TY - BOOK

T1 - Method versatility in analysing human attitudes towards technology

AU - Lezhnina, Olga

PY - 2023

Y1 - 2023

N2 - Various research domains are facing new challenges brought about by growing volumes of data. To make optimal use of them, and to increase the reproducibility of research findings, method versatility is required. Method versatility is the ability to flexibly apply widely varying data analytic methods depending on the study goal and the dataset characteristics. Method versatility is an essential characteristic of data science, but in other areas of research, such as educational science or psychology, its importance is yet to be fully accepted. Versatile methods can enrich the repertoire of specialists who validate psychometric instruments, conduct data analysis of large-scale educational surveys, and communicate their findings to the academic community, which corresponds to three stages of the research cycle: measurement, research per se, and communication. In this thesis, studies related to these stages have a common theme of human attitudes towards technology, as this topic becomes vitally important in our age of ever-increasing digitization. The thesis is based on four studies, in which method versatility is introduced in four different ways: the consecutive use of methods, the toolbox choice, the simultaneous use, and the range extension. In the first study, different methods of psychometric analysis are used consecutively to reassess psychometric properties of a recently developed scale measuring affinity for technology interaction. In the second, the random forest algorithm and hierarchical linear modeling, as tools from machine learning and statistical toolboxes, are applied to data analysis of a large-scale educational survey related to students’ attitudes to information and communication technology. In the third, the challenge of selecting the number of clusters in model-based clustering is addressed by the simultaneous use of model fit, cluster separation, and the stability of partition criteria, so that generalizable separable clusters can be selected in the data related to teachers’ attitudes towards technology. The fourth reports the development and evaluation of a scholarly knowledge graph-powered dashboard aimed at extending the range of scholarly communication means. The findings of the thesis can be helpful for increasing method versatility in various research areas. They can also facilitate methodological advancement of academic training in data analysis and aid further development of scholarly communication in accordance with open science principles.

AB - Various research domains are facing new challenges brought about by growing volumes of data. To make optimal use of them, and to increase the reproducibility of research findings, method versatility is required. Method versatility is the ability to flexibly apply widely varying data analytic methods depending on the study goal and the dataset characteristics. Method versatility is an essential characteristic of data science, but in other areas of research, such as educational science or psychology, its importance is yet to be fully accepted. Versatile methods can enrich the repertoire of specialists who validate psychometric instruments, conduct data analysis of large-scale educational surveys, and communicate their findings to the academic community, which corresponds to three stages of the research cycle: measurement, research per se, and communication. In this thesis, studies related to these stages have a common theme of human attitudes towards technology, as this topic becomes vitally important in our age of ever-increasing digitization. The thesis is based on four studies, in which method versatility is introduced in four different ways: the consecutive use of methods, the toolbox choice, the simultaneous use, and the range extension. In the first study, different methods of psychometric analysis are used consecutively to reassess psychometric properties of a recently developed scale measuring affinity for technology interaction. In the second, the random forest algorithm and hierarchical linear modeling, as tools from machine learning and statistical toolboxes, are applied to data analysis of a large-scale educational survey related to students’ attitudes to information and communication technology. In the third, the challenge of selecting the number of clusters in model-based clustering is addressed by the simultaneous use of model fit, cluster separation, and the stability of partition criteria, so that generalizable separable clusters can be selected in the data related to teachers’ attitudes towards technology. The fourth reports the development and evaluation of a scholarly knowledge graph-powered dashboard aimed at extending the range of scholarly communication means. The findings of the thesis can be helpful for increasing method versatility in various research areas. They can also facilitate methodological advancement of academic training in data analysis and aid further development of scholarly communication in accordance with open science principles.

U2 - 10.15488/14119

DO - 10.15488/14119

M3 - Doctoral thesis

CY - Hannover

ER -