Preliminary results of the analysis of various subsystems of society using the coefficients of paired correlation of Pea
Posted: Tue Apr 22, 2025 6:55 am
Research tasks of identifying relationships between quantitative/ordinal variables are often solved by the method of correlation analysis. Researchers are interested in the presence/absence of a relationship, and in identifying its strength, direction sign. Existing works devoted to the analysis of differences in the choice of a particular correlation coefficient for data analysis ultimately come down to mathematical variations on the theme of using parametric and nonparametric statistics tools, compensating for outliers that cause a sharp drop in Pearson's Phi values with certain sample sizes, etc. Substantive differences in measuring the distances between expert assessments (Pearson ) and the ranks of these assessments (Spearman ), on the one hand, and the number of permutations required for the coincidence of a number of expert assessments of the gradations of a social object among two experts (Kendall ), as a rule, are not touched upon. The authors of the report apply Tay - b -Kendall for typologies with a study of the values of the Internet audience in Donetsk in 2010The specifics of the results of applying different coefficients require presentation.
The empirical basis of the study was formed by the data [url=https://dbtodata.com/argentina-number-data/]argentina phone number data[/url of the projects "Public Opinion" "Pravda -1977", "Pravda -1990", Civil Expertise of the Sphere of Management 2012-2016, OSPI-2019-2022, Soviet-American studies of time budgets of 1986 (Pskov and Georgetown) and the Russian Federation 1993 (CESSI ), ESS - 2006-2021, research of the Department of Sociology of Family and Demography of Moscow State University, official reports of the UN, ILO.
The research methodology has a multi-level nature of analysis preparation. The studies were selected based on the content of two subject "fields " of social reality: factual and evaluative. The first is represented by the time budgets of the aforementioned studies and 9672 documents of the "Pravda -1990" study related to the food sphere (standard WHO questionnaire for identifying the frequency of consumption of 75 types of products). The evaluation series contain 5, 6, 7, 10-point assessments by each respondent of a wide range of various areas of public life (acute social problems, ways to solve them, assessments of the interaction of the population and the authorities, the work of politicians at all levels of government, life orientations and motives for self-realization, attitudes towards migrants, color preferences, etc., etc.). The individual assessments were based not only on standard point scales, but also on the semantic differential, M. Luscher color test cards for determining the degree of anxiety, and ranking. A range of not only sociological, but also psychological methods was used. “Removal ” of information has always been carried out in a standard way – by the method of multidimensional scaling of a certain social parameter (assessment of each of its gradations).
The resulting fields of individual assessments were standardized in processing. All arrays with a volume of more than 1000 respondents were reduced to a common "denominator " of 1000 documents * . In the factographic and assessment fields, distributions close to normal from the average assessment were found. This gave the right to build "triads " of layers according to the criterion ±1s. Subtracting the average from the value of each assessment of the parameter gradation for each respondent polarized the assessments of people's activity and attitudes toward the areas of social life being studied.
At the next level of structuring the object, all the obtained matrices of 1000 respondents with assessments of certain parameters were inverted. The variables became "cases " and the respondents became variables. The correlation coefficients of Pearson, Kendall, and Spearman were obtained between the coincidence of rows for each of the 1000 respondents with the remaining 999 respondents in each of the 75 measurements. In this way, 225 (75 by the number of parameters for each coefficient) correlation matrices were obtained with a single numbering of 499500 cells {[ (1000 x 1000) - 1000] / 2 = 499500}. The filling of the matrices with the values of the coefficients Phi , Tay - b , Po , taken both as a whole and in the 95 percent confidence interval, differed, but there were also patterns. The layers of a certain number of significant correlations were considered in terms of the polarization of the assessments of the state of affairs.
The empirical basis of the study was formed by the data [url=https://dbtodata.com/argentina-number-data/]argentina phone number data[/url of the projects "Public Opinion" "Pravda -1977", "Pravda -1990", Civil Expertise of the Sphere of Management 2012-2016, OSPI-2019-2022, Soviet-American studies of time budgets of 1986 (Pskov and Georgetown) and the Russian Federation 1993 (CESSI ), ESS - 2006-2021, research of the Department of Sociology of Family and Demography of Moscow State University, official reports of the UN, ILO.
The research methodology has a multi-level nature of analysis preparation. The studies were selected based on the content of two subject "fields " of social reality: factual and evaluative. The first is represented by the time budgets of the aforementioned studies and 9672 documents of the "Pravda -1990" study related to the food sphere (standard WHO questionnaire for identifying the frequency of consumption of 75 types of products). The evaluation series contain 5, 6, 7, 10-point assessments by each respondent of a wide range of various areas of public life (acute social problems, ways to solve them, assessments of the interaction of the population and the authorities, the work of politicians at all levels of government, life orientations and motives for self-realization, attitudes towards migrants, color preferences, etc., etc.). The individual assessments were based not only on standard point scales, but also on the semantic differential, M. Luscher color test cards for determining the degree of anxiety, and ranking. A range of not only sociological, but also psychological methods was used. “Removal ” of information has always been carried out in a standard way – by the method of multidimensional scaling of a certain social parameter (assessment of each of its gradations).
The resulting fields of individual assessments were standardized in processing. All arrays with a volume of more than 1000 respondents were reduced to a common "denominator " of 1000 documents * . In the factographic and assessment fields, distributions close to normal from the average assessment were found. This gave the right to build "triads " of layers according to the criterion ±1s. Subtracting the average from the value of each assessment of the parameter gradation for each respondent polarized the assessments of people's activity and attitudes toward the areas of social life being studied.
At the next level of structuring the object, all the obtained matrices of 1000 respondents with assessments of certain parameters were inverted. The variables became "cases " and the respondents became variables. The correlation coefficients of Pearson, Kendall, and Spearman were obtained between the coincidence of rows for each of the 1000 respondents with the remaining 999 respondents in each of the 75 measurements. In this way, 225 (75 by the number of parameters for each coefficient) correlation matrices were obtained with a single numbering of 499500 cells {[ (1000 x 1000) - 1000] / 2 = 499500}. The filling of the matrices with the values of the coefficients Phi , Tay - b , Po , taken both as a whole and in the 95 percent confidence interval, differed, but there were also patterns. The layers of a certain number of significant correlations were considered in terms of the polarization of the assessments of the state of affairs.