The Stata Journal: Promoting communications on statistics and Stata

Hazard ratios can be approximated by data extracted from published Kaplan–Meier curves. Recently, this curve approach has been extended beyond hazard-ratio approximation with the capability of constructing time-to-event data at the individual level. In this article, we introduce a command, ipdfc, to implement the reconstruction method to convert Kaplan–Meier curves to time-to-event data. We give examples to illustrate how to use the command.

Hazard ratios can be approximated by data extracted from published Kaplan–Meier curves. Recently, this curve approach has been extended beyond hazard-ratio approximation with the capability of constructing time-to-event data at the individual level. In this article, we introduce a command, ipdfc, to implement the reconstruction method to convert Kaplan–Meier curves to time-to-event data. We give examples to illustrate how to use the command.

In this article, I present a counterfactual model identifying average treatment effects by conditional mean independence when considering peer- or neighborhood-correlated effects, and I provide a new command, ntreatreg, that implements such models in practical applications. The model and its accompanying command provide an estimation of average treatment effects when the stable unit treatment-value assumption is relaxed under specific conditions. I present two instructional applications: the first is a simulation exercise that shows both model implementation and ntreatreg correctness; the second is an application to real data, aimed at measuring the effect of housing location on crime in the presence of social interactions. In the second application, results are compared with a no-interaction setting.

In this article, I present a counterfactual model identifying average treatment effects by conditional mean independence when considering peer- or neighborhood-correlated effects, and I provide a new command, ntreatreg, that implements such models in practical applications. The model and its accompanying command provide an estimation of average treatment effects when the stable unit treatment-value assumption is relaxed under specific conditions. I present two instructional applications: the first is a simulation exercise that shows both model implementation and ntreatreg correctness; the second is an application to real data, aimed at measuring the effect of housing location on crime in the presence of social interactions. In the second application, results are compared with a no-interaction setting.

The synthetic control methodology (Abadie and Gardeazabal, 2003, American Economic Review 93: 113–132; Abadie, Diamond, and Hainmueller, 2010, Journal of the American Statistical Association 105: 493–505) allows for a data-driven approach to small-sample comparative studies. synth_runner automates the process of running multiple synthetic control estimations using synth. It conducts placebo estimates in space (estimations for the same treatment period but on all the control units). Inference (p-values) is provided by comparing the estimated main effect with the distribution of placebo effects. It also allows several units to receive treatment, possibly at different time periods. It allows automatic generation of the outcome predictors and diagnostics by splitting the pretreatment into training and validation portions. Additionally, it provides diagnostics to assess fit and generates visualizations of results.

The synthetic control methodology (Abadie and Gardeazabal, 2003, American Economic Review 93: 113–132; Abadie, Diamond, and Hainmueller, 2010, Journal of the American Statistical Association 105: 493–505) allows for a data-driven approach to small-sample comparative studies. synth_runner automates the process of running multiple synthetic control estimations using synth. It conducts placebo estimates in space (estimations for the same treatment period but on all the control units). Inference (p-values) is provided by comparing the estimated main effect with the distribution of placebo effects. It also allows several units to receive treatment, possibly at different time periods. It allows automatic generation of the outcome predictors and diagnostics by splitting the pretreatment into training and validation portions. Additionally, it provides diagnostics to assess fit and generates visualizations of results.

In this article, we review methodologies to fix the size distortions of tests for forecast evaluation in the presence of instabilities. The methodologies implement tests for relative and absolute forecast evaluation that are robust to instabilities. We also introduce the giacross and rosssekh commands, which implement these procedures in Stata.

In this article, we review methodologies to fix the size distortions of tests for forecast evaluation in the presence of instabilities. The methodologies implement tests for relative and absolute forecast evaluation that are robust to instabilities. We also introduce the giacross and rosssekh commands, which implement these procedures in Stata.

Text mining is the process of turning free text into numerical variables and then analyzing them with statistical techniques. We introduce the command ngram, which implements the most common approach to text mining, the “bag of words”. An n-gram is a contiguous sequence of words in a text. Broadly speaking, ngram creates hundreds or thousands of variables, each recording how often the corresponding n-gram occurs in a given text. This is more useful than it sounds. We illustrate ngram with the categorization of text answers from two open-ended questions.

Text mining is the process of turning free text into numerical variables and then analyzing them with statistical techniques. We introduce the command ngram, which implements the most common approach to text mining, the “bag of words”. An n-gram is a contiguous sequence of words in a text. Broadly speaking, ngram creates hundreds or thousands of variables, each recording how often the corresponding n-gram occurs in a given text. This is more useful than it sounds. We illustrate ngram with the categorization of text answers from two open-ended questions.

In this article, I introduce a new package with five commands to perform econometric convergence analysis and club clustering as proposed by Phillips and Sul (2007, Econometrica 75: 1771–1855). The logtreg command performs the log t regression test. The psecta command implements the clustering algorithm to identify convergence clubs. The scheckmerge command conducts the log t regression test for all pairs of adjacent clubs. The imergeclub command tries to iteratively merge adjacent clubs. The pfilter command extracts the trend and cyclical components of a time series of each individual in panel data. I provide an example from Phillips and Sul (2009, Journal of Applied Econometrics 24: 1153–1185) to illustrate the use of these commands. Additionally, I use Monte Carlo simulations to exemplify the effectiveness of the clustering algorithm.

In this article, I introduce a new package with five commands to perform econometric convergence analysis and club clustering as proposed by Phillips and Sul (2007, Econometrica 75: 1771–1855). The logtreg command performs the log t regression test. The psecta command implements the clustering algorithm to identify convergence clubs. The scheckmerge command conducts the log t regression test for all pairs of adjacent clubs. The imergeclub command tries to iteratively merge adjacent clubs. The pfilter command extracts the trend and cyclical components of a time series of each individual in panel data. I provide an example from Phillips and Sul (2009, Journal of Applied Econometrics 24: 1153–1185) to illustrate the use of these commands. Additionally, I use Monte Carlo simulations to exemplify the effectiveness of the clustering algorithm.

In this article, we review some recent advances in testing for serial correlation, provide code for implementation, and illustrate this code's application to market risk forecast evaluation. We focus on the classic and widely used portmanteau tests and their data-driven versions. These tests are simple to implement for two reasons: First, the researcher does not need to specify the order of the tested autocorrelations, because the test automatically chooses this number. Second, its asymptotic null distribution is chi-squared with one degree of freedom, so there is no need to use a bootstrap procedure to estimate the critical values. We illustrate the wide applicability of this methodology with applications to forecast evaluation for market risk measures such as value-at-risk and expected shortfall.

In this article, we review some recent advances in testing for serial correlation, provide code for implementation, and illustrate this code's application to market risk forecast evaluation. We focus on the classic and widely used portmanteau tests and their data-driven versions. These tests are simple to implement for two reasons: First, the researcher does not need to specify the order of the tested autocorrelations, because the test automatically chooses this number. Second, its asymptotic null distribution is chi-squared with one degree of freedom, so there is no need to use a bootstrap procedure to estimate the critical values. We illustrate the wide applicability of this methodology with applications to forecast evaluation for market risk measures such as value-at-risk and expected shortfall.

Empirical econometric research often requires implementation of nonlinear models whose regressors include one or more endogenous variables—regressors that are correlated with the unobserved random component of the model. In such cases, conventional regression methods that ignore endogeneity will likely produce biased results that are not causally interpretable. Terza, Basu, and Rathouz (2008, Journal of Health Economics 27: 531–543) discuss a relatively simple estimation method (two-stage residual inclusion) that avoids endogeneity bias, is applicable in many nonlinear regression contexts, and can easily be implemented in Stata. In this article, I offer a step-by-step protocol to implement the two-stage residual inclusion method in Stata. I illustrate this protocol in the context of a real-data example. I also discuss other examples and pertinent Stata code.

Empirical econometric research often requires implementation of nonlinear models whose regressors include one or more endogenous variables—regressors that are correlated with the unobserved random component of the model. In such cases, conventional regression methods that ignore endogeneity will likely produce biased results that are not causally interpretable. Terza, Basu, and Rathouz (2008, Journal of Health Economics 27: 531–543) discuss a relatively simple estimation method (two-stage residual inclusion) that avoids endogeneity bias, is applicable in many nonlinear regression contexts, and can easily be implemented in Stata. In this article, I offer a step-by-step protocol to implement the two-stage residual inclusion method in Stata. I illustrate this protocol in the context of a real-data example. I also discuss other examples and pertinent Stata code.

Terza (2016b, Health Services Research 51: 1109–1113) gives the correct generic expression for the asymptotic standard errors of statistics formed as sample means of nonlinear data transformations. In this article, I assess the performance of the Stata margins command as a relatively simple alternative for calculating such standard errors. I note that margins is not available for all packaged nonlinear regression commands in Stata and cannot be implemented in conjunction with user-defined-and-coded nonlinear estimation protocols that do not make a predict command available. When margins is available, however, I establish (using a real-data example) that it produces standard errors that are asymptotically equivalent to those obtained from the formulations in Terza (2016b) and the appendix available with this article. This result favors using margins (with its relative coding simplicity) when available. In all other cases, use Mata to code the standard-error formulations in Terza (2016b). I discuss examples, and I give corresponding Stata do-files in appendices.

Terza (2016b, Health Services Research 51: 1109–1113) gives the correct generic expression for the asymptotic standard errors of statistics formed as sample means of nonlinear data transformations. In this article, I assess the performance of the Stata margins command as a relatively simple alternative for calculating such standard errors. I note that margins is not available for all packaged nonlinear regression commands in Stata and cannot be implemented in conjunction with user-defined-and-coded nonlinear estimation protocols that do not make a predict command available. When margins is available, however, I establish (using a real-data example) that it produces standard errors that are asymptotically equivalent to those obtained from the formulations in Terza (2016b) and the appendix available with this article. This result favors using margins (with its relative coding simplicity) when available. In all other cases, use Mata to code the standard-error formulations in Terza (2016b). I discuss examples, and I give corresponding Stata do-files in appendices.

Obtaining the routing distance between two addresses should not be a hassle with current technology. Unfortunately, this is more complicated than it first seems. Recently, several commands have been implemented for this purpose (traveltime, traveltime3, mqtime, osrmtime), but most of them became obsolete only a few months after their introduction or appear complicated to use. In this article, we introduce the community-contributed command georoute, which retrieves travel distance and travel time between two points defined either by their addresses or by their geographical coordinates. Compared with other existing commands, it is simple to use, efficient in terms of computational speed, and versatile regarding the information that can be provided as input.

Obtaining the routing distance between two addresses should not be a hassle with current technology. Unfortunately, this is more complicated than it first seems. Recently, several commands have been implemented for this purpose (traveltime, traveltime3, mqtime, osrmtime), but most of them became obsolete only a few months after their introduction or appear complicated to use. In this article, we introduce the community-contributed command georoute, which retrieves travel distance and travel time between two points defined either by their addresses or by their geographical coordinates. Compared with other existing commands, it is simple to use, efficient in terms of computational speed, and versatile regarding the information that can be provided as input.

With the development of large and long panel databases, the theory surrounding panel causality evolves quickly, and empirical researchers might find it difficult to run the most recent techniques developed in the literature. In this article, we present the community-contributed command xtgcause, which implements a procedure proposed by Dumitrescu and Hurlin (2012, Economic Modelling 29: 1450–1460) for detecting Granger causality in panel datasets. Thus, it constitutes an effort to help practitioners understand and apply the test. xtgcause offers the possibility of selecting the number of lags to include in the model by minimizing the Akaike information criterion, Bayesian information criterion, or Hannan–Quinn information criterion, and it offers the possibility to implement a bootstrap procedure to compute p-values and critical values.

With the development of large and long panel databases, the theory surrounding panel causality evolves quickly, and empirical researchers might find it difficult to run the most recent techniques developed in the literature. In this article, we present the community-contributed command xtgcause, which implements a procedure proposed by Dumitrescu and Hurlin (2012, Economic Modelling 29: 1450–1460) for detecting Granger causality in panel datasets. Thus, it constitutes an effort to help practitioners understand and apply the test. xtgcause offers the possibility of selecting the number of lags to include in the model by minimizing the Akaike information criterion, Bayesian information criterion, or Hannan–Quinn information criterion, and it offers the possibility to implement a bootstrap procedure to compute p-values and critical values.

In this article, we present response surface coefficients for a large range of quantiles of the Elliott, Rothenberg, and Stock (1996, Econometrica 64: 813–836) unit-root tests, for different combinations of number of observations, T, and lag order in the test regressions, p, where the latter can either be specified by the user or be endogenously determined. The critical values depend on the method used to select the number of lags. We present the command ersur and illustrate its use with an empirical example that tests the validity of the expectations hypothesis of the term structure of interest rates.

In this article, we present response surface coefficients for a large range of quantiles of the Elliott, Rothenberg, and Stock (1996, Econometrica 64: 813–836) unit-root tests, for different combinations of number of observations, T, and lag order in the test regressions, p, where the latter can either be specified by the user or be endogenously determined. The critical values depend on the method used to select the number of lags. We present the command ersur and illustrate its use with an empirical example that tests the validity of the expectations hypothesis of the term structure of interest rates.

The calibration belt is a graphical approach designed to evaluate the goodness of fit of binary outcome models such as logistic regression models. The calibration belt examines the relationship between estimated probabilities and observed outcome rates. Significant deviations from the perfect calibration can be spotted on the graph. The graphical approach is paired to a statistical test, synthesizing the calibration assessment in a standard hypothesis testing framework. In this article, we present the calibrationbelt command, which implements the calibration belt and its associated test in Stata.

The calibration belt is a graphical approach designed to evaluate the goodness of fit of binary outcome models such as logistic regression models. The calibration belt examines the relationship between estimated probabilities and observed outcome rates. Significant deviations from the perfect calibration can be spotted on the graph. The graphical approach is paired to a statistical test, synthesizing the calibration assessment in a standard hypothesis testing framework. In this article, we present the calibrationbelt command, which implements the calibration belt and its associated test in Stata.

Receiver operating characteristic (ROC) curves are an established method for assessing the predictive capacity of a continuous biomarker for a binary outcome. However, in some cases, outcomes are time dependent. Although the literature has proposed packages for performing ROC analysis of time-independent outcomes, a package is not yet available for analyzing the predictive capacity of continuous biomarkers when the binary outcome is time dependent. In this article, we present stroccurve, a new command for performing ROC analysis within a survival framework.

Receiver operating characteristic (ROC) curves are an established method for assessing the predictive capacity of a continuous biomarker for a binary outcome. However, in some cases, outcomes are time dependent. Although the literature has proposed packages for performing ROC analysis of time-independent outcomes, a package is not yet available for analyzing the predictive capacity of continuous biomarkers when the binary outcome is time dependent. In this article, we present stroccurve, a new command for performing ROC analysis within a survival framework.