Penalized likelihood methods for modeling count data

The paper considers parameter estimation in count data models using penalized likelihood methods. motivating consists of multiple independent variables with a moderate sample size per variable. were collected during the assessment oral reading fluency (ORF) school-aged children. A fourth-grade students given one ten available passages to read these differing length and difficulty. observed number words incorrectly (WRI) is used measure ORF. Three are considered for WRI scores, namely binomial, zero-inflated beta-binomial. We aim efficiently estimate passage difficulty, quantity expressed as function underlying model parameters. Two types penalty functions respective goals shrinking estimates closer zero or another. simulation study evaluates efficacy shrinkage Mean Square Error (MSE) metric. Big reductions MSE relative unpenalized maximum observed. concludes an analysis ORF data.