Analyzing tourists’ evaluations using item response models

Abstract

Item Response Theory is the statistical study, which focuses mainly on test scoring and the improvement of the test items. It determines the respondent’s probability of rating an item in relation to some considerations, such as the respondent’s ability, known as the trait level, and the item’s difficulty, item discrimination and guessing. Generally, it is used in the area of psychometrics, where there is a mathematical connection between the item responses, item difficulty and individual ability. The responses are usually either in binary or ordinal form, and these are used in connection with the individual test items to estimate the specifications mentioned before. Dichotomous Item Response Models have been developed to cater for two-category responses. The Rasch Model or One-Parameter Logistic model is the simplest model which establishes the probability of answering an item, with a specific difficulty, correctly by a person having a particular trait level. The Two-Parameter Logistic Model is a generalization of the Rasch model, where items are allowed to vary both in difficulty and ability. In addition, the Three-Parameter Logistic model is a generalization of the Two-Parameter Logistic Model, which includes also a guessing parameter. When the rating responses of items have more than two categories, Multichotomous IRT Models are used to analyze the data. The most common models are the Rating Scale Model (RSM) and the Partial Credit Model (PCM). A tourist questionnaire comprising eleven items measured on a 4-point Likert scale will be used as a data collection tool. The items describe factors which influence tourists when choosing to visit a particular destination. These include personal safety and security, accessibility, cleanliness, accommodation quality, climate conditions, local cuisine, hospitality, cultural/historical attractions, shopping and casino facilities and nightlife in general. The questionnaire will also include a number of demographic and socio-economic variables including age, gender, marital status, level of education, occupation and income. Different item response models will be fitted to the data using STATA’S subroutine gllamm.