文章基本信息

标题：Comparing manual and computational approaches to theme identification in online forums: A case study of a sex work special interest community
本地全文：下载
作者：Pelham Carter ; Matt Gee ; Hollie McIlhone 等
期刊名称：Methods in Psychology
印刷版ISSN：2590-2601
出版年度：2021
卷号：5
页码：100065
语种：English
出版社：Elsevier
摘要：Online forums afford individuals opportunities to take part in a community with shared interests and goals. This involves the sharing of experiences and advice (Attard and Coulson, 2012) and can lead to positive effects (Pendry and Salvatore, 2015). Online forums also afford access to rich sources of detailed data, personal experiences, and hard-to-reach or taboo communities. Such online research, though well-suited to qualitative analysis, leads to a number of practical problems in terms of range, depth, and ease of access to data. Even extensive data collection and manual analysis often only engage with a small percentage of the data available in online communities.In this article, we present a traditional manual collection and thematic analysis of data (2631 posts across 60 different threads, approximately 300,000 words) from forums where sex workers and men who pay for sex discuss matters relating to prostitution. This analysis revealed five themes of forum use: preference sharing, personal narrative sharing, practical advice, philosophical issues, and community maintenance. Further automated data collection and corpus analysis, such as keyness and topic modelling, are presented as a potential innovation within online qualitative research. This approach allowed for the analysis of a larger dataset of 255,891 posts, across 14,232 threads (16,472,006 words), revealing additional themes such as sexual hygiene, desire, legality, and ethnicity, as well as differences in the use of terms of address and slang by punters and sex workers. The automated methods presented allow for more comprehensive investigations of online communities than traditional approaches, but we also note that manual interpretation should still be incorporated into the analysis.