Page 155 - Never Too Far Away? The Roles of Social Network Sites in Sojourners’ Adjustment

grained and longer time-intervals in assessing developmental trajectories. The number of waves for the types of analyses performed in this current dissertation were sufficient. It is suggested, however, to consider increasing the number of assessments for greater flexibility in multilevel analytical procedures (Hamaker, 2018).
Across the various studies in this dissertation, I made use of self-report measures. There are several limitations to the validity of self-reports of media use and adjustment (see e.g., Araujo, Wonneberger, Neijens, & de Vreese, 2017; Boyle, 1985; Junco, 2012; Rosen, Whaling, Carrier, Cheever, & Rokkum, 2013; Stroebe et al., 2015a). Although it has been shown that self-reports highly correlate with actual time spent on SNS (as measured by Facebook use computer monitoring software) (Junco, 2012), it is important to consider a more comprehensive measure of SNS use in future studies (e.g., include actual time spent on SNS; comparable across studies) (Rosen et al., 2013). Along the same lines, I suggest more precise, as well as more comprehensive measures of depressive symptoms that can be used to validate (e.g., neurophysiological measures) and characterize (e.g., state-trait interactions) self-reports (see e.g., Boyle, 1985). This way, professional support providers are much more informed in the design of intervention programs for sojourners.
There were a few single-item measures used in this dissertation (e.g., frequency of social interactions, homesickness, perceived relationship stability, perceived relationship satisfaction). In the empirical chapters (Chapters 3 and 4), I argued that the use of these single-item measures was theoretically and methodologically justified; for example, the appropriateness of single item face- valid measures, the necessity of using shorter versions of the scales to keep the survey short, as well as to avoid overlaps of items in the validated scales used in the survey (Bergkvist & Rossiter, 2007; Konrath, Meier, & Bushman, 2014). However, future studies should investigate their comparability with the multi-item scale measurements of the same constructs (Konrath et al., 2014; Van der Linden & Rosenthal, 2016). The use of single-item measures limits the exploration of the possible multidimensionality of the constructs (Van der Linden & Rosenthal, 2016). Another limitation is that single-item measures do not allow for the investigation of methods variance. It is important to note, though, that these limitations were not deleterious in the current dissertation since the focus is on the causal relations of the variables based on the structural model (Little, Cunningham, Shahar, & Widaman, 2002).
General Discussion 153