are understood as synonyms. See [D] import sasxport.

6. xshell support for Mac. See [D] shell.

What's new in statistics (general)

1. Highlights, a. Contrasts. See [R] contrast and [R] margins, contrast.

b. Pairwise comparisons. See [R] pwmean, [R] pwcompare, and [R] margins, pwcompare.

c. Graphs of margins, marginal effects, contrasts, .... See [R] marginsplot.

d. ROC adjusted for covariates. See [R] rocreg and [R] rocregplot.

e. Estimation output improved: --Baseline odds now shown. --Implied zero coefficients now shown. --You can set displayed precision. See [R] set cformat and [R] estimation options. --Estimation commands now respect the width of the Results window. See [R] estimation options. --You can now set whether base levels, empty cells, and omitted are shown. See [R] set showbaselevels and [R] estimation options.

See What's new (highlights).

2. test with coefficient names not using _b[] notation is now allowed, even when the specified variables no longer exist in the current dataset. See [R] test.

3. areg now faster. areg is orders of magnitude faster when there are hundreds of absorption groups, even if you are not running Stata/MP. See [R] areg.

4. misstable summarize will now create summary variable recording the missing-values pattern. See new option generate() for summarize in [R] misstable.

5. margins command supports contrasts. See [R] margins, contrast and [R] contrast.

6. sfrancia uses better algorithm. sfrancia now uses an algorithm based on the log transformation for approximating the sampling distribution of the W' statistic for testing normality. The old algorithm, using the Box-Cox transformation, is available under version control or via the new boxcox option. Based on simulation, the new algorithm is more powerful for sample sizes greater than 1,000 and is comparable to the old algorithm for sample sizes less than 1,000. Also, similarly to swilk, sfrancia now allows you to suppress the treatment of ties when option noties is used. See [R] swilk.

7. logistic now allows option noconstant. See [R] logistic.

8. Probability predictions now available. predict after count-data models, such as poisson and nbreg, can now predict the probability of any count or count range. See [R] nbreg postestimation, [R] poisson postestimation, [R] tnbreg postestimation, [R] tpoisson postestimation, [R] zinb postestimation, and [R] zip postestimation.

9. Truncated count-data models now available. New estimation commands tpoisson and tnbreg fit models of count-data outcomes with any form of left truncation, including truncation that varies observation by observation. These new commands supersede ztp and ztnb. See [R] tpoisson and [R] tnbreg.

10. cnsreg checks for collinear variables prior to estimation and has new option collinear, which keeps the collinear variables instead of omitting them. The old behavior of always keeping collinear variables is preserved under version control. See [R] cnsreg.

11. ml improved,

a. ml now distinguishes the Hessian matrix produced by technique(nr) from the other techniques that compute a substitute for the Hessian matrix. This means that ml will compute the real Hessian matrix of second derivatives to determine convergence when all other convergence tolerances are satisfied and technique(bfgs), technique(bhhh), or technique(dfp) is in effect.

The old behavior was to use the nrtolerance() value with the H matrix associated with the technique() currently in effect to determine convergence; this behavior is preserved under version control.

b. ml has new option qtolerance() that distinguishes itself from nrtolerance() when technique(bfgs), technique(bhhh), or technique(dfp) is specified. Option qtolerance() replaces nrtolerance() when technique(bfgs), technique(bhhh), or technique(dfp) is in effect.

See [R] ml and [R] maximize.

12. margins has new option estimtolerance() for setting tolerance used to determine estimable functions. See [R] margins.

13. Option addplot() now places added graphs above or below. Commands that allow option addplot() can now place the added plots above or below the command's plots.

What's new in statistics (longitudinal/panel data)

1. Highlights,

a. MI support for panel-data and multilevel models including xtcloglog, xtgee, xtlogit, xtmelogit, xtmepoisson, xtmixed, xtnbreg, xtpoisson, xtprobit, xtrc, and xtreg. See [MI] estimation.

b. Survey feature support for linear multilevel models, xtmixed, including multilevel sampling weights and robust variance estimators. See [XT] xtmixed.

c. Documentation for xtmixed, xtmelogit, and xtmepoisson has been modified to adopt the standard "level" terminology from the literature on hierarchical models. For example, what in previous Stata versions was considered a one-level model is now called a two-level model with the observations now being counted as "level one"; see the Introduction section of Remarks in both [XT] xtmixed and [XT] xtmelogit for more details.

d. Contrasts available after most xt commands. See [R] contrast and [R] margins, contrast.

e. Pairwise comparisons available after most xt estimation commands. See [R] pwcompare and [R] margins, pwcompare.

f. Graphs of margins, marginal effects, contrasts, and pairwise comparisons available after all xt estimation commands. See [R] marginsplot.

g. xtmixed now uses maximum likelihood (ML) as the default method of estimation, where previously it used restricted maximum likelihood (REML). REML is still available with the reml option, and previous behavior is preserved under version control.

h. Estimation output improved. --Baseline odds now shown. --Implied zero coefficients now shown. --You can set displayed precision. See [R] set cformat and [R] estimation options. --Estimation commands now respect the width of the Results window. See [R] estimation options. --You can now set whether base levels, empty cells, and omitted are shown. See [R] set showbaselevels and [R] estimation options.

See What's new (highlights).

2. Robust and cluster-robust SEs after fixed-effects xtpoisson. See [XT] xtpoisson.

3. New residual covariance structures for multilevel models include exponential, banded, and Toeplitz. See [XT] xtmixed.

4. Probability predictions now available. predict after random-effects and population-averaged count-data models, such as xtpoisson and xtgee, can now predict the probability of any count or count range. See [XT] xtpoisson postestimation, [XT] xtgee postestimation, and [XT] xtnbreg postestimation.

5. Option addplot() now places added graphs above or below. Commands that allow option addplot() can now place the added plots above or below the command's plots. Affected is the command xtline; see [XT] xtline.

What's new in statistics (time series)

1. Highlights,

a. MGARCH. See [TS] mgarch.

b. UCM. See [TS] ucm.

c. ARFIMA. See [TS] arfima.

d. Filters for extracting business and seasonal cycles. See [TS] tsfilter.

e. Business dates. See [D] datetime business calendars.

f. Improved documentation for date and time variables. See [D] datetime, [D] datetime translation, and [D] datetime display formats.

g. Contrasts available after many time-series estimation commands. See [R] contrast and [R] margins, contrast.

h. Pairwise comparisons available after many time-series estimation commands. See [R] pwcompare and [R] margins, pwcompare.

i. Graphs of margins, marginal effects, contrasts, and pairwise comparisons available after most time-series estimation commands. See [R] marginsplot.

j. Estimation output improved. --Implied zero coefficients now shown. --You can set displayed precision. See [R] set cformat and [R] estimation options. --Estimation commands now respect the width of the Results window. See [R] estimation options. --You can now set whether base levels, empty cells, and omitted are shown. See [R] set showbaselevels and [R] estimation options.

See What's new (highlights).

2. Spectral densities from parametric models via new postestimation command psdensity lets you estimate using arfima, arima, and ucm and then obtain the implied spectral density. See [TS] psdensity.

3. dvech renamed mgarch dvech. The command for fitting the diagonal VECH model is now named mgarch dvech, and innovations may follow multivariate normal or Student's t distributions. See [TS] mgarch.

4. Loading data from Haver Analytics supported on all 64-bit Windows. See [TS] haver.

5. Option addplot() now places added graphs above or below. Graph commands that allow option addplot() can now place the added plots above or below the command's plots. Affected by this are the commands corrgram, cumsp, pergram, varstable, vecstable, wntestb, and xcorr.

What's new in statistics (survey)

1. Highlights,

a. Contrasts available after survey estimation. See [R] contrast and [R] margins, contrast.

b. Pairwise comparisons available after survey estimation. See [R] pwcompare and [R] pwcompare postestimation.

c. Graphs of margins, marginal effects, contrasts, and pairwise comparisons available after survey estimation. See [R] marginsplot.

d. Survey SDR weights. See [SVY] svy sdr.

e. Bootstrap standard errors for survey data. See [SVY] svy bootstrap.

f. Estimation output improved. --Baseline odds now shown. --Implied zero coefficients now shown. --You can set displayed precision. See [R] set cformat and [R] estimation options. --Estimation commands now respect the width of the Results window. See [R] estimation options. --You can now set whether base levels, empty cells, and omitted are shown. See [R] set showbaselevels and [R] estimation options.

See What's new (highlights).

2. Survey estimation may be combined with new SEM for structural equation modeling. See [SVY] svy estimation and [SEM] sem.

3. Survey goodness-of-fit available after logistic, logit, and probit with new command estat gof. See [SVY] estat.

4. Survey coefficient of variation (CV) available with new command estat cv. See [SVY] estat.

What's new in statistics (survival analysis)

1. Highlights,

a. Contrasts available after stcox, stcrreg, and streg. See [R] contrast and [R] margins, contrast.

b. Pairwise comparisons available after stcox, stcrreg, and streg. See [R] pwcompare and [R] margins, pwcompare.

c. Graphs of margins, marginal effects, contrasts, and pairwise comparisons available after stcox, stcrreg, and streg. See [R] marginsplot.

d. Estimation output improved. --Implied zero coefficients now shown. --You can set displayed precision. --Estimation commands now respect the width of the Results window. See [R] set cformat and [R] estimation options. --You can now set whether base levels, empty cells, and omitted are shown. See [R] set showbaselevels and [R] estimation options.

See What's new (highlights).

2. Gönen and Heller's K concordance coefficient available after Cox proportional hazards estimation. K is robust to censoring. See new option gheller for estat concordance in [ST] stcox postestimation.

3. Option addplot() now places added graphs above or below. Graph commands that allow option addplot() can now place the added plots above or below the command's plots. Affected by this are the commands ltable, stci, stcoxkm, stcurve, stphplot, strate, sts graph, and tabodds.

What's new in statistics (multivariate)

1. Highlights,

a. Structural equation modeling (SEM). See [SEM] sem.

b. Contrasts. See [R] contrast and [R] margins, contrast.

c. Pairwise comparisons. See [R] pwcompare and [R] margins, pwcompare.

d. Graphs of margins, marginal effects, contrasts, and pairwise comparisons. See [R] marginsplot.

See What's new (highlights).

2. Option addplot() now places added graphs above or below. Graph commands that allow option addplot() can now place the added plots above or below the command's plots. Affected by this are the commands screeplot and cluster dendrogram; see [MV] screeplot and [MV] cluster dendrogram.

What's new in statistics (multiple imputation)

1. Highlights,

a. Chained equations. See [MI] mi impute chained.

b. Four new imputation methods. See [MI] mi impute truncreg, [MI] mi impute intreg, [MI] mi impute poisson, and [MI] mi impute nbreg.

c. Conditional imputation. See Conditional imputation in [MI] mi impute and new option conditional() in the univariate imputation entries such as [MI] mi impute regress.

d. Imputation by groups. See new option by() in [MI] mi impute.

e. Imputation by drawing posterior estimates from bootstrapped samples. See new option bootstrap in the univariate imputation entries such as [MI] mi impute regress.

f. Panel-data and multilevel models are now supported. Included are xtcloglog, xtgee, xtlogit, xtmelogit, xtmepoisson, xtmixed, xtnbreg, xtpoisson, xtprobit, xtrc, and xtreg. See [MI] mi estimation.

g. Linear and nonlinear predictions after MI estimation. See [MI] mi estimate postestimation.

h. Monte Carlo jackknife error estimates. See new option mcerror in [MI] mi estimate.

i. Estimation output improved. --Baseline odds now shown. --Implied zero coefficients now shown. --You can set displayed precision. See [R] set cformat and [R] estimation options. --Estimation commands now respect the width of the Results window. See [R] estimation options. --You can now set whether base levels, empty cells, and omitted are shown. See [R] set showbaselevels and [R] estimation options.

See What's new (highlights).

2. Handling of perfect prediction during imputation of categorical data using logit, ologit, and mlogit. See The issue of perfect prediction during imputation of categorical data in [MI] mi impute and see new option augment in [MI] mi impute logit, [MI] mi impute ologit, and [MI] mi impute mlogit.

3. Faster imputation. mi impute no longer secretly converts to flongsep and back again.

4. mi estimate now supports total. See [MI] estimation.

5. misstable summarize will now create summary variables recording the missing-values pattern. See new option generate() for summarize in [R] misstable. Note that mi misstable does not have this new option. The new option is useful before data are imputed.

6. mi estimate and mi estimate using now use a small-sample adjustment when computing fractions of missing information and, subsequently, when computing relative efficiencies when the specified estimation command provides complete-data degrees of freedom. Before, these statistics were always computed assuming a large sample. Fractions of missing information and relative efficiencies are reported when the vartable option is used. The old behavior is available under version control.

7. mi impute monotone retains in the imputation model imputation variables that do not contain missing values in the imputation sample. Before, mi impute monotone omitted such variables from the imputation model, assuming independence between the variables being imputed and the variables being omitted. The old behavior is available under version control.

What's new in graphics

1. Highlights,

a. Graphs of margins, marginal effects, contrasts, .... See [R] marginsplot.

b. Contour plots. See [G-2] graph twoway contour and [G-2] graph twoway contourline.

c. PDF export for graphs and logs lets you directly create PDFs from your Stata graphs. See [G-2] graph export and [R] translate.

See What's new (highlights).

2. Time-series operators now supported by twoway lfit, twoway lfitci, twoway qfit, and twoway qfitci. See [G-2] graph twoway lfit, [G-2] graph twoway lfitci, [G-2] graph twoway qfit, and [G-2] graph twoway qfitci.

3. Graphs of marginal and covariate-specific ROC curves. New command rocregplot plots the fitted ROC curve after rocreg. See [R] rocregplot.

4. Option addplot() now places added graphs above or below. Graph commands that allow option addplot() can now place the added plots above or below the command's plots.

What's new in programming

1. Stored results r() and e() can be marked hidden or historical, which means they do not show when the user types return list or ereturn list unless the user also specifies option all. See [P] return.

2. Estimation commands now store in r() as well as e(). r() values are stored at estimation time and after replaying. Stored are

a. r(level), a scalar containing the confidence level for the CIs.

b. r(label#), a macro containing the label displayed with the #th coefficient, such as "(base)", "(omitted)", or "(empty)".

c. r(table), a matrix containing all the data displayed in the coefficient table. The matrix is the coefficient table, transposed; each column contains coefficients and associated statistics. To understand the matrix, do the following:

. sysuse auto, clear . regress mpg weight displ . matrix list r(table)

See [P] ereturn.

3. ereturn display offers new options for controlling the look of the coefficient table.

a. Options noomitted, vsquish, noemptycells, baselevels, and allbaselevels control row spacing and display of omitted variables and base and empty cells.

b. Formatting display options cformat(%fmt), pformat(%fmt), and sformat(%fmt) control the formats of numbers in the coefficient table.

c. ereturn display now respects the width of the Results window. This feature may be turned off by new display option nolstretch.

See [R] estimation options.

4. Matrices can be in tables with equation names only using new options coleqonly and roweqonly. See [P] matlist.

5. matrix accum allows option absorb() to accumulate deviations from the mean within groups. See [P] matrix accum.

6. Version control for random-number generators is now determined when the seed is set, not when the generator function is used; see [P] version. New creturn result c(version_rng) records the version number currently in effect for random-number generators; see [P] creturn.

7. fvrevar has new option stub(), which generates stub+index variables rather than temporary variables. See [R] fvrevar.

8. mprobit now posts base outcome equation to e(b). See [R] mprobit.

9. Default time for network timeouts was reduced. timeout1 has been reduced from 120 seconds to 30, and timeout2 has been reduced from 300 seconds to 180. See [R] netio.

What's new in Mata

1. New Stata commands getmata and putmata make it easy to transfer your data into Mata, manipulate them, and then transfer them back to Stata. getmata and putmata are especially designed for interactive use. See [D] putmata.

2. New functions imported from Stata,

a. Tukey's Studentized range, cumulative and inverse, tukeyprob() and invtukeyprob().

b. Dunnett's multiple range, cumulative and inverse, dunnettprob() and invdunnettprob().

c. New date conversion functions dofb() and bofd() convert between business dates and standard calendar dates. See [D] datetime business calendars.

See [FN] Functions by category, [M-5] normal(), and [M-5] date().

3. Support for hidden and historical saved results. Existing Mata functions st_global(), st_numscalar(), and st_matrix() now allow an optional third argument specifying the hidden or historical status. Three new functions -- st_global_hcat(), st_numscalar_hcat(), st_matrix_hcat() -- allow you to determine the saved hidden or historical status. See [M-5] st_global(), [M-5] st_numscalar(), and [M-5] st_matrix().

4. Support for new ml features. Stata's ml now distinguishes the Hessian matrix produced by technique(nr) from the other techniques that compute a substitute for the Hessian matrix. This means that ml will compute the real Hessian matrix of second derivatives to determine convergence when all other convergence tolerances are satisfied and technique(bfgs), technique(bhhh), or technique(dfp) is in effect.

Mata's commands optimize() and moptimize() have been similarly changed. See [M-5] optimize() and [M-5] moptimize().

What's more

We have not listed all the changes, but we have listed the important ones.

Stata is continually being updated, and those updates are available for free over the Internet. All you have to do is type

. update query

and follow the instructions.

To learn what has been added since this manual was printed, select Help > What's New? or type

. help whatsnew

We hope that you enjoy Stata 12.

Reference

Harvey, A. C. 1989. Forecasting, Structural Time Series Models and the Kalman Filter. Cambridge: Cambridge University Press.

-------- previous updates -----------------------------------------------------

See whatsnew11.

-------------------------------------------------------------------------------


© Copyright 1996–2018 StataCorp LLC   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index