A simple method for estimating a regression model for κ between a pair of raters

Agreement studies commonly occur in medical research, for example, in the review of X-rays by radiologists, blood tests by a panel of pathologists and the evaluation of psychopathology by a panel of raters. In these studies, often two observers rate the same subject for some characteristic with a discrete number of levels. The κ-coefficient is a popular measure of agreement between the two raters. The κ-coefficient may depend on covariates, i.e. characteristics of the raters and/or the subjects being rated. Our research was motivated by two agreement problems. The first is a study of agreement between a pastor and a co-ordinator of Christian education on whether they feel that the congregation puts enough emphasis on encouraging members to work for social justice (yes versus no). We wish to model the κ-coefficient as a function of covariates such as political orientation (liberal versus conservative) of the pastor and co-ordinator. The second example is a spousal education study, in which we wish to model the κ-coefficient as a function of covariates such as the highest degree of the father of the wife and the father of the husband. We propose a simple method to estimate the regression model for the κ-coefficient, which consists of two logistic (or multinomial logistic) regressions and one linear regression for binary data. The estimates can be easily obtained in any generalized linear model software program.     

A Meta-Analysis of Children's Hand-to-Mouth Frequency Data for Estimating Nondietary Ingestion Exposure

Because of their mouthing behaviors, children have a higher potential for exposure to available chemicals through the nondietary ingestion route; thus, frequency of hand-to-mouth activity is an important variable for exposure assessments. Such data are limited and difficult to collect. Few published studies report such information, and the studies that have been conducted used different data collection approaches (e.g., videography versus real-time observation), data analysis and reporting methods, ages of children, locations, and even definitions of "mouthing." For this article, hand-to-mouth frequency data were gathered from 9 available studies representing 429 subjects and more than 2,000 hours of behavior observation. A meta-analysis was conducted to study differences in hand-to-mouth frequency based on study, age group, gender, and location (indoor vs. outdoor), to fit variability and uncertainty distributions that can be used in probabilistic exposure assessments, and to identify any data gaps. Results of this analysis indicate that age and location are important for hand-to-mouth frequency, but study and gender are not. As age increases, both indoor and outdoor hand-to-mouth frequencies decrease. Hand-to-mouth behavior is significantly greater indoors than outdoors. For both indoor and outdoor hand-to-mouth frequencies, interpersonal, and intra-personal variability are approximately 60% and approximately 30%, respectively. The variance difference among different studies is much bigger than its mean, indicating that different studies with different methodologies have similar central values. Weibull distributions best fit the observed data for the different variables considered and are presented in this article by study, age group, and location. Average indoor hand-to-mouth behavior ranged from 6.7 to 28.0 contacts/hour, with the lowest value corresponding to the 6 to <11 year olds and the highest value corresponding to the 3 to <6 month olds. Average outdoor hand-to-mouth frequency ranged from 2.9 to 14.5 contacts/hour, with the lowest value corresponding to the 6 to <11 year olds and the highest value corresponding to the 6 to <12 month olds. The analysis highlights the need for additional hand-to-mouth data for the <3 months, 3 to <6 months, and 3 to <6 year age groups using standardized collection and analysis because of lack of data or high uncertainty in available data. This is the first publication to report Weibull distributions as the best fitting distribution for hand-to-mouth frequency; using the best fitting exposure factor distribution will help improve estimates of exposure. The analyses also represent a first comprehensive effort to fit hand-to-mouth frequency variability and uncertainty distributions by indoor/outdoor location and by age groups, using the new standard set of age groups recommended by the U.S. Environmental Protection Agency for assessing childhood exposures. Thus, the data presented in this article can be used to update the U.S. EPA's Child-Specific Exposure Factors Handbook and to improve estimates of nondietary ingestion in probabilistic exposure modeling.     

The human ecology of tornadoes

This paper offers an empirical test of the impact of human ecological patterns and other known correlates on tornado occurrence. It uses the National Severe Storms Forecast Center’s information on tornadoes from 1950 through 1990 and employs ecological data from the U.S. Bureau of the Census and the Environmental Protection Agency. The results show that metropolitan and other urban counties have higher odds of tornado occurrence than rural counties, and that the probability of occurrence of tornadoes increases with increases in the number of previous tornadoes. The paper assesses the meaning of this finding for demographers, atmospheric scientists, engineers, and disaster managers.     

A social indicator based on time allocation

The authors have developed a social indicator, called Lambda, that has as its base the manner in which individuals allocate their time among various life activities. This indicator does not suffer from the usual demand characteristics and experimenter bias associated with most subjective social indicators, i.e., those based on question-naires. It is not a single rate indicator in the sense of suicide rate. Lambda is a weighted sum social indicator with both subjective and objective aspects. The weights are decided by the population under consideration and not by the investigators or some other outside agents. The elements to be summed are the frequency functions for the discrepancies that arise when individuals state the amount of time they would like to spend in an activitiy versus the amount of time they actually spend at the activity. There are frequency functions for each activity. The properties of this indicator are discussed in detail. A pilot study comparing Lambda to another social indicator and some demographic variables was conducted with 1012 undergraduate students. The results of the pilot study show a high correlation between Lambda and the other social indicator. The results are presented. Lambda is currently being used by the Gallup organization.     

ARE THE DISABLED LESS LOSS AVERSE? EVIDENCE FROM A NATURAL POLICY EXPERIMENT

Research findings show that disabled persons often develop physical and psychological mechanisms to compensate for disabilities. Coping mechanisms may not be limited to the psychophysiological domain and may extend to cognitive bias and loss aversion. In this study, we apply unique microdata from a natural policy experiment to assess the role of loss aversion in home purchase among nondisabled and disabled households. Results of survival analysis indicate that the physically disabled are substantially less loss averse in home purchase. Furthermore, loss aversion varies with other population characteristics and attenuates with degree of disability. Findings provide new evidence of diminished cognitive bias and more rational economic decision‐making among the physically disabled. (JEL D03, C9, R38)     

Modelling Association Football Scores and Inefficiencies in the Football Betting Market

A parametric model is developed and fitted to English league and cup football data from 1992 to 1995. The model is motivated by an aim to exploit potential inefficiencies in the association football betting market, and this is examined using bookmakers' odds from 1995 to 1996. The technique is based on a Poisson regression model but is complicated by the data structure and the dynamic nature of teams' performances. Maximum likelihood estimates are shown to be computationally obtainable, and the model is shown to have a positive return when used as the basis of a betting strategy.