SUMMARY The technique set out in the paper, CHAID, is an offshoot of AID (Automatic Interaction Detection) designed for a categorized dependent variable. Some important modifications which are relevant to standard AID include: built-in significance testing with the consequence of using the most significant predictor (rather than the most explanatory), multi-way splits (in contrast to binary) and a new type of predictor which is especially useful in handling missing information.

An Exploratory Technique for Investigating Large Quantities of Categorical Data

SUMMARY This paper initiates an investigation into the statistical significance of results obtained by analysing data through the Automatic Interaction Detection technique. A test statistic which is asymptotically independent of the sample size is examined and critical values are obtained for a special case. The asymptotic results are compared with the exact distribution for varying sample size in a specific example. A brief discussion on the application of the results to a particular case study concludes the paper.

Significance Testing in Automatic Interaction Detection (A.I.D.)

The aim of this paper is twofold First we discuss the origin of tree methods Essentially we survey earlier methods that led to CHAID (Kass, 1980; Biggs et al, 1991) The second goal is then to explain in details the functioning of CHAID, especially the differences between the original method as described in Kass (1980) and the nowadays currently implemented extension that was proposed by Biggs et al (1991)

/pdf/chaid-and-earlier-supervised-tree-methods-5fpyed9119.pdf

CHAID and Earlier Supervised Tree Methods

The medical work of a 40 year birth cohort study is described in order to show the continuing associations of childhood factors with later life health, behaviour, illness and survival. Comparisons with the health of children born at later times show the importance of a sensitive monitoring of changes in child health. A summary of work on social and psychological topics is also given, together with a complete list of publications from the study.

Follow-up of the first national birth cohort: findings from the Medical Research Council National Survey of Health and Development.

Over the past ten to fifteen years, attempts have been made within various research institutes to develop mathematical models to help in educational planning. Recognition of the potential of models as planning tools has grown enormously over this period. In a very brief survey during 1969 of twenty member countries, the O.E.C.D. found 123 mathematical planning models with a status of either "finished" or "in progress"-and not all educational planning models within each country were listed (Kahn, 1970). This survey reflects the acceptance that the principle of the need to build models has achieved.

Mathematical Models Developed for Use in Educational Planning: A Review:

SUMMARY In this paper an attempt is made to identify factors which affect the behaviour of pupils in terms of academic survival. Structural features are included in the analysis. The data analysed are taken from a longitudinal national sample survey which makes it possible for each pupil to be classified by the educational criteria and by a number of different exogenous factors simultaneously. The method of analysis is based on a multivariate technique developed for initial exploration of simultaneous effects, by which no functional model is imposed and no additivity of effects assumed. 1.1. Purpose THE present study examines the ways in which factors exogenous to the formal structure of the education system affect the flow proportions of pupils from one stage of academic studies to the next. This investigation was undertaken as part of a larger study attempting to build a model of the system (Armitage et al., 1970). The basic model is being built with the forecasting of future trends and outputs in mind, and special attention is focused on the upper section of secondary education, from the statutory leaving age to higher education entry, in which opportunity and choice determine the type and level of education attained by individuals. A sub-model dealing in some detail with this section is worked on, the detail being in terms of school types, subject specialization, or bias, and sex. Although the use of model-building techniques in educational research is relatively new, the work done hitherto has necessitated a critical appraisal. Forecasts, even short-term ones, have turned out to be far from what actually happened in reality, suggesting that insufficient information concerning the underlying factors which affect pupils' behaviour and change over time is fed into the model. The attempt here is to identify some of these underlying factors by examining a number of personal, social and school features together with the structural divisions of the school system. It is a first step towards incorporating in a 15-19-year-olds model those factors which account most for differences in the behaviour of pupils, in order to increase its predictive power. Operationally, the following questions to be answered by the study were formulated:

Lea Orr

Papers

The Dependence of Transition Proportions in the Education System on Observed Social Factors and School Characteristics