Page 148 - Never Too Far Away? The Roles of Social Network Sites in Sojourners’ Adjustment

shared elements underlying these adjustment processes (Zhang & Goodson, 2011). It is indeed important to re-think the conceptualization of the two types of adjustment and explore possible frameworks that “would better reflect empirical evidence and open doors to re-integration of theories in explaining sojourner adjustment” (Zhang & Goodson, 2011, p. 625).
Methodological Contributions
By using a longitudinal panel design (Chapters 2 and 3) and comparative designs (Chapter 4 and 5), I was able to provide a scheme in teasing out the complex interplay of face-to-face and SNS interactions in international sojourners’ adjustment. In Chapters 2 and 3, I addressed the challenges and recommendations for future research articulated in extant literature that similarly called for the investigation of reciprocal causal associations between social interactions and subjective outcomes, and subjective outcomes and adjustment (Trepte & Scharkow, 2016; Meng et al., 2017; Verduyn et al., 2017; Stroebe et al., 2015a; Stroebe et al., 2015b; Thurber & Walton, 2005). Specifically, I tested a concurrent communication model by using cross-lagged and non-lagged reciprocal causality path analyses to account for the long-term and short-term reciprocal associations, respectively. The cross-lagged analysis assumes causal associations that happen over time; while the non-lagged analysis assumes causal effects that occur within a short span of time (Finkel, 1995; Kline, 2016). In essence, the non-lagged analysis disentangles the bidirectional causal association in correlations found in cross- sectional studies (Mathisen, et al., 2007). The use of a longitudinal panel design contributed to a more nuanced understanding of the reciprocal relations of social interactions, subjective outcomes, and adjustment. Although panel designs do not conclusively prove causality, they are useful in estimating reciprocal effects and assessing whether a set of results is consistent with a causal model (Finkel, 1995).
In Chapter 5, I conducted a multi-group confirmatory factor analysis, which included latent variable modelling and measurement invariance testing. Measurement invariance testing is an issue related to the current debate on methodological shortcomings in conducting research in the social sciences (Kühne, 2013). In comparative studies, testing for measurement invariance can be regarded as an additional step to ensure the integrity of measurement instruments and the comparability of the models. Despite its relevance, this methodological
146 Chapter 6