Measures

The six-item BSMAS assessed the level of PSMU (20). The six items corresponding to six core addiction components (i.e., salience, mood modification, tolerance, withdrawal, conflict, and relapse) proposed by Griffiths (57). Each BSMAS item evaluated how an individual experienced the use of social media during past month. A BSMAS sample item is “You feel an urge to use social media more and more” with the response options being “very rarely, rarely, sometimes, often, and very often.” All the items were rated on a five-point Likert scale with total scores ranging between 6 and 30. A higher BSMAS total score indicated a higher level of PSMU (58). Satisfactory psychometric properties of the BSMAS such as high internal consistency and confirmed unidimensional structure have been found in previous studies [e.g., (21, 59–61)]. Furthermore, the measurement invariance of the BSMAS has been supported across Hong Kong and Taiwan university students (62). The Cronbach's α of the BSMAS in the present study was 0.81 (baseline), 0.84 (first follow-up), 0.84 (second follow-up), and 0.84 (third follow-up).

The nine-item IGDS-SF9 assessed PG (63, 64) with each item corresponding to the nine IGD criteria described in the DSM-5 (65). Each IGDS-SF9 item evaluated how an individual experienced gaming during the past month. An IGDS-SF9 sample item is “Do you feel preoccupied with your gaming behavior?” with the response options being “never, rarely, sometimes, often, and very often.” All the items were rated on a five-point Likert scale with total scores ranging between 9 and 45. A higher IGDS-SF9 total score indicated a higher level of PG (66). Satisfactory psychometric properties of the IGDS-SF9 such as high internal consistency and confirmed unidimensional structure have been found in previous studies [e.g., (21, 61, 66–73)]. Furthermore, the measurement invariance of the IGDS-SF9 has been supported among Hong Kong and Taiwan university students (62). The Cronbach's α of the IGDS-SF9 in the present study was 0.92 (baseline), 0.91 (first follow-up), 0.91 (second follow-up), and 0.92 (third follow-up).

The 14-item HADS assessed two types of psychological distress (i.e., anxiety and depression) (74). Seven HADS items assessed anxiety and seven HADS items assessed depression during the past 2 weeks. A HADS anxiety sample item is “I feel tense or wound up” with the response options being “most of the time, a lot of the time, from time to time occasionally, and not at all.” A HADS depression sample item is “I feel as if I am slowed down” with the response options being “nearly all the time, very often, sometimes, and not at all.” All 14 items were rated on a four-point Likert scale with scores ranging between 0 and 21 for each type of psychological distress. A higher score on each subscale indicated a higher level of anxiety and/or depression, respectively (74). Satisfactory psychometric properties of the HADS such as high internal consistency and confirmed unidimensional structure have been found in previous studies [e.g., (74, 75)]. Furthermore, the measurement invariance of the HADS has been supported among Hong Kong and Taiwan university students (76). The Cronbach's α of the HADS anxiety subscale in the present study was 0.80 (baseline), 0.81 (first follow-up), 0.79 (second follow-up), and 0.81 (third follow-up); HADS depression subscale was 0.65 (baseline), 0.70 (first follow-up), 0.71 (second follow-up), and 0.79 (third follow-up).