A Psychological Approach to Big Data Analysis: The Implicit Symbolic Meanings in Online Content (ISMOC) Method

The broad and growing field of big data - dealing with the storing, retrieval, and analysis of large amounts of information - is recently offering several opportunities to psychological research. The relationship between user and content is one of the main foci of observation of current research in social network analysis and online content analysis (Sapountzi & Psannis, 2016). About the latter, several techniques are used for the content retrieval (e.g., API, crawler, scraping scripts or commercial monitoring tools) and for the subsequent content analysis depending on whether the aim of the analysis is toward thematic and topic detection analysis or sentiment analysis (e.g., Natural Language Processing, Machine Learning Techniques or Artificial Neural Networks). Specifically, scientific literature (e.g., Harlow & Oswald, 2016) has highlighted the relevance of social media and online tools to explore people’s emotions, attitudes and behaviours within real-world contexts, given the availability of very large datasets regarding a broad spectrum of the population and the easier access to geographically diverse respondent groups. As well, big data research promotes greater collaboration across disciplines such as social sciences, computer science and applied statistics, contributing to innovative hypotheses generation and model testing across different scenarios, which may be fruitful for both the research and applied advancements of psychological science. The present chapter aims at proposing an innovative research method, named Implicit Symbolic Meanings in Online Content (ISMOC), which allows the analysis of big data according to a mixed approach, taking into account both quantitative and qualitative information and integrating computer science techniques with psychological insights. If a complex challenge for professionals and scholars working in the field is to make sense of this extensive amount of information, the unique feature of the ISMOC method allows the interpretation of hidden and symbolic meanings within the internet-based collective discourse, beyond thematic content classification or general sentiment analysis. Indeed, the ISMOC method is grounded on a theoretical framework derived from socio-constructivist and psychodynamic models, looking at discursive practices as an expression of shared sense-making processes. From such a perspective, language can be conceived as a means to anchor and objectify new phenomena in order to construct a socially represented knowledge, consistently with Social Representation Theory (Moscovici, 1984). Besides this, by relying on the “double reference” principle - both lexical and symbolic - implicitly connected to the language (Fornari, 1979), it is possible to capture the affective and symbolic dimensions running through discourses, apart from their intentional structuring or cognitive sense. Indeed, through the use of multivariate text analysis techniques, lexical associations with the research object are detected that may reveal implicit symbolic patterns, regulated by the generalization and symmetry principles of the unconscious thought according to a bi-logic theory of mind (Matte Blanco, 1975). The chapter illustrates the different steps and procedures of the ISMOC method (from research plan to data interpretation); as well, some practical (e.g., ethical considerations) and technical issues (e.g., problems of reliability and validity) are discussed. Two research studies are then presented, respectively focused on migration-related issues and dieting-related eating behaviours, as examples in which the method is applied to real data and may provide interesting insights about the implicit discourse existing behind the social media messages and the online representation of these phenomena, adding a new dimension to the knowledge of such themes with direct implications for public awareness and social and health policies or interventions.