Le Centre pour l’étude de la citoyenneté démocratique est heureux d’organiser:
7e École d’hiver de méthodologie du CÉCD
Olivier Jacques, Panel data analysis*
* l’atelier sera donné en français
Quand: 9 février 2021, de 13h à 15h.
Inscriptions: si vous souhaitez participer à un ou plusieurs des ateliers de l’École d’hiver, vous devez vous inscrire. Veuillez vous inscrire ici au moins 2 jours avant l’atelier.
À propos: The workshop will address issues in modelling time-series cross-sectional data, otherwise known as panel datasets. This type of data is very common in political science as they involve that the data has multiple temporal observations for each units of analysis (multiple individuals overserved overtime, multiple countries observed overtime, etc.). The workshop will first deal with the issue of whether the data can be “pooled”, meaning that we can ignore the fact that some observations are gathered from the same unit (country, individual). Then it will introduce the concept of random and fixed effects to control for unexplained unit-level differences in outcomes across units.
Then, the workshop will move towards dealing with stickiness overtime and the fact that the data in one year almost certainly depends on the data in the year before (autocorrelation), which is the main issue in time series analysis. This will involve giving tips to detect autocorrelation of the error term as well as fixing and modelling autocorrelation, notably with the use dynamic models with lagged dependent variables. Finally, the workshop will the deal with the issue of stationarity in the time series and show how to detect a unit root time series, as regressing two-unit root series together often involves a spurious regression. I will introduce the concept of first-differencing the data to deal with the unit root process
Programme complet et détails sur les autres ateliers de l’École d’hiver ici.