Abstract: In order to illustrate a relatively new approach to research methods and data analysis, most commonly referred to as facet theory (Gratch, 1973), the present paper describes three examples of the approach in use. The range of data types and variety of modes of analysis which can be accommodated by this approach are exemplified, starting with a simple example dealing with energy conservation in universities, then moving on to examine differences between prisoners and staff in their evaluation of prison buildings, and finally considering the more complex methodological issues involved in establishing the structure of housing satisfaction. However, whilst this paper essentially concerns research procedures, especially the use of multivariate statistics, I wish to make the case that the facet approach holds particular potential for applied social psychology.

Abstract: In the classics of quantitative sociological analysis, such as The People's Choice (Lazarsfeld et al, 1948) or The American Soldier (Stouffer et al, 1949), percentage tables were the principal tools of statistical analysis Using this simple instrument, the researchers managed to shed empirical light on interesting theoretical ideas, such as cross-pressures or relative deprivation, presenting the results in a way that gave even the lay reader an opportunity to follow the steps in the statistical analysis Even today, most students are introduced to the logic and procedures of quantitative analysis by working with cross-tables, and many research reports still present their findings in the form of percentage tables The varying quality of these analyses indicates the critical importance of the imagination and skills of the researcher Over the last decade, however, more elaborate statistical techniques, going a long way towards routinising the analysis, have gradually taken over in articles in the leading journals as well as in advanced courses in statistics Such techniques provide systematic procedures for reaching precise answers to highly complex questions However, it may be asked whether the price paid for these advances in statistical sophistication i n terms of hard labour for students, a widening gap between the quantitatively oriented and the rest of the sociological community, and a virtual breakdown in communication with a broader public--has so far been offset by gains in the substantive interest of the results reached For a sociologist with an average training in statistics who tries to keep pace with the latest developments within this field, the highly technical presentation is frustrating, and the disagreements among the specialists concerning what seem to be factual questions are confusing This article does not engage in skirmishes along the frontier of advances in statistics; here, the

Abstract: When Pascal wanted to test one of the implications of Torricelli's hypothesis, he made his brother-in-law P6rier climb the Puy-de-D6me, a mountain some 4800 feet high, with a mercury barometer in his baggage. Torricelli believed the Earth to be surrounded by a sea of air, which, owing to its weight, exerts pressure upon the surface below. Torricelli had demonstrated that mercury could not be pumped up as high as water. Since mercury's specific gravity is - 14 times that of water, he expected mercury to reach 1/14 of the maximum height reached by water when pumped up by a suction pump. And just this happened. The fact that a suction pump which draws water from a well will lift water only - 34 feet induced Torricelli to formulate his hypothesis of a sea of air. And indeed, Pascal's brother-in-law P6rier became immortal in finding a mercury column at the top of the Puy-de D6me-mountain to be three inches shorter than when at the bottom. Air really seems to be less heavy and cause less pressure at the top of a mountain than at the bottom. This implies that a lower temperature would be needed to boil eggs on Mount Everest than in Maggie's Bed & Breakfast in downtown London. Moreover, Torricelli's idea immediately brought about the construction of the barometer and the altimeter. I read this story in Carl G. Hempel's Philosphy of Natural Science (1966, p. 9). We used this book as an introductory text for a seminar held at the University of Munich in Spring 1980. This seminar was dedicated to "Metaeconomics". Unfortunately, the participants of the seminar did not stick to natural science. Most participants had a rather limited knowledge of the natural sciences and were therefore embedded in large fields of speculation and discussion. However, they discovered the concept of rationality and some got caught in its trap. Rationality was identified as the gravity of

Abstract: Goodman (1972a) has introduced a measure of goodness of fit for log-linear models that is analogous to R 2, the squared multiple correlation coefficient. The measure has been used to advantage in several studies; one particularly good example is Rosenfeld's (1978) study of female occupational mobility.

Abstract: In this paper a number of relations are considered that hold between the individual members of a small organisation: relations of cognition or acquaintance, formal and informal communication, the affective relation corresponding to like or dislike, and the relation of direct superiority or subordinacy. For each of these relations a similarity or proximity relation is constructed. Two individuals will be related under the proximity relation derived from the informal communicat ion relation, say, when they are similarly embedded in the communicat ion structure. The proximity relations so constructed are not, in general, transitive: that is, individuals A and B may be proximately related, and so may B and C, with A and C unrelated. A proximity relation is therefore not an equivalence relation, and cannot be described by disjoint equivalence classes of individuals. However, maximally connected subgroups, or cliques, can be constructed, although these subsets will in general intersect one with another. The number of such cliques generated by each proximity relation gives an indication of the complexity of structure associated with the relation. For example, a simple proximity structure would be one in which every individual was proximately related to every other individual. A clique program based on the procedure of Harary and Ross (1957) and presented as an algorithm by Doreian (1970) was used to generate the clique structure associated with each proximity relation (see Alt and Schofield (1975, 1978) for details of this program). A proximity relation can be regarded as a representation of some aspect of the deep social structure of an organisation: from this follows the notion of embedding proximity relations in one another, to determine whether they are different reflections of essentially the same underlying structure. The intersection between two proximity relations can be used to determine the coefficient of scalability between the two relations. The coefficients of

Abstract: The question of sample size—how much is enough—has no simple answer. Magical numbers do not exist.