Q1. Psychometric characterization of a Chinese version of the NFA
The 10-item Chinese short scale based on (20) showed a test-retest correlation and internal consistency of r = .943, p < .001, ICC = .942, p < .001, and Cronbach’s α = .829, F(9,10485) = 54.580, p < .001. Consistent with Kuang, Shi (22), the Chinese Need for Cognition (NFC) scale had an internal consistency above .850, Cronbach’s α = .866, F(17,19805) = 99.992, p < .001. The correlation between the Chinese NFA and NFC was .221 (p < .001), comparable with Appel et al.’s (20) (r = .170) finding in the convergent validity testing of NFA.
The results of the confirmatory factor analysis and ESEM demonstrated a satisfactory fit of 10-item NFA’s two-factor model in both Chinse and American samples (Table 1)
Table 1
Goodness-of-fit statistics of using Exploratory Structural Equation Modeling (ESEM) and Confirmatory Factor Analysis (CFA) for the Appel et al.’s 10-item NFA scale in the 4 subsamples. Splitting samples in CFA is recommended by Fokkema and Greiff (2017) to minimize model overfit.
Model tested | λ2 | df | p | CFI | TLI | NFI | RMSEA | AIC | ECVI |
ESEM | Chinese subsample 1 | 79.325 | 22 | < .001 | 0.975 | 0.95 | 0.967 | 0.067 | 165.325 | 0.285 |
Chinese subsample 2 | 90.266 | 22 | < .001 | 0.971 | 0.94 | 0.962 | 0.073 | 176.266 | 0.302 |
American subsample 1 | 40.86 | 22 | < .001 | 0.988 | 0.969 | ..974 | 0.041 | 126.86 | 0.251 |
American subsample 2 | 52.681 | 22 | < .001 | 0.976 | 0.94 | 0.96 | 0.054 | 138.681 | 0.294 |
CFA | Chinese subsample 1 | 83.327 | 15 | < .001 | 0.971 | 0.912 | 0.965 | 0.089 | 187.327 | 0.323 |
Chinese subsample 2 | 86.597 | 15 | < .001 | 0.969 | 0.908 | 0.964 | 0.09 | 190.597 | 0.326 |
American subsample 1 | 24.259 | 15 | < .001 | 0.994 | 0.978 | 0.985 | 0.035 | 124.259 | 0.246 |
American subsample 2 | 38.27 | 15 | < .001 | 0.982 | 0.933 | 0.971 | 0.057 | 138.27 | 0.293 |
Criterion for goodness of fit | | | | ≥ 0.9 | ≥ 0.9 | ≥ 0.9 | ≤ 0.1 | | |
The factorial invariance of the NFA structure was tested following a stepwise procedure as in (26). Although the baseline model implying Configural Invariance reported a significant chi-square (χ2(60) = 271.21, p < .01), other indices pointed to good fit from a practical standpoint (27), i.e., comparative fit index (CFI) = .972, normed fit index (NFI) = .965, mean square error of approximation (RMSEA) = .041. This result implied that the measurement structure with two latent factors was invariant among groups. When all factor loadings were constrained, multigroup analysis showed reasonable fitting, χ2(84) = 414.489, p < .01, CFI = .956, NFI = .946, RMSEA = .042 (see supplement for more detailed invariance test statistics). The above results supported that Appel et al.’s (20) 10-item NFA scale is sufficient as a culture-fair assessment of affective intrinsic motivation.
Q2. Cultural Differences In Nfa And Nfc
The mean NFA approach scores and avoidance scores were computed and compared between the Chinese sample and American sample, and also contrasted to the 4 independent European samples from Appel et al.’s (20) study (see Fig. 1). As shown in Table 2, the effect sizes in terms of Hedges’s g values and the 95% CIs revealed that the Chinese general public sample had significantly lower NFA approach scores than the American and European samples, and significantly higher NFA avoidance scores than people in other cultures. Compared to the European samples, the American adults had significantly lower NFA approach scores and higher avoidance scores. The Chinese general public had significantly lower NFC than their American peers.
Table 2
The mean, standard deviation, sample size of NFA average subscale scores and NFC total scores in multinational samples, and the effect size indices of the scores comparisons across cultures.
| Sample 1 | Sample 2 | Mean 1 | Mean 2 | N 1 | N 2 | SD 1 | SD 2 | Cohen's d | Hedges's g | SEg | 95% CI Lower limit | 95% CI Upper limit |
Approach | Chinese adults | American adults | .55 | .91 | 1186 | 980 | 1.54 | 0.92 | -0.28 | -0.28 | 0.04 | -0.36 | -0.19 |
German/Austria students | 1.28 | 1160 | 0.96 | -0.57 | -0.57 | 0.04 | -0.65 | -0.48 |
German Adults | 1.29 | 627 | 0.92 | -0.54 | -0.54 | 0.05 | -0.64 | -0.45 |
Austria Couples | 1.15 | 126 | 1.07 | -0.40 | -0.40 | 0.09 | -0.58 | -0.22 |
UK Adults | 1.02 | 236 | 1 | -0.32 | -0.32 | 0.07 | -0.46 | -0.18 |
American adults | German/Austria students | .91 | 1.28 | 980 | 1160 | .92 | 0.96 | -0.39 | -0.39 | 0.04 | -0.48 | -0.31 |
German Adults | 1.29 | 627 | 0.92 | -0.41 | -0.41 | 0.05 | -0.51 | -0.31 |
Austria Couples | 1.15 | 126 | 1.07 | -0.26 | -0.26 | 0.09 | -0.44 | -0.07 |
UK Adults | 1.02 | 236 | 1 | -0.12 | -0.12 | 0.07 | -0.26 | 0.02 |
Avoidance | Chinese adults | American adults | − .05 | − .91 | 1186 | 980 | 1.29 | 1.15 | 0.70 | 0.70 | 0.04 | 0.61 | 0.79 |
German/Austria students | -1.39 | 1160 | 1.12 | 1.11 | 1.11 | 0.04 | 1.02 | 1.19 |
German Adults | -1.06 | 627 | 1.18 | 0.81 | 0.81 | 0.05 | 0.71 | 0.91 |
Austria Couples | -1.5 | 126 | 1.07 | 1.14 | 1.14 | 0.10 | 0.95 | 1.33 |
UK Adults | − .55 | 236 | 1.2 | 0.39 | 0.39 | 0.07 | 0.25 | 0.53 |
American adults | German/Austria students | − .91 | -1.39 | 980 | 1160 | -1.39 | 1.12 | 0.42 | 0.42 | 0.04 | 0.34 | 0.51 |
German Adults | -1.06 | 627 | 1.18 | 0.13 | 0.13 | 0.05 | 0.03 | 0.23 |
Austria Couples | -1.5 | 126 | 1.07 | 0.52 | 0.52 | 0.10 | 0.33 | 0.70 |
UK Adults | − .55 | 236 | 1.2 | -0.31 | -0.31 | 0.07 | -0.45 | -0.17 |
NFC | Chinese | American | 44.90 | 55.28 | 1186 | 980 | 11.70 | 11.87 | -0.88 | -0.88 | 0.05 | -0.97 | -0.79 |
The correlation between the NFA approach and avoidance score in the Chinese sample was significantly positive, r = .44, p < .01, in contrast to the negative correlations in the American sample (r = − .30, p < .01), Z = 18.08, p<. 001 (see Fig. 2), German/Austria students (r = − .34, p < .01), Z = 19.79, p < .001, German adults (r = − .40, p < .01), Z = 18.11, p<. 001, Austria couples (r = − .46, p < .01), Z = 10.23, p<. 001, and UK adults (r = − .44, p < .01), Z = 13.18, p<. 001. There was no significant difference on the approach-avoidance correlation between the American sample and the German/Austria students (Z = 1.03, p=. 30) and Austria couples (Z = 1.96, p=. 05) respectively, However, the correlation was significantly smaller in magnitude in the American sample than in the German adults, Z = 2.23, p=. 030, and the UK adults, Z = 2.25, p=. 025.
Q3. Cultural features of NFA and NFC in a population targeted by affective-motivation public health campaigns
The difference on NFA observed between the Chinese and American general public sample was also present in the older participants (≥ 60 yrs), with significantly lower approach in Chinese seniors (M = − .57, SD = 1.28) than American seniors (M = .93, SD = .81), F(1,586) = 167.33, p < .001, ηp2 = .22, and significant higher avoidance in Chinese seniors (M = − .42, SD = 1.29) than American seniors (M = − .93, SD = 1.09), F(1,586) = 16.93, p < .001, ηp2 = .03.
Hearing loss occurred in 62.6% of seniors in the current study. As shown in Fig. 3, compared to seniors without hearing loss, those who had hearing loss had significantly lower NFA approach scores, F(3,482) = 9.00, p < .01, ηp2 = .05, significantly lower avoidance scores, F(3,482) = 26.13, p < .01, ηp2 = .05, and significantly lower NFC scores, F(3,482) = 14.66, p < .01, ηp2 = .08. There was a small group of individuals who reported having normal hearing but failed the pure tone audiometric test, n = 34, who had similar levels of NFA and NFC (see Table 3) with normal hearing group.
Table 3
The mean and SD of NFA and NFC scores in the Chinese senior participants.
| Hearing loss |
| Normal hearing | Mild hearing loss | Moderate-to-severe hearing loss | Hearing loss deniers |
NFA approach | -0.62 ± 1.02 | -1.08 ± 1.05 | -0.94 ± .88 | -0.34 ± 1.31 |
NFA avoidance | -0.01 ± 1.16 | -0.94 ± 1.10 | -0.93 ± 1.01 | -0.23 ± 1.27 |
NFC | 54.98 ± 11.94 | 47.49 ± 13.54 | 45.98 ± 13.94 | 52.88 ± 11.60 |
Intention to avail themselves of early hearing intervention did not differ between groups,
F(3,482) = 1.01,
p = .39,
ηp2 = .01. Regardless of hearing status, participants tended to choose “when I have severe hearing loss and have difficulty communicating with others” in response to the question “When do you think you will consider wearing hearing aids and take serious action?” Of the 486 community-based participants, only 6.8% and 13.8% reported willingness to seek early intervention when they have mild and moderate hearing loss, respectively. Intention to engage in early hearing care seeking did not significantly correlate with either NFA approach,
r=-.01,
p = .89 or NFA avoidance,
r = .06,
p = .22, but was marginally associated with NFC,
r = .09,
p = .04.