Dominique Haughton

Direct marketing modeling with CART and CHAID

In a recent article in the Journal of Direct Marketing, R. Thrasher gave an introduction to the Classification and Regression Trees (CART) methodology from the point of view of direct marketing. Following up on this work, with the help of a case study, this article investigates further technical aspects of response modeling with CART and Chi-Square Automatic Interaction Detector (CHAID) of practical relevance to direct marketing professionals and researchers and compares the performance of CART and CHAID models built on the same analysis file, finding that the response lifts are very close for both types of models.

Son preference in Vietnam.

This article assesses the strength of son preference in Vietnam, as reflected in fertility behavior. It formulates and estimates a proportional hazards model applied to birth intervals, and a contraceptive prevalence model, using household survey data from 2,636 ever-married women aged 15-49 with at least one living child who were interviewed for the Vietnam Living Standards Survey 1992-1993. Son preference is found to be strong by world standards, but nevertheless, it has a minor effect on fertility; in its absence, the total fertility rate would fall by roughly 10 percent from the current level of about 3.2 children per woman of reproductive age.

Review of Three Latent Class Cluster Analysis Packages: Latent Gold, poLCA, and MCLUST

This article reviews three software packages that can be used to perform latent class cluster analysis, namely, Latent Gold®, MCLUST, and poLCA. Latent Gold® is a product of Statistical Innovations whereas MCLUST and poLCA are packages written in R and are available through the web site http://www.r-project.org. We use a single dataset and apply each software package to develop a latent class cluster analysis for the data. This allows us to compare the features and the resulting clusters from each software package. Each software package has its strengths and weaknesses and we compare the software from the perspectives of usability, cost, data characteristics, and performance. Whereas each software package utilizes the same methodology, we show that each results in a different cluster solution and suggest some rationales for deciding which package to use.

Ethnic minority development in Vietnam

Abstract This study examines the disparities in living standards between and among the different ethnic groups in Vietnam. Using data from the Vietnam Living Standards Surveys and 1999 Census, we show that ‘majority’ Kinh and Hoa households have substantially higher living standards than ‘minority’ households from Vietnams 52 other ethnic groups. While the Kinh, Hoa, Khmer and Northern Highland Minorities benefited from economic growth in the 1990s, the position of the Central Highland Minorities stagnated. Decompositions show that even if minority households had the same endowments as Kinh households, this would close no more than a third of the gap in their per capita expenditures. While some ethnic minorities seem to be doing well out of a strategy of assimilating with the Kinh-Hoa majority, others groups are attempting to integrate economically while retaining distinct cultural identities, and a third group is largely being left behind by the growth process.

Informational complexity criteria for regression models

This paper pursues three objectives in the context of multiple regression models: 1) To give a rationale for model selection criteria which combine a badness of fit term (such as minus twice the log likelihood) with a measure of complexity of a model. 2) To investigate the asymptotic consistency properties of the class of ICOMP criteria first in the case when one of the models considered is the true model and to introduce and establish a consistency property for the case when none of the models is the true model. 3) To investigate the finite sample behavior of ICOMP criteria by means of a simulation study where none of the models considered is the true model.

Explaining Child Nutrition in Vietnam

UNICEF has written that widespread malnutrition in Vietnam stems not from the insufficient production of food but from problems of availability distribution and demand. The authors estimated two models of child nutrition using data from a representative and relatively large sample of Vietnamese households surveyed in 1992-93. No evidence was found of gender bias in nutrition for either stunting or wasting. These findings are consistent with the lack of gender bias in Vietnam in providing health care and education to preteen children. Birth order and mother and fathers education are however important with the effect of education remaining even when income is included in the regression equations. The effect of education upon nutrition therefore appears to work directly rather than through its influence upon income. Families in the northern regions rural households and families belonging to ethnic minority groups have significantly higher levels of malnutrition than do those elsewhere in the country.

Identifying Groups: A Comparison of Methodologies

This paper describes and compares three clustering techniques: traditional clustering methods, Kohonen maps and latent class models. The paper also proposes some novel measures of the quality of a clustering. To the best of our knowledge, this is the first contribution in the literature to compare these three techniques in a context where the classes are not known in advance.

A Review of Software Packages for Data Mining

We present to the statistical community an overview of five data mining packages with the intent of leaving the reader with a sense of the different capabilities, the ease or difficulty of use, and the user interface of each package. We are not attempting to perform a controlled comparison of the algorithms in each package to decide which has the strongest predictive power, but instead hope to give an idea of the approach to predictive modeling used in each of them. The packages are compared in the areas of descriptive statistics and graphics, predictive models, and association (market basket) analysis. As expected, the packages affiliated with the most popular statistical software packages (SAS and SPSS) provide the broadest range of features with remarkably similar modeling and interface approaches, whereas the other packages all have their special sets of features and specific target audiences whom we believe each of the packages will serve well. It is essential that an organization considering the purchase of a data mining package carefully evaluate the available options and choose the one that provides the best fit with its particular needs.

Information and other criteria in structural equation model selection

This article presents the results of a simulation study evaluating information criteria in conjunction with other well-known criteria for model selection in structural equation modeling (SEM). Two sets of simulation experiments were performed. In both sets, sample sizes of n = 100,400,1000,6000 were used and the performance of 18 criteria was assessed by the frequency with which each of five analytic models was selected as best by each criterion in 500 replications. In the first set of experiments correctly specified analytic models (noncentrality parameter 0) were entertained in combination with misspecified ones, while in the second set all five models were misspecified. In both sets of experiments, we found that the information criteria perform better than the other criteria overall, but that Cudeck and Brownes cross-validation index ( CVI) remains an attractive option. Within the class of information criteria, Akaikes information criterion (AIC) is found to show some overfitting tendency. We demonst...

Gender differences in CEO compensation: evidence from the USA

Purpose – To provide insights into the experience of women aspiring to the CEO position, particularly regarding qualifications and compensation expectations.Design/methodology/approach – The ExecuComp database of executives at 1,500 large US corporations from 1992 to 2004 was used to identify women CEOs and to examine gender differences in compensation of executives over that period. Additional information about the backgrounds of female CEOs was collected from company press releases and regulatory filings.Findings – Women are not as highly compensated as men before becoming CEO but the few who reach the CEO position receive similar compensation as men. While women CEOs are younger on average than men, they have impressive work experience and education.Research limitations/implications – The study covers relatively large US companies that are publicly traded; thus, smaller firms and privately‐held firms are not included.Practical implications – Impressive work experience, usually from within the company, ...