Dienstag, 27.10.2015, 12-13 Uhr - Raum: W9-109
Prof. Dr. Roland Langrock
Universität Bielefeld
Statistical modelling of temporal ecological processes
Observations of ecological processes often lead to data with temporal structure. Examples of such data include movement tracks, capture-recapture data, line transect surveys and regular censuses, to name but a few. A common feature of such data is that the observations (e.g. movement speeds) are driven by underlying non-observable states (e.g. behavioural states such as feeding or travelling), and in many cases it is reasonable to assume that there is a finite number of such states. This naturally leads to the consideration of state-switching models, in particular hidden Markov models (HMMs), for making inference on the corresponding ecological processes and their drivers. After motivating the consideration of HMMs, I will demonstrate how the associated powerful inferential machinery can be exploited to make inference tractable in fairly complex scenarios, including several extensions of the basic model formulation tailored to the particular ecological problem at hand. I will demonstrate the usefulness of these models in ecology by sketching several real data applications featuring, inter alia, blue whales, sheep, white sharks, elk and the legendary wild haggis animal. While the talk will focus on ecological problems, most of the methods discussed are in fact much more widely applicable.
Dienstag, 10.11.2015, 12-13 Uhr - Raum: W9-109
Prof. Dr. Fridtjof Nussbeck
Universität Bielefeld
Multitrait Multimethod Analysis
Psychological measures have to fulfil the three main criteria of objectivity, reliability and validity. Objectivity and reliability are well defined and already well operationalized concepts. Inspecting the validity of a psychological measure is more tedious and there is no single coefficient indicating wether a measure is valid or not.
In order to inspect the validity of measures Campbell and Fiske introduced the Multitrait-Multimethod (MTMM) analysis, which allows for the inspection of convergent and discriminant validity. Modern MTMM models are based on Confirmatory Factor Analysis (CFA) and additionally allow for the specification of a theoretical measurement model separating true-scores from measurement error and for a model based decision if the supposed model fits the data. In this presentation, the base rationale of Campbell and Fiske will be introduced and modern approaches of MTMM models will be presented.
Dienstag, 24.11.2015, 12-13 Uhr - Raum: W9-109
Christian Heinze
Universität Bielefeld
An oblique projection approach to consumer price level prediction
Consumer price index data suitable for spatial comparison at a high resolution is scarce in many countries but of great interest to regional economists. We focus on German counties and present predictions of suitable consumer price indexes for the years 1993–2012. For this case, the available price index data comprise a small cross-section for 1993 and annual inflation rates for a subset of the German states. In a first step, we construct an auxiliary panel of pseudo price indexes in form of predictions from a low-complexity regression model. Therein, parameter estimates are based on the above mentioned data and corresponding covariate observations. Secondly, the pseudo index panel is used to estimate a VAR transition matrix - using nuclear norm regularization to enforce a low rank - alongside an innovation covariance matrix. Finally, the implied spatio-temporal covariance structure allows Kridge-type prediction - implemented as Kalman filtering - of the missing price indexes based on the available price indexes. This approach amounts to replacing the covariances of an (almost) unobserved process by those of a related and observed (pseudo-)process. This has an interpretation as an oblique projection. Therefore, the angles between certain subspaces characterize its inferiority to the corresponding
orthogonal projection. We also present an expression in terms of the two processes.
Dienstag, 08.12.2015, 12-13 Uhr - Raum: W9-109
Dr. Christian Schellhase
Universität Bielefeld
D-Vines Estimation for Mixed Data using Penalized Splines
The paper presents the estimation of nonparametric vines for mixed data using penalized B-splines, extending the nonparametric estimation of pair-copula construction in Kauermann and Schellhase (2014, Flexible Pair-Copula Estimation in D-vines using Bivariate Penalized SplinesStatistics and Computing) to support also ordinary variables. We estimate each pair-copula density in a flexible, that is semi-parametric way by refraining from any strong parametric assumptions on the structure of the pairs, using penalized B-splines, as spline bases in each knot of the vine throughout each level. The estimation of probability mass functions in the context of copulas is easily done using constant B-splines, whose nodes are placed at the jump points of the empirical cumulative distribution function of some ordinary data. The independence of the constant B-splines from each other ensures the applicability. The entire approach allows to fit log-linear graphical models. The estimation of distributions for mixed data sets is especially interesting for calculating counterfactual distributions. As application we present results of an analysis of the rent survey of Munich (Schellhase, C., Schnurbus J. and Kauermann, G., 2015, Mixed copulae for Munich rent survey, working paper), calculating counterfactual distributions analysing whether the rent increase for German major cities is merely a matter of rising demand that is exploited by flat owners. First, the rent increase caused by an improvement of the flats and second, the increase in terms of equivalent flats simply getting more expensive.
Dienstag, 12.01.2016, 12-13 Uhr - Raum: W9-109
Prof. Dr. Christian Deutscher
Universität Bielefeld
Intermediate Information, Loss Aversion and Effort: Empirical Evidence
The present study empirically explores the impact of intermediate information on contestants’ effort. Data involving substituted soccer players of the German Bundesliga indicate only weak evidence of a negative effect of ex ante heterogeneity on effort; in contrast, intermediate information, measured by goal difference at the time of substitution, significantly affects effort. Players exert the greatest effort when their team is leading by one goal and reduce their effort when it is trailing. When intermediate information suggests the contest is already decided, players from both teams reduce effort. This behavior reflects loss aversion, such that players weight potential losses more than potential gains and adjust their effort accordingly.
Dienstag, 26.01.2016, 12-13 Uhr - Raum: W9-109
Aktuelle Forschungsbereiche im ZeSt