Home  /  Stata Conferences  /  2020 Switzerland

The third Swiss Stata Conference took place virtually on 19 November 2020.

Proceedings

10:30–11:00 randregret: A command for fitting random regret minimization models using Stata Abstract: In this presentation, we describe the randregret command, which implements a variety of random regret minimization (RRM) models. The command allows the user to apply the classic RRM model (Chorus 2010), the generalized RRM model (Chorus 2014), and also the mu-RRM and pure RRM models (Van Cranenburgh, Guevara, and Chorus 2015).
... (Read more)
We illustrate the usage of the randregret command using stated choice data on route preferences. The command offers robust and cluster standard-error correction using analytical expressions of the score functions. It also offers likelihood ratio tests, which can be used to assess the relevance of a given model specification. Finally, predicted probabilities from each model can be easily computed using the randregretpred postestimation command.

References:

Chorus, C. G. 2010. A new model of random regret minimization. European Journal of Transport and Infrastructure Research 10(2).

Chorus, C. G. 2014. A generalized random regret minimization model. Transportation Research Part B: Methodological 68: 224–238.

Van Cranenburgh, S., C. A. Guevara, and C. G. Chorus. 2015. New insights on random regret minimization models. Transportation Research Part A: Policy and Practice 74: 91–109.

Contributors:
Michel Meulders
Martina Vandebroek
KU Leuven
(Read less)

Additional information:
Switzerland20_Gutiérrez-Vargas.pdf

Álvaro A. Gutiérrez Vargas
KU Leuven
11:00–11:30 How to assess the fit of multilevel logit models with Stata? A project in progress Abstract: Stata 16 includes the multilevel model for binary (melogit) and ordinal logits (meologit). Unfortunately, except for the global Wald test of the estimated fixed effects, both models do not provide any fit measure to assess its practical significance. Therefore, I developed an ado-file to calculate McFadden's and McKelvey and Zavoina's pseudo-R². It estimates the
... (Read more)
intraclass correlation (ICC) of the dependent variable for the actual sample to assess the maximum of the contextual effect, too. Since the early 90s, a lot of Monte Carlo simulation studies (Hagle and Mitchell 1992; Veall and Zimmermann 1992, 1993, 1994); Windmeijer 1995; DeMaris 2002) proved that McKelvey and Zavoina's pseudo-R² is the best one to assess the fit of binary and ordinal logit models. My ado-file calculates this fit measure in two complementary ways: for the fixed effects only and for the fixed and random effects together. The estimation of McFadden's pseudo-R² uses two different zero models: the random-intercept-only model (RIOM) knowing the contextual units and the fixed-intercept-only model (FIOM) ignoring the contextual units completely. For each of them, it calculates the global likelihood-ratio χ² test statistic 1 whether all fixed effects or all fixed and random effects are zero in the population. It performs two global likelihood-ratio χ² tests for all fixed effects and all fixed and random effects using a probability weight for each level. An empirical study of drug consumption in European countries demonstrates the usefulness of my fit_meologit_2lev.ado or fit_meologit_3lev.ado files for multilevel binary and ordinal logit models.
(Read less)

Additional information:
Switzerland20_Langer.pdf

Wolfgang Langer
Martin Luther Universität Halle-Wittenberg
11:30–12:00 xtbreak: estimating and testing break points in panel data in Stata Abstract: I present the new community-contributed command xtbreak, which allows the empirical analysis of multiple structural changes in panel data. The panels may be strongly cross-sectionally dependent, where cross-sectional dependence takes the form of a multifactor structure in the errors. The number of break points and their dates are treated as unknown.
... (Read more)
The command includes the following two hypothesis tests. First, a test of the null hypothesis of no breaks against a user-specified number of breaks. Second, a test of the null hypothesis of k numbers of breaks against the alternative of k+1 number of breaks. Once the number of breaks has been specified, the command allows the estimation of the break dates and the construction of confidence intervals for these dates. The syntax and options of xtbreak are explained. Examples for the estimation of the breakpoints and testing those are given.
(Read less)

Additional information:
Switzerland20_Ditzen.pdf

Jan Ditzen
Heriot Watt University
1:00–1:30 acreg: Arbitrary correlation regression Abstract: In this presentation, we present acreg, a new Stata command that implements the standard error correction proposed in Colella et al. (2019). This command computes regression coefficients' standard errors, accounting for dependence across arbitrary correlation structures in a flexible way. Arbitrary here refers to the way units are correlated with
... (Read more)
each other: we impose no restrictions so that our approach can be used with a wide range of data. In particular, this command allows the estimation of OLS and 2SLS coefficient correcting standard errors across units in three environments: in a spatial setting using objects' coordinates or distance between units, in a network setting starting from the adjacency matrix, and in a multi-way cluster framework taking multiple clustering variables as input. This command suits both cross-sectional and panel databases. Distance and time cutoffs can be specified by the user and linear decay in time and space are also optional.

Contributors:
Rafael Lalive
Seyhun Orcan Sakalli
Mathias Thoenig
Université de Lausanne
(Read less)

Additional information:
Switzerland20_Colella,pdf

Fabrizio Colella
Université de Lausanne
1:30–2:00 A new implementation of relative distribution methods in Stata Abstract: A relative distribution is defined as the distribution of the relative ranks that the outcomes of one distribution take on in another distribution. An example is the relative positions that women's wages take on in the distribution of men's wages. In this presentation, I will discuss a new implementation of relative distribution methods in Stata. The new
... (Read more)
command called reldistcan be used to estimate and plot the relative density function (relative PDF), a histogram of the relative distribution, or the relative distribution function (relative CDF), or to estimate polarization indices and descriptive statistics of the relative data. The command also supports decomposition of the relative distribution by adjusting differences in location, scale, and shape or differences in covariate distributions (using reweighting). Standard errors are estimated based on influence functions and support complex survey data.
(Read less)

Additional information:
Switzerland20_Jann.pdf

Ben Jann
Universität Bern
2:00–2:30 Quantile and distribution regression in Stata: Algorithms, pointwise and functional inference Abstract: This presentation and the related commands offer fast estimation and inference procedures for the linear quantile and distribution regression models. First, qrprocess and drprocess implement new algorithms that are much quicker than the built-in Stata commands, especially when a large number of regressions or bootstrap replications must be
... (Read more)
estimated. Second, the commands provide analytical estimates of the variance–covariance matrix of the coefficients for several regressions allowing for weights, clustering, and stratification. Third, in addition to traditional pointwise confidence intervals, these commands also provide functional confidence bands and tests of functional hypotheses. Fourth, predict—called after either qrprocess or drprocess—can generate monotone estimates of the conditional quantile and distribution functions obtained by rearrangement. Fifth, plotprocess conveniently plots the estimated coefficients with their confidence intervals and uniform bands.
(Read less)

Additional information:
Switzerland20_Melly.pdf

Blaise Melly
Universität Bern
3:00–3:30 arhomme: A Stata implementation of the Arrellano/Bonhomme (2017) estimator for quantile regression with selection correction Abstract: Despite constituting a major theoretical breakthrough, the quantile selection model of Arrellano and Bonhomme (2017, Econometrica, Vol. 85, pp. 1–28) based on copulas has not found its way into many empirical applications. We introduce the command arhomme, which implements different variants of the estimator along with standard errors based on subsampling. We illustrate the command by replicating the empirical application in the original article and a related application in Arrellano and Bonhomme (2017, Handbook of Quantile Regression, Ch. 13).
... (Read more)

Contributor:
Pascal Erhardt
University of Tübingen
(Read less)

Additional information:
Switzerland20_Biewen.pdf

Martin Biewen
University of Tübingen
3:30–4:00 Conditional likelihood models for distributional regression analysis Abstract: This presentation illustrates the use of conditional likelihood models for "distributional regression" analysis. Conditional likelihood models impose specific parametric functional forms for an outcome variable of interest—for example, normality, log-normality, Gamma distribution, or more elaborate shapes—but allow the shape parameters to
... (Read more)
vary with covariates; for example, assuming normality and allowing both the mean and the standard deviation to vary with a vector of explanatory variables. Drawing on examples from income distribution analyses, I show how such models are fit in Stata, relying on a range of community-contributed commands, and how predictions and counterfactual simulations for both the conditional and the marginalized (unconditional) distributions can be easily obtained from the fitted models. I also discuss the pros and cons of such conditional likelihood models in comparisons with quantile regression or distribution regression approaches.
(Read less)

Additional information:
Switzerland20_Van_Kerm.pdf

Phillippe Van Kerm
Universität von Luxemburg
4:00–4:30 Econodistplot: distribution box plots, economist-style Abstract: I present a new Stata package called conodistplot, which produces simple box plots showing median and interquantile ranges defaulting to the 10% and 90% percentiles. These plots have been introduced by The Economist (www.economist.com/united-states/2019/06/29/willtransparent-pricing-make-americas-health-care-cheaper)
... (Read more)
and slightly adapted by Bob Rudis for the R platform (cinc.rud.is/w eb/packages/ggeconodist). The package presented here reproduces the look of Rudis' graphs, with the motivation of creating an easily legible and visually appealing data summary.
(Read less)

Additional information:
Switzerland20_Gamma.pdf

Alex Gamma
Universität Zürich
5:00–6:00 Introduction to Bayesian Statistics Using Stata Abstract: Bayesian analysis has become a popular tool for many statistical applications. Yet many data analysts have little training in the theory of Bayesian analysis and software used to fit Bayesian models. This presentation will provide an intuitive introduction to the concepts of Bayesian analysis and will demonstrate how to fit Bayesian models using Stata.
... (Read more)
No prior knowledge of Bayesian analysis is necessary. Some specific topics I will cover are the relationship between likelihood functions, prior distributions, and posterior distributions; Markov chain Monte Carlo (MCMC) using the Metropolis—Hastings algorithm; and how to use Stata's bayes prefix to fit Bayesian models.
(Read less)

Additional information:
Switzerland20_Sánchez.pdf

Gustavo Sánchez
StataCorp
6:00–6:30
Open panel discussion with Stata developers
StataCorp

Scientific committee

Ben Jann
Universität Bern
Blaise Melly
Universität Bern

Logistics organizer

The logistics organizer for the 2020 Swiss Stata Conference is Universität Bern and Ritme, scientific solutions, the official distributor of Stata in Belgium and Switzerland.

View the proceedings of previous Stata Conferences and Users Group meetings.