Decision trees are a method of classification and prediction, widely applied in sociological research. It is unchangingly popular due to its flexibility and simplicity of interpretation. Choosing the most appropriate decision tree algorithm is not an easy task for several reasons: (a) there is already over a hundred algorithms with different strengths, weaknesses and logic of growth, (b) literature on the topic is fragmented and optimized versions of existing algorithms are often presented as entirely new types of trees and (c) the statistical software is equally as fragmented as literature. As a result, researchers often apply algorithms that are available in their preferred statistical package or rely on one of the old, imperfect methods. The review aims to reveal and describe decision tree algorithms and their application, discussed over the last five years. Bibliographic analysis and a network of keywords are applied to reveal the path of current scientific discussion on the topic.
The article studies educational trajectories of schoolchildren in Yaroslavl Oblast. Conclusions point out that schoolchild’s educational achievements, educational plans of his/her friends and the level of education of his/her father are key predictors of a decision about continuing education. Thanks to this information it is possible to know which schoolchildren are at risk of not continuing their studies. In the course of the research comparative advantages of the logistic regression and the discriminant analysis in the case of binary dependent variables were examined. With the necessary prerequisites for the use of methods fulfilled, both strategies work well classifying schoolchildren.
We used texts published in web blogs after the Domodedovo terrorist attack in January 2011 in order to demonstrate the possibility of reconstruction of actor categories, which are blamed for this unfavorable event. A system of actor/action categories is constructed, indicators of responsibility are defined. We also conducted a pilot study in order to analyze the intercoder reliability that provides evidence of the applicability of the constructed categories in further empirical research.
Development of an inclusive society challenges researchers to study
representatives of various social groups, including those with unique
cognitive and communicative characteristics. One of these groups is the
deaf and hard of hearing, which differs significantly from other people with
disabilities in cultural and linguistic aspects. This article deals with the
difficulties that sociologists face interviewing the deaf and hard of hearing.
The authors consider the main cognitive and communicative features of this
group that influence the course of questioning, the role of sign language
in the perception of a sociological survey (including in writing form),
analyze the specifics of question-and-answer communication with deaf
respondents, which consists, first of all, of a difficult semantic and contextual
understanding of questions. The authors also describe the specifics of
information perception by respondents with hearing impairments using
specific examples from the researcher practice. Based on expert interviews
and methodological reflections on the results of several empirical studies,
the authors suggest possible ways to improve the quality of communication
between the hearing researcher and the deaf respondent, as well as increase
the validity of the data obtained during the survey.
This paper is devoted to the detailed description of pictorial technique and steps of pictorial data coding in the empirical study on social representations of prestigious job. The procedure of interpreting images, specifically the procedure of pictorial sociological data is proposed in addition to the existing in research practice approaches to the pictorial data analysis. This procedure is based on the 6-vector image analysis, and also contains a number of rules and recommendations for working with pictorial data. The study is accompanied by a large number of illustrations and methodological explanations.
This article describes the expierence of studying factors influencing the social well-being of educational migrants as mesured by means of a psychological well-being scale (A. Perrudet-Badoux, G.A. Mendelsohn, J.Chiche, 1988) previously adapted for Russian by M.V. Sokolova. A statistical analysis of the scale's reliability is performed. Trends in dynamics of subjective well-being are indentified on the basis the correlations analysis between the condbtbions of adaptation and its success rate, and potential mechanisms for developing subjective well-being among student migrants living in student hostels are described. Particular attention is paid to commuting as a factor of adaptation.
Two approaches to the selection of sources for a systematic literature review are compared — expert approach and algorithmic approach, based on an analysis of citation networks. Comparative analysis is conducted on the example of the research field of studying mass behavior. A comparison of two approaches according to formal and substantive criteria allows us to draw conclusions about significant difference, but also complementarity of the obtained results. Sources selected by the algorithmic approach do not reproduce the history of development of the research field described through sources selected by the expert approach. Adding expert elements to the algorithmic approach leads to a result that gives an idea of the links between classical and modern publications and groups of modern works that are interesting for study. The data obtained allow to draw a conclusion about the main reasons for the inconsistency of the results and formulate recommendations for the implementation of the proposed methodology in other areas.
This article investigates the preconditions and underestimated consequences of the emotional experience of fieldwork. The lack of theoretical and methodological discussion on the subject of researcher’s “emotional work” in Russian science misrepresents specifics of interview and ethnography. As opposed to that, the perfection of conceptual schemes and research tools, as well as formal ethical procedures, is highly emphasized. Therefore misperception and unpreparedness of those who are up for their first empirical material can lead to a painful encounter with reality. The epistemological, ethical and ontological consequences of such “trauma” can greatly influence both the researcher's strategy and his or her personality. The article briefly summarizes some foundations of “qualitative” methods, and compares different role of emotions in disciplinary fields of sociology and anthropology, and investigates complications and risks of emotions during fieldwork.
Multiple imputation is an approach to missing data elimination created by Donald Rubin. The purpose of multiple imputation is to reconstruct the initial structure of data, i.e. to generate the answers as close as possible to hypothetical complete dataset. However, the original algorithm of multiple imputation is complicated and demands a major amount of effort to accomplish. In the study simpler alternative approach –averaging of imputed values – was experimentally tested against Rubin’s rule in a number of common research situations. We compared two approaches to multiple imputation results aggregation – Rubin’s rule and averaging of imputed values – considering given analytical tools, share of missing values and type of the variable that contains missing values. The results were summed up in a set of recommendations describing a pertinent approach to aggregation for each research situation.