The effectiveness and mechanisms of a brief online best-possible-self intervention among young adults from mainland China

ABSTRACT

The best-possible-self (BPS) intervention has been shown to promote well-being, optimism, and positive affect. However, most research was conducted in Western countries, and its effectiveness in people from Eastern countries has been questioned. In this study, we examined the effect of BPS on young adults from mainland China. We also investigated the moderating effect of cultural orientation, self-efficacy, and trait optimism and the mediating effect of state optimism and goal-related cognitions. Seventy participants were randomly assigned to the BPS or the control condition. In contrast to previous studies, the BPS did not lead to higher positive affect or well-being in Chinese, but rather to decreased negative affect and negative future expectations. No significant moderating or mediating effect was found. This study indicated that the effects of BPS might differ across cultures. Future studies may consider a longer session and qualitative analysis.

KEYWORDS:

• Best possible self
• positive-psychological intervention
• online intervention
• cultural differences
• China

The ‘Best Possible Self’ (= BPS) intervention requires participants to write about and imagine what they will achieve or who they will be in their best possible futures (King, 2001; Peters et al., 2013). The effectiveness of the BPS intervention was established across different populations, including children (Owens & Patterson, 2013), students (Carrillo et al., 2019), adults (Loveday et al., 2018a), chronic pain patients (Flink et al., 2015), diabetes patients (Gibson et al., 2021), and depressive individuals (Pietrowsky & Mikutta, 2012). Most studies applied BPS in the format of writing (Heekerens & Eid, 2021), but Harrist et al. (2007) found that BPS can also be used orally. The BPS can be delivered online or in person. Layous et al. (2012) found no significant difference between participants who completed it online or in person. This finding is consistent with the literature review and meta-analysis (Carrillo et al., 2019; Loveday et al., 2018a). The BPS can be used in different dosages, ranging from one session to 28 daily sessions (Loveday et al., 2018a). Previous studies have indicated the effectiveness of BPS on different outcomes, including depression (Auyeung & Mo, 2019; Liau et al., 2016), subjective well-being (Liau et al., 2016; Manthey et al., 2016), life satisfaction (Boehm et al., 2011), optimism (Enrique et al., 2017; Heekerens & Eid, 2021; Meevissen et al., 2011; Peters et al., 2010), positive affect (Altintas et al., 2020; Peters et al., 2010; Renner et al., 2014), flourishing (Auyeung & Mo, 2019). In addition, physical outcomes such as cortisol response (Nicolson et al., 2020; Teismann et al., 2014) or behavior outcomes such as health center visits (Harrist et al., 2007) have been found. Some researchers have also applied the BPS to enhance academic motivation and commitment (Altintas et al., 2020) and leadership behavior (Jennings et al., 2022).

Most evidence was derived from Western countries where an individualistic cultural norm prevails. Individualism and collectivism are two distinct cultural characteristics. Individualism emphasizes personal aspects and independence, and prioritizes personal goals, while collectivism focuses more on collective aspects and interdependence, and places more value on communal relationships and collective goals (Triandis & Gelfand, 1998). Researchers have been questioning the effect of positive psychological interventions in Eastern samples (Hendriks et al., 2018; Ng & Ong, 2022). As an intervention that requires writing on self-relevant aspects, BPS is more in line with the value orientation that emphasizes the self in individualism (Ng & Ong, 2022). However, China is a typical collectivistic country, and Chinese people value the group’s harmony more than their self-improvement (Triandis, 1995). Boehm et al. (2011) found that Anglo Americans showed a larger increase in life satisfaction compared to Asian Americans after the BPS and argued that the different levels of self-focus may explain the result. In addition, the BPS requires a future-orientation perspective, but Chinese people tend to be past-oriented (Guo et al., 2012). Encouraging Chinese people to orient their temporal focus on the future may not accord with their thinking style. Therefore, whether the BPS can benefit Chinese participants can be questioned.

Some evidence also supports the effect of the BPS in Chinese. First, Chinese people are less optimistic than Western samples (Lee & Seligman, 1997). Peters et al. (2010) found that people with a lower baseline trait optimism showed a higher increase in future expectations after the BPS. Therefore, being less optimistic may leave more room for the BPS to be effective. Second, Guo et al. (2012) found that Chinese people value the future more than the past when they are induced to think about and focus on the future. Therefore, practicing the BPS may help the Chinese gain more future-oriented thinking.

However, as far as the authors know, only one study showed that practicing the BPS increased positive affect and flourishing and decreased depression was conducted in Hong Kong students (Auyeung & Mo, 2019). Hong Kong is more westernized and individualistic compared to mainland China, thus direct empirical evidence supporting the effect of the BPS in mainland Chinese samples is needed.

A few studies have applied the BPS in other Asian countries that are also assumed to have a collectivistic culture. Liau et al. (2016) found a decrease in negative affect after the BPS in Singapore students, while positive affect, life satisfaction, and optimism did not change. Similarly, the BPS intervention decreased anxiety and depression in Japanese students, but no positive state was assessed in this study (Yogo & Fujihara, 2008). The above two studies all found the BPS effective in negative outcomes instead of positive ones. However, Ng (2016) found the BPS effective in promoting happiness in another Singapore sample. One possible reason for the inconsistency is that the levels of individualism and collectivism may vary between countries and even individuals, despite their collectivistic orientation. Hence, the actual orientation of cultural norms needs to be considered. Therefore, individualism and collectivism were included as moderators of the BPS’s effectiveness in this study.

We also examined self-efficacy and trait optimism as additional moderators and state optimism and goal-related variables as mediators. Previous studies have indicated that people with higher affective and social self-efficacy tend to have more positive future expectations (Caprara et al., 2006). In the BPS practice, people with a higher level of general self-efficacy are likely to have a more vivid imagination and hold more confidence about the likelihood of them coming true. Therefore, the BPS could be more effective in highly self-efficacious people. Previous studies have examined the moderating role of optimism with inconclusive results. Peters et al. (2010) found that trait optimism was related to a lesser increase in state optimism after both the BPS and a control intervention, possibly because of a ceiling effect, while a subsequent study did not find a moderating role for trait optimism (Peters et al., 2016). The moderating role of trait optimism was therefore be revisited.

Understanding the mechanism of the BPS is meaningful in promoting the generalization of the BPS and increasing its effectiveness. In this study, we tested several potential mediators. First, as proposed by Heekerens and Eid (2021), we examined whether state optimism could be an underlying factor leading to changes in well-being outcomes. The BPS has been found to promote optimistic thinking in different populations (Peters et al., 2010), while optimism is associated with higher levels of current and future subjective well-being (Forgeard & Seligman, 2012). Moreover, following the recommendations of King (2001), we examined whether the BPS might work through goal-related cognitions. Writing about life goals could help people to examine a previously unexamined part of themselves, generate clear goals, reduce goal conflict, and build confidence about attaining their goals. Therefore, we included perceived goal-clarity, goal-motivation, goal-commitment, and goal-confidence as potential mediators.

In sum, this study aimed to investigate the effectiveness of BPS in a Chinese adult sample on their mental well-beings and explore the moderating and mediating mechanisms. This study will contribute to future application and modification of the BPS in the Chinese community and help understand the mechanisms explaining how and for whom the BPS will work. As this is the first study with the BPS in mainland China, we start by examining the short-term effect of a single session of BPS. Based on previous findings and theoretical considerations, we have the following hypotheses.

Our first hypothesis concerns the overall effect of the BPS intervention. We hypothesize that participants in the BPS condition will report a larger increase in positive affect and satisfaction with life, and a larger decrease in depression, anxiety, and negative affect from before to after the interventions than participants in the control condition.

Our second hypothesis concerns the potential moderators. We hypothesize that participants with a higher level of individualism and general self-efficacy, and a lower level of collectivism and trait optimism, will report a larger increase in satisfaction with life and positive affect, and a larger decrease in state depression, state anxiety, and negative affect.

Our third hypothesis concerns the potential mediators. We hypothesize that participants benefit from the BPS via increased positive future expectations and decreased negative future expectation, and increased goal-clarity, goal-commitment, goal-confidence, and goal-motivation.

Method

Participants

The result from Gpower 3.1 indicated that at least 54 participants were needed to reach a medium effect size of 0.25 to obtain a power of 0.95 for our primary analyses. Considering that moderation and mediation analyses require a larger sample to obtain sufficient power, we aimed to include at least 70 participants. Participants were recruited through online advertising via social media, including WeChat and Tencent, from 1 September 2021 to 13 September 2021. The inclusion criteria were Chinese adults aged between 18–30 years old. In total, 102 participants were approached, of whom 85 agreed to participate in the study and gave their informed consent, and 74 completed the survey. Four of those completing the survey did not pass the manipulation check of their writing contents. Specifically, two were excluded because of coping from the Internet, and two were excluded because they did not write according to instructions. Finally, 70 participants were included in the statistical analyses, 35 in the BPS condition and 35 in the control condition. The mean age of these participants was 21.57 years old (SD = 2.20 years; range 18–28), 28 were men, and 42 were women. All participants were born and living in China at the time of assessment. Most participants had attained an educational level of high school or equivalent (n = 53, 75.7%). Forty-five (64.3%) participants were currently pursuing a bachelor’s degree, 11 (15.7%) were pursuing a master’s degree, and 14 (20%) were employed or job-seeking. Almost all participants were Han (n = 68, 97.1%), with two belonging to the Man minority.

Design and procedures

This study applied a 2 × 2 mixed design, with condition (BPS vs. TD) as a between-subject variable and time (pre-test and post-test) as a within-subject variable. All materials in this study were in Mandarin. After giving their informed consent, participants received a link to Qualtrics. Qualtrics is an online survey tool that allows researchers to build and distribute surveys (Cushman et al., 2021). In this study, participants completed the survey and the BPS or control intervention on Qualtrics following the procedure below. First, participants filled out several questionnaires as baseline measurement (T0), including socio-demographic variables, state depression, state anxiety, satisfaction with life, mood, state and trait optimism, individualism and collectivism, general self-efficacy, and four goal-related variables (goal-clarity, goal-commitment, goal-confidence, and goal-motivation). Afterward, participants were randomly assigned to Best-possible-self (BPS) or typical day (TD) condition. Written instructions were displayed online (Seen in the Appendix). Participants in the BPS condition were asked first to think about their best possible selves for one minute, then to write for 15 minutes about what they had just thought, followed by 5 minutes of visualizing what they wrote. Participants in the TD condition were asked to do a similar practice focusing on their typical days. Each step was timed automatically, and participants could only proceed after the predefined time had elapsed. In the writing part, participants had to write at least 200 Chinese characters to be able to proceed. After finishing their visualization, participants were forwarded to the post-intervention assessment (T1). This assessment included all the above questionnaires except the demographic and moderator variables. The whole procedure took about 50 minutes. After finishing the post-intervention assessment, participants were thanked, debriefed, and compensated with a ¥50 voucher. The study was approved by the Ethics Review Committee Psychology and Neuroscience of Maastricht University (ERCPN number: 240_111_07_2021), and pre-registered in ASPREDICTED (#73475).

Measures

Socio-demographic variables

Participants were asked to report their age, gender, education level, occupation, birthplace (China or not), living place (China or not), and ethnicity (i.e., Han or minority).

Dependent variables

State depression

State depression was measured by the state subscale of the 20-item State-Trait Depression Scale (STDS) (Spielberger et al., 2003). The STDS excludes somatic symptoms, which are more relevant in clinical depression, and concentrates on cognitive-affective symptoms (Krohne et al., 2002). The state subscale of STDS comprises five positively framed items and five negatively framed items. All items refer to the present state and are answered on a 4-point scale ranging from 1 (not at all) to 4 (very strong). The total score ranges from 10 to 40 and is calculated by summing all items after reversing the positive items. A higher total score indicates a higher state depression. The STDS has been validated and shown internal reliability of 0.91 among the Chinese sample (Lei et al., 2011). In this study, the reliability of the state subscale for the current sample was 0.90 at T0 and 0.91 at T1, pointing to excellent internal reliability (George & Mallery, 2003).

State anxiety

State anxiety was measured by the six-item short form of the State-Trait Anxiety Inventory (STAI-6) (Marteau & Bekker, 1992). The six items are answered from 1 (not at all) to 4 (very much). The total score ranges from 6 to 24 and is calculated by summing all six items after reversing the three negatively formulated items. A higher score indicates higher state anxiety. The Chinese version of STAI-6 has been validated and shown internal reliability of 0.90 among the Chinese community (Shek, 1988). Cronbach’s alpha in the current sample was 0.85 at T0 and 0.83 at T1, indicating good internal reliability.

Satisfaction with life

Satisfaction with life was measured by the five-item Satisfaction With Life Scale (SWLS) (Diener et al., 1985). The five items are answered from 1 (strongly disagree) to 7 (strongly agree). The total score ranges from 5–35 and is calculated by summing all items. A higher total score indicates a higher level of satisfaction with life. The Chinese version of SWLS has been validated and showed good internal reliability of 0.92 in Chinese respondents (Bai et al., 2011). Cronbach’s alpha value in the current sample was 0.90 at T0 and 0.93 at T1, indicating excellent internal reliability.

Positive affect and negative affect

Positive and negative affect were measured by the 20-item Positive and Negative Affect Scale (PANAS) (Watson et al., 1988). The PANAS consists of 20 items describing positive and negative feelings that are scored from 1 (very slightly) to 5 (extremely). The total score of each subscale ranges from 10–50 and is calculated by summing all items in the subscale. A higher score indicates a higher level of positive or negative affect. The Chinese version of PANAS has been validated and showed internal reliability of 0.82 in Chinese people (Huang et al., 2003). Cronbach’s alpha in the current sample was 0.89 for positive affect and 0.93 for negative affect at T0, 0.89 for positive affect and 0.94 for negative affect at T1, indicating good to excellent internal reliability.

Moderators

Individualism and collectivism

Participants’ levels of individualism and collectivism were measured by the 16-item Individualism and Collectivism Scale (IC) (Triandis & Gelfand, 1998). Although the original questionnaire comprised 27 items measuring four dimensions (i.e. vertical and horizontal individualism and vertical and horizontal collectivism), most studies have applied a 16-item version that includes the four items with the highest factor loadings on each of the four dimensions. To be consistent with other recent studies, we used this 16-item version and calculated a total score for individualism and collectivism separately. All items are answered from 1 (absolutely disagree) to 9 (absolutely agree). The total scores of individualism and collectivism range from 8–72, with a higher total score indicating a higher level of individualism or collectivism. The IC was used in a Chinese sample and showed internal reliability ranging from 0.66 to 0.87 for each subscale (Chan & Snape, 2013). Cronbach’s alphas in the current sample were 0.71 for the individualism subscale and 0.78 for the collectivism subscale, indicating acceptable internal reliability.

Trait optimism

Trait optimism was measured by the 10-item Life Orientation Test-Revised (LOT-R) (Scheier et al., 1995). The LOT-R consists of three positively framed items, three negatively framed items, and four filler items. All items are answered from 0 (strongly disagree) to 4 (strongly agree). After reversing the negatively framed items, the total score, which ranges from 0 to 24, is calculated by summing the six items. A higher total score indicates higher trait optimism. The LOT-R has shown internal reliability of 0.70 in a Chinese sample (Lai et al., 1998). Cronbach’s alpha in the current sample was 0.79, indicating acceptable internal reliability.

General self-efficacy

General self-efficacy was measured by the 10-item Generalized Self-Efficacy Scale (GSES) (Schwarzer & Jerusalem, 1995). The GSES comprises ten items answered from 1 (not at all true) to 4 (exactly true). The total score ranges from 4–40 and is calculated by summing all items, with a higher total score indicating a higher level of general self-efficacy. The GSES has been validated in a Chinese sample and showed internal reliability of 0.91 (Zeng et al., 2022). Cronbach’s alpha in the current sample was 0.88, indicating good internal reliability.

Mediators

State optimism

State optimism was measured by the Questionnaire for Future Expectation (FEX) (Hanssen et al., 2012), translated into Chinese by the first author. The FEX consists of 20 items, of which half refer to positive future outcomes and the other half to negative future outcomes. Items are answered from 1 (not likely at all to occur) to 7 (extremely likely to occur). The ten items referring to positive future outcomes are added up to reflect positive expectations (FEX-Pos), while the other ten items are added up to reflect negative expectations (FEX-Neg). The total score of FEX-Pos and FEX-Neg ranges from 10–70. A higher score indicates a higher level of positive/negative future expectations. Cronbach’s alpha in the current sample was 0.84 and 0.87 for the positive and negative subscale at T0, 0.80 and 0.87 at T1, respectively, indicating good internal reliability.

Goal-related variables

We measured goal-clarity, goal-commitment, goal-confidence, and goal-motivation with self-constructed single items. Items were formulated to reflect the corresponding content, i.e., How easy is it to come up with your life goals, such as what to accomplish in the future or where you want to be (goal-clarity), answered from 1 (very easy) to 7 (very hard); How willing are you to give 100% of your effort for your life goals and not stop until you reach them (goal-commitment), answered from 1 (very much I do) to 7 (not at all); How confident are you that you will reach your life goals as long as you work hard (goal-confidence), answered from 1 (surely I do) to 7 (not at all confident); How motivated are you to work hard to reach your life goals (goal-motivation), answered from 1 (not motivated at all) to 7 (very motivated). A higher score indicated a higher level of each variable after the scores of goal-clarity, goal-commitment, and goal-confidence were reversed.

Data analysis

Associations between variables measured at T0 were calculated with bivariate Pearson’s correlation analyses. The primary research question on the effectiveness of the BPS was tested through a series of 2 (condition: BPS, TD) x 2 (time: pre-test, post-test) repeated-measures ANOVAs. Depression, anxiety, satisfaction with life, positive affect, negative affect, positive and negative future expectations, and the four goal-related variables were used as dependent variables in separate analyses. The condition-by-time interaction effect is the effect of interest. Significant interaction effects were followed by paired sample t-tests for pre-test versus post-test in both conditions separately.

We used hierarchical regression analyses to examine the moderation effect. The predictors were entered in blocks, with each block representing one step in the hierarchical model. The pre-test scores of the outcome variable, condition, and moderator variable were included in the first step, while the condition-moderator interaction was added in the second step. For each outcome, separate models were specified for the moderators of individualism, collectivism, self-efficacy, and trait optimism.

The SPSS process macro was used to test the mediating role of state optimism and the four goal-related variables (Hayes et al., 2017). The condition was used as the independent variable (X) that coded as 1 (BPS) and 0 (TD). The post-test score (T1) of potential mediators, i.e. positive and negative future expectations, goal-clarity, goal-commitment, goal-confidence, and goal-motivation were used as mediating variables (M). The post-test scores (T1) of the outcomes were used as dependent variables (Y), with their respective pre-test scores (T0) being used as covariates. All analyses were performed in SPSS 26.0.

Results

Descriptive and correlational analyses

The mean levels of the variables of interest at the pre-test (T0) and their inter-correlations are shown in Table 1. There is no significant difference between the two groups regarding all variables of interest at baseline (all p’s > 0.05). Positive psychological outcomes (satisfaction with life, positive affect) were significantly positively related to positive future expectations, trait optimism, collectivism, and general self-efficacy (r’s ranging from 0.34 to 0.63, p’s <0.01), and significantly negatively related to negative future expectations (r’s ranging from 0.51 to 0.58, p’s <0.01). Negative psychological outcomes (depression, anxiety, negative affect) were significantly positively related to negative future expectations (r’s ranging from −0.48 to −0.51, p’s <0.01) and significantly negatively related to positive future expectations, general self-efficacy, and trait optimism (r’s ranging from −0.26 to −0.60, p’s <0.05). Collectivism was significantly negatively related to depression and anxiety (r’s ranged from −0.43 to −0.44, p’s <0.01). Individualism did not significantly correlate with any psychological outcome.

Table 1. Mean levels and inter-correlations between variables of interest at pre-test (T0).

	M	SD	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15
1. Depression	17.01	5.27	-
2. Anxiety	11.41	3.62	.88**	-
3. Subjective Well-being	19.80	6.46	−.43**	−.50**	-
4. Positive Affect	32.11	7.24	−.62**	−.57**	.65**	-
5. Negative Affect	18.96	7.79	.59**	.69**	−.34**	−.21	-
6. Positive Future Expectations	53.34	7.28	−.59**	−.60**	.54**	.61**	−.50**	-
7. Negative Future Expectations	33.94	9.97	.45**	.51**	−.51**	−.58**	.50**	−.65**	-
8. Trait optimism	14.32	4.32	−.51**	−.60**	.55**	.62**	−.51**	.61**	−.68**	-
9. Individualism	52.83	8.14	−.01	.05	−.09	.00	.18	.18	.09	−.04	-
10. Collectivism	49.71	8.91	−.43**	−.44**	.34**	.48**	−.23	.50**	−.42**	.44**	.24*	-
11. General Efficacy	29.44	4.67	−.47**	−.40**	.44**	.63**	−.26*	.62**	−.55**	.65**	.20	.53**	-
12. Goal-clarity	4.49	1.54	−.26*	−.28*	.32**	.30*	−.24*	.49**	−.40**	.42**	.23	.26*	.32**	-
13. Goal-Commitment	5.66	1.23	−.25*	−.27*	.18	.37**	−.29*	.47**	−.44**	.38**	.20	.25*	.39**	.39**	-
14. Goal-Confidence	5.19	1.43	−.32**	−.38**	.33**	.40**	−.36**	.64**	−.43**	.53**	.32**	.36**	.55**	.37**	.49**	-
15. Goal-Motivation	5.70	1.08	−.44**	−.50**	.35**	.52**	−.34**	.65**	−.44**	.52**	.31**	.46	.55**	.45**	.60**	.67**	-

Note. ** p < 0.01; * p < 0.05.

Effects of the BPS

Pre- and post-test scores of dependent and mediator variables are displayed in Table 2. Significant interaction effects between condition and time were found for state depression (F(1, 68) = 4.936, p = 0.030, ${\eta }_{\mathrm{P}}^2$ = 0.068); state anxiety (F(1, 68) = 4.539, p = 0.037, ${\eta }_{\mathrm{P}}^2$ = 0.063); negative future expectations (F(1, 68) = 7.299, p = 0.009,${\eta }_{\mathrm{P}}^2$ = 0.097), and goal-clarity (F(1, 68) = 14.76, p < 0.001, ${\eta }_{\mathrm{P}}^2$ = 0.178). Paired-sample t-tests within conditions showed that the decrease in negative future expectations in the BPS condition reached significance. Depression significantly increased in the TD condition and showed a non-significant decrease in the BPS condition. Anxiety decreased in the BPS condition and increased in the TD condition, although neither reached significance. Goal clarity significantly increased in the BPS condition but not in the TD condition. No significant interaction effects were found for satisfaction with life (F(1, 68) = 2.232, p = 0.140, ${\eta }_{\mathrm{P}}^2$ = 0.032); positive affect (F(1, 68) = 3.113, p = 0.082, ${\eta }_{\mathrm{P}}^2$ = 0.044); negative affect (F(1, 68) = 1.659, p = 0.202, ${\eta }_{\mathrm{P}}^2$ = 0.024); positive future expectations (F(1, 68) = 0.841, p = 0.362, ${\eta }_{\mathrm{P}}^2$ = 0.012), goal-commitment (F(1, 68) = 2.808, p = 0.098, ${\eta }_{\mathrm{P}}^2$ = 0.040); goal-confidence (F(1, 68) = 1.267, p = 0.264, ${\eta }_{\mathrm{P}}^2$ = 0.018); and goal-motivation (F(1, 68) = 3.164, p = 0.080, ${\eta }_{\mathrm{P}}^2$ = 0.044). Negative affect showed a significant decrease in the BPS condition but not in the TD condition.

Table 2. Means of variables of interest in each group at pre-test (T0) and post-test (T1).

	BPS		p	Cohen’s d	TD		p	Cohen’s d
	Pre-test	Post-test			Pre-test	Post-test
1. Depression	17.54 (5.20)	16.85 (5.02)	0.534	0.106	16.48 (5.35)	18.91 (6.72)	0.009	0.466
2. Anxiety	11.94 (3.44)	11.14 (3.68)	0.187	0.228	10.89 (3.77)	11.91 (3.90)	0.106	0.281
3. Subjective well-being	18.80 (6.18)	19.57 (7.54)	0.172	0.236	20.80 (6.66)	20.43 (7.37)	0.487	0.119
4. Positive Affect	31.34 (7.29)	32.94 (6.60)	0.147	0.251	32.89 (7.22)	32.17 (8.44)	0.347	0.161
5. Negative Affect	19.74 (8.18)	17.57 (6.90）	0.004	0.520	18.17 (7.41)	17.57 (6.95)	0.551	0.102
6. Positive Future Expectations	53.03 (6.87)	54.37 (7.75)	0.206	0.218	53.66 (7.75)	53.91 (8.13)	0.952	0.010
7. Negative Future Expectations	35.37 (9.32)	31.14 (8.11)	<.001	0.692	32.51 (10.53)	31.77 (10.15)	0.344	0.162
8. Goal-Clarity	4.37 (1.42)	5.40 (1.12)	<.001	0.824	4.60 (1.67)	4.74 (1.54)	0.134	0.260
9. Goal-Commitment	5.31 (1.23)	5.54 (1.20)	0.254	0.196	6.00 (1.13)	5.86 (1.14)	0.169	0.238
10. Goal-Confidence	5.43 (1.33)	5.46 (1.15)	0.856	0.031	4.94 (1.49)	5.23 (1.44)	0.096	0.289
11. Goal-Motivation	5.54 (1.07)	5.83 (0.89)	0.067	0.320	5.86 (1.09)	5.77 (1.17)	0.556	0.100

Moderation effects

We examined the moderating effect of individualism, collectivism, general self-efficacy, and trait optimism on the five outcome variables (depression, anxiety, satisfaction with life, and positive and negative affect). Twenty models were specified for each of the outcomes and moderators separately. None of the models showed any evidence of moderation. The results of all models can be seen in the Appendix.

Mediation effects

Based on the result of the repeated measures ANOVA that indicated significant effects of the BPS on anxiety and depression, we examined the mediating effect of positive and negative future expectations, goal-clarity, goal-commitment, goal-confidence, and goal-motivation on these two outcomes. In total, we examined 12 mediation models, one for each dependent variable and the mediator separately. None of the models showed a significant indirect effect of the proposed mediators. Detailed results can be seen in the Appendix.

Discussion

The BPS intervention was found effective in inducing positivity, but most evidence stems from Western populations. Moreover, it is yet unclear how the BPS intervention works. This study’s results partly confirmed the BPS’s effectiveness in the Chinese adult sample but not the mechanisms as hypothesized.

We first investigated the effectiveness of the BPS intervention on mental well-beings. The repeated measures ANOVA results indicated a significant interaction effect between condition and time for depression, anxiety, and negative future expectations. Specifically, according to the post hoc paired t-test, depression significantly increased in the TD condition but not in the BPS condition; anxiety decreased in the BPS condition and increased in the TD condition but neither reached significance; negative future expectations decreased significantly in the BPS but not in the TD condition. In addition, the negative affect significantly decreased in the BPS condition but not in the TD condition, and no significant interaction effect was found.

These results confirmed the hypothesis that in Chinese adults, the BPS practice would lead to less depression, anxiety, and negative future expectations compared to a control condition with small effect sizes (Ferguson, 2009). Despite being relatively small, the effect sizes are comparable to results in previous studies. For example, Peters et al. (2010) reported an effect of 0.04 on negative future expectations; Boselie et al. (2023) reported an effect of 0.08 on negative future expectations; Auyeung and Mo (2019) reported an effect of 0.04 for depression. In addition, the result partly confirmed the hypothesis that the BPS practice would lead to less negative affect, but only in the within-condition analyses. We found no significant effect of the BPS intervention on positive psychological outcomes. The above findings concur with Liau et al. (2016)’s study. However, this is inconsistent with previous meta-analyses indicating that BPS leads to more changes in positive psychological outcomes than in negative ones (Carrillo et al., 2019; Heekerens & Eid, 2021).

One possible explanation for this discrepancy might be the cultural difference in the content and specificity of the imagery. Imagining positive scenarios more vividly may have a greater impact on mood (Ji et al., 2016), but Chinese people produced fewer specific details when imagining their future than Euro-Americans (Wang et al., 2011). In addition, Chinese adolescents from Hong Kong and mainland China mentioned future careers a lot in their possible selves (Chen et al., 2015; Zhu & Tse, 2015). Focusing on future careers might not lead to as much positive feedback as imagery related to leisure activities and work-life balance (Loveday et al., 2018b).

Unexpectedly, the significant interaction effect regarding depression was not due to the BPS causing a decrease in state depression, but the TD exercise leading to a significant increase. The typical day (TD) practice is a frequently used method as a control condition in BPS studies due to its assumed neutral nature (Enrique et al., 2017). However, participants showed a significant increase in depression after the TD exercise in this study. One possible explanation is the ongoing COVID-19 at the time of the survey. A recent study found that 67.5% of Chinese college students reported moderate to severe stress during the COVID-19 pandemic (Zhan et al., 2021). Therefore, writing and imagining their typical days may have triggered responses to stressful daily routines and led to an increase in depressive feelings.

We did not find a moderation effect of collectivism and individualism. In this study, participants showed a somewhat higher level of individualism than collectivism, consistent with a recent study that young people in China endorse more individualistic values (Zeng & Greenfield, 2015). Though holding a higher individualistic value, participants in the current study still show no significant effect in positive outcomes from the BPS. Therefore, other cultural differences besides orientation may influence the effect of the BPS. For example, Chinese people cherish the ‘zhong yong’ principle, or the Doctrine of the Mean, which advocates a moral ideal where people’s affect, cognition, and behavior should be expressed in moderation (Ji & Chan, 2017). In that case, Chinese people may not express positive affect from imagining a best future self. Similarly, a study found that drawing a picture of a good future brings Chinese people a less positive mood than Americans (Yap et al., 2022). In addition, Chinese people’s dialectical thinking may prompt them to remain cautious when experiencing good fortune (Ji et al., 2012). Ji et al. (2021) found that the Chinese tend to generate more negative outcomes for positive events. Therefore, imagining a good future may not trigger a change in positive future expectations for Chinese participants.

We did not find the moderation effect of trait optimism and general self-efficacy. The result of trait optimism replicated previous studies (Meevissen et al., 2011; Peters et al., 2016). The results may indicate that people, regardless of their trait optimism and general self-efficacy level, can profit from the BPS intervention. However, no previous study has looked into the moderating effect of general self-efficacy. Thus, future studies may investigate this again in samples from similar or different cultural backgrounds.

We did not find a significant mediating effect of state optimism. These results replicate the previous finding by Heekerens et al. (2020). It is worth mentioning that in this previous study, only the mediating effect of positive future expectations was examined. Our study found a decrease in negative future expectations but did not find a significant mediating effect of it. Since this is the first study that investigated the effect of the BPS on state optimism in participants from a collectivistic culture, both the effect of the BPS on state optimism and the mediating effect of state optimism need further replication in other samples.

We also examined whether goal-related variables mediated the effects of the BPS. Participants in the BPS condition showed a significant increase in goal-clarity, but not in goal-commitment, goal-confidence, and goal-motivation. The increase in goal-clarity is in line with King’s assumption that imagining future can help people generate clearer goals. However, goal-clarity at post-intervention did not mediate the effect of the BPS on outcomes. This finding is partly consistent with the study of Heekerens et al. (2020). It should be noted that goal-ambivalence and goal-clarity are distinct variables measured differently. Therefore, future replications are needed.

The findings of this study should be interpreted in the context of several limitations. First, participants may have a more negative attitude toward their daily lives than in pre-pandemic times. For the Chinese sample, writing about a typical day during the COVID-19 situation may not be as neutral as expected. However, no information was collected regarding the extent to which participants feel influenced by COVID-19. Second, demographic information about rural or urban backgrounds is lacking. The difference between China’s rural and urban areas may influence the effect of the BPS and should be included in future studies. Third, the potential mechanism investigated in this study was based on previous studies from western samples. Researchers may investigate other variables that better reflect collectivistic cultural characteristics. Fourth, the manipulation check only examined if participants implemented the intervention following the instruction, not the subjective feeling towards the intervention. Therefore, a more comprehensive manipulation check should be considered.

In conclusion, this study supports the effectiveness of a single session of BPS writing and imagery in lowering depression, anxiety, and negative future expectation in Chinese participants. However, no effect on positive outcomes was found. Therefore, the original BPS may be better used as a buffering tactic in the Chinese sample. It should be noted that participants may not gain enough benefit from just one session. Thus, future studies should investigate the longer-term effect of BPS on Chinese participants. The result is partly consistent with previous studies in other collectivistic samples but inconsistent with those in western samples, indicating possible cultural sensitivity of the BPS. It concurs with previous arguments that PPIs developed in western culture should be modified before being implemented in eastern cultures (Hendriks & Graafsma, 2019). Future studies should explore the proper modification from a cultural perspective for the BPS to be used in Chinese society. In addition, qualitative analyses should be considered to investigate the influence of different writing content.

Author contributions

Liyang Wu: design of the study, data collection, and analysis, writing the original draft; Marjolein Hanssen: collaboration in the study design and data analysis, reviewing and editing the draft, supervision; Madelon Peters: collaboration in the study design, reviewing and editing the draft, supervision. All authors approved the final version of this manuscript.

Open scholarship

This article has earned the Center for Open Science badge for Preregistered. The materials are openly accessible at https://aspredicted.org/es8dh.pdf

Acknowledgments

The first author of this study received a scholarship from the Chinese Scholar Council (CSC), file NO. 202006870016. The authors would like to thank all the participants for their contribution.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data that support the findings of this study has been uploaded and is available from the authors upon reasonable request.

Additional information

Funding

This study received financial support from the program of China Scholarships Council.

Appendix

The written instruction of the Best-Possible-Self condition

You have been randomly assigned to an exercise in which you are going to think about your ‘best possible self’. Thinking about your ‘best possible self’ means imagining yourself in a future in which everything has turned out as good as possible. You have worked hard and you have managed to realize all your life goals. You can envision it as the realization of all your life dreams and the development of your best potential. In a moment, you are going to think for 1 minute about the best possible ways in which your life could develop. Afterward, we ask you to write down your thoughts about this for 15 minutes and finally, to visualize for 5 minutes the things you have been writing about. This exercise is timed automatically, so please make sure to follow the given time frame. In addition, we ask you to do this exercise alone, and not discuss the content of your story with your partner. Please turn to the next page to start with the exercise.

Figure A1. The mediating effect of different possible mediator between condition and state depression.

PFE = positive future expectations; NFE = negative future expectations.

Figure A2. The mediating effect of different possible mediator between condition and state anxiety.

PFE = positive future expectations; NFE = negative future expectations.

The written instruction of the Typical Day condition

For the next 15 minutes, please write about a typical day in your life. This means that you write about the ordinary details of your day that you usually don’t think about. These might include particular classes or meetings you attend to, people you meet, things you do, etc., during the day. Think of this as moving through your typical day, hour after hour. Thus, you identify what a typical day looks like for you.

The result of detecting the moderation effect using hierarchical regression analysis

Table

	Individualism				Collectivism				General self-efficacy				Optimism
	B	SE B	β	p	B	SE B	β	p	B	SE B	β	p	B	SE B	β	p
Step 1
Constant	8.13	4.86		0.099	15.81	5.38		0.005	10.93	6.11		0.078	14.60	4.25		0.001
Depression (pre-test)	0.51	0.12	0.45	0.000	0.42	0.13	0.37	0.002	0.48	0.14	0.42	0.001	0.38	0.14	0.34	0.008
Condition	2.60	1.29	0.22	0.048	2.61	1.26	0.22	0.043	2.62	1.28	0.22	0.045	2.71	1.25	0.23	0.035
Moderator	−0.01	0.08	−0.01	0.945	−0.13	0.08	−0.19	0.106	−0.09	0.16	−0.07	0.586	−0.32	0.17	−0.23	0.064
Step 2
Constant	9.75	6.38		0.131	10.61	6.66		0.116	7.16	6.71		0.290	12.97	4.76		0.008
Depression (pre-test)	0.51	0.13	0.45	0.000	0.44	0.13	0.39	0.002	0.45	0.14	0.39	0.002	0.38	0.14	0.34	0.008
Condition	−0.82	8.75	−0.07	0.926	11.89	7.18	1.00	0.102	13.55	8.35	1.14	0.110	5.94	4.38	0.50	0.180
Moderator	−0.03	0.11	−0.05	0.756	−0.03	0.11	−0.44	0.787	0.06	0.19	0.05	0.740	−0.20	0.23	−0.14	0.390
Interaction	0.07	0.16	0.30	0.694	−0.19	0.14	−0.81	0.193	−0.37	0.28	−0.95	0.190	−0.23	0.29	0.31	0.444

Note: Interaction = Condition x Moderator.

Table

	Individualism				Collectivism				General self-efficacy				Optimism
	B	SE B	β	p	B	SE B	β	p	B	SE B	β	p	B	SE B	β	p
Step 1
Constant	5.10	2.90		0.083	9.59	3.30		0.005	6.80	3.52		0.058	9.07	2.87		0.002
Anxiety (pre-test)	0.56	0.11	0.53	0.000	0.47	0.12	0.45	0.000	0.52	0.12	0.50	0.000	0.41	0.13	0.39	0.003
Condition	1.35	0.79	0.18	0.092	1.33	0.77	0.18	0.089	1.37	0.79	0.18	0.086	1.36	0.77	0.18	0.081
Moderator	−0.01	0.05	−0.03	0.812	−0.08	0.05	−0.19	0.094	−0.07	0.09	−0.08	0.478	−0.20	0.11	−0.23	0.073
Step 2
Constant	7.44	3.67		0.047	9.94	3.96		0.015	6.52	3.91		0.100	8.98	3.11		0.005
Anxiety (pre-test)	0.55	0.11	0.53	0.000	0.47	0.12	0.45	0.000	0.52	0.12	0.50	0.000	0.41	0.14	0.39	0.003
Condition	−4.02	5.22	−0.53	0.444	0.63	4.39	0.08	0.886	2.22	5.21	0.30	0.672	1.56	2.69	0.21	0.565
Moderator	−0.06	0.07	−0.12	0.393	−0.09	0.07	−0.21	0.179	−0.05	0.12	−0.10	0.641	−0.20	0.14	−0.22	0.180
Interaction	0.10	0.10	0.72	0.302	0.01	0.09	0.10	0.872	−0.03	0.18	−0.12	0.870	−0.01	0.18	−0.03	0.940

Note: Interaction = Condition x Moderator.

Table

	Individualism				Collectivism				General self-efficacy				Optimism
	B	SE B	β	p	B	SE B	β	p	B	SE B	β	p	B	SE B	β	p
Step 1
Constant	−2.43	2.93		0.409	−1.43	2.26		0.527	1.64	2.48		0.509	−0.88	1.47		0.553
SWL (pre-test)	1.05	0.06	0.92	0.000	1.03	0.07	0.90	0.000	1.07	0.07	0.94	0.000	1.01	0.07	0.88	0.000
Condition	−1.21	0.78	−0.08	0.125	−1.24	0.78	−0.08	0.117	−1.24	0.78	−0.08	0.117	−1.25	0.78	−0.9	0.114
Moderator	0.04	0.05	0.05	0.390	0.03	0.05	0.04	0.495	−0.08	0.09	−0.05	0.403	0.10	0.11	0.06	0.341
Step 2
Constant	−2.49	3.63		0.496	−0.96	3.14		0.761	2.10	3.33		0.532	−1.35	1.98		0.499
SWL (pre-test)	1.05	0.06	0.92	0.000	1.04	0.07	0.90	0.000	1.07	0.07	0.93	0.000	1.01	0.07	0.88	0.000
Condition	−1.08	5.20	−0.07	0.837	−2.21	4.46	−0.15	0.622	−2.25	5.05	−0.15	0.657	−0.32	2.72	−0.02	0.907
Moderator	0.04	0.07	0.05	0.506	0.02	0.07	0.03	0.743	−0.10	0.12	−0.06	0.434	0.14	0.15	0.08	0.351
Interaction	−.003	0.10	−0.01	0.979	0.02	0.09	0.07	0.826	0.03	0.17	0.07	0.839	−0.07	0.18	−0.07	0.723

Note: SWL = satisfaction with life; Interaction = Condition x Moderator.

Table

	Individualism				Collectivism				General self-efficacy				Optimism
	B	SE B	β	p	B	SE B	β	p	B	SE B	β	p	B	SE B	β	p
Step 1
Constant	8.94	5.06		0.082	4.36	3.73		0.247	6.29	4.06		0.126	9.21	2.94		0.003
PA (pre-test)	0.76	0.09	0.73	0.000	0.67	0.10	0.64	0.000	0.69	0.11	0.66	0.000	0.77	0.11	0.74	0.000
Condition	−1.94	1.27	−0.13	0.132	−1.93	1.23	−0.13	0.123	−1.96	1.26	−0.13	0.125	−1.94	1.27	−0.13	0.132
Moderator	0.004	0.08	0.004	0.962	0.16	0.08	0.18	0.054	0.18	0.18	0.11	0.321	−0.03	0.19	−0.02	0.889
Step 2
Constant	7.85	6.21		0.211	6.89	5.07		0.180	12.96	5.28		0.017	8.82	3.74		0.022
PA (pre-test)	0.76	0.09	0.73	0.000	0.67	0.10	0.65	0.000	0.69	0.11	0.66	0.000	0.77	0.11	0.74	0.000
Condition	0.65	8.51	0.04	0.939	−7.04	7.05	−0.47	0.322	−17.12	7.98	−1.14	0.036	−1.21	4.46	−0.08	0.787
Moderator	0.03	0.10	0.03	0.813	0.10	0.11	0.12	0.345	−0.05	0.21	−0.03	0.805	0.00	0.24	0.00	0.999
Interaction	−0.05	0.16	−0.18	0.759	0.10	0.14	0.36	0.464	0.51	0.27	1.05	0.059	−0.05	0.30	−0.05	0.865

Note: PA = positive affect; Interaction = Condition x Moderator.

Table

	Individualism				Collectivism				General self-efficacy				Optimism
	B	SE B	β	p	B	SE B	β	p	B	SE B	β	p	B	SE B	β	p
Step 1
Constant	2.30	3.72		0.539	5.37	3.75		0.157	5.36	4.19		0.205	5.63	3.25		0.087
NA (pre-test)	0.67	0.07	0.76	0.000	0.67	0.07	0.76	0.000	0.67	0.07	0.76	0.000	0.66	0.08	0.75	0.000
Condition	1.09	1.08	0.08	0.317	1.08	1.09	0.08	0.324	1.09	1.09	0.08	0.321	1.10	1.08	0.08	0.317
Moderator	0.04	0.07	0.04	0.586	−0.02	0.06	−0.03	0.720	−0.04	0.12	−0.03	0.755	−0.08	0.15	−0.05	0.602
Step 2
Constant	3.90	4.86		0.425	4.62	4.90		0.350	5.72	5.04		0.261	5.23	3.95		0.190
NA (pre-test)	0.67	0.07	0.76	0.000	0.68	0.07	0.76	0.000	0.68	0.07	0.77	0.000	0.66	0.08	0.75	0.000
Condition	−2.61	7.25	−0.19	0.720	2.55	6.21	0.19	0.683	0.18	7.12	0.01	0.980	1.75	3.84	0.13	0.649
Moderator	0.01	0.09	0.01	0.935	−0.01	0.09	−0.01	0.933	−0.05	0.16	−0.04	0.749	−0.05	0.21	−0.03	0.809
Interaction	0.07	0.14	0.28	0.607	−0.03	0.123	−0.11	0.810	0.03	0.24	0.07	0.898	−0.05	0.26	−0.05	0.859

Note: NA = negative affect; Interaction = Condition x Moderator.

The result of detecting mediating effect using process macro in SPSS22.0

Based on the result of repeated measures ANOVA that indicated significant effects of the BPS on anxiety and depression, we examined the mediating effect of positive and negative future expectations, goal-clarity, goal-commitment, goal-confidence, and goal-motivation on these two outcomes. In total, we examined 12 mediation models, one for each dependent variable and mediator separately. None of the models showed a significant indirect effect of the proposed mediators. Detailed results can be seen below.