Correlates of subjective well-being of the elderly were scrutinized in a national representative sample of elderly community residents. A stratified multistage random sampling yielded 2,240 eligible persons, and 80.8% of them were successfully interviewed. In the present study, data of 1,510 subjects responded by themselves without missing observations were analyzed.
     Covariance structure model analysis (structural equation model) was performed with a model consisted of 3 exogenous and 9 endogenous variables. The model was improved through a hierarchical deletion of parameters until Chi-square/df ratio reached 2.0. The endogenous outcome variable was life satisfaction measured by the Life Satisfaction Index K.
     Three variables having significant direct effects upon life satisfaction were identified. The variables were functional health status, income, and coresidence with married child(ren). The effects of functional health status and income were positive, and the subjects living together with married child(ren) in the same household showed higher life satisfaction than those without co-residing child(ren). Social networks did not have direct effects upon life satisfaction.
     These results seem to confirm the findings of previous studies. However, given the lack of direct effects of social networks and the small value of multiple correlation coefficient of life satisfaction (R=.28), new instruments to measure social environment of elderly persons need to be developed.

View full abstract

     A procedure of using MDPREF to analyze paired comparison or dominance data on items and to disclose between and within group differences when subjects are consisted of several different groups is presented. The procedure consists of four phases. (a) To cluster subjects of each group by applying some of cluster analysis methods. (b) To derive an inner product matrix among items for each cluster. (c) To analyze these inner product matrices by INDSCAL, and (d) To derive subject configuration (subject vectors) of MDPREF by an external MDPREF analysis based on the common object (item) configuration given by the INDSCAL analysis. The procedure inherited the uniquely oriented dimensions of the object (item) configuration from INDSCAL. The procedure is applied to the survey data comparing eating habits among three cities.

View full abstract

     The implications and applicability of Hermitian from model are discussed for the analysis of asymmetric structure observed frequently in our daily life between objects such as stimuli, persons, regions and nations. The Hermitian form model includes the finite-dimensional complex (f. d. c.) Hilbert space model, the indefinite metric model for asymmetric multidimensional scaling in psychometrics, and the Hermitian canonical model for two-way contingency tables. It is shown that the f. d. c. Hilbert space model is necessary to argue the holistic geometrical structure contained in an observed similarity or dissimilarity matrix, whose elements are composed of the degrees of proximity from object to object. It is also shown that the f. d. c. Hilbert space model has an interesting property. That is, two points in a relatively central region (i. e., near the origin) of the space have a smaller similarity measure than two points of equal interpoint distance but located in a peripheral region (i. e., far from the origin) of the space. It is suggested that we shall sometimes encounter the situation in which the f. d. c. Hilbert space structure is destroyed, and that in such a case the indefinite metric model might serve well. Finally, the possibility of adopting the Hermitian canonical model is suggested in cases where the usual canonical model is not appropriate for the analysis of two-way contingency tables.

View full abstract

     It becomes easier to use multivariate analyses than before when personal computers and statistical software were not available. Social and behavioral sciences also benefited by them which stimulated various kinds of researches in these fields. But the stimulation induced negative effect on these fields as well, i. e. inappropriate application of multivariate analyses. They can be classified into two types. One is not directly related with the analysis but caused by the earlier which precede the analysis. The other is in the analysis itself. Although personal computers and statistical software packages have accelerated the second type of the inappropriate applications of multivariate analyses, there are many kinds of inappropriate applications of multivariate analyses which have no relationship with personal computers and statistical software packages but caused by the researchers themselves. In the present paper we focus our attention onto the second type of inappropriate applications of multivariate analyses. Typical examples of inappropriate applications of multiple regression analyses, factor analyses, cluster analyses, and multidimensional scaling analyses are presented and discussed. Problems which cause inappropriate applications of statistical or quantitative analysis procedures are talked over, and some measures to overcome them are proposed.

View full abstract

     Data analyses in sociology deal with concerning micro (or individual) variables and data concerning macro variables. The latter consists of two types of data: One is “aggregative data” computed by aggregating data obtained from each single individual, while the other is “collective data” directly reflecting attributes of a collectivity as a whole. The present study classified the “collective data” into four categories, and reviewed the methods for measuring data in each category. The four categories are (1) data obtained by more sophisticated aggregation than those by simple aggregation for “aggregative data,” (2) data deduced from a matrix summarizing relations among members of a collectivity, (3) data collected by direct measurement of characteristics of a collectivity, and (4) data obtained by indirect measurement of characteristics of a collectivity. The present study demonstrated that only a few techniques had been developed in the categories (3) and (4). Further methods for measuring macro-variables, which may be distinctive form field to field, should be explored by conducting field research from the viewpoint of micro-macro linkage.

View full abstract

     Traditional theories of power premised the existence of it. However, in this essay, I try to prove that free actors have power relation, and insist that power as action does not exist. We can use the term “power” when we describe social relation which has particular features. However this does not mean that power as action exist. So, if the concept of power implies rejection of actor's freedom, it is not only redundant but also false. Because “to be free” and “to be in power relation” are compatible.

View full abstract

     Using the data of an Australian national survey in 1973-74, this study investigates the social stratification of the Australian society in terms of the status consistency and inconsistency. Three status variables (educational attainment, occupational prestige, and income) were analysed by the method of cluster analysis, and the analysis has revealed the following;
     (1) The cluster analysis produced three status consistent clusters, and four status inconsistent clusters. While the former accounted for 49.1 per cent, the latter formed 43.6 per cent.
     (2) The following mechanisms which distributed the the social resources were found. As a result of high schools, universities, and colleges of advanced education popularising quite recently, younger people tended to be highly educated. Occupational prestige of farmers was too high for their educational attainment. Also income of the manual employees tended to decrease due to lack of the seniority system. when they reached around forty years of age. However, the self-employed manual workers obtained much higher income than the manual employees. The different language and culture overseas immigrants in such a disadvantageous position that they could not get a job meeting the standards of their educational attainment. As a result, their income tended to be lower.
     (3) Although status inconsistency had no effect on class self-identification and voting behaviour, the status variables did.

View full abstract

     We present a thesis which has a negative implication to the power theory of Ehara. We examine the significance of the thesis. First, we define “existence of power” by amplifying Ehara's discussion. We formalize “willing choice” in her theory. Secondly, we demonstrate impossibility of escape from “Power” defined by Ehara, by invoking Sen's “liberal paradox.” We guess that Sen's condition “unlimitation of preference” is not really satisfied and that the limitation of preference is related to “context”. Finally, we cast a about on “power” theory approaches in feminism.

View full abstract

     Several research groups are currently studying the distribution of power in social exchange networks. This paper first outlines the history of the dispute, which stimulated the current interests in network exchange theory, between two research groups-Cook, Yamagishi and their associates who have been trying to extend the power/ dependence theory originally developed by Emerson beyond its original dyadic formulations, on the one hand, and Willer, Markovsky and their associates, on the other-concerning the proper method for predicting the distribution of power across network positions. We then present a new equidependence algorithm (EDA) that was developed to overcome the limitations of the old algorithm based on the power/ dependence formulations. The predicitions by the new algorithm were compared with those by GPI proposed by Willer, Markovsky and their associates against simulation results. These comparisons demonstrate the superiority of EDA over GPI.

View full abstract

     Let us assume a group divided into two sub-groups in conflict. We also assume that each combination of two members of the group has friendship if they belong to the same sub-group, while it has hostility if they belong to the different sub-groups. Let $N$ be the number of all members in the group. Let $x$ be the number of all members belonging to a sub-group. Then, the total number of hostility between the members in the group is $x(N-x)$. Let $y$ be the hostilities ratio to all the “relations” in the group, then
     y=¥frac{x(N-x)}{_{N}C_{2}}=¥frac{2x(N-x)}{N(N-1)}
Only when $¥frac{N}{2}-¥sqrt{N}<x<¥frac{N}{2}+¥sqrt{N}$, $y$ is more than 1/2. If $N$ is 10,000, $y$ is more than 1/2 only when 4,900<x<5,100. That is to say, if $N$ is sufficiently large, $y$ is more than 1/2 only when the group is divided into almost right halves. We may expect that in most cases there are the majority and the minority when a group is divided into two sub-groups. Thus, we assume that $x$ will emerge with equal probability between 0 and $N$. Let $P$ be the expected ratio of hostilities of the relations, then,
     P=¥frac{1}{N+1}¥sum_{x=0}^N ¥frac{2x(N-x)}{N(N-1)}=¥frac{1}{3}.
Therefore, we may assume that in most cases we will find more friendship than hostilities in a social group.

View full abstract

     Umino's commons dilemma model aims to formulate “the tragedy of the commons”. However, Misumi pointed out that (1) this model failed to formulate “the tragedy of the commons”, and that (2) this model was also inappropriate for a model of social dilemma. This paper re-examines Umino model on the second point above. At first, the discussion by Misumi as to (2) is inappropriate. Finally, I fundamentally agree with the conclusion by Misumi again. However, I simultaneously clarify that it is possible for Umino model to formulate the dilemma situation where costs are equal among actors, and moreover the redistribution system is introduced.

View full abstract