1 item has been added to your cart.

In this talk, I introduce the margins command, used to estimate the partial effects at the mean and the mean of the partial effects. I discuss both the Stata syntax and the underlying statistical methods of margins. I also briefly discuss factor variables.

Additional information
margins_sp.pdf

La visualización de resultados exige a menudo la preparación de una variedad tablas y/o graficos. Sin embargo, el proceso de realizar estas tablas de resultados a mano puede ser laborioso, tomándonos mucho tiempo, y vulnerable al error. Presentaremos como combinar de una manera sencilla diversos comandos oficiales de Stata, que combina el uso eficiente de bucles sobre items (foreach) para guardar las estimaciones disponibled después de cualquier análisis con Stata (postestimation) en matrices de resultados (matrix). Mostramos, a través de diversos ejemplos, las ventajas de la automatización de estas tareas así como la flexibilidad para guardar, listar y graficar los resultados de cualquier análisis.

Additional information
presentacion_1_tobias_bleda.ppt

La representación gráfica es, a menudo, la manera más conveniente de presentar los datos. El objetivo de los gráficos estadísticos es mostrar las características básicas, sustituyendo en ocasiones a las tablas y constituyendo por sí mismos una poderosa herramienta para el análisis. A pesar de haber experimentado una mejora sustancial en la elaboración de los gráficos, existen figuras que en Stata requieren un manejo de datos y una sintaxis que en ocasiones puede ser complicada. Se presentan a continuación dos ado-files que realizan gráficos muy útiles para la presentación de variables cuantitativas y que son muy usados en inmunología:

bubbles realiza un scatter plot de los valores observados en varios grupos, con los símbolos ponderados por la frecuencia. Tiene opciones para mostrar la mediana y el IQR o la media (aritmética o geométrica) y el 95%CI, para mostrar el test de compa ración de los grupos y para manipulación de la leyenda.

recdf (reverse empirical cumulative distribution function) realiza gráficos de distribución de los valores en los grupos estudiados, opcionalmente estratificando por otra variable. Tiene opciones para mostrar el p-valor de la comparación de los grup os y realizar modificaciones requeridas por el usuario.

Additional information
presentacion_2_bubbles_recdf.zip

Se presenta la macro kappa2 que implementa dos generalizaciones del índice kappa que permiten ponderar y usar una definición explícita de acuerdo para múltiples observadores usando clasificaciones multinomiales y diseños incompletos según la prop uesta de Abraira (1997, Tesis Doctoral, http://www.ucm.es/eprints/4168/) y Abraira y Pérez de Vargas (1999, Qüestilló 23: 561–571). En el caso particular de no usar ponderación, kappa2 se reduce al comando kappa de Stata.

kappa2 calcula el índice kappa cuando existen dos, o más, evaluadores o evaluaciones, usando clasificaciones bi- o multinomiales. Cada evaluación es una variable en el archivo de datos. Para facilitar el uso de este comando, su sintaxis es una extensión de la del comando kappa, incorporando algunas opciones nuevas, como majority para indicar que calcule el índice kappa para una definición de acuerdo por mayoría, o jackknife para pedir que se estime su error estándar y el intervalo de confianza al 95% por el método de jackknife.

Additional information
presentacion_3_kappa2.zip
presentacion_3_zamora_abraira_lazaro.pptx

Spatial analysis is gaining more importance in social sciences. Though spatial analysis is still developing in Stata, several contributions have been made in recent years. In this presentation, I briefly describe the current state of the art of spatial tools in Stata, and then I describe several implementations that are complementary to other standard social-sciences software. In particular, I present a data-management tool for using spatial matrices under Stata, Geoda, R, and Matlab; a procedure to build LISA maps and LISA significance maps using Stata; and last, some improvements to Maurizio Pisati's article “Tools for spatial data analysis” (2001, Stata Technical Bulletin 60: 21–37), to use spatial econometric techniques for cross-section regression analysis.

In this presentation, I illustrate the potentiality of these tools using a classic example from the literature: crime data in Columbus, Ohio, on 49 contiguous planning neighborhoods (L. Anselin, 1988, Spatial Econometrics: Methods and Models [Kluwer Academic]).

Additional information
presentacion_4_ramos_royuela.ppt

Air quality is one of the major concerns in big cities, and so the proper evaluation of air pollution is important. We will explain how air quality is incorporated in transaction prices in downtown Madrid. For that purpose, we use multilevel models because our sample is hierarchically organized into three levels. The first level consists of 5,080 house prices; the second level consists of 759 census tracts; and the third level consists of 43 neighborhoods. We have variables available for each level, individual characteristics for the first level, and various socioeconomic data for the second and third levels. First, we combine a set of noise and air pollutants measured at several monitoring stations available for each census tract. Second, we apply kriging to match the monitoring station records to the census data. Third, we estimate hedonic models to measure the marginal willingness to pay for air quality in downtown Madrid. While the conventional approach to estimate hedonic models is to use ordinary least squares, we exploit the hierarchical nature of our data and instead estimate multilevel models. Multilevel models allow for a more reliable statistical inference.

Additional information
presentacion_4_spatial.zip

A common feature of financial time series is their strong persistence. Yet long memory may just be the spurious effect of either structural breaks or slow switching regimes. So far, five testing procedures have been proposed to distinguish between true and spurious long memory. The tests are constructed under the null of true long memory versus the alternative of spurious long memory due to level shifts or breaks. We compare the tests in an extensive Monte Carlo experiment and analyze their overall performances. Finally, we study credit and credit default swap (CDS) spreads, and we find evidence of genuine long memory.

This paper is our attempt to answer the question of whether households in Brazil with higher lifetime income save a larger fraction of their income. The Brazilian Family Budget survey (POF) provides us with rich and unexplored data with which we can construct different lifetime income estimates. Our empirical analysis, using least-absolute-deviation estimators, suggests that the estimated relationship between saving rates and lifetime incomes is sensitive to the instrument used to proxy lifetime income. Nevertheless, when it is controlled for preference for savings, our estimatives indicate that the poorer households save more than the rich ones.

Additional information
presentacion_5_henrique_gomes_salvato.pptx

In this half-day class, I introduce the econometric analysis of linear panel-data models using Stata. A general familiarity with Stata is assumed. I present an overview of the econometric/statistical theory and show how to use Stata to perform the discussed methods.

The topics I will cover are the following:

1. Panel data giveth and panel data taketh away
2. Estimators for linear models that require random effects
3. Fixed-effect estimators for linear models
4. Instrumental-variables estimators
5. The Hausman–Taylor estimator
6. Models with endogenous variables
7. Dynamic panel-data models

The class will be in lecture format with participants asking questions as the class progresses. The class will go from 2:30 PM–6:30 PM with a break from 4:30 PM to 4:45 PM. The slides will be in English, but the presentation will be in Spanish.