Bayesian Interval Estimation for Predictive Values from Case-Control Studies

Positive predictive and negative predictive values (PPV and NPV) are often used to assess the accuracy of binary diagnostic tests. Unlike sensitivity and specificity, PPV and NPV are functions of the accuracy of the test and the overall prevalence of the disease in the population. In many studies of performance of estimators of PPV and NPV the population prevalence is assumed known. We allow for uncertainty in the estimate of the population prevalence and via simulation explore the impact of deviations from the assumed value.     

‘The terrorist in my home’: teenagers' violence towards parents – constructions of parent experiences in public online message boards

Teenagers' violence towards parents is a hidden and underexplored problem, particularly within the UK, and the stigma attached to such experiences makes research access difficult. In this study, two online message boards which featured parents' posted accounts of their teenagers' violence towards them were analysed. Using discourse analysis, three consistent discursive themes were identified: the emotional terrain of such experiences, the psychologisation of the child‐as‐‘perpetrator’ and parental responses to these complex experiences. Overall, these three themes were weaved together to produce overarching narratives of powerlessness and loss of hope. The theoretical and practical implications of this analysis are discussed, including a consideration of how such online message boards enable and delimit parental agency.     

THE COVARIANCE ANALYSIS OF SOME CENSORED SURVIVAL DATA FROM A LARGE SCALE STUDY OF MELANOMA

Regression models of the proportional hazard type are applied to investigate the effect of certain covariates on the survival function of some cancer patients. The censored survival data for these patients were taken from a survey conducted by the Queensland Melanoma Project through which there are available histories of all malignant melanoma cases in Queensland for the period 1963–1969; follow-up data on these cases were considered up to mid-1976. One of the results of the survival analysis shows that the well-known empirical observation that women have a better prognosis than men can be explained by the existence of a significant interaction between sex and the anatomical site of the lesion.     

EXPERIMENTAL TESTS OF RICARDIAN EQUIVALENCE

This study employs laboratory experiments to test aspects of Ricardian equivalence. An overlapping generations model forms the basis of the experimental design. Whenever Ricardian equivalence predicts a positive bequest, outcomes close to those predicted occur, after allowance for learning. Observed deviations from Ricardian equilibrium are small, but biased in a Keynesian direction: changes in bequests do not fully offset changes in government debt. The imposition of a binding nonnegativity constraint on bequests produces the predicted neoclassical behavior. The cost to the players of deviations from the theoretical equilibrium is apparently an important determinant of the variability of observed deviations.     

Cybernetic Hasidism: An Essay on Social and Religious Change*

This paper employs a cybernetic model of a religious symbol system to explain the competition among Rabbinists, Hasidim, and Messianists in the Jewish communities of eighteenth-century Poland. Hasidism's ability to attract adherents is explained by the great flexibility of its constitutive belief, “communion with God”when compared with Rabbinic Torah or the Messianic belief in the “lifting of sparks.” This flexibility is traced through all levels of the cybernetic hierarchy.     

The Social and Economic Underpinning of Paternalism and Violence on the Maize Farms of the South-Western Transvaal, 1900–1950*

Abstract All social systems - including highly repressive ones - manifest features of accommodation as well as those of coercion. These are often seen as discrete entities and, in the case of modern South Africa, the pervasiveness of racial oppression has given rise to a literature which fails to explore more accommodatory ideologies and social practices such as paternalism. In this essay it is argued that paternalism and violence are not necessarily exclusive phenomena. Indeed, it is suggested that the very presence of paternalistic relationships can - at various historic junctures - exacerbate the propensity to violence.     

POVERTY AND ECONOMIC GROWTH: TRICKLE DOWN PETERS OUT

Previous studies have found that economic growth tends to "trickle down" to poor families. This study investigates whether use of a relative definition of poverty eliminates the impact of economic growth on the incidence of poverty of various demographic groups and whether a decrease in the impact of growth has occurred since 1964. With the exception of families with male heads, economic growth no longer affects poverty irrespective of whether a relative or absolute definition of poverty is used. Our findings indicate that the contribution of growth has been overstated, that much of the past success has been illusory, and that poverty will be more intractable in the future.     

TAXATION, GOVERNMENT SPENDING, AND LABOR SUPPLY A DIAGRAMMATIC EXPOSITION

The effects of a change in the level of taxation on labor supply are examined under the assumption that the government budget balances. The government spends tax revenues either on redistributions (which are assumed to have an income effect on labor supply) or on other government goods (which are assumed not to influence labor supply). In order for a rise in the tax rate to increase the quantity of labor supplied, it must be that (i) the labor-supply function bends backward, and (ii) sufficiently little of the increased tax revenues are redistributed. The quantity of labor supplied must fall if all marginal tax revenues are redistributed.     

Non‐parametric Regression for Circular Responses

Regression with a circular response is a topic of current interest. We introduce non‐parametric smoothing for this problem. Simple adaptations of a weight function enable a unified formulation for both real‐line and circular predictors, whereas these cases have been tackled by quite distinct parametric methods. Additionally, we discuss various methodological extensions, obtaining a number of promising techniques – totally new in circular statistics – such as confidence intervals for the value of a circular regression and non‐parametric autoregression in circular time series. The findings are also illustrated through real data examples.