1,721,036 research outputs found
Missing data and parameters estimates in multidimensional item response models
Full text linkStatistical analyses of data based on surveys usually face the problem
of missing data. However, some statistical methods require a complete data
matrix to be applicable, hence the need to cope with such missingness. Literature
on imputation abounds with contributions concerning quantitative responses, but
seems to be poor with respect to the handling of categorical data. The present
work aims at evaluating the impact of different imputation methods on
multidimensional IRT models estimation for dichotomous data
Simulating Ordinal Data
No full textThe increasing use of ordinal variables in different fields has led to the introduction of new statistical methods for their analysis. The performance of these methods needs to be investigated under a number of experimental conditions. Procedures to simulate from ordinal variables are then required. In this article, we deal with simulation from multivariate ordinal random variables. We propose a new procedure for generating samples from ordinal random variables with a prespecified correlation matrix and marginal distributions. Its features are examined and compared with those of its main competitors. A software implementation in R is also provided along with examples of its application
An extension and a new interpretation of the rank-based concordance index
No full textIn applied research many data sets contain observations from ordinal variables rather than continuous ones. In such situations, the study of dependence relationship among variables represents an interesting issue, since ordinal variables are not specified according to a metric scale. Recently, a novel dependence measure called “Rank-based Concordance Index” (RCI), that can contribute to solve this problem, was introduced. In this paper an extension and a statistical interpretation of this index in terms of dependence relationship between a real-valued dependent variable and a quantitative or ordinal independent one is provided. The adequacy and robustness of RCI in the new context are discussed and validated by a simulation study
Nonlinear Principal Component Analysis as a Tool for the Evaluation of Customer Satisfaction
Full text linkIn this paper we examine the problem of setting-up a suitable indicator for the assessment of customer satisfaction. The proposed indicator is based on the nonlinear principal component analysis technique. Its properties are examined, and further analysis concerning its application to real data, the treatment of missing values and comparisons with other competitors is presented. Finally, findings with
regard to data from an opinion survey are presented and discussed
Simulation of correlated Poisson variables
No full textGenerating correlated Poisson random variables is fundamental in many applications in the management and engineering fields, and in many others where multivariate count data arise. Multivariate Poisson data are often approximately simulated by either independent univariate Poisson or multivariate Normal data, whose implementation is provided by the most common statistical software packages such as R. However, such simulated data are often not satisfactory. Alternatively, methods for simulating multivariate Poisson data can be used, but they are adversely affected by limitations ranging from computational complexity to restrictions on the correlation matrix, which dramatically reduce their practical applicability. In this work, we propose a new method that is highly accurate and computationally efficient and can be usefully employed even by non-expert users in generating correlated Poisson data (and, more generally, any discrete variable), with assigned marginal distributions and correlation matrix
An R package for the simulation of correlated discrete variables
No full textA package for the stochastic simulation of discrete variables with assigned marginal distributions and correlation matrix is presented and discussed. The simulating mechanism relies upon the Gaussian copula, linking the discrete distributions together, and an iterative scheme recovering the correlation matrix for the copula that ensures the desired correlations among the discrete variables. Examples of its use are provided as well as three possible applications (related to probability, sampling, and inference), which illustrate the utility of the package as an efficient and easy-to-use tool both in statistical research and for didactic purposes
Generation of multivariate discrete data
No full textOver the recent years, great interest has been addressed to ordinal variables and to the development of multivariate statistical techniques for their analysis. The empirical comparison of such techniques, either exploratory or inferential, often requires simulating ordinal data under different
experimental conditions. Several methods for generating multidimensional data from continuous or discrete variables have been proposed. This paper focuses on an algorithm for generating ordinal data with assigned marginal distributions and correlation matrix. The procedure consists of two steps: the first one aims at setting up the desired experimental conditions, employing a straightforward discretization procedure from a standard multinormal variable, whose correlation matrix is computed through an iterative algorithm in order to achieve the target correlation matrix for ordinal
data. The second step actually implements the sampling under the experimental conditions and allows performing a Monte Carlo simulation study. The algorithm does not suffer from some drawbacks encountered by other existing techniques and has a large application. It is implemented in R through a function which allows the user to choose the sample of size, the number of variables, their distribution and the target correlation matrix which is also checked for its feasibility. Two examples of application are provided
Multidimensional item response theory models for dichotomous data in customer satisfaction evaluation
Full text linkIn this paper, multidimensional item response theory models for dichotomous data, developed in the fields of psychometrics and ability assessment, are discussed in connection with the problem of evaluating customer satisfaction. These models allow us to take into account latent constructs at various degrees of complexity and provide interesting new perspectives for services quality assessment. Markov chain Monte Carlo techniques are considered for estimation. An application to a real data set is also presente
GenOrd: Simulation of ordinal and discrete variables with given correlation matrix and marginal distributions (v. 1.4.0)
No full textA gaussian copula based procedure for generating samples from discrete random variables with prescribed correlation matrix and marginal distribution
- …
