The seminar gives an introduction into statistical models with latent variables. Models with latent variables explicitly take measurement error into account, which cannot be avoided in psychological measurements, and thus make it possible to avoid many methodological problems that can be caused by the measurement error. First, an overview of the basic ideas of models with latent variables (factor analysis, item-response theory, latent class analysis, latent profile analysis) is given and it is shown which problems and methodological artifacts can be caused by not considering measurement error in data analysis. Students in the first semester learn how to define models of confirmatory factor analysis using the example of multimethod research and how to estimate the parameters of the model using the R-Packet lavaan and to prove the fit of a model. Using real data sets with multiple raters (self-report, parent report, peer report), the question of how validation studies can be conducted taking measurement error into account, how it can be verified whether the same psychological construct is measured by self- and other reports, and what is meant by method effects will be investigated. By giving an answer to these questions, students learn to apply and interpret both basal and more complex linear structural equation models. The further content of the seminar, especially for the second semester, will be determined together with the students. These can relate to the following topics, among others: (1) mediator analysis (2) multigroup analysis (e.g., comparative cultural psychology), (3) measurement of change with latent variables, (4) analysis of typologies (latent class analysis, latent profile analysis), (5) multidimensional test models (e.g., bifactor models), (6) models of item response theory. Apart from interest in psychometric issues, the seminar does not require in-depth knowledge of methods. The methods are learned and practiced step by step, and there are extensive opportunities to ask questions. The seminar in the first semester must be offered as an online seminar. Due to very positive experiences in the last semester, the seminar will have the following structure: 1. A teaching video will be posted every week at the latest on Tuesdays at 8:15 a.m., which will be available from then on independently of time. The video can, but does not have to be watched at this time. 2. Students receive a data set to practice the contents independently. 3. Four obligatory online seminar (Tuesday 8:30 - 10 am) take place. The first serves to get to know each other and to explain the procedure. Two further sessions (e.g. 6th and 11th session) serve to clarify open questions and problems (questions can always be asked in the discussion forum). The fourth session serves to conclude the seminar, in particular also to jointly determine the topics for the second semester, which will then hopefully be taught in a face-to-face session. Since the seminar will also be open to students of the module "Data Science in the Social Sciences" of the master program Data Science in the winter semester, the seminar will be held in English in the winter semester (but not in the summer semester). Questions can always be asked in German. close