My main current research interests lie in penalised likelihood methods, especially with applications to the improvement of the properties of the maximum likelihood estimators in multivariate generalised linear and non-linear models for discrete and categorical data.

Some other general areas I am interested on are pseudo and composite likelihood methods, asymptotic theory, distribution theory and simulation, and clustering methods and applications.

• Improved estimation in cumulative link models. Kosmidis, I. (2012). ArXiv e-prints, arXiv:1204.0105 (Accepted at the Journal of the Royal Statistical Society: Series B).
• Extended Beta Regression in R: Shaken, Stirred, Mixed, and Partitioned. Grün, B., Kosmidis, I. and Zeileis, A. (2012). Journal of Statistical Software, 48, 11.
• Multinomial logit bias reduction via Poisson log-linear model. Kosmidis, I., Firth, D. (2011). Biometrika, 98, 755-759.
• Simulating Events of Unknown Probabilities via Reverse Time Martingales. Latuszynski, K., Kosmidis, I., Papaspiliopoulos, O. and Roberts, G. O. (2011). Random Structures and Algorithms, 38, 441-452.
• A generic algorithm for reducing bias in parametric estimation. Kosmidis, I., Firth, D. (2010). Electronic Journal of Statistics, 4, 1097-1112. Relevant R Code and an example are available here .
• Supervised sampling for clustering large data sets. Kosmidis, I., Karlis, D. (2010). CRiSM working paper 10-10.
• On iterative adjustment of responses for the reduction of bias in binary regression models. Kosmidis, I. (2009). CRiSM working paper 09-36.
• Bias Reduction in Exponential Family Nonlinear Models. Kosmidis, I., Firth, D. (2009). Biometrika, 96, 793-804.
• The profileModel R package: Profiling objectives for models with linear predictors. Kosmidis, I. (2008). R News, 8/2, 12-18, R Foundation for Statistical Computing.

• Bias reduction in generalized nonlinear models . Joint Statistical Meetings 2009, Washington, DC, 2009.
• On iterative adjustment of the responses for the reduction of bias in binary regression models . 24th International Workshop on Statistical Modelling , Ithaka, NY, July 2009.
• Reduction of bias in exponential family models with emphasis on models for categorical responses . University of Padua, Italy, May 2009.
• Profiling the parameters of models with linear predictors . useR 2008, Dortmund, Germany, August 2008.
• Penalized likelihood for a non-linear Rasch model . 22nd International Workshop on Statistical Modelling, Barcelona, Spain, July 2007.

• Bias Reduction in Exponential Family Nonlinear Models (August 2007, Department of Statistics, University of Warwick)
• Errata