1.  Introduction

The digital economy’s rapid growth is reshaping global economic and employment landscapes. Per CNNIC data, by 2024, China’s digital economy exceeded 50 trillion yuan (over 40% of GDP), with the platform economy (a core driver) spawning a "gig, decentralized, online" labor model. Short video, knowledge payment, and live e-commerce platforms have built a new content ecosystem, drawing many to work as knowledge-based content creators who monetize professional knowledge or skills, forming a large new group.

Platform labor’s low threshold and flexibility break traditional occupational spatio-temporal limits, offering diverse careers [1]: education bloggers profit from online courses, tech creators from technical analysis. This "everyone can create" model boosts knowledge flow and industry growth [2]. Yet, platform labor differs from traditional organizations: relations shift to lose cooperation, management to AI-driven algorithmic management, weakening creators’ identity/security and interpersonal interaction, and increasing participation difficulty [3,4]. Knowledge-based creators face high-vitality and high-risk careers: top creators earn via traffic and IP, but most face three challenges: (1) Algorithm dependence (homogeneous competition for exposure; [5] (2) Unstable careers (opaque algorithms, overwork; [6]); (3) AI-driven skill pressure (shift to creative planning or replacement; [7]). Nearly half quit within three years, limiting industry sustainability.

Career sustainability, emphasized long-term competitiveness and adaptability balance [8], is critical for creators, manifesting in stable quality content, adaptability to changes, income, and identity gains [9]. It eases individual anxiety [10] and fosters industry quality, but existing research ignores platform labor’s career sustainability, especially how platform support acts via psychological mechanisms.

This study explores platform support’s impact on creators’ career sustainability and tests creative self-efficacy’s mediating role. It builds a "platform support–creative self-efficacy–career sustainability" model, tests it via PLS-SEM, and adapts mature scales to clarify paths and link external support with individual psychology. Theoretically, it extends career sustainability research to platforms, integrates social exchange/social cognitive theories, and enriches career theory. Practically, it guides platforms to optimize support, boosting creators’ self-efficacy and industry sustainability.

The structure of this paper is as follows: the second part introduces the theoretical foundation and hypothesis construction of this study; the third part describes the research methodology; the fourth part presents the data analysis and results; the fifth part is a discussion and research revelation, and points out the limitations of the study and the direction of future research.

2.  Literature review and theoretical foundations

2.1.  Sense of platform support and career development of knowledge-based content creators

In the digital economy’s reshaped career ecology, knowledge-based content creators’ career development is increasingly detached from traditional organizational paths and relies on platform support systems. Platform support, an extension of organizational support in the platform context, emphasizes creators’ subjective perception of platform-provided support in resources, rules, and emotions. Derived from [11] organizational support theory, rooted in social exchange theory, it reflects individuals’ perception of whether the platform values their interests [12]; per the theory, individual-environment interaction follows the norm of reciprocity [13], so when platforms prioritize creators’ development, creators gain responsibility and identity, leading to higher commitment and initiative [14].

Given platform labor’s traits of de-employment and weak contractual, platform support operates via three dimensions [15]: resource support lowers creation thresholds and reduces creative interruption risks; rule support boosts sustained platform investment by improving system transparency and benefit certainty; emotional support strengthens platform community belonging and professional identity to help cope with uncertainty [16].

Existing studies confirm platform support’s positive impact on career development: stable profit-sharing helps mid-tier creators maintain update frequency [17]; positive atmosphere perception enhances skill iteration and content quality [18]; WeChat official account creators transition from creators to entrepreneurs via the platform ecosystem [19]. Thus, this study proposes:

H1: Sense of platform support has a significant positive effect on the career sustainability of knowledge-based content creators.

2.2.  Sense of platform support and creative self-efficacy

The sense of platform support is a key environmental cue in the digital platform ecology [20]. four-factor model of social cognitive theory, platform support shapes individuals’ efficacy judgments via direct experience, alternative experience, verbal persuasion, and emotional arousal, enhancing their coping ability and creative motivation in uncertain creative settings.

Direct experience: Platforms’ low-risk trial-and-error mechanisms (e.g., content preview) cut content failure costs, helping novice creators build feedback confidence [21]. Alternative experience: Sharing others’ success stories (e.g., guiding with cutting-edge tech logic) reduces innovation uncertainty, boosting creators’ efficacy [22]. Verbal persuasion: Platform labels (e.g., quality creators) and user feedback (likes or comments) convey positive evaluations. For instance, Douyin’s “DOU+” traffic boost strengthens creators’ recognition of their creative value [23]. Emotional arousal: Platform tools (Canva templates, AIGC plug-ins) ease from-scratch anxiety and increase willingness to try new content forms. Thus, this study proposes:

H2: A sense of platform support has a significant positive effect on the creative self-efficacy of knowledge-based content creators.

2.3.  Relationship between creative self-efficacy and career sustainability

Creative self-efficacy, a core social cognitive theory construct, is an individual’s belief in their creative thinking and action ability [24]. In platform environments, it includes content novelty judgments and adaptability to ecosystem changes, and boosts sustained innovation by enhancing goal-setting and frustration tolerance [25]. Career sustainability emphasizes a career’s continuity, adaptability, and meaning in dynamic environments (covering health, happiness, productivity;. For platform-based knowledge content creators, it manifests in three competencies: content vitality, career resilience, and developmental extension [26]. Prior studies link creative self-efficacy to career sustainability: high efficacy drives proactive skill learning for content sustainability; continuous innovative output forms an “efficacy-behavior-accomplishment” cycle to enhance career resilience [27]; efficacy also pushes creators to break content boundaries for career expansion. Thus, this study proposes:

H3: Creative self-efficacy has a significant positive effect on the career sustainability of knowledge-based content creators.

H4: Creative self-efficacy positively mediates the relationship between perceptions of platform support and knowledge-based content creators' career sustainability.

Based on this, the study uses social exchange theory and social cognitive theory to construct a theoretical model, systematically analysing the mechanism of platform-enabled career development for knowledge-based content creators (see Figure 1).

3.  Research design

3.1.  Research methodology and sample

This study uses partial least squares structural equation modeling for data analysis. Compared with covariance-based SEM, PLS-SEM is more suitable for medium sample sizes, complex model structures, and variable measurement prediction needs [28]; as this study focuses on the path mechanism among sense of platform support, creative self-efficacy, and career sustainability, PLS-SEM can handle both measurement and structural models, boasting strong estimation efficiency and applicability. The anonymous self-administered questionnaire included five parts: subject description, basic information, platform support scale, creative self-efficacy scale, and career sustainability scale. To ensure logic, clarity, and adaptability, 50 knowledge-based content creators were pre-tested—based on feedback, some question wording was optimized and items with poor reliability and validity were deleted. Surveys targeted knowledge-based content creators active on mainstream digital platforms, using "snowball sampling" combined with creator community recruitment to collect samples. To meet PLS-SEM’s sample size requirement, the target sample size was set at over 800, with 774 valid questionnaires finally returned (criteria: complete answers and reasonable response time).

3.2.  Variable measurement and scale source

This study’s core variables, platform support, creative self-efficacy, and career sustainability, are adapted from validated literature scales (moderately contextualized for platforms) and measured on a 7-point Likert. Perceived Platform Support (PPS) is contextually adapted from [29] Perceived Organizational Support Scale (POS Scale, widely used in employment research to measure subjective perceptions of organizational importance and support). Aligned with digital platform labor traits, items are reconstructed—6 items total. Creative Self-Efficacy (CSE) uses the CSE subscale from [30] short Creative Self-Efficacy Scale, which assesses confidence in creative problem-solving under complexity and uncertainty—6 items total. Career Sustainability adopts [31] Career Sustainability Scale, covering 4 dimensions (Resources, Flexibility, Regeneration, Integration) with 2 items each (8 total). Sample items include “My career makes me feel that I have a bright future”, comprehensively measuring perceptions of career growth and continuity in the platform context.

3.3.  Analysing strategy

This study used SPSS 26.0 and SmartPLS 4.0 for questionnaire data analysis and structural modeling. First, SPSS pre-processed data, eliminating invalid questionnaires, handling missing values and reverse questions, and conducting descriptive statistics on key variables to present mean, standard deviation, and other basic distribution characteristics. Next, PLS-SEM analysis was performed: the measurement model’s reliability and validity were assessed to ensure the measurement structure’s rationality. In structural modeling analysis, each path hypothesis’ significance and direction were tested. Model fit and explanatory power were judged via indexes, and the Bootstrapping method (5000 repeated samples) constructed confidence intervals to verify the mediating effect’s significance. Finally, sensitivity analysis was conducted to enhance conclusion robustness: control variables (e.g., creators’ field, platform experience) were introduced, and model path stability was tested across sub-samples to improve results’ external validity and generalization value.

4.  Analysis of empirical results

4.1.  Measurement model assessment

This study treats all constructs as reflective variables. Following PLS methodology, Cronbach's alpha, rho_a, and composite reliability (rho_c) were calculated to assess the measurement model’s reliability, i.e. whether questionnaire items stably and consistently reflect latent variables [32]. As shown in Table 1, all latent variables had Cronbach's alpha and rho_c above the 0.7 threshold [33], indicating excellent scale internal consistency. Additionally, all factor loadings exceeded 0.7, and average variance extracted was above 0.5, confirming convergent validity [34]., HTMT ratios were below 0.85, and each latent variable’s AVE square root (Table 2 diagonal) exceeded its correlations with other variables, supporting discriminant validity. All indicators’ variance inflation factor (VIF) was <3, so multicollinearity was negligible .

Table 2 presents variable descriptive statistics and correlations: on a 7-point scale, platform support perception, creative self-efficacy, and career sustainability showed creators’ positive perceptions of platform support and personal competence, plus high career sustainability. Correlation analysis revealed significant positive correlations between PPS and CSand CSE and CS, preliminarily supporting hypotheses. Notably, platform activity and CS had the strongest correlation, consistent with [35] findings on user engagement’s impact on career development. However, age and content area had no significant correlations with core variables, suggesting limited roles in the model. Overall, the measurement model passed reliability and validity tests, with core variables showing good psychometric properties, supporting subsequent structural modeling and providing a basis for exploring variable causal relationships.

Note: Diagonal: square root of AVE.

4.2.  Structural model test

This study systematically tested path relationships in the theoretical model (Table 3 reports path coefficients and significance). First, sense of platform support exerted a significant positive effect on career sustainability (β=0.378, p<0.001), supporting H1, validating digital platforms’ role as external support for creators’ careers (consistent with [36]) with a medium effect size (f²=0.208; [37]). Perceived platform support had a particularly significant positive effect on creative self-efficacy (β=0.596, p<0.001), supporting H2; the large effect size (f²=0.552) confirms platform support as a key external factor boosting creators’ self-efficacy, deepening [38] social cognitive theory application in digital contexts. Creative self-efficacy also significantly positively affected career sustainability (β=0.352, p<0.001), supporting H3; though its effect size (f²=0.178) was slightly lower than the direct path, it remained practically significant, corroborating [39] findings that creators’ innovation confidence drives long-term career sustainability.

Mediation tests showed creative self-efficacy partially mediated the link between perceived platform support and career sustainability (β=0.210, p<0.001), supporting H4, indicating platform support acts on careers both directly and indirectly via enhanced self-efficacy (a dual-path mechanism for understanding platform empowerment).

In model explanatory power (Figure 2), sense of platform support explained 35.6% of creative self-efficacy variance (R²=0.356), while together with creative self-efficacy, it explained 56.4% of career sustainability variance (R²=0.564). Control variable analysis found platform activity positively affected career sustainability (β=0.233, p<0.001), gender had a weak positive effect (β=0.101, p<0.05), age and content area had no significant effects (p>0.05)—aligning with descriptive statistics and indicating limited predictive power of demographics after controlling core variables.

4.3.  Robustness tests

Further, we conducted robustness tests by removing control variables from the study. This is mainly based on methodological rigor considerations. According to standard procedures in structural equation modeling research , this approach serves three main purposes: first, to ensure the purity of the relationship between the core variables by removing potentially interfering variables; and second, to verify that the model is not overcomplicated by the introduction of extraneous variables, in line with the principle of theoretical parsimony. This type of testing is widely recommended by leading scholars in the fields of management and psychology, and is effective in enhancing the reliability and generalizability of research findings, especially when the control variables lack clear associations with the core theoretical constructs . By systematically assessing the stability of the model in different settings, researchers can more accurately grasp the essential relationships between variables. As shown in Figure 3 and Table 4, the core path relationships remained stable and some of the effect values were enhanced after excluding the effects of control variables.

Therefore, the theoretical model constructed in this study has robust internal validity, and the relationship between core variables is not disturbed by the setting of control variables. This, on the one hand, confirms the theoretical advantages of the sense of platform support and creative self-efficacy in explaining career sustainability, and, on the other hand, offers the possibility of simplifying the model for subsequent research.

5.  Discussion and implications

5.1.  Main findings

Using 774 knowledge-based content creators as samples, this study empirically explored via PLS-SEM how perceived platform support affects career sustainability and analysed creative self-efficacy’s mediating role. Findings identify dual paths that external platform support and internal psychological mechanisms drive creators’ career sustainability. Perceived platform support significantly positively impacts career sustainability, as creators build careers through platform resources, rules, and community support, making it key to their commitment and identity. It also notably affects creative self-efficacy. Creative self-efficacy itself positively impacts career sustainability, highlighting psychological mechanisms’ central role. Overall, creative self-efficacy partially mediates the link: platform support drives career development directly and indirectly by shaping self-efficacy.

5.2.  Theoretical contributions

This study has three theoretical contributions. First, it extends career sustainability research to digital platform labor, existing literature focuses more on traditional organizational careers, overlooks platform labor as a new form, and by examining knowledge-based content creators, responds to digital economy career transitions. Second, it integrates social exchange theory and social cognitive theory to verify how perceived platform support affects career sustainability through creative self-efficacy, expanding career development research boundaries. Third, it deepens creative self-efficacy understanding: unlike prior studies emphasizing its innovation role, it reveals that in platform labor, creative self-efficacy not only directly drives career sustainability but also mediates, enriching efficacy theory’s digital labor applicability.

5.3.  Practical implications

At the practical level, this study has implications for the development of both platform operators and creators themselves. For platforms, the findings suggest that their support systems are directly related to creators' career stability and development potential. Platforms should actively promote sustainable support mechanisms to enhance creators' sense of security and professional identity. This will not only help reduce creator turnover, but also promote the formation of a healthy content ecosystem. For creators, improving creative self-efficacy is the key to long-term career development. Creators need to proactively update their skills and innovate their content, while maintaining resilience and a positive mindset in the midst of uncertainty, in order to maintain their career advantage in the face of fierce competition.

5.4.  Limitations of the study and outlook for future research

Despite robust results, this study has limitations. First, samples were mainly from mainstream Chinese digital platforms, so findings’ cross-cultural or cross-platform applicability needs verification; future research could conduct international comparisons to explore platform support’s role in career development under different institutional environments. Second, cross-sectional data was used, requiring caution in causal inferences; longitudinal studies or experiments can further verify dynamic variable relationships.