Dienstag, 16.04.2019, 12-13 Uhr - Raum: W9-109David Winkelmann
An empirical analysis of the hot hand effect in shooting sports
The recent literature reveals the existence of a small but statistically evident hot hand effect in different sports. This effect refers to the presence of phases with exceptional high success probabilities of players. In this talk it is examined whether an analysis using continuous-valued state space models confirms the presence in shooting sports, which provides appropriate conditions for the analysis and was not investigated so far. The results of 74 players in 84 world cups for the air rifle and air pistol are considered. Modelling player heterogeneity by random effects and applying a Poisson as well as a negative binomial distribution to account for overdispersion in the data does not give rise to acknowledge the presence of a hot hand effect in this setting. Nevertheless, the structure of competitions and results as well as the small sample size, whereby the impact on the accuracy of an analysis is underlined by a simulation study, exacerbate the opportunity to detect the hot hand in this data set.
Dienstag, 30.04.2019,12-13 Uhr - Raum: W9-109
Prof. Dr. Jost Reinecke
Continuous Time Modeling of Panel Data: Applications with Mplus, OpenMX and ctsem
Panel studies, in which the same subjects are repeatedly observed at multiple time points, are among the most popular longitudinal designs in the social sciences. Meanwhile, there exists a wide range of different methods to analyze such data, for example autoregressive and cross-lagged models. Unfortunately, in these models time is only considered implicitly, making it difficult to account for unequally spaced measurement occasions or to compare parameter estimates across studies that are based on different time intervals. Stochastic differential equations offer a solution to this problem by relating the discrete time model to its underlying model in continuous time. It can be shown how continuous time parameters can be obtained via structural equation modeling. Two different empirical examples are used to illustrate the approach: One on the relationship between authoritarianism and anomia (GFE-study) and one on the relationship between Identity with Germany and Intention to stay in Germany (SOEP study).
Dienstag, 14.05.2019, 12-13 Uhr - Raum: W9-109
Dienstag, 28.05.2019, 12-13 Uhr - Raum: W9-109
We investigate the prevalence and sources of reporting errors in hypothesis tests in three top economic journals. Reporting errors are defined as inconsistencies between reported significance levels and statistical values such as coefficients and standard errors. We analyze 30,993 tests from 370 articles and find that 34% of the articles contain at least one reporting error. Survey responses from the respective authors, replications and regression analyses suggest some simple solutions to mitigate the prevalence of reporting errors in future research. Open data and software code policies in line with a vivid replication culture seem to be most important.
Dienstag, 11.06.2019, 12-13 Uhr - Raum: W9-109
Prof. Dr. Göran Kauermann
Institut für Statistik der Ludwig-Maximilians-Universität München
Statistical Modeling for (real) large scale data – Estimating price elasticity and segmenting customers based on airline booking data
We use booking data of a major European airline to estimate price elasticity of passengers. Massive booking data are available that allow to fit complex Poisson models for the demand of air travelling. We incorporate different customer segments in the model, which differ by their price elasticity. The data allow to estimate the composition of these customer segments, where the segment proportions changes over time and are different for the different departure times over the day. We also demonstrate the need to incorporate an instrumental variable to account for endogeneity effects. The problem tackled in the paper serves as blue print for applied statistical modeling in large scale data. It demonstrates the usefulness of statistical models, in particular generalized additive models, and emphasizes the necessity to incorporate instrumental variables. The talk also demonstrates the hurdles and requirements, which were necessary to turn the collaboration project with the airline to come to a successful end.
Dienstag, 25.06.2019, 12-13 Uhr - Raum: W9-109
J.-Prof. Dr. Nicola Bilstein
Dienstag, 09.07.2019, 12-13 Uhr - Raum: W9-109