Browsing by Subject "http://www.yso.fi/onto/yso/p7135"

Societal norms guide the discussion on the transition to retirement. The norms dictate how and when one should retire and how one should spend retirement days. Norms are expressed in ways to speak about retirement, that is, in the narratives of retirement. The normative narratives guide individual choices and define who has retired successfully. If individual’s retirement does not meet the criteria described in the narratives, they may feel they have failed. In this Master’s thesis, the retirement narratives are searched from Finland’s largest online discussion forum Finland24, and the narratives are compared with the narratives found in previous studies. The data are the posts, mentioning the word retirement or pension, that were written on the forum between years 2001 and 2016. As the data consists of over 300 000 conversations threads, it is, first, thematically grouped using a machine learning method called Latent dirichlet allocation topic modeling. With the help of the model, it is possible to choose from the data only the conversations that contain relevant information on retirement narratives. Because computational topic modeling and internet’s big data have yet but few applications in the Social Science research, the second research question of the thesis is, how they can be applied in the research of this discipline. There is not yet consensus on the best practices of the method’s usage, hence the analytical choices made in this thesis are described in detail. An attempt is also made to develop the interpretation of the model’s results: a system is created for labelling the modelled topics and for finding the key themes among all the topics the model outputs. As a result of the modeling, five retirement specific themes were found from the forum: Social issues, Social security system, Social development, Retirement transition, and Life and feelings. Of these themes, Retirement transition was selected for the qualitative content analysis. The discussions within the theme reveal the typically normative nature of retirement discussion on the forum. What unifies the discussions is the view that choices and chances during career define how well one succeeds in retirement. This common narrative is called in the thesis the Retirement game. Before retirement, one has to work and pay pension payments. The most widely accepted reason for retirement is achieving pension eligibility age, and those who continue working after this are seen as cheating the game, threatening the younger workers. The winner of the game is the one who survives in paid work all the way to the pension eligibility age, manages to accumulate enough pension and enjoys their freedom to control how to use their time. By combining computational topic modeling and qualitative analysis, the thesis found retirement narratives that supplement the existing knowledge of them. The expressed norms were stronger in the discussion forum, but on the other hand, there were ways to retire that were completely against the norm. The results show that using similar data and methods it is possible to find new perspectives to existing scientific knowledge of Social Sciences’ research objectives. However, topic modeling and other computational methods require interdisciplinary expertise, and further research on their best practices and application possibilities is needed.