The mean aggregate RPAS and RSAS scores for the final sample (N = 3,435), after exclusion of participants with incomplete questionnaires or who met the exclusion criterion (i.e. mental disorder diagnoses), were 14.089 (95%CI, 13.837–14.341) and 8.376 (95%CI, 8.186–8.566), respectively. The mean total aggregate score for the uniformed RPAS&RSAS was 22.465 (95%CI, 22.079–22.851). The coefficients of correlation (r values) obtained between each RPAS&RSAS item and the total score are listed in Table S2 in supplementary material.
Reliability: Internal consistency and test-retest reliability
The Cronbach’s α values obtained for the RPAS&RSAS, RPAS, and RSAS were 0.884, 0.835, and 0.827, respectively, which indicated good reliability with respect to internal consistency. The Pearson’s r values obtained for the RPAS&RSAS, RPAS, and RSAS test-retest scores were 0.644 (p < 0.001), 0.572 (p < 0.001), and 0.602 (p < 0.001) in a subsample of 197 participants, indicating good test-retest reliability.
Construct validity: RPAS/RSAS factor analysis
After the elimination of six items for culture inappropriateness or poor relevance (see Methods), EFA of the RPAS (56 items) and RSAS (39 items), involving half of the participants, was conducted with a WLSMV estimator. The model fit indexes obtained for the RPAS and RSAS are presented in Table 1 and Table 2, respectively. The RPAS fit well with a 4-factor structure model (CFI = 0.924, TLI = 0.912, SRMR = 0.053, RMSEA = 0.024) and the RSAS fit well with a 2-factor structure model (CFI = 0.941, TLI = 0.934, SRMR = 0.063, RMSEA = 0.028). The factor loadings of each item are reported in Table S3 and Table S4 in the supplementary material.
In the RPAS, the physical anhedonia items segregated into four factors as follows: Factor 1 contains items 5, 6, 8, 10, 11, 12, 13, 14, 16, 17, 18, 20, 22, 23, 26, 27, 28, 29, 32, 33, 43, 44, 48, 50, 51, 52, 55, 56, and 57; Factor 2 contains items_2, 3, 19, 21, 24, 25, 30, 31, 34, 35, 36, 37, 39, 41, 45, 46, 47, 49, 54, 58, 59, 60, and 61; Factor 3 contains items 7 and 15; and Factor 4 contains items 38 and 42. As can be seen in Table S4, Factors 3 and Factor 4 each have only two items, and all four of these items had relatively high loadings (loadings > 0.350) in the first two factors. Thus, we spread these four items into the first two factors according to loading parameters. Accordingly, item 5 was placed with Factor 1 and items 7, 38, and 42 were placed with Factor 2 ( Table S3 in supplementary material). Although the fitness indexes of the resultant 2-factor model were not as desirable as those obtained for the 4-factor model, the conceptualization of anhedonia and the associations of these items’ contents with the first two factors support a 2-factor model.
The RPAS items gathered in Factor 1 (P1) were related to consummatory physical pleasure whereas the RPAS items gathered in Factor 2 (P2) were related to anticipatory physical pleasure. Regarding items gathered in Factors 3 and 4, the content of item 15 (There aren't many things I really like to do) associated well with “liking”, thus relating it to consummatory physical anhedonia (Factor 1). In contrast, the content of items 7 (The taste of food has always been important tome), 38 (The beautiful scenery can make me feel delighted). and 42 ( I seldom have the idea of singing in the bath) associated well with “wanting”, thus relating them to anticipatory physical anhedonia (Factor 2).
The RSAS items gathered in Factor 1 (S1)—that is, items 1, 2, 3, 6, 10, 13, 14, 17, 21, 22, 23, 26, 27, 28, 29, 32, 34, 35, 37, 38, 39, and 40—were related to the consummatory social pleasure. The RSAS items gathered in Factor 2 (S2)—that is, items 4, 5, 7, 8, 9, 11, 12, 15, 16, 18, 19, 20, 24, 25, 30, 31, and 36—were related to anticipatory social pleasure.
For CFA, we parceled randomized four items and simplified the scales because the RPAS/RSAS items seemed excessive and scattered. There were 14 parcels in the RPAS (7 in P1 and 7 in P2) and 9 parcels in the RSAS (5 in S1 and 4 in S2). As is shown in Table 3, fit index values supported a good fit with a 2-factor model for the RSAS (CFI = 0.967, TLI = 0.957, SRMR = 0.037, and RMSEA = 0.048) and a very good fit with a 2-factor model for the RPAS (CFI = 0.947, TLI = 0.936, SRMR = 0.039, RMSEA = 0.052) in CFA, consistent with our EFA results.
Factor analysis of the RPAS&RSAS
Based on the good fits of the 2-factor models for the two individual scales and good first-order model fitness results for the RPAS&RSAS (reported above), CFA confirmed a novel second-order model for the RPAS&RSAS as a unified scale. The acceptable model fit indexes (CFI = 0.901, TLI = 0.899, RMSEA = 0.055, and SRMR = 0.086) support a simpler construct of anhedonia, as is shown in Figure 1. Although the fit indexes of the second-order model are not as good as those of the first-order model for the RSAS&RPAS (Table 4), the advantages of the second-order model, including a lower number of degrees of freedom and a more simple structure make the second-order model desirable. Additionally, the second-order model supports the view of multiple dimensions of anhedonia, which combines observable behavioral symptoms and underlying biological mechanisms.