Multivariate Ordered Probit Model Research Articles

Composite likelihood approach to the regression analysis of spatial multivariate ordinal data and spatial compositional data with exact zero values

In many environmental and ecological studies, it is of interest to model compositional data. One approach is to consider positive random vectors that are subject to a unit-sum constraint. In landscape ecological studies, it is common that compositional data are also sampled in space with some elements of the composition absent at certain sampling sites. In this paper, we first propose a practical spatial multivariate ordered probit model for multivariate ordinal data, where the response variables can be viewed as the discretized non-negative compositions without the unit-sum constraint. We then propose a novel two-stage spatial mixture Dirichlet regression model. The first stage models the spatial dependence and the presence of exact zero values, and the second stage models all the non-zero compositional data. A maximum composite likelihood approach is developed for parameter estimation and inference in both the spatial multivariate ordered probit model and the two-stage spatial mixture Dirichlet regression model. The standard errors of the parameter estimates are computed by an estimate of the Godambe information matrix. A simulation study is conducted to evaluate the performance of the proposed models and methods. A land cover data example in landscape ecology further illustrates that accounting for spatial dependence can improve the accuracy in the prediction of presence/absence of different land covers as well as the magnitude of land cover compositions.

Multidimensional inequality for current status of Japanese private companies’ employees

In this paper, we show a method for measuring multidimensional inequality. We use a Bayesian method for a multivariate ordered probit model to obtain a single index, which we refer to as a probability associated with an individual’s overall status, from a multivariate distribution representing a number of attributes. We evaluate multidimensional inequality by the Gini coefficient and the generalized entropy measure of inequality using this index. This method is used for the employee data of Japanese private companies, which include income and attributes related to subjective perception such as job satisfaction.

“Same, Same—but Different!” On Consumers’ Use of Corporate PR Media in Sports

Organizations around the globe have numerous avenues to share information with their target groups and communicate directly without any intermediaries such as journalists. Particularly, sports organizations like professional sports teams make frequent use of e-mail newsletters, (online) club TV channels, stadium magazines, and Internet platforms. In addition, they frequently share information using social networks like Twitter and Facebook. Surprisingly, however, very little is known about the factors influencing consumers’ use of these different communication channels. This paper is the first to analyze simultaneously the factors associated with consumers’ use of different public relations (PR) media by using representative data from club members of one of the biggest professional soccer clubs in Germany and employing a multivariate ordered probit model. Results suggest that decisions on the use of different PR media are closely related, though sociodemographic and membership characteristics have a media-specific impact on the frequency of use.

Electricity saving behavior of households by making efforts, replacing appliances, and renovations: empirical analysis using a multivariate ordered probit model

AbstractThis study empirically explores the determinants of household electricity saving behavior based on a sample of Japanese households. Our study makes four contributions. First, we examine the short‐term saving action of making efforts, the medium‐term saving action of replacing appliances, and the long‐term saving action of undertaking renovations. Second, we consider not only the physical characteristics of households but also behavioral characteristics, such as their understanding of the importance of energy saving and willingness to save energy. Third, in our analysis, we assume that appliances and housing facilities have the electricity‐saving capability and service‐providing capability. While previous studies focus on the former, consumers in reality consider both capabilities. Fourth, we consider the optimization by households and include broad types of appliances and renovations. Using a multivariate ordered probit model with data on 518 Japanese households in 2012, we obtain the following results. First, in addition to the demand response, the effort level is determined by behavioral, rather than by physical characteristics. Second, the motivation behind appliance replacement is durability, rather than electricity‐saving and service‐providing capabilities. Third, appliance replacement is enhanced by an understanding of the importance of energy saving, rather than by the willingness to save energy. Households can replace appliances because of social pressure, even if they are not so concerned about saving energy. Fourth, households are willing to renovate housing facilities even though they pay a higher cost for the greater saving capability rather than for the service‐providing capability. However, such renovation is not influenced by internal incentives, but by physical aspects of the house, such as size, age, and type.

Analysis of Job Satisfaction: The Case ofJapanese Private Companies

AbstractThe scope of economic theory expands if we consider that working provides people with both pecuniary compensation for its disutility and job satisfaction. This study empirically analyzes the job satisfaction of employees inJapanese private companies using a multivariate ordered probit model. In particular, we examine the effects of a big economic shock, such as theLehman shock, on job satisfaction, as well as the effects of differences in gender and employment status. We estimate the model using aBayesian analysis and a multivariate ordered probit model using theMarkov chainMonteCarlo method. We show that non‐pecuniary aspects of jobs play an important role in job satisfaction.

Prediction of Patient-Reported Outcome Measures Via Multivariate Ordered Probit Models

SummaryThe assessment of patient-reported outcome measures (PROMs) is of central importance in many areas of research and public policy. Unfortunately, it is quite common for clinical studies to employ different PROMs, thus limiting the comparability of the evidence base that they contribute to. This issue is exacerbated by the fact that some national agencies are now explicit about which PROMs must be used to generate evidence in support of claims for reimbursement. The National Institute for Health and Care Excellence for England and Wales, for instance, has identified in EuroQoL-5D, EQ-5D, the PROM of choice, while accepting the use of a ‘mapping’ approach to predict EQ-5D from other PROMs when EQ-5D data have not been collected. Here we consider the problem of directly predicting EQ-5D responses from ‘Short form 12', while recognizing both the likely dependence between the five dimensions of the EQ-5D responses at the patient level, and the fact that the levels of each health dimension are naturally ordered. We carry out the analysis within a Bayesian framework. We also address the key problem of choosing an appropriate summary measure of agreement between predicted and actual results when analysing PROMs, with particular attention devoted to scoring rules.

The determinants of the intention to purchase animal welfare-friendly meat products in Spain

AbstractThe aim of this paper is to analyse the intention to purchase animal welfare-friendly meat products and determine the factors explaining this intention. Additionally, a model of the intention to purchase animal welfare-friendly meat products has been developed based on the Theory of Planned Behaviour (TPB). This model has been specified as a two-equation multivariate ordered probit model and estimated using data from a survey conducted in Spain in 2008. Results indicate that one of the most important factors associated with the intention to purchase animal welfare-friendly meat products was consumer self-identification with ethical issues. Second, findings suggested that, as the Theory of Planned Behaviour states, other factors related to the intention to purchase these products were attitudes, subjective norms and perceived behavioural control.

Self‐Assessed Social Position and Poverty

In this article, we use microlevel data extracted from the 2006 Japanese General Social Surveys to analyze the relationships between self‐assessed social position and socioeconomic factors such as income and poverty. We provide the posterior results of the estimation of the Bayesian multivariate ordered probit model and propose an inequality measure for self‐assessed social position on the basis of the posterior results. We call the inequality measure “regret” and show that the distributions of regret differ for people above and below the poverty line.

Analyzing tourists' satisfaction: A multivariate ordered probit approach

This article considers a Bayesian estimation of the multivariate ordered probit model using a Markov chain Monte Carlo (MCMC) method. The method is applied to unit record data on the satisfaction experienced by tourists. The data were obtained from the Annual Report on the Survey of Tourists' Satisfaction 2002, conducted by the Department of Economic Affairs of the Hokkaido government. Furthermore, using the posterior results of the Bayesian analysis, indices of the relationship between the overall satisfaction derived from the trip and the satisfaction derived from specific aspects of the trip are constructed. The results revealed that the satisfaction derived from the scenery and meals has the largest influence on the overall satisfaction.

Direct interactions medical school faculty members have with professionals and managers working in public and private sector organizations: A cross-sectional study

The research questions are as follows: to what extent do Canadian medical school faculty members have person-to-person interactions with individuals working in public and private sector organizations? What are the characteristics of Canadian medical school faculty members who interact with individuals working in these work settings? Are these different network patterns complementary or substitute? The data used for this study are from a cross-sectional survey of Canadian medical school faculty members (n = 907). Structural multivariate ordered probit models were estimated to explore the characteristics of faculty members with different network patterns and to see if these network patterns are complementary or substitute. Study results suggest that the different network patterns considered in the study are not conflicting, but that some patterns correspond to different faculty member profiles.

State statutory reforms and retention of prescribed fire liability laws on U.S. forest land

Escalating wildfires on U.S. forest lands in recent years have resulted in statutory reforms in redefining the liability for landowners using prescribed fires. This study summarized these reforms in recent years. While in some states there is still strict tort liability for damages from prescribed fires, 18 states have reduced the liability burdens on landowners with simple negligence rules, and furthermore, 4 states with gross negligence rules. A multivariate ordered probit model was estimated to examine factors that have influenced the retention of certain liability rules across states. Demand for prescribed fires from industrial and nonindustrial private forest landowners turned out to be the key driving force behind these statutory changes related to prescribed fires.

Family background, cohort and education: A French–German comparison based on a multivariate ordered probit model of educational attainment

This paper analyses the impact of family background, cohort and gender on educational attainment in France and Germany, on the basis of a theoretical model in which the educational process is broken down into school and post-school achievement. Empirically, the model can be estimated as a multivariate ordered probit. The results show that, in spite of huge differences in the educational distribution, France and Germany prove surprisingly similar regarding the impact of family background and cohorts. However, there are significant dissimilarities depending on the stage observed in the educational career, in particular with respect to gender differences.

Family Background, Cohort and Education: A French-German Comparison

This paper analyses the impact of family background, gender and cohort on educational attainment in France and Germany, relying on a theoretical model imbedded in the human capital theory. In a second step, the educational process is decomposed into school and post-school achievement. The same conceptual framework applies at both stages, but a correlation is permitted between them. Empirically, this boils down to estimating a multivariate ordered probit model. The results show that in spite of huge differences in the distribution of education in France and Germany, these countries prove surprisingly similar with respect to the impact of family background and cohorts. However, there are significant dissimilarities depending on the stage observed in the educational career, in particular with respect to gender differences.

A Model of Educational Attainment Application to the German Case

This paper aims at formulating a theoretical model of educational attainment which is directly implementable empirically. The conceptual framework relies on the human capital theory, where individuals weigh the costs and the returns of education to decide on the optimal amount of education they want to invest in. In an extension of the model, the educational process is decomposed into two stages: achievement at school and post-school achievement. The same conceptual framework applies at both stages, but a correlation is permitted between them. Empirically, this boils down to estimate a multivariate ordered probit model. An illustration based on German data shows how this model functions in practice. Thus, the model confers an economic foundation on the empirical estimation of the impact of certain factors on educational outcomes.