Data Science meets Institutionalism : How to measure jobs' persistence over time using a text mining algorithm on job advertisements

In a process of institutionalization meanings increasingly produce shared patterns of behavior that become generalized and diffused until they are fully detached from their creators and transmitted across generations (Berger & Luckmann, 1967; Zucker, 1977). On basis of this theoretical premise, six jobs could be ranked according to their degree of institutionalization in descending order (call-center employee, kitchen helper, plant electrician, IT system engineer, product manager and personnel developer). This study introduces machines learning and large-scale quantification to a field that has hitherto been dominated by hermeneutical approaches such as close-reading of interview scripts or ethnographic notes (Mohr, 1998). Strongly guided by a theoretical framework, this study shows how to process and extract value from Big Data. In fact, the process of institutionalization is translated to a step-by-step cultural analysis. Meanings that remain dormant among relations and patterns in texts are captured through an algorithmic text mining tool, aggregated to a higher analytical level and processed further for quantitative and qualitative analysis. This procedure has been applied to online job advertisements released in Austria in 2019. Through the distinct theoretical lens that I employ, they are treated as knowledge domains and provide insights into meaning-making processes embedded in larger socio-political opportunities and constraints (R. Meyer & Höllerer, 2010). Thus, the study informs the co-constitution of cultural meanings, social and institutional processes. In that, it has provided a first guideline to measure the degree of institutionalization of jobs.
Arbeit an der Bibliothek noch nicht eingelangt - Daten nicht geprüft
Innsbruck, Univ., Masterarb., 2021