Comparability of scores between culturally and socially different is always a problematic question. It is even more questionable when the scores of latent variables are compared. Latent variables are frequently measured with several indicators, and the structure of indicators may differ between groups, resulting in the scores of the latent variables that may turn to be very different in terms of configuration and scale. This problem was labeled measurement invariance (MI) and became a necessary part of the studies with latent constructs. The lack of MI between groups may lead to biased or wrong conclusions. MI is an issue especially in cross-cultural studies, in which cultural differences as well as translation of questionnaire may evolve differences in latent construct structures. The paper discusses different methods of assessing MI and uses multiple group confirmatory factor analysis to test MI of four Schwartz higher order values among four populations surveyed during 4 and 5th rounds of European Social Survey. Since our interest is the sources and the extent of MI, only samples surveyed in Russian language were selected. The results show only partial scalar invariance that allows for comparison of means across groups. However, full scalar invariance is not achieved due to the differences in translation to Russian between four countries.
This article deals with the science citation indexes and touches upon the problem of transformation of practices of their application. The history of the first citation index is retraced: the main goal of making this index was initially to put in order the growing output of scientific literature and to improve a bibliographic search. The emergence of another way to use – as a tool for evaluation of scientists’ productiveness – is connected with the specifics of bureaucratic control over professional activity. An administrator has to orient himself mainly to the system of formal indicators. The article describes the existing practices of application of citation indexes and latest innovations in this sphere in modern Russia.
Throughout most of their history, sociologists have sought to study unstructured organic texts: newspaper materials, diaries, memoirs, letters, documents, and, more recently, messages, publications and other texts on various online platforms. This article discusses how modern techniques of text mining can improve classical sociological approaches to the analysis of this type of data. The article is structured according to the following plan. First, examples of classical quantitative content analysis and its limitations are discussed that could be solved with the help of text mining. Then I discuss how text mining is applied in contemporary social science research with topic modeling and text classification. Finally, I conclude with a discussion of some of the current approaches to text analysis using deep learning, as well as theoretical issues related to the application of text mining
In the social sciences, researchers are often faced with the task of classifying objects into meaningful groups, including in education. However, homogeneous groups can be extracted basing on various conceptual foundations leading to the identification of different group structures which might influence the interpretation of the results. The purpose of this study is to compare the structure of groups identified by the methods of cluster analysis and cut-off classification taking into account possible interpretations of results. The data analyzed are drawn from the START large-scale assessment study. This is a large-scale assessment project that involved 2645 first-grade students from the Krasnoyarsk region in 2018. Three cases of students’ grouping were investigated by k-means cluster analysis, latent profile analysis, and cut-off method. The results of the study show that the k-means method cluster structure is similar to the cut-off method group structure while the results of the analysis of latent profiles are different. The results of the study show which of these methods is applicable depending on the hypothesis and the purpose of eliciting groups.
In this paper we analyze social stratification measuring methods including both “subjective” and “objective” approaches. We focused on widely spread procedures based on professional positions of social actors. Three big groups of methods are described: occupational prestige scales, socioeconomic index, and social class typology. Detailed description of procedures and obtained results are included.
The article problematizes the methodological assumption that in participant observation achieving a role of high degree of involvement indispensably needs a long-term intensive interaction as a unique condition, which allows researcher to become familiar for the observed people while negotiating a primary role as observer. Based on the data of a short-term participant observation over 8 Russian voluntary organizations, it is demonstrated that some roles with high degree of involvement can be achieved in a short-term participant observation, if the observer manages to adapt to the rules and norms of the studied community, and to the general context of the participation. Particularly five facilitating factors are described: 1) the observer's insider stance, 2) high degree of openness of the observed community for outsiders, 3) performing non-specific functions by the insiders, 4) institutionalization of roles with intensive involvement, available for outsiders, 5) personal characteristics of the observer.
The focus of this article is the methodological aspect of political activism determinants identifying; specifically variants of handling with categorical predictors which hypothetically explain the level of activism. When using regression for explaining the issue, one may transform such predictors into dummy variables. Such a popular solution makes the model bulky and causes troubles with assessing this model’s quality. Moreover, if a researcher wants to consider interaction effects of the mentioned predictors, the supernumerary combinations of the mentioned predictors values are pended because regression modeling does not take into account the degree of similarity of the mentioned predictors values’ effects. The article authors proposed CHAID as the alternative to the mentioned solution. The research’s aim was i) a comparison of the two mentioned methods leaning on their a priori known properties; ii) arguing CHAID’s some theoretical advantages comparing to logistic regression, iii) parallel implementing the two methods, iv) a comparison of gained empirical results and v) arguing that it is useful to examine multiple interaction effects when developing a predictive model. The raw data were extracted from ESS 2012. The dependent variable was Political activism; the hypothetical predictors belonged to the socio-economic bloc of the Panel.