Hierarchical Log-linear Models Research Articles

Stepping and kicking behaviour during milking in relation to response in human–animal interaction test and clinical health in loose housed dairy cows

Associations between kicking and stepping behaviour during milking and response to humans, getting up behaviour as well as clinical health status in loose housed dairy cows were investigated. During a 12-month period behavioural and clinical data from 10 commercial Danish dairy herds were collected including in total 1196 individual cows. Hierarchical log-linear models for contingency tables were applied for an analysis of conditional associations among the included parameters. Results showed a general herd effect on occurrence of stepping and kicking behaviour during milking ( P<0.001). In general, stepping during milking was associated positively to daily milk yield ( P<0.05). However, low yielding cows showing avoidance behaviour at the clinical examination were also more likely to step during milking. Furthermore, stepping during milking was positively associated to avoidance in a human approach test ( P<0.001) as well as occurrence of ticks ( P<0.05), and cows with tick lesions that further avoided human approach were most likely to step during milking. Cows with teat lesions were more likely to kick during milking ( P<0.001), especially if they did not avoid human approach. The analysis showed no relation between lameness and kicking or stepping behaviour during milking.

Associations between lying-down behaviour problems and body condition, limb disorders and skin lesions of lactating sows housed in farrowing crates in commercial sow herds

Health problems, difficulties in lying down and skin lesions are relevant indicators of poor welfare in sows confined in farrowing crates. This study investigated limb disorders, injuries, body condition and lying-down behaviour in 10 commercial sow herds. Five hundred seventy lactating sows were examined clinically, and 555 sows were further observed when lying down. Forty-one percent of the sows showed some difficulties in lying down and 15% were lame. Skin lesions were most frequently observed on lateral hind feet or shoulders, each in approximately 20% of the sows. The data were analysed for conditional associations among the parameters applying hierarchical log-linear models for contingency tables. Slipping and interruptions while lying down were associated ( P<0.05), but other difficulties in lying down were conditionally independent within the herd. Shoulder wounds were more common in thin or lame sows (odds ratio 5.5 and 2.5, respectively), and lameness was further related to hock lesions (OR 11.9, severely lame sows), hind feet lesions (OR 2.5–4.3 depending on the severity of lameness) and uncontrolled lying-down behaviour (OR 1.8–5.8). Fat sows were more likely to show stepping behaviour when lying down (OR 2.7), slipping was associated with hind feet lesions (OR 2.2) and overgrown hooves were associated with abnormal lying-down behaviour (OR 11.5). The specific difficulties in lying down may provide relevant support for the clinical findings.

Knowledge and Productivity in Technical Support Work

In this paper, we examine the process of technical support work and the role of knowledge in enhancing the productivity of such work. We develop the concepts of problem-solving tasks and moves to describe technical support work, while using call resolution time and problem escalation as measures of productivity. Using hierarchical log-linear modeling, we establish the link between problem-solving moves and productivity. We find that the mix of moves exercised in technical support strongly depends on the formulation of tasks by those requesting support. Because the formulation of tasks is performed by users, knowledge management initiatives must target users as well as support providers to have the desired impact on productivity.

Estrategia para la selección de modelos gráficos en tablas de contingencia no estructuradas

Los modelos loglineales se utilizan con frecuencia con el objetivo de analizar los patrones de asociación entre las variables que conforman una tabla de contingencia multidimensional. Recientemente se ha popularizado la utilización de los modelos gráficos discretos, que son un subconjunto de los modelos loglineales jerárquicos cuya estructura de independencia condicional se representa de manera única por un grafo, lo que facilita la interpretación de los mismos. En este trabajo se describen y comparan cuatro métodos de selección de modelos gráficos discretos: hacia atrás, hacia delante, paso a paso y automático EH. Asimismo, se propone una estrategia general de selección de modelos.

Professional and managerial migration from core to periphery: the case of English migration to Scottish cities

AbstractThe population of English‐born persons living in Scotland has risen rapidly since 1971. The dynamics of the group are theorised in this paper relative to the escalator region hypothesis, in terms of the role of professional and managerial staff linking global cities with lower‐order urban centres within their hinterland. A hierarchical log‐linear modelling approach is taken in explaining the highly structured geography of English migration to Scotland's four largest urban centres. Links between place of birth, recent mobility, social class and car ownership are explored. It is argued that service class migration by the English to Scottish cities is representative of a much wider process found in other countries where economic and political integration has stimulated parallel types of skilled migration from core to peripheral urban locations. Copyright © 2002 John Wiley &amp; Sons, Ltd.

Individual and Situational Interest: The Role of Gender and Skill

The purpose of this study was to examine individual and situational interests in learning motor skills as associated with gender and skill. Individual and situational interests and motor skill were measured in middle school girls (n = 109) and boys (n = 82). A correlation analysis for the entire sample showed a moderate correlation between skill and individual interest (r = .63) and weak relationships between individual and situational interest (r < .10) and between situational interest and skill (r < .18). A MANOVA analysis revealed a difference between the boys and girls in situational interest at a borderline significance level (p = .05). However, when individual interest, skill, and gender were compared between students with high or low situational interest in a hierarchical log linear model, it was found that both groups did not differ in number of boys and girls (p = .98). But the high situational interest group had more students with high skill (p = .001) and high individual interest (p = .02). The results suggest that discrepancies in acquired skill accounted for the gender difference and that acquired skill is associated with individual interest and high situational interest in learning motor skills.

Counting Guns in Early America

Probate inventories, though perhaps the best prevailing source for determining ownership patterns in early America, are incomplete and fallible. In this Article, the authors suggest that inferences about who owned guns can be improved by using multivariate techniques and control variables of other common objects. To determine gun ownership from probate inventories, the authors examine three databases in detail-Alice Hanson Jones's national sample of 919 inventories (1774), 149 inventories from Providence, Rhode Island (1679-1726), and Gunston Hall Plantation's sample of 325 inventories from Maryland and Virginia (1740-1810). Also discussed are a sample of 59 probate inventories from Essex County, Massachusetts (1636-1650), Gloria L. Main's study of 604 Maryland estates (1657-1719), Anna Hawley's study of 221 Surry County, Virginia estates (1690-1715), a sample of 289 male inventories from Vermont (1773-1790), and Judith A. McGaw's study of 250 estates in New Jersey and Pennsylvania (1714-1789). Guns are found in 50-73% of the male estates in each of the eight databases and in 6-38% of the female estates in each of the first four databases. Gun ownership is particularly high compared to other common items. For example, in 813 itemized male inventories from the 1774 Jones national database, guns are listed in 54% of estates, compared to only 30% of estates listing any cash, 14% listing swords or edged weapons, 25% listing Bibles, 62% listing any book, and 79% listing any clothes. Using hierarchical loglinear modeling, the authors show that guns are more common in early American inventories where the decedent was male, Southern, rural, slave-owning, or above the lowest social class - or where the inventories were more detailed. The picture of gun ownership that emerges from these analyses substantially contradicts the assertions of Michael Bellesiles in Arming America: The Origins of a National Gun Culture (Arming America). Contrary to Arming America's claims about probate inventories in seventeenth and eighteenth-century America, there were high numbers of guns, guns were much more common than swords or other edged weapons, women in 1774 owned guns at rates (18%) higher than Bellesiles claimed men did in 1765-1790 (14.7%), and 87-91% of gun-owning estates listed at least one gun that was not old or broken. The authors replicated portions of Bellesiles's published study in which he both counted guns in probate inventories and cited sources containing inventories. They conclude that Bellesiles appears to have substantially misrecorded the seventeenth and eighteenth century probate data he presents. For the Providence probate data (1679-1726), Bellesiles has misclassified over 60% of the inventories he examined. He repeatedly counted women as men, counted about a hundred wills that never existed, and claimed that the inventories evaluated more than half of the guns as old or broken when fewer than 10% were so listed. Nationally, for the 1765-1790 period, the average percentage of estates listing guns that Bellesiles reports (14.7%) is not mathematically possible, given the regional averages he reports and known minimum sample sizes. Last, an archive of probate inventories from San Francisco in which Bellesiles claims to have counted guns apparently does not exist. By all accounts, the entire archive before 1860 was destroyed in the San Francisco earthquake and subsequent fire of 1906. Neither part of his study of seventeenth and eighteenth-century probate data is replicable, nor is his study of probate data from the 1840s and 1850s.

Religious Attendance and Life Satisfaction in the Finnish General Population

In this study we investigated whether there were associations between religious attendance and life satisfaction in the Finnish general population. Furthermore, we examined whether social support could explain these associations. The nationwide sample included 869 women and 773 men. The respondents were asked during telephone interviews about sociodemographic variables, their household category, social and family support, and the frequencies of religious attendance and other social contacts. Life satisfaction was assessed by means of a scale developed by the Survey Research Center of the University of Michigan. Women were religious attenders more often than men (62% versus 50%). There was no gender difference in life satisfaction. Positive associations were found between life satisfaction and both social support and religious attendance. In multivariate logistic regression analysis an independent positive association was found between religious attendance and the absence of severe life dissatisfaction. Hierarchical loglinear modeling suggested that social support did not mediate the association between religious attendance and life satisfaction.

Rectal constipation and clinical decision-making: multiple correspondence analysis of defecographic findings

The aim of this study was to establish the impact of defecography on the management of patients with chronic constipation. The defecographic series of 581 patients (426 women, 155 men, mean age 51 ± 17 years) with evacuation dysfunctions was reviewed. Proper utilization of defecography was investigated by examining the structural interaction between presenting symptoms using a hierarchical log-linear model analysis. Then, to assess the impact of defecography on clinical decision-making, the multiple correspondence analysis was employed to calculate the relationship between eight radiographic findings (rectal prolapse/intussusception, rectocele, perineal descent, outlet obstruction, barium loss, delayed emptying, residue and rectal enlargement) and six clinical features (fecal blockade, provoked evacuation, tenesmus, feeling of prolapse, incontinence and mucus discharge) which occurred most frequently in the population under study. At the hierarchical log-linear analysis no more than one second-order three-symptom (i.e. fecal blockade, provoked evacuation and mucus discharge) and six first-order two-symptom interactions (including all other symptoms combined in various ways) were obtained, giving no evidence of overutilization of defecography in our constipated group. A close association (inertia, 34.4%) was found at the multiple correspondence analysis between delayed emptying, residue, outlet obstruction and rectal enlargement (relative weight 27.1%, 20.2%, 19.6% and 8.5%, respectively). Our results suggest labeling the first subdimension as the need for a conservative approach, while the second subdimension is surgery. The main value of defecography in chronic constipation is as a simple exploratory method for selecting patients who require admission to biofeedback and pelvic floor re-education.

Categorical Variables in Developmental Research: Methods of Analysis

Preface. Measurement and Repeated Observations of Categorical Data: D. Andrich, Measurement Criteria for Choosing Among Models with Graded Responses. B.O. Muthn, Growth Modeling with Binary Responses. G. Arminger, Probit Models for the Analysis of Limited-Dependent Panel Data. Catastrophe Theory: H.L.J. van der Maas and P.C.M. Molenaar, Catastrophe Analysis of Discontinuous Development. P.C.M. Molenaar and P. Hartelman, Catastrophe Theory of Stage Transitions in Metrical and Discrete Stochastic Systems. Latent Class and Log-Linear Models: L.M. Collins, P.L. Fidler, and S.E. Wugalter, Some Practical Issues Related to Estimation of Latent Class and Latent Transition Parameters. T.D. Wickens, Contingency Tables and Between-Subject Variability. C.C. Clogg and W. Manning, Assessing Reliability of Categorical Measurements Using Latent Class Models. D. Rindskopf, Partitioning Chi-square: Something Old, Something New, Something Borrowed, but Nothing BLUE (Just ML). A. von Eye and C. Spiel, Nonstandard Log-Linear Models for Measuring Change in Categorical Variables. H.J. Khamis, Application of the Multigraph Representation of Hierarchical Log-Linear Models. Applications: M.J. Rovine and A. von Eye, Correlation and Categorization under a Matching Hypothesis. S.L. Hershberger, Residualized Categorical Phenotypes and Behavioral Genetic Modeling. Chapter References. Subject Index.

Seasonal influences on admissions for affective disorder and schizophrenia in Ireland: a comparison of first and readmissions

Although genetic and environmental factors operating before or around the time of birth have been demonstrated to be relevant to the aetiology of the major psychoses, a seasonal variation in the rates of admission of such patients has long been recognised. Few studies have compared first and readmissions. This study examined for seasonal variation of admission in the major psychoses, and compared diagnostic categories by admission status. Patients admitted to Irish psychiatric inpatient facilities between 1989 and 1994 with an ICD-9/10 diagnosis of schizophrenia or affective disorder were identified from the National Psychiatric Inpatient Reporting System (NPIRS). The data were analysed using a hierarchical log linear model, the chi-square test, a Kolmogorov-Smirnov (KS) type statistic, and the method of Walter and Elwood. The hierarchical log linear model demonstrated significant interactions between the month of admission and admission order (change in scaled deviance 28.77, df = 11, P < 0.003). Both first admissions with mania, and readmissions with bipolar affective disorder exhibited significant seasonality. In contrast, only first admissions with schizophrenia showed significant seasonal effects. Although first admissions with mania and readmissions with bipolar disorder both show seasonality, seasonal influences appear to be more relevant to onset of schizophrenia than subsequent relapse.

Activated microvessels express vascular endothelial growth factor and integrin alpha(v)beta3 during focal cerebral ischemia.

Both vascular endothelial growth factor (VEGF) and integrin alpha(v)beta3 play roles in angiogenesis. In noncerebral vascular systems, VEGF can induce endothelial integrin alpha(v)beta3 expression. However, it is unknown whether VEGF, like integrin alpha(v)beta3, appears in the initial response of microvessels to focal brain ischemia. Their coordinate expression in microvessels of the basal ganglia after middle cerebral artery occlusion (MCAO) in the nonhuman primate model was examined quantitatively. Cells incorporating deoxyuridine triphosphate (dUTP+) by the polymerase I reaction at 1 hour (n = 3), 2 hours (n = 3), and 7 days (n = 4) after MCAO defined the ischemic core (Ic) and peripheral regions. Both VEGF and integrin alpha(v)beta3 were expressed by activated noncapillary (7.5- to 30.0-microm diameter) microvessels in the Ic region at 1 and 2 hours after MCAO. At 7 days after MCAO, the number of VEGF+, integrin alpha(v)beta3+, or proliferating cell nuclear antigen-positive microvessels had decreased within the Ic region. The expressions of VEGF, integrin alpha(v)beta3, and proliferating cell nuclear antigen were highly correlated on the same microvessels using hierarchical log-linear statistical models. Also, VEGF and subunit alpha(v) messenger ribonucleic acids were coexpressed on selected microvessels. Here, noncapillary microvessels are activated specifically early during a focal cerebral ischemic insult and rapidly express VEGF and integrin alpha(v)beta3 together.

Effect Analysis in Loglinear Model Approach to Path Analysis of Categorical Variables

This paper considers a method for measuring causal effects of factors in recursive causal systems described by hierarchical loglinear models. The total, direct and indirect effects of factors are defined by loglinear model parameters. A numerical example is also given to illustrate the present discussion.

How Do I Love Thee? Let Me Consider My Options Cognition, Verbal Strategies, and the Escalation of Intimacy

Previous research has revealed that individuals have expectations for the development of romance based on personal experience and cultural images. A series of research questions and hypotheses was generated and a sequence of hierarchical log-linear analysis models axis developed to test (a) the effect of gender and knowledge of relational stage and relationship type on prediction of next-occurring actions in the development of a romance, (b) the use of compliance-gaining strategies intended to advance a relationship to the next stage, and (c) justification of secondary goals in which participants reported why they chose the strategies they did as opposed to other alternatives. The data revealed effects for relational stage progression on prediction of ensuing actions. Ingratiation, explanation, and direct requests were popular strategies for developing intimacy. Gender differences in terms of secondary goals were reported. Results are discussed in terms of relational memory structures for the development of intimacy.

Sequential analysis using hierarchical log-linear modeling: Caretaker response to infant distress in different ecological niches

Sequential analysis using hierarchical log-linear modeling: Caretaker response to infant distress in different ecological niches

Generalized Linear Models: Loglinear Modeling in Managerial Research

In the past three decades we have seen the emergence of the generalized linear model (GLM) techniques for analyzing discrete multivariate data when the independent and dependent variables are categorical, ordinal, or mixed. The primary statistical techniques are loglinear modeling, probit, and logistic regression. The purpose of this article is to (a) briefly describe the emergence of these discrete multivariate techniques in the medical and social sciences, (b) disclose their relationship to one another, and (c) demonstrate the utility of hierarchical loglinear modeling in managerial research

Symptomatic cholelithiasis: a different disease in men? Connecticut Laparoscopic Cholesystectomy Registry.

To determine the importance of gender in the clinical presentation and subsequent clinical outcome (risk of conversion from laparoscopic to open technique and risk of postoperative mortality) for patients undergoing cholecystectomy. Age and clinical presentation have consistently been found to be important predictors of cholecystectomy outcomes; male gender has been cited in disparate studies as possibly having prognostic significance. A statewide cholecystectomy registry (30,145 cases between 1989-1993) was analyzed. Hierarchical log-linear modeling was used to identify associations between characteristics of clinical presentation. Multivariate logistic regression analysis was used to determine predictions of conversion and mortality. Male gender was associated with twice the expected incidence of acute cholecystitis and pancreatitis in the elderly (> or = 65 years). Males had a significantly increased risk for conversion to open technique, but this decreased during the time frame of the study. Mortality was twice as high among males (confidence interval, 1.4-2.9, p = 0.0001). Males presenting for cholecystectomy are more likely to have severe disease. Independent of clinical presentation, they face increased risks of conversion to open technique and of postoperative mortality.

Cultural Variations in Cognitive Style

Research by Herman A. Witkin reports that one aspect of cognitive style, namely field dependence/field independence, is affected significantly by socialization and child-rearing practices. The present cross-cultural project, using the Children's Embedded Figures Test, examined the cognitive style of 100 African American and 100 Black South African males and females in grades 4 and 5. Hierarchical loglinear models were utilized to uncover the complex relationships between gender (males and females), culture (African Americans and Black South Africans) and learning style (field dependence and field independence). The results were significant and suggest that African Americans are more likely to be using a learning style characterized by field independence and Black South Africans are more likely to be using a learning style characterized by field dependence. No differences were found relative to gender for either group or for the two groups combined. Although no formal cross-cultural research has previously been done relative to cognitive style in the context ofthese two groups, the findings are consistent with the overall theoretical and historical research relative to field dependence and field independence.

Iterative proportional fitting for nonhierarchical log-linear models

The iterative proportional fitting (IPF) algorithm is simple to use for fi~ting hierarchical log-linear models of any size whether their maximum likelihood estimates are given in a closed form or not. This paper shows that the algorithm can be extended toward a elass of nonhierarchical log-linear models.

Loglinear Symmetry and Quasi‐Symmetry Models for the Analysis of Change

AbstractLoglinear symmetry and quasi‐symmetry models are proposed as tools for investigating various hypotheses about change. First, a survey of model representations is provided, including model specification in terms of hierarchical loglinear models and in design matrix notation. Secondly, the range of symmetry and quasi‐symmetry models is extended to the joint analysis of several groups. Parameter constraints are discussed which allow one to test specific hypotheses about group differences in symmetric frequency distributions. Finally, symmetry and quasi‐symmetry models are considered for multiway contigency tables. In this context, loglinear total score models are proposed for the analysis of symmetry in several marginal distributions. The proposed models reflect cross‐sectional as well as longitudinal facets of development.