Algorithmic Feedback Framing and Platform Dependency: Experimental Evidence on Risk Perception and Decision-Making in the Creator Economy
演算法回饋框架與平台依賴性：創作者經濟中風險感知與決策的實驗證據

Abstract
摘要

Algorithmic feedback is a defining feature of the creator economy, imposing a significant psychological burden and testing the emotional resilience of digital workers. To understand how platform design impacts creators' psychosocial well-being, this study investigates how feedback framing (positive vs. negative) and platform dependency interact to influence risk perception and career decision-making. Across three experimental studies (N = 622), we manipulated feedback valence and information transparency using simulated platform interfaces. Results show that negative feedback framing exacerbates the psychological cost, significantly increasing perceived risk and reducing entrepreneurial intention, particularly for highly platform-dependent creators. Conversely, transparent algorithmic signals can foster emotional resilience by enhancing platform trust and decision confidence. Platform dependency consistently amplifies these psychological effects, highlighting the vulnerability of reliant creators. These findings underscore the urgent need for more transparent and psychologically supportive feedback systems, offering evidence-based insights for platform designers, policymakers, and mental health professionals aiming to mitigate psychological burdens and support the psychosocial well-being and career sustainability of digital workers.
演算法回饋是創作者經濟的核心特徵，對數位工作者施加了顯著的心理負擔，並考驗其情緒韌性。為了理解平台設計如何影響創作者的心理社會福祉，本研究探討回饋框架（正向與負向）與平台依賴性如何交互作用，進而影響風險感知與職業決策。透過三項實驗研究（N = 622），我們利用模擬平台介面操控回饋價值與資訊透明度。結果顯示，負向回饋框架加劇心理成本，顯著提升風險感知並降低創業意願，尤其對高度依賴平台的創作者影響更大。相反地，透明的演算法訊號能透過增強平台信任與決策信心，促進情緒韌性。平台依賴性持續放大這些心理效應，凸顯依賴型創作者的脆弱性。 這些發現強調了建立更透明且具心理支持性的反饋系統的迫切需求，為平台設計者、政策制定者及心理健康專業人士提供了基於證據的見解，旨在減輕心理負擔並支持數位工作者的心理社會福祉與職業可持續性。

Keywords: algorithmic feedback, risk perception, platform dependency, decision-making, creator economy, digital labor, transparency, psychosocial well-being
關鍵詞：演算法反饋、風險知覺、平台依賴、決策制定、創作者經濟、數位勞動、透明度、心理社會福祉

Introduction
引言

The creator economy has become a core component of the socio-economic structure of modern cyberspace, where platform-driven entrepreneurship is profoundly reshaping career trajectories for younger generations (Edeling & Wies, 2024; Kerrigan et al., 2022). Globally, this has transformed content creation from a niche pursuit into a mainstream professional aspiration (Ma et al., 2023; Rospitasari et al., 2023). Yet, unlike traditional professions, careers in this new digital landscape unfold in environments dominated by algorithmic opacity, shifting platform policies, and unpredictable feedback loops (Duggan et al., 2023; Maric et al., 2024; Möhlmannn et al., 2023).
創作者經濟已成為現代網路空間社會經濟結構的核心組成部分，平台驅動的創業精神深刻地重塑了年輕世代的職業軌跡（Edeling & Wies, 2024；Kerrigan et al., 2022）。在全球範圍內，這使內容創作從一項小眾活動轉變為主流的職業志向（Ma et al., 2023；Rospitasari et al., 2023）。然而，與傳統職業不同，這一新數位領域的職業發展環境充斥著演算法的不透明性、平台政策的變動以及不可預測的反饋迴路（Duggan et al., 2023；Maric et al., 2024；Möhlmannn et al., 2023）。

Within these platform-mediated environments, signals such as view counts, engagement rates, and recommendation scores serve as central forces shaping creators’ perceptions of risk, opportunity, and professional viability (Bleier et al., 2022). These signals not only reflect creators’ performance but also function as active cues that influence daily strategy and long-term career planning. Unlike employees in traditional sectors, creators must constantly interpret shifting signals from complex, often opaque algorithmic systems (Duggan et al., 2023; Maric et al., 2024). The volatility and unpredictability of such feedback—frequently delivered with little transparency—can make it challenging to understand changes in reach, visibility, or earnings. As a result, creators develop adaptive heuristics to manage feedback and sustain momentum in an uncertain environment (Appio & Corso, 2023; Meng, 2024; Ye et al., 2024).
在這些由平台媒介的環境中，觀看次數、互動率及推薦分數等訊號，成為塑造創作者對風險、機會及職業可行性認知的核心力量（Bleier et al., 2022）。這些訊號不僅反映創作者的表現，亦作為積極的提示，影響日常策略及長期職涯規劃。與傳統產業的員工不同，創作者必須持續解讀來自複雜且常常不透明的演算法系統中不斷變動的訊號（Duggan et al., 2023；Maric et al., 2024）。此類反饋的波動性與不可預測性——且常以低透明度方式呈現——使得理解觸及率、能見度或收益的變化變得困難。因此，創作者發展出適應性啟發式方法，以管理反饋並在不確定的環境中維持動能（Appio & Corso, 2023；Meng, 2024；Ye et al., 2024）。

The Creator Economy and Platform Signals
創作者經濟與平台訊號

As digital platforms have grown in reach and sophistication, their role as arbiters of opportunity and success has intensified. Platforms such as YouTube, Instagram, and TikTok not only provide the infrastructure for creative expression and audience engagement, but also act as gatekeepers through their design logic and data-driven feedback systems (Bleier et al., 2022). Platform-generated signals—including algorithmic recommendations, view and engagement metrics, and visibility scores—serve as powerful cues that inform creators’ day-to-day decisions and long-term career strategies.
隨著數位平台的影響力與複雜度日益增長，其作為機會與成功仲裁者的角色也愈加重要。YouTube、Instagram 及 TikTok 等平台不僅提供創意表達與觀眾互動的基礎設施，還透過其設計邏輯與數據驅動的反饋系統擔任守門人角色（Bleier et al., 2022）。平台生成的訊號——包括演算法推薦、觀看與互動指標，以及能見度分數——成為強而有力的線索，指引創作者的日常決策與長期職涯策略。

Within the creator economy, this continuous need to interpret and adapt to shifting signals from opaque algorithms marks a key difference from traditional employment (Duggan et al., 2023; Maric et al., 2024). As a result, creators are frequently required to develop heuristics and adaptive strategies to make sense of algorithmic feedback and maintain their professional momentum.
在創作者經濟中，持續解讀並適應來自不透明演算法的變動訊號，是與傳統就業模式的關鍵差異（Duggan et al., 2023；Maric et al., 2024）。因此，創作者經常需要發展啟發式方法與適應策略，以理解演算法反饋並維持其專業動能。

Rather than viewing creators as passive recipients, the current perspective highlights their active role in decoding, filtering, and engaging with the informational cues generated by platforms (Appio & Corso, 2023; Meng, 2024; Ye et al., 2024). This interpretive process is central to creators’ psychological resilience, shaping their perceptions of opportunity, risk, and long-term viability. Thus, understanding how platform signals function as both constraints and enablers is crucial for explaining the emergence and sustainability of creative digital careers.
當代觀點不將創作者視為被動接受者，而是強調他們在解讀、篩選及互動平台所產生資訊線索中的主動角色（Appio & Corso, 2023；Meng, 2024；Ye 等，2024）。這一詮釋過程是創作者心理韌性的核心，塑造他們對機會、風險及長期可行性的認知。因此，理解平台訊號如何同時作為限制與促進因素，對於解釋創意數位職涯的興起與持續性至關重要。

Theoretical Integration: Signaling, Framing, and Risk Perception
理論整合：訊號傳遞、框架設定與風險感知

To illuminate how creators navigate these dynamic environments, this study integrates signaling theory, framing theory, and risk perception perspectives. Signaling theory posits that observable cues—such as algorithmic analytics—reduce uncertainty in contexts of information asymmetry, guiding decisions about value and opportunity (Spence, 1973; Connelly et al., 2011; Hunter et al., 2022). However, unlike traditional signals, platform feedback is highly fluid, ambiguous, and context-dependent. Framing theory demonstrates that the manner in which feedback is presented—whether as a temporary fluctuation, a warning, or an algorithmic update—critically shapes interpretation and emotional response (Kahneman & Tversky, 1984; Janson & Dickhäuser, 2023; Murphy & Knowlton, 2022; Nefedov, 2022). Negative or unclear framing can trigger loss aversion and anxiety, while positive or transparent framing can bolster engagement and resilience.
為了闡明創作者如何在這些動態環境中進行導航，本研究整合了訊號理論、框架理論與風險知覺觀點。訊號理論主張，可觀察的線索——如演算法分析——在資訊不對稱的情境中降低不確定性，指引價值與機會的決策（Spence, 1973；Connelly 等，2011；Hunter 等，2022）。然而，與傳統訊號不同，平台回饋高度流動、模糊且依賴情境。框架理論則指出，回饋呈現的方式——無論是暫時波動、警示或演算法更新——關鍵影響詮釋與情緒反應（Kahneman & Tversky, 1984；Janson & Dickhäuser, 2023；Murphy & Knowlton, 2022；Nefedov, 2022）。負面或不明確的框架可能引發損失規避與焦慮，而正面或透明的框架則能增強參與度與韌性。

Risk perception theory further underscores that creators’ sense of professional risk arises not only from objective volatility but also from subjective appraisal of signals and platform narratives (Slovic, 1987; Cabeza-Ramírez et al., 2022). When algorithmic rationales are lacking, creators may attribute poor performance to randomness or bias, heightening uncertainty and undermining trust (Eslami et al., 2019; Pavleska & Jerman Blažič, 2017). Conversely, transparent and supportive feedback can mitigate negative psychological effects and support sustained engagement.
風險知覺理論進一步強調，創作者的專業風險感不僅源自客觀的波動性，還來自對訊號和平台敘事的主觀評估（Slovic, 1987；Cabeza-Ramírez 等，2022）。當缺乏演算法理據時，創作者可能將表現不佳歸因於隨機性或偏見，進而加劇不確定性並削弱信任（Eslami 等，2019；Pavleska & Jerman Blažič, 2017）。相反地，透明且支持性的回饋能減輕負面心理影響，並促進持續參與。

However, platform signals differ from traditional signals in that they are highly fluid, often ambiguous, and deeply contextual. Here, framing theory (Kahneman & Tversky, 1984; Janson & Dickhäuser, 2023) becomes essential. The framing of platform feedback—whether a metric drop is presented as an “algorithmic update,” a “temporary fluctuation,” or a “decline in engagement”—significantly affects how creators interpret and emotionally react to changes. Studies show that negative or ambiguous frames can intensify loss aversion, anxiety, and avoidance behaviors, while positive or transparent frames can sustain engagement and adaptive persistence (Murphy & Knowlton, 2022; Nefedov, 2022).
然而，平台訊號與傳統訊號不同，因其高度流動性、常帶有模糊性且深具情境性。在此，框架理論（Kahneman & Tversky, 1984；Janson & Dickhäuser, 2023）變得至關重要。平台反饋的框架——無論是將指標下降呈現為「演算法更新」、「暫時波動」或「參與度下降」——都顯著影響創作者對變化的解讀及情緒反應。研究顯示，負面或模糊的框架會加劇損失厭惡、焦慮及迴避行為，而正面或透明的框架則能維持參與度及適應性堅持（Murphy & Knowlton, 2022；Nefedov, 2022）。

Central to both signaling and framing is the concept of risk perception (Slovic, 1987; Cabeza-Ramírez et al., 2022). Creators’ sense of professional risk is shaped not just by objective volatility, but by their subjective appraisal of signals and the narratives provided by platform interfaces. When rationales for algorithmic changes are lacking, creators may attribute poor performance to randomness or bias, increasing uncertainty and eroding trust (Eslami et al., 2019; Pavleska & Jerman Blažič, 2017). Conversely, clear and supportive feedback can buffer negative psychological effects and foster resilience.
風險知覺的概念是訊號傳遞與框架設定的核心（Slovic, 1987；Cabeza-Ramírez 等，2022）。創作者對專業風險的感知，不僅受客觀波動性的影響，更取決於他們對訊號的主觀評估以及平台介面所提供的敘事。當缺乏對演算法變動的合理解釋時，創作者可能將表現不佳歸因於隨機性或偏見，進而增加不確定性並削弱信任（Eslami 等，2019；Pavleska & Jerman Blažič, 2017）。相反地，清晰且具支持性的回饋能緩衝負面心理影響，並促進韌性。

By combining these perspectives, we conceptualize creators as active interpreters whose reactions depend on signal design, framing, and perceived platform trust. This theoretical integration sets the stage for a nuanced investigation of how signal features—framing, transparency, and modality—interact to influence creators’ cognition, risk tolerance, and strategic behavior in algorithmic environments.
結合上述觀點，我們將創作者概念化為積極的詮釋者，其反應取決於訊號設計、框架設定及對平台信任的感知。此理論整合為細緻探討訊號特徵——框架、透明度與模態——如何交互影響創作者在演算法環境中的認知、風險容忍度與策略行為奠定基礎。

Individual Differences: Platform Dependency and Autonomy
個體差異：平台依賴性與自主性

While creators share a common reliance on digital platforms, their experiences diverge sharply depending on their degree of platform dependency—the extent to which income, professional identity, and career opportunities hinge on a single platform (Guiñez-Cabrera & Aqueveque, 2021; Cutolo & Kenney, 2021). Highly dependent creators, such as full-time influencers or streamers who rely exclusively on platforms like YouTube or TikTok, are especially vulnerable to sudden algorithmic changes, policy shifts, or monetization disruptions (Lee & Jin, 2024; Are & Briggs, 2023). Even small drops in engagement can trigger acute anxiety, risk aversion, and identity threats, as their livelihoods and social capital are tightly coupled to platform stability (Guo et al., 2022). By contrast, creators who diversify income streams or cultivate audiences across multiple channels tend to demonstrate greater resilience, perceiving negative signals as challenges rather than existential threats (Appio & Corso, 2023; Ye et al., 2024). Yet, much prior research has treated creators as a homogeneous group, overlooking crucial differences in autonomy, risk tolerance, and adaptive strategy (Verwiebe et al., 2024). Recognizing and systematically investigating these individual differences is essential for advancing theory and designing support mechanisms that enhance creator well-being and career sustainability in the platform economy.
儘管創作者普遍依賴數位平台，但其經驗因平台依賴程度的不同而大相逕庭——即收入、專業身份及職業機會在多大程度上依賴單一平台（Guiñez-Cabrera & Aqueveque, 2021；Cutolo & Kenney, 2021）。高度依賴的平台創作者，如全職網紅或僅依賴 YouTube 或 TikTok 等平台的直播主，特別容易受到突如其來的演算法變動、政策調整或貨幣化中斷的影響（Lee & Jin, 2024；Are & Briggs, 2023）。即使是輕微的互動率下降，也可能引發急性焦慮、風險規避及身份威脅，因為他們的生計與社會資本緊密綁定於平台的穩定性（Guo et al., 2022）。相較之下，多元化收入來源或在多個頻道培養觀眾的創作者，往往展現出較強的韌性，將負面訊號視為挑戰而非生存威脅（Appio & Corso, 2023；Ye et al., 2024）。然而，過去多數研究將創作者視為同質群體，忽略了其自主性、風險容忍度及適應策略上的關鍵差異（Verwiebe et al., 2024）。 認識並系統性地調查這些個體差異，對於推進理論發展及設計提升創作者福祉與職業永續性的支持機制，在平台經濟中至關重要。

The Current Study
本研究現況

Despite growing attention to platform labor, key gaps remain in understanding how creators actively interpret algorithmic signals and how these processes affect risk perception, trust, and career decisions (Choi et al., 2023; Verwiebe et al., 2024). Prior work has primarily focused on downstream outcomes, such as audience engagement or branding, while underexploring upstream cognitive and emotional mechanisms. Moreover, the dominance of cross-sectional and correlational research has limited insight into causal mechanisms or the interplay between signal framing, transparency, and individual differences.
儘管平台勞動日益受到關注，對於創作者如何主動解讀演算法訊號，以及這些過程如何影響風險感知、信任與職業決策，仍存在重要認知缺口（Choi et al., 2023；Verwiebe et al., 2024）。先前研究主要聚焦於下游結果，如觀眾互動或品牌建構，卻較少探討上游的認知與情感機制。此外，橫斷面與相關性研究的主導地位限制了對因果機制或訊號框架、透明度與個體差異間交互作用的深入理解。

To address these issues, the present study employs a multi-study experimental approach to investigate the psychological mechanisms linking platform signal design to creators’ risk perception, trust, and entrepreneurial intention. Guided by signaling theory, framing, and risk perception frameworks, we examine how feedback valence (positive vs. negative framing), transparency, and signal modality (numeric, visual, narrative) interact to shape creators’ cognitive and behavioral responses. Platform dependency is treated as a key moderator.
為了解決這些問題，本研究採用多階段實驗方法，探討平台訊號設計與創作者風險感知、信任及創業意圖之間的心理機制。以訊號理論、框架理論及風險感知架構為指導，我們檢視回饋價值（正向與負向框架）、透明度及訊號模式（數字、視覺、敘事）如何交互作用，形塑創作者的認知與行為反應。平台依賴性被視為關鍵調節變項。

Study 1 examines the effects of risk-related signal framing, Study 2 explores algorithmic transparency, and Study 3 investigates presentation modality. Hypotheses are detailed below.
研究一探討風險相關訊號框架的影響，研究二探討演算法透明度，研究三則調查呈現模式。假說詳述如下。

Based on the above theoretical rationale and empirical objectives, the following hypotheses are proposed:
基於上述理論基礎與實證目標，提出以下假說：

H1: Negative feedback framing will increase risk perception, thereby decreasing entrepreneurial intention.
H1：負向回饋框架將提升風險感知，進而降低創業意圖。

H2: Greater transparency will enhance platform trust and decision confidence, thereby increasing entrepreneurial intention.
H2：更高的透明度將提升平台信任與決策信心，從而增加創業意圖。

H3: Signal modality (visual or narrative vs. numeric) will reduce cognitive load and increase decision confidence, promoting entrepreneurial intention.
H3：訊號模式（視覺或敘事與數字）將降低認知負荷並提升決策信心，促進創業意圖。

H4: Platform dependency will moderate the effects of signal design:
H4：平台依賴性將調節訊號設計的效果：

H4a: Platform dependency moderates the effect of feedback framing on risk perception.
H4a：平台依賴性調節回饋框架對風險感知的影響。

H4b: Platform dependency moderates the effect of transparency on platform trust and decision confidence.
H4b：平台依賴性調節透明度對平台信任與決策信心的影響。

H4c: Platform dependency moderates the effect of signal modality on cognitive load.
H4c：平台依賴性調節訊號模態對認知負荷的影響。

By integrating theory-driven experimental manipulations with ecologically valid platform simulations, this research advances both theoretical and practical understanding of digital labor. The findings aim to elucidate the psychosocial mechanisms underlying adaptive and maladaptive responses to algorithmic feedback, inform the design of psychologically supportive feedback systems, and offer actionable insights for platform governance, policy, and creator support initiatives.
透過結合理論驅動的實驗操作與具生態效度的平台模擬，本研究推進了數位勞動的理論與實務理解。研究結果旨在闡明適應性與非適應性回應算法反饋的心理社會機制，指導心理支持性反饋系統的設計，並為平台治理、政策制定及創作者支持計畫提供可行的見解。

Methods
方法

Study 1: Framing Effects of Algorithmic Risk Signals
研究一：演算法風險訊號的框架效應

This study examines how the framing of algorithmic risk signals influences creators’ perceptions and intentions. Based on cognitive framing and signaling theories, we posit that risk information—when framed negatively versus positively—triggers different psychological responses. We further explore how platform dependency moderates these effects, as creators more reliant on platforms may react more strongly to threatening signals.
本研究探討演算法風險訊號的框架如何影響創作者的感知與意圖。基於認知框架理論與訊號理論，我們假設風險資訊在負向與正向框架下會引發不同的心理反應。我們進一步探討平台依賴性如何調節這些效應，因為對平台依賴較高的創作者可能對威脅性訊號反應更強烈。

By manipulating signal framing and measuring creators’ dependency levels, this study aims to uncover how feedback interpretation affects risk perception, trust, and entrepreneurial intention in algorithmic environments.
透過操控訊號框架並測量創作者的依賴程度，本研究旨在揭示反饋解讀如何影響演算法環境中的風險感知、信任及創業意圖。

Research Design
研究設計

We employed a 2 (signal framing: negative vs. positive) × 2 (platform dependency: high vs. low) between-subjects design. Participants received a scenario in which their content performance declined. In the negative frame, the decline was emphasized as a threat (e.g., demonetization risk); in the positive frame, the same data was presented as manageable fluctuation with growth potential.
本研究採用 2（訊號框架：負向 vs. 正向）× 2（平台依賴：高 vs. 低）之受試者間設計。參與者接收到一個內容表現下滑的情境。在負向框架中，該下滑被強調為威脅（例如，貨幣化風險）；在正向框架中，則將相同數據呈現為可控波動且具成長潛力。

Platform dependency was measured via a validated scale adapted from prior digital labor research (Alasoini et al., 2023).
平台依賴性透過一個經過驗證的量表進行衡量，該量表改編自先前的數位勞動研究（Alasoini et al., 2023）。

We hypothesize that negative framing increases perceived risk and lowers entrepreneurial intention (H1).
我們假設負面框架會增加風險感知並降低創業意圖（H1）。

Experimental Procedure and Materials
實驗程序與材料

Participants were recruited from university students and early-stage creators familiar with digital platforms. After providing consent, they completed a brief background survey, including demographics, prior experience, and platform dependency, using a standardized scale adapted from creator economy studies (Bhargava, 2022; Edeling & Wies, 2024).
參與者招募自大學生及熟悉數位平台的初期創作者。經同意後，他們完成了一份簡短的背景調查，包括人口統計資料、先前經驗及平台依賴度，調查使用了改編自創作者經濟研究的標準化量表（Bhargava, 2022；Edeling & Wies, 2024）。

Participants were randomly assigned to one of two framing conditions. In the negative frame, the simulated platform dashboard notified creators of a “significant decline” in content performance, warning of potential audience loss and demonetization. In the positive frame, the same performance data was reframed as normal fluctuation with growth opportunities.
參與者隨機分配至兩種框架條件之一。在負面框架中，模擬平台儀表板通知創作者其內容表現「顯著下滑」，並警告可能流失觀眾及失去變現機會。正面框架則將相同的表現數據重新詮釋為正常波動並伴隨成長機會。

Each scenario used realistic interface elements—charts, widgets, pop-ups—mimicking platforms like YouTube or Instagram. This enhanced ecological validity and immersion.
每個情境均使用真實介面元素——圖表、小工具、彈出視窗——模擬如 YouTube 或 Instagram 等平台，提升生態效度與沉浸感。

After viewing the simulated dashboard, participants completed validated scales measuring perceived risk, emotional response (e.g., anxiety, optimism), and entrepreneurial intention. Manipulation checks ensured participants perceived the intended framing.
觀看模擬儀表板後，參與者完成經驗證的量表，測量感知風險、情緒反應（如焦慮、樂觀）及創業意圖。操控檢查確保參與者感知到預期的框架。

The entire procedure took approximately 20 minutes. Pilot testing with a small creator sample confirmed clarity and effective framing manipulation. Data from inattentive participants or failed checks were excluded from analysis.
整個程序約需 20 分鐘。以少量創作者樣本進行的先導測試確認清晰度及框架操控的有效性。對不專心的參與者或未通過檢查者之數據予以排除分析。

Measures and Analytical Approach
測量與分析方法

This study measured four key constructs—perceived risk, entrepreneurial intention, emotional response, and platform dependency—using validated multi-item scales adapted for digital content creators. Perceived risk was assessed through items capturing participants’ sense of uncertainty, vulnerability, and anticipated negative consequences associated with algorithmic changes. These items reflected subjective appraisals of platform-related threats and were adapted from established risk perception measures in digital labor research (Beese et al., 2021). Entrepreneurial intention was evaluated using a combination of Likert-type items and scenario-based prompts, focusing on participants’ likelihood of continuing content creation, willingness to allocate additional resources, and commitment to platform-based career development. This operationalization followed previous approaches in entrepreneurial motivation studies (Edeling & Wies, 2024).
本研究採用經驗證的多項目量表，針對數位內容創作者調整，衡量四個關鍵構念——感知風險、創業意圖、情緒反應及平台依賴性。感知風險透過捕捉參與者對演算法變動所帶來的不確定性、脆弱感及預期負面後果的項目進行評估。這些項目反映了對平台相關威脅的主觀評價，並改編自數位勞動研究中既有的風險感知量表（Beese et al., 2021）。創業意圖則結合李克特量表項目與情境式提示進行評估，聚焦於參與者持續創作內容的可能性、願意投入額外資源的意願，以及對平台基礎職涯發展的承諾。此操作化方式遵循先前創業動機研究的方法（Edeling & Wies, 2024）。

Emotional responses—specifically anxiety and optimism—were measured using standardized affective scales that distinguish between high-arousal negative emotions (e.g., worry, fear) and positive states (e.g., hopefulness, confidence). These measures aimed to capture participants’ immediate affective reactions to the feedback manipulation (Shin et al., 2024). Platform dependency was assessed during the pre-experimental phase using items reflecting reliance on the platform for financial stability, professional identity, and daily structure. A composite dependency score was computed to categorize participants into high and low dependency groups, following procedures in recent research on algorithmic labor and creator economies (Alasoini et al., 2023).
情緒反應——特別是焦慮與樂觀——透過標準化的情感量表進行測量，該量表區分高喚起的負面情緒（例如，擔憂、恐懼）與正向狀態（例如，希望感、自信）。這些測量旨在捕捉參與者對反饋操作的即時情感反應（Shin et al., 2024）。平台依賴性則在實驗前階段透過反映對平台在財務穩定、專業身份及日常結構依賴的項目進行評估。依據近期關於演算法勞動與創作者經濟的研究程序（Alasoini et al., 2023），計算綜合依賴分數以將參與者分類為高依賴與低依賴群組。

Data analysis began with checks for normality, scale reliability, and manipulation validity. Internal consistency for each construct was examined using Cronbach’s alpha. Manipulation checks confirmed that participants interpreted the feedback framing as intended. Primary hypotheses were tested using 2 (framing: positive vs. negative) × 2 (platform dependency: high vs. low) factorial ANOVAs, with perceived risk, emotional response, and entrepreneurial intention as dependent variables. Where significant interaction effects were found, simple slope analyses and post hoc comparisons were conducted to interpret the conditional effects.
資料分析首先進行常態性檢驗、量表信度及操作效度檢查。各構念的內部一致性使用 Cronbach’s α係數進行檢驗。操作檢查確認參與者對回饋框架的詮釋符合預期。主要假設透過 2（框架：正向 vs. 負向）× 2（平台依賴性：高 vs. 低）因子變異數分析（ANOVA）進行檢驗，依變項包括風險知覺、情緒反應及創業意圖。當發現顯著交互作用時，進一步進行簡單斜率分析及事後比較以解釋條件效應。

All models included demographic covariates such as age, gender, and prior content creation experience to control for potential confounds. In addition to statistical significance (p-values), effect sizes (η²) and 95% confidence intervals were reported to assess the magnitude and practical relevance of observed effects. This analytical approach was selected to balance statistical rigor, reproducibility, and interpretability in uncovering how creators respond to algorithmic signals under varying informational conditions.
所有模型皆納入年齡、性別及先前內容創作經驗等人口統計共變數，以控制潛在混淆變項。除統計顯著性（p 值）外，亦報告效應量（η²）及 95%信賴區間，以評估觀察效應的大小及實務意義。此分析方法旨在平衡統計嚴謹性、重現性及解釋性，揭示創作者在不同資訊條件下對演算法訊號的反應。

Analyses included 2 (framing: negative/positive) × 2 (platform dependency: high/low, median split) factorial ANOVAs, controlling for age, gender, and platform experience. Reliability and manipulation check results are reported in the Results section.
分析採用 2（框架：負向／正向）× 2（平台依賴性：高／低，中位數分割）因子變異數分析，並控制年齡、性別及平台經驗。信度與操作檢查結果詳見結果部分。

Study 2: Platform Transparency and Trust
研究二：平台透明度與信任

Building upon the framing effects examined in Study 1, Study 2 shifts focus to the role of transparency in shaping trust and decision-making. In platform-based careers, opaque communication has been linked to greater uncertainty and anxiety, while transparency fosters trust and retention. Yet, most prior research has centered on consumers or employed correlational methods, leaving a gap in experimental evidence concerning creators.
在第一項研究所探討的框架效應基礎上，第二項研究轉而聚焦於透明度在塑造信任與決策中的角色。在平台型職業中，不透明的溝通常與較高的不確定性和焦慮感相關，而透明度則促進信任與留存。然而，過去大多數研究集中於消費者或採用相關性方法，缺乏針對創作者的實驗性證據。

This study experimentally manipulates the level of platform transparency and examines its causal effects on platform trust, perceived risk, and entrepreneurial intention. Special attention is given to creators' platform dependency as a potential moderator.
本研究透過實驗操控平台透明度的程度，檢視其對平台信任、感知風險及創業意圖的因果影響。特別關注創作者對平台依賴性作為潛在的調節變項。

Research Design and Hypotheses
研究設計與假說

Study 2 employs a 2 (transparency: high vs. low) × 2 (platform dependency: high vs. low) between-subjects design. Participants are randomly assigned to one of two transparency conditions. The high-transparency condition includes detailed algorithmic explanations, predictive analytics, and clear rationales for visibility changes. The low-transparency condition features vague, minimal feedback without context.
研究二採用 2（透明度：高 vs. 低）× 2（平台依賴性：高 vs. 低）之受試者間設計。參與者隨機分配至兩種透明度條件之一。高透明度條件包含詳細的演算法說明、預測分析及清楚的能見度變動理由。低透明度條件則提供模糊且最小的回饋，且缺乏背景說明。

We hypothesize that high transparency will enhance trust, reduce risk perception, and boost entrepreneurial intention (H2). Measures are adapted from Study 1, with additional items assessing perceived transparency and trust dimensions (fairness, competence, care).
我們假設高透明度將提升信任、降低風險感知，並促進創業意圖（H2）。衡量指標改編自第一項研究，並新增評估感知透明度及信任維度（公平性、能力、關懷）的項目。

This design allows for testing both main effects and interaction effects, extending the experimental inquiry into how platforms can reduce uncertainty and support creators through transparent feedback mechanisms.
此設計允許檢驗主效應與交互效應，進一步擴展實驗探討平台如何透過透明的反饋機制減少不確定性並支持創作者。

Experimental Procedure and Materials
實驗程序與材料

Participants were recruited from the same pool as in Study 1—students and early-career digital creators familiar with platform analytics. After completing a baseline survey measuring demographics, platform experience, and platform dependency, participants were randomly assigned to one of two conditions: high or low transparency.
參與者招募自與研究一相同的族群——熟悉平台分析的學生及早期數位創作者。完成基線調查以測量人口統計資料、平台經驗及平台依賴度後，參與者隨機分配至高透明度或低透明度兩種條件之一。

In the high-transparency condition, the simulated platform notification included a detailed explanation of algorithmic factors affecting recent content visibility (e.g., engagement thresholds, audience behavior shifts), along with predictive analytics indicating projected performance. The message emphasized fairness and provided step-by-step rationales. In the low-transparency condition, the notification simply stated that content visibility had changed, offering no reasons, data, or supporting explanation. Both conditions featured consistent visual design based on real-world platforms (e.g., YouTube Studio), ensuring differences in response were attributable to message content, not aesthetics.
在高透明度條件下，模擬平台通知包含了影響近期內容可見性的演算法因素詳細說明（例如，互動門檻、觀眾行為變化），並附有預測分析以顯示預期表現。該訊息強調公平性，並提供逐步的理據說明。在低透明度條件下，通知僅簡單說明內容可見性已變更，未提供任何原因、數據或支持性解釋。兩種條件均採用基於真實平台（如 YouTube Studio）的一致視覺設計，確保反應差異歸因於訊息內容而非美學。

After exposure to the notification, participants completed post-test measures on platform trust, perceived risk, entrepreneurial intention, and perceived transparency (used as a manipulation check). All items used 7-point Likert scales. Attention checks were embedded to ensure data quality.
接觸通知後，參與者完成了平台信任、風險感知、創業意圖及感知透明度（用作操作檢查）的後測量表。所有項目均採用 7 點李克特量表。研究中嵌入注意力檢查以確保資料品質。

To enhance ecological validity, the simulated interfaces were reviewed by platform communication experts and pilot-tested with a subset of actual creators. The full session, including baseline, scenario exposure, and post-measures, took approximately 20–25 minutes per participant. Only those passing attention and manipulation checks were included in the final dataset.
為提升生態效度，模擬介面經由平台溝通專家審查，並在部分實際創作者中進行先導測試。整個實驗流程，包括基線測量、情境暴露及後測，約需每位參與者 20 至 25 分鐘。最終資料集中僅納入通過注意力及操作檢核者。

Measures and Analytical Approach
測量與分析方法

Study 2 employed multi-item, 7-point Likert scales to assess four core constructs: platform trust, perceived risk, entrepreneurial intention, and perceived transparency (as a manipulation check). Platform trust was measured using items adapted from digital trust and service fairness literature (Cheng et al., 2024), capturing participants’ perceptions of the platform’s competence, fairness, predictability, and concern for user interests. Perceived risk and entrepreneurial intention were measured using the same validated scales from Study 1 (Beese et al., 2021; Edeling & Wies, 2024), reflecting creators’ sense of uncertainty and commitment to ongoing platform engagement. Perceived transparency was assessed by asking participants to rate the clarity, informativeness, and understandability of the platform’s feedback message (Abitbol et al., 2022).
研究二採用多項目 7 點李克特量表評估四個核心構念：平台信任、感知風險、創業意圖及感知透明度（作為操作檢核）。平台信任的測量項目改編自數位信任與服務公平性文獻（Cheng et al., 2024），涵蓋參與者對平台能力、公平性、可預測性及關注用戶利益的感知。感知風險與創業意圖則使用研究一中經驗驗證的量表（Beese et al., 2021；Edeling & Wies, 2024），反映創作者對不確定性的感受及持續參與平台的承諾。感知透明度透過請參與者評分平台回饋訊息的清晰度、資訊豐富度及易理解性來衡量（Abitbol et al., 2022）。

To evaluate the experimental effects, a 2 (transparency: high vs. low) × 2 (platform dependency: high vs. low) factorial ANOVA was conducted for each outcome variable. Main effects and interaction terms were examined, with effect sizes (η²) reported. Simple effects and post hoc comparisons were conducted when interaction effects were significant.
為評估實驗效應，針對每個結果變項進行 2（透明度：高 vs. 低）× 2（平台依賴性：高 vs. 低）因子變異數分析（ANOVA）。檢視主效應及交互作用項，並報告效應量（η²）。當交互作用顯著時，進行簡單效應分析及事後比較。

In addition, mediation analysis was performed to examine whether perceived transparency and platform trust served as mediators between the transparency manipulation and entrepreneurial intention. This followed established bootstrapping procedures for testing indirect effects. All statistical models controlled for demographic covariates, and the significance threshold was set at p < 0.05.
此外，進行中介分析以檢驗感知透明度與平台信任是否作為透明度操作與創業意圖之間的中介變項。此分析遵循既定的自助法（bootstrapping）程序以檢測間接效應。所有統計模型均控制人口統計學協變數，顯著水準設定為 p < 0.05。

Study 3: Signal Presentation and Decision Confidence
研究三：訊號呈現與決策信心

Building on the findings from Study 1 and Study 2—which respectively demonstrated the importance of signal framing and algorithmic transparency in shaping creators’ risk perceptions and intentions—Study 3 turns to a third critical dimension of platform feedback: modality. In digital platforms, the way feedback is presented—whether as raw numbers, visual dashboards, or narrative summaries—may further influence how creators process information, assess risk, and make career decisions. Prior research in cognitive psychology and human-computer interaction shows that information format affects comprehension, cognitive load, and judgment quality (Damman et al., 2016; Key-DeLyria et al., 2019). While creators often encounter diverse feedback types (raw numbers, charts, summaries), their effects on risk perception, trust, and entrepreneurial intention are underexplored, particularly in relation to platform dependency and digital experience.
基於研究一與研究二的發現——分別證明了訊號框架與演算法透明度在塑造創作者風險感知與意圖上的重要性——研究三轉向平台回饋的第三個關鍵維度：呈現方式。在數位平台中，回饋的呈現方式——無論是原始數字、視覺儀表板，或是敘述性摘要——可能進一步影響創作者如何處理資訊、評估風險及做出職涯決策。認知心理學與人機互動領域的先前研究顯示，資訊格式會影響理解力、認知負荷與判斷品質（Damman et al., 2016；Key-DeLyria et al., 2019）。儘管創作者經常接觸多樣的回饋類型（原始數字、圖表、摘要），其對風險感知、信任及創業意圖的影響仍未被充分探討，尤其是在平台依賴性與數位經驗的關聯上。

Study 3 addresses this gap by experimentally manipulating feedback modality and measuring its impact on decision confidence, cognitive load, and behavioral intention. By comparing numeric, visual, and narrative formats, the study clarifies how different presentation styles influence effective decision-making and offers insights for both platform designers and creator education in data-rich contexts.
研究三透過實驗操控回饋的呈現方式，並測量其對決策信心、認知負荷及行為意圖的影響，以填補此一研究空白。透過比較數字、視覺與敘事三種格式，本研究釐清不同呈現風格如何影響有效決策，並為平台設計者及創作者在數據豐富環境中的教育提供洞見。

Research Design and Hypotheses
研究設計與假說

Study 3 uses a between-subjects design: participants are randomly assigned to one of three feedback modalities (numeric, visual, narrative), all presenting identical performance data but in different formats. The hypothesis is that modality will significantly affect cognitive and behavioral outcomes. Visual and narrative feedback are expected to enhance comprehension and confidence, while numeric formats may increase cognitive load and risk of misinterpretation (H3) (Chen et al., 2020; Metoyer et al., 2018). It is further hypothesized that cognitive load mediates the relationship between signal format and entrepreneurial intention.
研究 3 採用受試者間設計：參與者被隨機分配至三種回饋模式之一（數字、視覺、敘事），所有模式均呈現相同的績效數據，但格式不同。假設回饋模式將顯著影響認知及行為結果。預期視覺與敘事回饋能提升理解力與信心，而數字格式可能增加認知負荷及誤解風險（H3）（Chen et al., 2020；Metoyer et al., 2018）。進一步假設認知負荷在訊號格式與創業意圖之間具有中介作用。

Validated measures for cognitive load, decision confidence, perceived risk, and entrepreneurial intention are used, along with manipulation checks for engagement and format perception. This design enables robust assessment of both direct and indirect effects of information modality on creator psychology and decision outcomes.
採用經驗證的認知負荷、決策信心、風險感知及創業意圖量表，並進行參與度與格式感知的操控檢核。此設計能夠穩健評估資訊模式對創作者心理與決策結果的直接及間接影響。

Experimental Procedure and Materials
實驗程序與材料

Participants were drawn from the same university and digital creator pools as previous studies, ensuring consistent familiarity with online platforms and analytics. Upon consent, individuals were randomly assigned to receive feedback in one of three modalities: numeric, visual, or narrative.
參與者來自與先前研究相同的大學及數位創作者群，確保對線上平台及分析工具具有一致的熟悉度。經同意後，個體被隨機分配接受三種回饋模式之一：數字、視覺或敘事。

All participants completed a brief pre-test survey covering demographics, digital literacy, and prior experience with analytics tools, which were later used as covariates and to verify balanced group assignment. In the numeric condition, participants viewed a simulated dashboard with tables of raw performance data—daily views, engagement, and subscriber growth—presented without interpretation. The visual condition displayed the same data using infographics, charts, and color-coded dashboards to highlight trends and anomalies, reflecting best practices in information visualization. The narrative condition provided a written summary interpreting quantitative results, identifying key patterns, and offering qualitative recommendations in an encouraging, accessible style.
所有參與者皆完成一份涵蓋人口統計、數位素養及先前分析工具使用經驗的簡短前測問卷，該資料後續用作共變數並驗證組別分配的平衡性。在數字條件下，參與者觀看模擬儀表板，呈現未經解釋的原始績效數據表格——每日瀏覽量、互動率及訂閱者成長。視覺條件則以資訊圖表、圖表及色彩編碼的儀表板展示相同數據，強調趨勢與異常，反映資訊視覺化的最佳實務。敘事條件則提供書面摘要，詮釋量化結果，識別關鍵模式，並以鼓勵且易於理解的風格提出質性建議。

After exposure to the assigned feedback, participants completed validated scales measuring decision confidence, cognitive load (e.g., NASA-TLX), perceived risk, and entrepreneurial intention. Manipulation checks assessed engagement with and differentiation among modalities.
在接收到指定的回饋後，參與者完成了經驗證的量表，測量決策信心、認知負荷（例如 NASA-TLX）、感知風險及創業意向。操作檢查則評估了對不同模式的參與度及區辨能力。

All materials and interface elements were pilot tested with a subset of creators and reviewed by platform professionals to ensure ecological validity. The full procedure, including pre-survey, scenario exposure, and post-survey, was completed within 25–30 minutes per participant. Materials, including example dashboards and survey items, are documented in Appendix 3 for replication.
所有材料與介面元素均先由部分創作者進行試點測試，並由平台專業人士審核，以確保生態效度。整個程序，包括前測調查、情境暴露及後測調查，每位參與者約需 25 至 30 分鐘完成。材料內容，包括範例儀表板及調查項目，詳載於附錄三以供複製使用。

Measures and Analytical Approach
測量與分析方法

To capture the cognitive and psychological impact of feedback modality, Study 3 employed several validated multi-item scales. Decision confidence was assessed using items adapted from prior research on self-efficacy and decision assurance in digital environments (Lisi et al., 2021). Cognitive load was measured via a short-form NASA Task Load Index (NASA-TLX), covering mental demand, effort, frustration, and complexity, with higher scores indicating greater processing burden (Cao et al., 2009). Perceived risk and entrepreneurial intention were evaluated using the same instruments as in prior studies, ensuring consistency across the research phases (Beese et al., 2021; Edeling & Wies, 2024).
為捕捉回饋模式對認知與心理的影響，研究三採用了多項經驗驗證的多題量表。決策信心透過改編自先前數位環境中自我效能與決策保證研究的題項進行評估（Lisi et al., 2021）。認知負荷則使用簡短版 NASA 任務負荷指數（NASA-TLX）測量，涵蓋心理需求、努力程度、挫折感與複雜度，分數越高代表處理負擔越大（Cao et al., 2009）。感知風險與創業意圖則採用與先前研究相同的量表，確保各研究階段間的一致性（Beese et al., 2021；Edeling & Wies, 2024）。

Manipulation checks required participants to rate the clarity, format, and informativeness of the feedback, verifying the perceived differences among the three modalities.
操作檢核要求參與者評分回饋的清晰度、格式與資訊量，以驗證三種模式間的感知差異。

Analyses began with manipulation checks and descriptive statistics. One-way ANOVA was used to test the effect of modality (numeric, visual, narrative) on all main outcomes. When significant effects were found, Tukey’s HSD post hoc tests examined pairwise group differences. To explore whether cognitive load mediated the link between feedback format and entrepreneurial intention, mediation analysis was performed using the PROCESS macro (Hayes, 2013). Covariates such as digital literacy, prior analytic experience, and baseline risk tolerance were included to control for confounds.
分析首先從操作檢查和描述性統計開始。採用單因子變異數分析（ANOVA）檢驗模態（數字、視覺、敘事）對所有主要結果的影響。當發現顯著效應時，使用 Tukey 的 HSD 事後檢定檢視組間兩兩差異。為探討認知負荷是否中介回饋格式與創業意圖之間的關係，採用 PROCESS 巨集（Hayes, 2013）進行中介分析。並納入數位素養、先前分析經驗及基線風險容忍度等共變數以控制混淆因素。

Effect sizes (partial η² for ANOVA, standardized indirect effects for mediation) and confidence intervals were reported, following best practices for experimental research (Chen & Peng, 2015; Dudgeon, 2016). This comprehensive strategy allowed rigorous testing of both direct and indirect hypotheses concerning the psychological consequences of signal presentation modality.
報告效應量（ANOVA 採用部分η²，中介分析採用標準化間接效應）及信賴區間，遵循實驗研究最佳實務（Chen & Peng, 2015；Dudgeon, 2016）。此綜合策略允許嚴謹檢驗訊號呈現模態對心理影響的直接及間接假說。

Results
結果

Study 1: Framing Effects of Algorithmic Risk Signals
研究一：演算法風險訊號的框架效應

A total of 204 participants completed Study 1, with equal numbers randomly assigned to the negative framing (n = 102) and positive framing (n = 102) conditions. Participants were further classified as highly dependent (n = 104) or less dependent (n = 100) on platforms, based on a median split of dependency scores. Age, gender, and prior content creation experience were comparable across all groups. All key measures demonstrated acceptable internal consistency (Cronbach’s alpha: perceived risk = 0.86, entrepreneurial intention = 0.81, anxiety = 0.79, optimism = 0.82).
共有 204 名參與者完成了研究一，並隨機平均分配至負向框架組（n = 102）與正向框架組（n = 102）。參與者依據依賴度分數的中位數分割，進一步分類為高度依賴組（n = 104）或低度依賴組（n = 100）。各組在年齡、性別及先前內容創作經驗方面均無顯著差異。所有主要測量指標皆展現出可接受的內部一致性（Cronbach’s α：風險知覺 = 0.86，創業意圖 = 0.81，焦慮 = 0.79，樂觀 = 0.82）。

Reliability analyses demonstrated satisfactory internal consistency for all scales: perceived risk (α = 0.86), entrepreneurial intention (α = 0.81), anxiety (α = 0.79), and optimism (α = 0.82). Manipulation checks confirmed that the framing conditions were perceived as intended. Participants in the negative framing condition rated the feedback as significantly more threatening (M = 5.9, SD = 1.1) compared to those in the positive framing condition (M = 3.4, SD = 1.0), t(202) = 17.12, p < 0.001.
信度分析顯示所有量表皆具令人滿意的內部一致性：風險知覺（α = 0.86）、創業意圖（α = 0.81）、焦慮（α = 0.79）及樂觀（α = 0.82）。操作檢核確認框架條件如預期被感知。負向框架條件的參與者對回饋的威脅感評分顯著高於正向框架條件（負向：M = 5.9，SD = 1.1；正向：M = 3.4，SD = 1.0），t(202) = 17.12，p < 0.001。

A series of 2 (framing: negative vs. positive) × 2 (platform dependency: high vs. low) factorial ANOVAs were conducted for each dependent variable. For perceived risk, both the main effect of framing (F(1, 200) = 34.75, p < 0.001, η² = 0.15) and the main effect of platform dependency (F(1, 200) = 8.94, p = 0.003, η² = 0.04) were significant. Participants in the negative framing condition reported higher perceived risk (M = 4.82) than those in the positive framing condition (M = 3.58), and highly dependent creators perceived more risk overall.
針對每個依變項進行了一系列 2（框架：負向 vs. 正向）× 2（平台依賴性：高 vs. 低）因子變異數分析。對於風險知覺，框架的主效應（F(1, 200) = 34.75, p < 0.001, η² = 0.15）及平台依賴性的主效應（F(1, 200) = 8.94, p = 0.003, η² = 0.04）皆達顯著。負向框架組的參與者報告的風險知覺較高（M = 4.82），相較於正向框架組（M = 3.58），且高度依賴平台的創作者整體感知風險較高。

Importantly, the interaction effect was also significant (F(1, 200) = 7.19, p = 0.008, η² = 0.03). Among highly dependent participants, negative framing resulted in substantially higher risk perception (M = 5.28) compared to positive framing (M = 3.61), t(102) = 8.13, p < 0.001. For less dependent participants, the difference was less pronounced (negative: M = 4.35; positive: M = 3.56), t(98) = 3.42, p = 0.001. This interaction is visualized in Figure 1, which illustrates the magnified risk effect under negative framing for highly dependent creators. Table 1 presents the means and standard deviations for perceived risk and entrepreneurial intention across all experimental groups (negative/positive framing × high/low platform dependency).
值得注意的是，交互作用亦達顯著水準（F(1, 200) = 7.19，p = 0.008，η² = 0.03）。在高度依賴平台的參與者中，負面框架導致風險感知顯著較高（M = 5.28），相較於正面框架（M = 3.61），t(102) = 8.13，p < 0.001。對於依賴度較低的參與者，差異較不明顯（負面：M = 4.35；正面：M = 3.56），t(98) = 3.42，p = 0.001。此交互作用如圖 1 所示，該圖展示了在負面框架下，高度依賴平台的創作者風險感知的放大效應。表 1 則呈現所有實驗組（負面/正面框架 × 高/低平台依賴）中風險感知與創業意向的平均值與標準差。

For entrepreneurial intention, a significant main effect of framing emerged (F(1, 200) = 21.16, p < 0.001, η² = 0.10), with participants in the positive framing condition reporting greater entrepreneurial intention (M = 5.17) than those in the negative condition (M = 4.22). A significant interaction effect (F(1, 200) = 5.64, p = 0.018, η² = 0.03) indicated that the reduction in entrepreneurial intention under negative framing was larger for highly dependent creators (ΔM = 1.12) than for less dependent ones (ΔM = 0.65).
在創業意圖方面，框架效應呈現顯著主效應（F(1, 200) = 21.16，p < 0.001，η² = 0.10），正向框架條件下的參與者報告的創業意圖較高（M = 5.17），而負向條件下則較低（M = 4.22）。顯著的交互作用效應（F(1, 200) = 5.64，p = 0.018，η² = 0.03）顯示，負向框架下創業意圖的降低在高度依賴的創作者中較大（ΔM = 1.12），而在依賴較低的創作者中則較小（ΔM = 0.65）。

Emotional responses followed a consistent pattern. Negative framing increased anxiety (M = 4.88) relative to positive framing (M = 3.54), F(1, 200) = 28.13, p < 0.001. Conversely, optimism was significantly higher in the positive framing group (M = 5.02) than the negative group (M = 3.79), F(1, 200) = 24.07, p < 0.001.
情緒反應呈現一致的模式。與正向框架（M = 3.54）相比，負向框架增加了焦慮感（M = 4.88），F(1, 200) = 28.13，p < 0.001。相反地，樂觀情緒在正向框架組（M = 5.02）顯著高於負向組（M = 3.79），F(1, 200) = 24.07，p < 0.001。

Control variables such as age, gender, and prior experience were not significantly associated with the outcome measures. Overall, these findings offer strong support for H1 and H4a: negatively framed signals elevate risk perception and reduce entrepreneurial motivation, especially among creators with high platform dependency, while positive framing exerts a buffering effect on psychological outcomes.
控制變項如年齡、性別及先前經驗與結果變項間未呈現顯著關聯。整體而言，這些發現強力支持假說 H1 與 H4a：負面框架訊號提升風險感知並降低創業動機，尤其在高度依賴平台的創作者中更為明顯，而正面框架則對心理結果具有緩衝效果。

(See Appendix 1 for full text of scenario manipulations and survey items.)
（情境操作及問卷題目全文請見附錄 1。）

Study 2: Platform Transparency and Trust
研究二：平台透明度與信任

In Study 2, 208 participants took part, distributed between high-transparency (n = 106) and low-transparency (n = 102) feedback conditions. Using pre-survey measures, 110 individuals were identified as highly platform-dependent and 98 as less dependent. Demographic characteristics and prior platform experience were balanced across conditions, ensuring group equivalence. All measures demonstrated strong internal consistency (Cronbach’s α: trust = 0.89; risk = 0.85; intention = 0.83; transparency = 0.88). The dataset was screened for missing values, outliers, and failed attention checks to ensure robust data quality.
在研究二中，共有 208 名參與者參與，分布於高透明度（n = 106）與低透明度（n = 102）回饋條件。透過前測調查量表，識別出 110 名高度依賴平台者與 98 名較少依賴者。各條件間的人口統計特徵及先前平台經驗均衡，確保組間等同性。所有量表皆展現出良好的內部一致性（Cronbach’s α：信任 = 0.89；風險 = 0.85；意圖 = 0.83；透明度 = 0.88）。資料集經過遺漏值、離群值及注意力檢查失敗的篩選，以確保資料品質穩健。

Manipulation checks confirmed the effectiveness of the transparency treatment. Participants in the high-transparency condition rated the feedback as significantly more clear and informative (M = 6.13, SD = 0.82) than those in the low-transparency group (M = 3.05, SD = 1.21), t(206) = 22.08, p < 0.001 (see Appendix 2 for full wording of manipulations and manipulation check items).
操作檢查確認透明度處理的有效性。高透明度條件的參與者對回饋的評價顯著較為清晰且具資訊性（平均數 M = 6.13，標準差 SD = 0.82），而低透明度組則為（平均數 M = 3.05，標準差 SD = 1.21），t(206) = 22.08，p < 0.001（完整操作說明及操作檢查項目請見附錄 2）。

Table 2 summarizes means and standard deviations for key outcome variables across experimental conditions. Factorial ANOVA revealed significant main effects of transparency on all three outcomes. Participants exposed to high transparency reported higher platform trust (M = 5.77, SD = 0.84) compared to low transparency (M = 4.51, SD = 1.04), F(1, 204) = 46.28, p < 0.001, η² = 0.19. They also reported lower perceived risk (M = 3.40 vs. 4.45), F(1, 204) = 32.10, p < 0.001, η² = 0.14, and stronger entrepreneurial intention (M = 5.28 vs. 4.34), F(1, 204) = 21.95, p < 0.001, η² = 0.10.
表 2 彙整了各實驗條件下主要結果變項的平均數與標準差。因子變異數分析顯示透明度對三項結果均有顯著主效應。接受高透明度資訊的參與者報告較高的平台信任度（平均數 M = 5.77，標準差 SD = 0.84），相較於低透明度組（M = 4.51，SD = 1.04），F(1, 204) = 46.28，p < 0.001，η² = 0.19。高透明度組同時報告較低的風險感知（M = 3.40 對比 4.45），F(1, 204) = 32.10，p < 0.001，η² = 0.14，以及較強的創業意圖（M = 5.28 對比 4.34），F(1, 204) = 21.95，p < 0.001，η² = 0.10。

Significant interaction effects between transparency and platform dependency emerged for both perceived risk and entrepreneurial intention. For perceived risk, F(1, 204) = 8.74, p = 0.004, η² = 0.04: highly dependent creators experienced markedly lower risk under high transparency (M = 3.68) than low transparency (M = 4.97), t(108) = 7.45, p < 0.001. Among less dependent creators, transparency also reduced perceived risk, but to a lesser extent (M = 3.12 vs. 3.94), t(96) = 4.13, p < 0.001 (see Figure 2 for interaction effect visualization).
透明度與平台依賴性之間對於風險知覺與創業意圖均呈現顯著交互作用。風險知覺方面，F(1, 204) = 8.74, p = 0.004, η² = 0.04：高度依賴平台的創作者在高透明度條件下感受到的風險顯著較低（M = 3.68）相較於低透明度（M = 4.97），t(108) = 7.45, p < 0.001。較少依賴平台的創作者中，透明度同樣降低了風險知覺，但幅度較小（M = 3.12 對 3.94），t(96) = 4.13, p < 0.001（交互作用效果視覺化請見圖 2）。

For entrepreneurial intention, the interaction was also significant, F(1, 204) = 6.22, p = 0.013, η² = 0.03. The positive effect of transparency was stronger among highly dependent participants (ΔM = 1.10) than those less dependent (ΔM = 0.78).
對於創業意圖，交互作用亦顯著，F(1, 204) = 6.22，p = 0.013，η² = 0.03。透明度的正向效應在高度依賴的參與者中較強（ΔM = 1.10），而在依賴較低者中則較弱（ΔM = 0.78）。

Mediation analysis further showed that the effects of transparency on entrepreneurial intention were partially mediated by increased platform trust and decreased perceived risk. Bootstrap estimates confirmed significant indirect effects (95% CI did not include zero), supporting H2 and H4b.
中介分析進一步顯示，透明度對創業意圖的影響部分透過提升平台信任及降低感知風險來中介。自助法估計確認了顯著的間接效應（95% 信賴區間不包含零），支持假設 H2 與 H4b。

Study 3: Signal Presentation and Decision Confidence
研究三：訊號呈現與決策信心

Study 3 included 210 participants, who were randomly allocated to receive feedback in one of three modalities: numeric, visual, or narrative. The distribution of digital literacy, analytic experience, and core demographic variables was consistent across groups, confirming successful randomization. All scales demonstrated satisfactory reliability (Cronbach’s alpha: decision confidence = 0.85, cognitive load = 0.81, perceived risk = 0.84, entrepreneurial intention = 0.82).
研究三包含 210 名參與者，隨機分配接受三種回饋模式之一：數字、視覺或敘事。數位素養、分析經驗及核心人口統計變項在各組間分布一致，確認隨機分配成功。所有量表皆展現令人滿意的信度（Cronbach’s α：決策信心 = 0.85，認知負荷 = 0.81，感知風險 = 0.84，創業意圖 = 0.82）。

Manipulation checks confirmed that participants accurately identified the assigned feedback format, and that modalities were perceived as distinct (mean differentiation rating = 6.48/7, SD = 0.54; see Appendix 3 for manipulation wording and materials).
操控檢查確認參與者能準確識別所分配的回饋格式，且各感官模式被感知為不同（平均區分評分 = 6.48/7，標準差 = 0.54；操控措辭與材料詳見附錄 3）。

Table 3 presents means and standard deviations for all outcome variables. One-way ANOVA revealed significant main effects of modality on decision confidence, cognitive load, perceived risk, and entrepreneurial intention. For decision confidence, the effect was robust (F(2, 207) = 29.24, p < 0.001, partial η² = 0.22); post hoc comparisons showed that both visual (M = 5.37, SD = 0.93) and narrative (M = 5.58, SD = 0.85) groups reported higher confidence than the numeric group (M = 4.12, SD = 0.88), both p < 0.001, with no difference between visual and narrative (p = 0.29).
表 3 呈現所有結果變項的平均值與標準差。單因子變異數分析顯示感官模式對決策信心、認知負荷、風險感知及創業意圖具有顯著主效應。決策信心方面，效應顯著且穩健（F(2, 207) = 29.24, p < 0.001, 部分η² = 0.22）；事後比較顯示視覺組（M = 5.37, SD = 0.93）與敘事組（M = 5.58, SD = 0.85）均報告出比數字組（M = 4.12, SD = 0.88）更高的信心，兩者皆 p < 0.001，視覺組與敘事組間無顯著差異（p = 0.29）。

For cognitive load, the effect was significant (F(2, 207) = 24.81, p < 0.001, partial η² = 0.19); numeric feedback imposed greater load (M = 5.02, SD = 1.01) than both visual (M = 3.72, SD = 0.96) and narrative (M = 3.51, SD = 1.03) conditions, both p < 0.001.
認知負荷方面，效應顯著（F(2, 207) = 24.81, p < 0.001, 部分η² = 0.19）；數字回饋組（M = 5.02, SD = 1.01）所承受的負荷顯著高於視覺組（M = 3.72, SD = 0.96）及敘事組（M = 3.51, SD = 1.03），兩者皆 p < 0.001。

Perceived risk also differed by modality (F(2, 207) = 13.69, p < 0.001, partial η² = 0.12); numeric group participants reported higher risk (M = 4.41, SD = 1.11) than visual (M = 3.60, SD = 1.02, p = 0.002) and narrative (M = 3.58, SD = 1.09, p = 0.001) groups.
感知風險亦因呈現方式而異（F(2, 207) = 13.69，p < 0.001，部分η² = 0.12）；數字組參與者報告的風險較高（平均值 = 4.41，標準差 = 1.11），高於視覺組（平均值 = 3.60，標準差 = 1.02，p = 0.002）及敘事組（平均值 = 3.58，標準差 = 1.09，p = 0.001）。

For entrepreneurial intention, the main effect was significant (F(2, 207) = 17.34, p < 0.001, partial η² = 0.14). Visual (M = 5.28, SD = 1.05) and narrative (M = 5.43, SD = 1.02) groups scored higher than numeric (M = 4.51, SD = 1.13), both p < 0.001.
在創業意向方面，主效應顯著（F(2, 207) = 17.34，p < 0.001，部分η² = 0.14）。視覺組（平均值 = 5.28，標準差 = 1.05）及敘事組（平均值 = 5.43，標準差 = 1.02）得分均高於數字組（平均值 = 4.51，標準差 = 1.13），兩者 p 值均小於 0.001。

Figure 3 illustrates cognitive load and decision confidence across modalities. Mediation analysis showed that cognitive load partially mediated the effect of modality on entrepreneurial intention (indirect effect = 0.38, 95% CI [0.16, 0.67]), confirming that formats which reduce cognitive burden foster greater intention to persist as a creator.
圖 3 說明了不同模式下的認知負荷與決策信心。中介分析顯示認知負荷部分中介了模式對創業意圖的影響（間接效果 = 0.38，95% 信賴區間 [0.16, 0.67]），確認減輕認知負擔的格式能促進創作者持續創作的意願。

To examine whether the effects of feedback modality varied according to creators’ platform dependency, additional moderation analyses were conducted (two-way ANOVA and regression with modality × platform dependency interaction terms). The interaction effects for cognitive load, decision confidence, and entrepreneurial intention were significant (all p < 0.00). Thus, the impact of signal modality appeared to be robust across levels of platform dependency. This finding is consistent with the significant moderation observed for framing and transparency in Studies 1 and 2.
為檢驗反饋模式的效果是否因創作者的平台依賴性而異，進行了額外的調節分析（雙因子變異數分析及包含模式 × 平台依賴性交互作用項的迴歸分析）。認知負荷、決策信心及創業意圖的交互作用效果均達顯著水準（全部 p < 0.00）。因此，訊號模式的影響在不同平台依賴性層級中均具穩健性。此結果與研究 1 及研究 2 中對框架與透明度的顯著調節效應相符。

These results strongly support H3, indicating that visual and narrative feedback enhance comprehension, reduce risk and cognitive load, and promote higher decision confidence and entrepreneurial motivation compared to numeric-only data. H4c was also supported, as platform dependency did not significantly moderate the effects of signal modality in this context.
這些結果強烈支持假設三，顯示視覺與敘事反饋相較於僅有數字資料，能增進理解、降低風險與認知負荷，並促進更高的決策信心與創業動機。假設四 c 亦獲支持，因為平台依賴性在此情境下並未顯著調節訊號模式的效果。

(See Appendix 3 for full dashboard and narrative materials.)
（完整儀表板與敘述材料請參見附錄三。）

Discussion
討論

The evidence from Study 1 indicates that not only does the framing of algorithmic feedback directly affect creators’ perceived risk and motivation, but these effects are amplified for those who are highly dependent on the platform. When negative frames were presented, creators experienced greater anxiety and diminished entrepreneurial intention (Cutolo & Kenney, 2021), whereas positive framing offered a protective, motivational buffer.
第一項研究的證據顯示，演算法反饋的框架不僅直接影響創作者的風險感知與動機，且這些效應在高度依賴平台的使用者中更為顯著。當呈現負面框架時，創作者經歷較大的焦慮與降低的創業意圖（Cutolo & Kenney, 2021），而正面框架則提供了一種保護性且具激勵作用的緩衝。

These findings highlight that algorithmic signals are not neutral indicators but psychologically potent messages whose interpretation depends on both content and context (Kruikemeier et al., 2021; Ostinelli et al., 2024). The framing of such feedback activates cognitive and emotional processes that can either reinforce or undermine creators’ engagement with platform-based work. In particular, the heightened sensitivity of highly dependent creators underscores the emotional costs of algorithmic opacity and volatility in digital labor markets.
這些發現強調，演算法訊號並非中立的指標，而是具有心理效力的訊息，其解讀取決於內容與情境（Kruikemeier et al., 2021；Ostinelli et al., 2024）。此類反饋的框架激活認知與情感過程，可能強化或削弱創作者對平台工作之投入。特別是，高度依賴平台的創作者敏感度提升，凸顯了數位勞動市場中演算法不透明性與波動性的情感成本。

Theoretically, this study supports an integrated model where signaling theory and cognitive framing jointly shape risk-related decision-making in platform environments. By showing that the same information, when differently framed, can lead to divergent outcomes—especially when moderated by individual dependency—this research adds nuance to the understanding of how creators navigate uncertainty. These insights contribute to the emerging literature on the psychosocial dynamics governing the socio-economic structures of modern cyberspace. In this context, creators are not merely subjects of algorithmic labor, but are reflexive agents navigating a new form of professional and economic existence (Jago & Carroll, 2023; Verwiebe et al., 2024).
理論上，本研究支持一個整合模型，該模型結合訊號理論與認知框架，共同形塑平台環境中的風險相關決策。透過展示相同資訊在不同框架下可能導致截然不同的結果——尤其在個人依賴性調節下——本研究為理解創作者如何應對不確定性增添了細緻的見解。這些洞察有助於新興文獻中對現代網路空間社會經濟結構心理社會動態的探討。在此脈絡中，創作者不僅是演算法勞動的主體，更是反思性行動者，駕馭一種全新的專業與經濟存在形式（Jago & Carroll, 2023；Verwiebe et al., 2024）。

Practically, the implications are significant for platform designers and policy-makers. Minor shifts in the language used in performance notifications can produce disproportionate effects on user confidence and retention. Platforms aiming to foster sustainable creator ecosystems—which are vital for the health of this new socio-economic structure in cyberspace—should consider implementing feedback systems that promote clarity and motivation. This is particularly critical for creators whose livelihood and identity are tightly linked to platform performance, as they are most vulnerable to the psychological toll of negatively framed signals.
實務上，這些發現對平台設計者與政策制定者具有重要意涵。績效通知中語言的細微變化，可能對使用者信心與留存率產生不成比例的影響。旨在培育永續創作者生態系統的平台——這對於網路空間中新興社會經濟結構的健康至關重要——應考慮實施促進清晰度與動機的回饋系統。此點對於生計與身份緊密依賴平台績效的創作者尤為關鍵，因為他們最容易受到負面框架訊號所帶來的心理負擔影響。

Finally, the results of Study 1 set the stage for further inquiry into other signal characteristics beyond framing. The next studies extend this line of investigation by examining how transparency (Study 2) and information modality (Study 3) further influence creator responses, aiming to provide a comprehensive understanding of how signal design can support healthier, more resilient creator-platform relationships.
最後，第一項研究的結果為進一步探討除框架外的其他訊號特性奠定基礎。接下來的研究將延續此方向，檢視透明度（第二項研究）與資訊模態（第三項研究）如何進一步影響創作者的反應，旨在提供訊號設計如何支持更健康、更具韌性的創作者與平台關係的全面理解。

Building on these framing effects, Study 2 extends these insights by demonstrating that algorithmic transparency fundamentally strengthens creators’ trust and commitment (Verwiebe et al., 2024). Transparent feedback not only reduced perceived risk, but also enhanced entrepreneurial intention, particularly among those whose careers are closely tied to platform outcomes.
基於這些框架效應，第二項研究進一步擴展了這些見解，證明演算法透明度從根本上增強了創作者的信任與承諾（Verwiebe et al., 2024）。透明的回饋不僅降低了感知風險，還提升了創業意圖，尤其是在那些職業與平台成果密切相關的創作者中更為顯著。

The results extend prior work on signaling and trust by showing that the clarity of signals—not just their content—determines psychological outcomes. Trust is built through communication processes, not just reputational metrics (Song, 2023). The observed interaction effect also underscores the importance of tailoring platform strategies to the needs of dependent users, who are more vulnerable to information asymmetry.
結果擴展了先前關於訊號與信任的研究，顯示訊號的清晰度——不僅僅是其內容——決定了心理結果。信任是透過溝通過程建立的，而非僅依賴聲譽指標（Song, 2023）。觀察到的交互作用效果也強調了針對依賴性使用者需求調整平台策略的重要性，因為這些使用者對資訊不對稱更為脆弱。

For platform designers and policymakers, the findings suggest actionable strategies to support creators. These include offering detailed feedback explanations, predictive insights, and transparent rationale for content decisions. Transparent communication not only fosters trust but also enhances creator retention, resilience, and strategic autonomy.
對於平台設計者與政策制定者而言，研究結果提出了支持創作者的可行策略，包括提供詳細的反饋說明、預測性洞見以及內容決策的透明理由。透明的溝通不僅促進信任，還能提升創作者的留存率、韌性與策略自主性。

Future research could explore how transparency interacts with other platform features (e.g., community norms, monetization structures) and investigate long-term behavioral outcomes. These directions are further developed in Study 3.
未來研究可探討透明度如何與其他平台功能（例如社群規範、貨幣化結構）互動，並調查長期行為結果。這些方向在第三項研究中有更進一步的發展。

Importantly, Study 3 reveals that the format of feedback presentation plays a pivotal role in shaping creators’ psychological engagement. Compared with numeric-only feedback, both visual and narrative formats alleviated cognitive load and increased decision confidence, fostering greater intention to persist as content creators across experience levels (Moffett et al., 2021).
重要的是，研究三揭示了回饋呈現格式在塑造創作者心理參與度方面扮演關鍵角色。與僅有數字回饋相比，視覺與敘事格式皆減輕了認知負荷並提升決策信心，促進不同經驗層級的創作者持續創作的意願（Moffett et al., 2021）。

These findings reinforce the importance of information format in platform communication. Taken together, they set the stage for a holistic discussion on how feedback design shapes psychological and strategic outcomes within the socio-economic architecture of modern cyberspace.
這些發現強化了資訊格式在平台溝通中的重要性。綜合而言，這些結果為全面討論回饋設計如何在現代網路空間的社會經濟架構中形塑心理與策略結果奠定了基礎。

Integration of Findings
整合研究結果

Collectively, these findings demonstrate that creators’ perceptions and decisions are not shaped by isolated platform signals, but rather by the complex interplay among signal framing, feedback transparency, and presentation modality. Study 1 highlighted the psychological vulnerability of highly platform-dependent creators to negatively framed feedback. Study 2 underscored the trust-building and risk-reducing value of transparent algorithmic communication, especially for those with high dependency. Study 3 confirmed that how information is presented—visually or narratively—can further alleviate mental burden and promote adaptive engagement. These multi-layered effects underscore the need for holistic platform design strategies that address both the content and delivery of feedback in supporting creator well-being and sustained participation.
綜合而言，這些發現顯示創作者的感知與決策並非由孤立的平台訊號所塑造，而是由訊號框架、回饋透明度與呈現方式之間的複雜交互作用所影響。研究一強調高度依賴平台的創作者在面對負向框架回饋時的心理脆弱性。研究二則凸顯透明演算法溝通在建立信任與降低風險感方面的價值，尤其對高度依賴者更為顯著。研究三證實資訊的呈現方式—無論是視覺化或敘事化—能進一步減輕心理負擔並促進適應性參與。這些多層次的效應強調了整體性平台設計策略的必要性，該策略需同時關注回饋內容與傳遞方式，以支持創作者的福祉及持續參與。

Table 4 presents a summary of these integrative findings, illustrating how framing, transparency, and modality each exert both direct and mediated effects on risk perception, platform trust, cognitive load, decision confidence, and entrepreneurial intention. Notably, platform dependency consistently moderates these relationships, amplifying sensitivity to negative or ambiguous feedback and enhancing the positive effects of transparent, user-friendly communication.
表 4 彙整了這些綜合性發現，說明框架設定、透明度與模式各自如何對風險感知、平台信任、認知負荷、決策信心及創業意圖產生直接及中介效應。值得注意的是，平台依賴性持續調節這些關係，放大對負面或模糊回饋的敏感度，並強化透明且使用者友善溝通的正面效應。

This integrative framework demonstrates that optimizing digital platform communication requires more than simply increasing feedback volume or automating processes; it demands thoughtful attention to how information is framed, conveyed, and formatted. By targeting these dimensions, platform managers can directly influence creators’ psychological well-being, strategic adaptation, and career sustainability.
此整合性架構顯示，優化數位平台溝通不僅僅是增加反饋量或自動化流程；更需謹慎關注資訊的框架設定、傳達方式及格式。透過針對這些面向，平台管理者能直接影響創作者的心理健康、策略調適及職涯永續性。

Overall, this research contributes a nuanced, evidence-based understanding of creator–platform dynamics and provides a foundation for developing next-generation digital labor platforms that empower and sustain creators.
整體而言，本研究提供對創作者與平台互動動態的細緻且具實證基礎的理解，並為開發新一代賦能且支持創作者的數位勞動平台奠定基礎。

Theoretical Contributions
理論貢獻

This research makes several important theoretical contributions to the study of digital labor, platform economies, and entrepreneurial psychology.
本研究對數位勞動、平台經濟及創業心理學的研究做出多項重要理論貢獻。
First, by integrating signaling theory with cognitive framing and risk perception frameworks, this work advances beyond traditional models that treat platform signals as static or unidirectional. Instead, it conceptualizes creators as active interpreters who strategically decode and respond to algorithmic cues within dynamic digital ecosystems. This interpretive agency is essential for understanding both career persistence and the development of adaptive strategies in uncertain informational environments (Jeitschko & Tremblay, 2020; Sherratt & O’Neill, 2023).
首先，透過整合訊號理論與認知框架及風險感知架構，本研究超越了將平台訊號視為靜態或單向的傳統模型。相反地，本研究將創作者概念化為積極的詮釋者，能在動態的數位生態系中策略性地解讀並回應演算法提示。這種詮釋能動性對於理解職涯持續性及在不確定資訊環境中發展適應性策略至關重要（Jeitschko & Tremblay, 2020；Sherratt & O’Neill, 2023）。

Second, further enriching this integrated framework, the findings highlight the nuanced role of platform dependency as a moderator. Whereas prior research often assumed homogeneity among platform workers, this study empirically demonstrates that creators’ sensitivity to informational design and feedback is significantly shaped by their reliance on platform income and identity. By systematically modeling and validating this moderating effect, the research extends theories of digital entrepreneurship and self-employment, providing new insights into vulnerability, resilience, and strategic adaptation among creators (Ye et al., 2022).
其次，進一步豐富此整合框架，研究結果突顯了平台依賴作為調節變項的細緻角色。過去的研究常假設平台工作者的同質性，本研究則實證顯示創作者對資訊設計與回饋的敏感度，顯著受到其對平台收入與身份認同的依賴所影響。透過系統性建模與驗證此調節效果，本研究擴展了數位創業與自僱理論，提供了關於創作者脆弱性、韌性及策略性調適的新見解（Ye et al., 2022）。

Third, to explain the mechanisms underlying these interactions, the experimental results advance models of information processing and cognitive load theory. The evidence shows that while numeric feedback offers precision, it imposes greater mental burden and elevates risk perception, potentially undermining creators’ motivation. In contrast, visualizations and narrative summaries promote pattern recognition, comprehension, and adaptive strategies, empowering creators to interpret platform signals more effectively (Choudhry et al., 2021; Guo et al., 2018). These findings support recent calls to integrate human-computer interaction and information visualization principles into the study of algorithmic management (Elmalech, 2023; Yang & Zhang, 2024).
第三，為了解釋這些互動背後的機制，實驗結果推進了資訊處理模型與認知負荷理論。證據顯示，雖然數字反饋提供了精確性，但它帶來更大的心理負擔並提升風險感知，可能削弱創作者的動機。相較之下，視覺化和敘事摘要促進了模式識別、理解力及適應策略，使創作者能更有效地解讀平台訊號（Choudhry et al., 2021；Guo et al., 2018）。這些發現支持近期將人機互動與資訊視覺化原則整合進演算法管理研究的呼籲（Elmalech, 2023；Yang & Zhang, 2024）。

Fourth, the use of a multi-study experimental methodology provides rare causal evidence for the complex relationships among platform signals, psychological mechanisms, and entrepreneurial outcomes. The combination of scenario-based manipulations, realistic interfaces, and robust statistical analysis offers a methodological benchmark for future studies in digital labor and platform communication.
第四，採用多研究實驗方法提供了平台訊號、心理機制與創業成果之間複雜關係的罕見因果證據。情境操作、真實介面與嚴謹統計分析的結合，為未來數位勞動與平台溝通研究樹立了方法論標竿。

Finally, the conceptual model developed in this work serves as an integrative framework for future research on digital career formation, algorithmic management, and the behavioral economics of online work. This model invites further inquiry into additional moderators and mediators, as well as comparative studies across diverse cultural, technological, and regulatory contexts.
最後，本研究所建立的概念模型作為未來數位職涯形成、演算法管理及線上工作行為經濟學研究的整合框架。此模型邀請進一步探討更多調節變項與中介變項，並進行跨文化、技術及法規環境的比較研究。

In sum, by building a multi-layered model that integrates signal design, psychological mechanisms, and individual dependency, this research deepens the theoretical understanding of how human agency operates within the socio-economic structures of modern cyberspace. It not only opens new avenues for the psychology of entrepreneurship but also provides a crucial framework for the future governance of digital platforms.
總之，透過構建一個整合訊號設計、心理機制與個人依賴性的多層次模型，本研究深化了對人類能動性如何在現代網路空間社會經濟結構中運作的理論理解。此研究不僅為創業心理學開啟了新途徑，也為未來數位平台治理提供了關鍵框架。

Practical Implications
實務意涵

The findings from this multi-study investigation offer significant practical guidance for platform managers, policymakers, and organizations seeking to support a healthier and more sustainable creator ecosystem. By clarifying the psychological mechanisms through which algorithmic signals influence creators’ decisions, this research provides an evidence-based roadmap for designing interventions, governance frameworks, and educational resources that empower creators and mitigate risks of disengagement, burnout, or maladaptation. This section distills actionable recommendations and highlights directions for future research.
本多重研究調查的發現為平台管理者、政策制定者及致力於支持更健康且永續創作者生態系統的組織提供了重要的實務指引。透過釐清演算法訊號影響創作者決策的心理機制，本研究提供了一條以實證為基礎的路徑，協助設計介入措施、治理架構及教育資源，賦能創作者並減少其脫離、倦怠或不良適應的風險。本節萃取可行的建議，並強調未來研究方向。

Digital platforms should move beyond viewing creators as passive data points, instead recognizing them as strategic agents whose sustained engagement relies on the quality of the informational environment shaped by platform design. Several recommendations emerge:
數位平台應超越將創作者視為被動數據點的觀念，轉而認識他們為策略性行動者，其持續參與依賴於由平台設計所塑造的資訊環境品質。由此產生數項建議：

Platforms must prioritize transparent, user-oriented communication. Providing clear explanations of algorithmic logic, advance notification of policy changes, and accessible predictive analytics can significantly increase creator trust.
平台必須優先考量透明且以使用者為導向的溝通。提供清晰的演算法邏輯說明、政策變動的事前通知，以及易於理解的預測分析，皆能顯著提升創作者的信任度。

Designing Transparent and Trustworthy Communication. Platforms must prioritize transparent, user-oriented communication. Providing clear explanations of algorithmic logic, advance notification of policy changes, and accessible predictive analytics can significantly increase creator trust and reduce risk perception, especially for those highly dependent on the platform for income or identity. Transparent communication protocols should be standardized and regularly updated to address evolving system logic and user needs (Chen, 2022).
設計透明且值得信賴的溝通。平台必須優先考量透明且以使用者為導向的溝通。提供清晰的演算法邏輯說明、政策變更的事前通知，以及易於理解的預測分析，能顯著提升創作者的信任度並降低風險感知，尤其對於高度依賴平台作為收入或身份認同來源的創作者而言更為重要。透明的溝通協議應標準化並定期更新，以因應系統邏輯與使用者需求的演變（Chen, 2022）。

Optimizing Feedback Framing and Modality. Signal framing and modality should be adapted to the needs of diverse creator segments (Polk & Diver, 2020). Platforms should avoid alarmist or overly negative language, particularly for feedback on routine fluctuations. Multi-modal feedback options—including intuitive visualizations and narrative summaries—enable creators with varying levels of data literacy to access actionable information in forms that foster confidence and strategic planning.
優化回饋框架與傳達模式。訊號的框架與傳達模式應根據不同創作者群體的需求進行調整（Polk & Diver, 2020）。平台應避免使用危言聳聽或過於負面的語言，特別是在回饋日常波動時。多模式的回饋選項——包括直觀的視覺化呈現與敘事式摘要——使具不同數據素養程度的創作者能以促進信心與策略規劃的形式，獲取可行的資訊。

Investing in Creator Education and Psychological Support. Investment in creator education and support is essential. Platforms should provide resources, training, and peer networks to help creators interpret feedback, manage risk, and develop adaptive strategies (Lee et al., 2024). These initiatives help reduce cognitive overload and support agency, particularly for those with less digital literacy or experience.
投資於創作者教育與心理支持。投資於創作者教育與支持至關重要。平台應提供資源、培訓及同儕網絡，協助創作者解讀回饋、管理風險並發展適應策略（Lee et al., 2024）。這些措施有助於減輕認知負荷並支持行動能力，特別是對於數位素養或經驗較少者。

Policy and Governance Recommendations. From a policy standpoint, regulators and industry groups should advocate for greater transparency, fairness, and accountability in platform governance. This may include setting standards for algorithmic disclosure, ensuring creators’ right to clear information, and establishing independent oversight mechanisms for dispute resolution. Collaboration between platforms and policymakers can help build an innovative, resilient, and equitable creator economy (Alauddin et al., 2024; Choi et al., 2023).
政策與治理建議。從政策角度來看，監管機構與產業團體應倡導平台治理的更高透明度、公平性與問責制。這可能包括制定演算法揭露標準、確保創作者獲得明確資訊的權利，以及建立獨立的爭議解決監督機制。平台與政策制定者之間的合作，有助於打造創新、具韌性且公平的創作者經濟（Alauddin et al., 2024；Choi et al., 2023）。

Together, these recommendations underscore the potential for thoughtful platform communication and governance to enhance creator well-being and the need for ongoing dialogue among platforms, creators, and policymakers as the digital labor landscape evolves.
綜合而言，這些建議強調了深思熟慮的平台溝通與治理對提升創作者福祉的潛力，以及隨著數位勞動環境演變，平台、創作者與政策制定者之間持續對話的必要性。

The findings highlight the importance of robust support systems and targeted educational initiatives to equip creators with the skills, resilience, and strategic capacity needed for long-term success in algorithmically mediated environments. Interpreting platform signals—amid algorithmic fluctuations and ambiguous feedback—places significant demands on creators’ cognitive and emotional resources. Without adequate support, creators risk misinterpreting feedback, experiencing overwhelm, or disengaging from the platform altogether (Lee et al., 2024; Stegeman et al., 2024).
研究結果強調了建立健全支援系統及針對性教育計畫的重要性，以裝備創作者具備在演算法媒介環境中長期成功所需的技能、韌性及策略能力。在演算法波動與模糊反饋中解讀平台訊號，對創作者的認知與情緒資源提出了重大挑戰。若缺乏足夠支援，創作者可能誤解反饋、感到不堪負荷，甚至完全退出平台（Lee et al., 2024；Stegeman et al., 2024）。

To address these challenges, platforms and allied organizations should invest in comprehensive creator education programs. Onboarding workshops, online courses, and modular learning tracks that demystify algorithmic logic, clarify key performance indicators, and promote data literacy can foster creators’ understanding of performance analytics and reduce uncertainty. These initiatives help creators distinguish between normal fluctuations and problematic feedback, thereby enhancing strategic agency.
為應對這些挑戰，平台及相關組織應投資於全面性的創作者教育計畫。入門工作坊、線上課程及模組化學習路徑，透過解密演算法邏輯、釐清關鍵績效指標並促進數據素養，能增進創作者對績效分析的理解並降低不確定性。這些措施有助於創作者分辨正常波動與問題性回饋，從而提升策略能動性。

Psychological resilience training should also be integral to creator support. Educational content focused on stress management, adaptive coping, and decision-making under uncertainty can mitigate the adverse effects of negative feedback and perceived risk. Peer mentoring, coaching, and community-based support further provide social resources, enabling creators to navigate setbacks and celebrate achievements, reducing feelings of isolation common in digital labor (Kotturi et al., 2024).
心理韌性訓練亦應成為創作者支援的核心部分。聚焦於壓力管理、適應性因應及不確定性下決策的教育內容，能減輕負面反饋與風險感知的負面影響。朋輩指導、教練輔導及社群支持進一步提供社會資源，使創作者能夠應對挫折並慶祝成就，減少數位勞動中常見的孤立感（Kotturi et al., 2024）。

Platforms might also develop customizable feedback systems, allowing creators to tailor the format and frequency of notifications to their preferences and learning styles. Such user-centered design not only increases satisfaction and engagement but also aligns feedback delivery with creators’ developmental and psychological needs.
平台亦可開發可自訂的反饋系統，允許創作者根據個人偏好和學習風格調整通知的格式與頻率。此類以使用者為中心的設計不僅提升滿意度與參與度，亦使反饋傳遞與創作者的發展及心理需求相契合。

Collaboration among platforms, educational institutions, and industry associations is essential for advancing best practices in creator education. This may include joint certification programs, public awareness campaigns, and knowledge hubs where creators access current resources and share experiential insights. Proactive investment in education and support strengthens both individual creator well-being and the sustainability of vibrant, adaptive creator communities.
平台、教育機構與產業協會之間的合作，對推動創作者教育的最佳實踐至關重要。這可能包括聯合認證計畫、公共宣導活動以及創作者可取得最新資源並分享經驗見解的知識中心。積極投資於教育與支持，不僅強化個別創作者的福祉，也促進充滿活力且具適應性的創作者社群的永續發展。

In summary, empowering creators through support and education is not only vital for individual success, but also a strategic imperative for platforms seeking to sustain engaged, innovative, and resilient digital labor ecosystems.
總結而言，透過支持與教育賦能創作者，不僅對個人成功至關重要，也是平台維持積極參與、創新且具韌性數位勞動生態系的策略性必須。

Limitations
限制

Despite its strengths, this research is subject to several limitations. First, the experimental studies relied primarily on scenario-based manipulations and simulated platform interfaces. Although ecological validity was enhanced through realistic materials and pilot testing, such simulations cannot fully capture the complexity and unpredictability of real-world platform dynamics. Actual creator behavior may be influenced by additional contextual factors—such as financial stakes, social relationships, and long-term feedback cycles—that are difficult to replicate in laboratory settings (Are & Briggs, 2023; Glatt, 2023).
儘管本研究具有其優勢，但仍存在若干限制。首先，實驗研究主要依賴情境式操控與模擬平台介面。雖然透過真實材料與先導測試提升了生態效度，但此類模擬無法完全捕捉真實世界平台動態的複雜性與不可預測性。實際創作者行為可能受到更多情境因素影響——例如財務風險、社會關係及長期回饋循環——這些因素難以在實驗室環境中複製（Are & Briggs, 2023；Glatt, 2023）。

Second, the participant samples largely consisted of university students and early-career creators recruited from educational and professional networks. While relevant to the emerging creator economy, these groups may not reflect the full diversity of creator experiences, particularly in terms of age, cultural background, platform specialization, or entrepreneurial tenure. As such, the generalizability of results to more established or highly specialized creators warrants further investigation.
其次，參與者樣本主要由大學生及早期職涯創作者組成，這些受試者多從教育及專業網絡招募。雖然與新興創作者經濟相關，但這些群體可能無法反映創作者經驗的全部多樣性，特別是在年齡、文化背景、平台專業化或創業資歷方面。因此，研究結果對於更成熟或高度專業化創作者的普遍適用性仍需進一步探討。

Third, the research focused on three specific aspects of platform signal design—framing, transparency, and modality—while holding other potential influences constant. In practice, creators encounter a much broader array of signals, policies, social cues, and market forces, which may interact in complex and unpredictable ways. The proposed conceptual model, while integrative, may only partially account for the diverse influences shaping creator decision-making.
第三，本研究聚焦於平台訊號設計的三個特定面向——框架設定、透明度與模態——同時將其他潛在影響因素保持不變。實務上，創作者面臨的訊號、政策、社會線索及市場力量遠比此更為多元，且可能以複雜且難以預測的方式相互作用。所提出的概念模型雖具整合性，但可能僅部分解釋塑造創作者決策的多元影響因素。

Fourth, the cross-sectional nature of the experiments limits conclusions regarding the long-term or cumulative effects of platform signals on creator behavior and psychological well-being. Longitudinal or field-based studies would be valuable for examining how creators adapt to sustained algorithmic feedback, changing platform policies, or negative experiences over time.
第四，實驗的橫斷面性質限制了對平台訊號對創作者行為及心理福祉長期或累積效應的結論。縱向或實地研究將有助於探討創作者如何隨時間適應持續的演算法反饋、變動的平台政策或負面經驗。

Finally, this research was conducted within a specific digital and cultural context. Differences in platform governance, regulatory environments, and user norms across regions may moderate the observed effects or introduce new dynamics not captured in this study. Comparative research across platforms, industries, and countries is needed to further enrich both theoretical understanding and practical relevance.
最後，本研究是在特定的數位與文化脈絡中進行。不同地區的平台治理、監管環境及用戶規範差異，可能會調節所觀察到的效應，或引入本研究未涵蓋的新動態。跨平台、產業及國家的比較研究，對於進一步豐富理論理解與實務相關性具有必要性。

Given these limitations, caution is warranted when generalizing the findings. Nonetheless, the experimental rigor, conceptual integration, and empirical insights of this research lay a solid foundation for ongoing inquiry into the psychological and strategic dynamics of creator work in the algorithmic age.
鑑於這些限制，對於研究結果的推論需謹慎。然而，本研究的實驗嚴謹性、概念整合性及實證洞見，為持續探究演算法時代創作者工作之心理與策略動態奠定了堅實基礎。

Directions for Future Research
未來研究方向

While this research advances our understanding of psychological and strategic processes underlying creator decision-making in algorithmic environments, it also points to several avenues for future inquiry. First, longitudinal studies are needed to explore how the effects of signal framing, transparency, and information modality evolve over time. Investigating creators’ adaptive responses to sustained or repeated feedback will deepen insight into long-term engagement, burnout risk, and the development of resilience or maladaptive strategies (Choi et al., 2023).
雖然本研究增進了我們對演算法環境中創作者決策背後心理與策略過程的理解，但同時也指出了未來研究的若干方向。首先，需要進行縱向研究，以探討訊號框架、透明度及資訊模式的效應如何隨時間演變。調查創作者對持續或重複反饋的適應性反應，將深化對長期參與、倦怠風險以及韌性或不良適應策略發展的理解（Choi et al., 2023）。

Second, future research should broaden the scope of platforms and creator diversity. Comparative studies across various digital platforms, creative industries, and cultural contexts can clarify how structural and normative differences shape the impact of algorithmic signals. Research focusing on established, professional, or niche creators would also determine whether these mechanisms generalize across experience levels and economic models.
其次，未來研究應擴大平台範圍及創作者多樣性。跨不同數位平台、創意產業及文化脈絡的比較研究，能釐清結構性與規範性差異如何影響演算法訊號的作用。聚焦於成熟、專業或利基創作者的研究，也將有助於判斷這些機制是否能跨越經驗層級與經濟模式普遍適用。

Third, while this study centered on informational design, other factors—such as community feedback, social capital, collaboration networks, and external shocks (e.g., policy changes, economic crises)—deserve systematic investigation. Integrating these elements through experimental or mixed-method approaches will provide a more holistic understanding of the ecosystem-level dynamics influencing creator careers.
第三，本研究雖聚焦於資訊設計，但其他因素如社群回饋、社會資本、合作網絡及外部衝擊（例如政策變動、經濟危機）亦值得系統性探討。透過實驗或混合方法整合這些元素，將有助於更全面理解影響創作者職涯的生態系統層級動態。

Fourth, interdisciplinary collaboration with fields such as human–computer interaction, behavioral economics, and organizational psychology can enrich theoretical development and methodological rigor. Complementing experiments with qualitative interviews, ethnographies, or field studies will better capture the complexity of real-world creator experiences (Dumont, 2024; Salamon, 2025).
第四，與人機互動、行為經濟學及組織心理學等領域的跨學科合作，能夠豐富理論發展與方法論嚴謹性。結合實驗與質性訪談、民族誌或田野研究，將更全面捕捉現實世界中創作者經驗的複雜性（Dumont, 2024；Salamon, 2025）。

Finally, future studies should design and test new interventions aimed at reducing the negative effects of algorithmic opacity, risk amplification, or information overload. Evaluating the efficacy of novel platform features, educational tools, or policy innovations will be crucial for translating theoretical insights into practical improvements for the creator economy (Bhargava, 2022; Ren, 2024).
最後，未來的研究應設計並測試旨在減少演算法不透明性、風險放大或資訊過載負面影響的新干預措施。評估新穎平台功能、教育工具或政策創新的效能，對於將理論見解轉化為創作者經濟的實際改進至關重要（Bhargava, 2022；Ren, 2024）。

By pursuing these directions, future research can advance both knowledge and practice in the rapidly evolving field of digital creative labor.
透過追求這些方向，未來的研究能夠推動數位創意勞動這一快速發展領域的知識與實務進步。

Conclusion
結論

This research offers an integrated, evidence-based account of how creators navigate the contemporary platform economy. By demonstrating that the design of algorithmic signals directly influences creators’ risk perception, trust, and resilience, our findings advance both psychological theory and the practical governance of digital labor. Ultimately, by mapping the psychological pathways from signal to decision, this study illuminates the new fundamentals of human cognition and behavior in the algorithmic age.
本研究提供了一個整合且基於實證的說明，探討創作者如何在當代平台經濟中進行導航。透過證明演算法訊號的設計直接影響創作者的風險感知、信任與韌性，我們的發現推進了心理學理論與數位勞動的實務治理。最終，透過描繪從訊號到決策的心理路徑，本研究揭示了演算法時代人類認知與行為的新基礎。

References
參考文獻

Abitbol, A. S., Lee, N. M., & VanDyke, M. S. (2022). Examining the perceived transparency of DTC genetic testing company communication and its impact on consumer trust, attitude and behavioral intentions. Journal of Communication Management, 26(3), 315–330. https://doi.org/10.1108/jcom-01-2022-0006
Abitbol, A. S., Lee, N. M., & VanDyke, M. S. (2022)。檢視直接面向消費者基因檢測公司溝通的感知透明度及其對消費者信任、態度與行為意圖的影響。《傳播管理期刊》，26(3)，315–330。https://doi.org/10.1108/jcom-01-2022-0006

Alasoini, T., Immonen, J., Seppänen, L., & Känsälä, M. (2023). Platform workers and digital agency: Making out on three types of labor platforms. Frontiers in Sociology, 8. https://doi.org/10.3389/fsoc.2023.1063613
Alasoini, T., Immonen, J., Seppänen, L., & Känsälä, M. (2023)。平台工作者與數位能動性：在三種類型的勞動平台上的生存之道。《社會學前沿》，8。https://doi.org/10.3389/fsoc.2023.1063613

Alauddin, F. D., Aman, A., Ghazali, M. F., & Daud, S. (2024). The Influence of Digital Platforms on Gig Workers: A Systematic Literature Review. Heliyon, 11(1), e41491. https://doi.org/10.1016/j.heliyon.2024.e41491
Alauddin, F. D., Aman, A., Ghazali, M. F., & Daud, S. (2024)。數位平台對零工工作者的影響：系統性文獻回顧。《Heliyon》，11(1)，e41491。https://doi.org/10.1016/j.heliyon.2024.e41491

Appio, F. P., & Corso, M. (2023). From mutualism to commensalism: Assessing the evolving relationship between complementors and digital platforms. Information Systems Journal. https://doi.org/10.1111/isj.12491
Appio, F. P., & Corso, M. (2023)。從互利共生到片利共生：評估補充者與數位平台之間不斷演變的關係。《資訊系統期刊》。https://doi.org/10.1111/isj.12491

Are, C., & Briggs, P. (2023). The Emotional and Financial Impact of De-Platforming on Creators at the Margins. Social Media and Society, 9(1), 205630512311551. https://doi.org/10.1177/20563051231155103
Are, C., & Briggs, P. (2023)。邊緣創作者被平台下架的情感與財務影響。《社群媒體與社會》，9(1)，205630512311551。https://doi.org/10.1177/20563051231155103

Beese, N. O., Rodriguez, F. S., Spilski, J., & Lachmann, T. (2021). Development of a Digital Video-Based Occupational Risk Assessment Method. Frontiers in Public Health, 9, 683850. https://doi.org/10.3389/FPUBH.2021.683850
Beese, N. O., Rodriguez, F. S., Spilski, J., & Lachmann, T. (2021)。數位影片基礎職業風險評估方法之開發。《公共衛生前沿》，9，683850。https://doi.org/10.3389/FPUBH.2021.683850

Bhargava, H. K. (2022). The Creator Economy: Managing Ecosystem Supply, Revenue Sharing, and Platform Design. (2022). Management Science, 68(7), 5233–5251. https://doi.org/10.1287/mnsc.2021.4126
Bhargava, H. K. (2022)。創作者經濟：生態系統供給、收益分配與平台設計管理。（2022）。管理科學，68(7)，5233–5251。https://doi.org/10.1287/mnsc.2021.4126

Bleier, A., Goldfarb, A., & Tucker, C. (2020). Consumer Privacy and the Future of Data-Based Innovation and Marketing. Social Science Research Network. https://doi.org/10.2139/SSRN.3570156
Bleier, A., Goldfarb, A., & Tucker, C. (2020)。消費者隱私與基於數據的創新與行銷未來。社會科學研究網絡。https://doi.org/10.2139/SSRN.3570156

Cabeza-Ramírez, L. J., Fuentes-Garcia, F. J., Cano-Vicente, M. C., & González-Mohino, M. (2022). How Generation X and Millennials Perceive Influencers’ Recommendations: Perceived Trustworthiness, Product Involvement, and Perceived Risk. Journal of Theoretical and Applied Electronic Commerce Research, 17(4), 1431–1449. https://doi.org/10.3390/jtaer17040072
Cabeza-Ramírez, L. J., Fuentes-Garcia, F. J., Cano-Vicente, M. C., & González-Mohino, M. (2022)。X 世代與千禧世代如何看待影響者的推薦：感知可信度、產品參與度與感知風險。《理論與應用電子商務研究期刊》，17(4)，1431–1449。https://doi.org/10.3390/jtaer17040072

Cao, A., Chintamani, K., Pandya, A., & Ellis, R. D. (2009). NASA TLX: software for assessing subjective mental workload. Behavior Research Methods, 41(1), 113–117. https://doi.org/10.3758/BRM.41.1.113
Cao, A., Chintamani, K., Pandya, A., & Ellis, R. D. (2009)。NASA TLX：評估主觀心理負荷的軟體。《行為研究方法》，41(1)，113–117。https://doi.org/10.3758/BRM.41.1.113

Chen, J. Y. C. (2022). Transparent Human–Agent Communications. International Journal of Human-Computer Interaction, 38(18–20), 1737–1738. https://doi.org/10.1080/10447318.2022.2120173
Chen, J. Y. C. (2022)。透明的人機代理溝通。國際人機互動期刊，38(18–20)，1737–1738。https://doi.org/10.1080/10447318.2022.2120173

Chen, L.-T., & Peng, C.-Y. J. (2015). The sensitivity of three methods to nonnormality and unequal variances in interval estimation of effect sizes. Behavior Research Methods, 47(1), 107–126. https://doi.org/10.3758/S13428-014-0461-3
陳立堂、彭志遠（2015）。三種方法對非常態性及變異數不等於區間估計效果量之敏感度。《行為研究方法》，47(1)，107–126。https://doi.org/10.3758/S13428-014-0461-3

Chen, S., Li, J., Andrienko, G., Andrienko, N., Wang, Y., Nguyen, P. H., & Turkay, C. (2020). Supporting Story Synthesis: Bridging the Gap between Visual Analytics and Storytelling. IEEE Transactions on Visualization and Computer Graphics, 26(7), 2499–2516. https://doi.org/10.1109/TVCG.2018.2889054
陳思、李傑、Andrienko, G.、Andrienko, N.、王洋、Nguyen, P. H.、Turkay, C.（2020）。支持故事綜合：彌合視覺分析與敘事之間的鴻溝。《IEEE 視覺化與電腦圖學匯刊》，26(7)，2499–2516。https://doi.org/10.1109/TVCG.2018.2889054

Cheng, X., Huang, X., Yang, B., Chen, S., & Yan, Y. (2024). How perceived justice leads to stickiness to short-term rental platforms: Unveiling the effect of relationship commitment and trust. Electronic Commerce Research and Applications, 66, 101422. https://doi.org/10.1016/j.elerap.2024.101422
程翔、黃曉、楊斌、陳思、嚴毅（2024）。感知正義如何導致對短期租賃平台的黏著性：揭示關係承諾與信任的影響。《電子商務研究與應用》，66，101422。https://doi.org/10.1016/j.elerap.2024.101422

Choi, Y. O., Kang, E. J., Lee, M. K., & Kim, J. (2023). Creator-friendly Algorithms: Behaviors, Challenges, and Design Opportunities in Algorithmic Platforms. International Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3544548.3581386
崔永奧、姜恩珠、李明姬、金正（2023）。創作者友善的演算法：演算法平台中的行為、挑戰與設計機會。國際人因計算系統會議。https://doi.org/10.1145/3544548.3581386

Choudhry, A., Sharma, M., Chundury, P., Kapler, T., Gray, D. W. S., Ramakrishnan, N., & Elmqvist, N. (2021). Once Upon A Time In Visualization: Understanding the Use of Textual Narratives for Causality. IEEE Transactions on Visualization and Computer Graphics, 27(2), 1332–1342. https://doi.org/10.1109/TVCG.2020.3030358
Choudhry, A., Sharma, M., Chundury, P., Kapler, T., Gray, D. W. S., Ramakrishnan, N., & Elmqvist, N. (2021)。〈視覺化中的故事敘述：理解文本敘事在因果關係中的運用〉。《IEEE 視覺化與電腦圖學匯刊》，27(2)，1332–1342。https://doi.org/10.1109/TVCG.2020.3030358

Cutolo, D., & Kenney, M. (2021). Platform-Dependent Entrepreneurs: Power Asymmetries, Risks, and Strategies in the Platform Economy. Academy of Management Perspectives, 35(4), 584–605. https://doi.org/10.5465/AMP.2019.0103
Cutolo, D., & Kenney, M. (2021)。平台依賴型創業者：平台經濟中的權力不對稱、風險與策略。管理學院觀點，35(4)，584–605。https://doi.org/10.5465/AMP.2019.0103

Connelly, B. L., Certo, S. T., Ireland, R. D., & Reutzel, C. R. (2011). Signaling Theory: A Review and Assessment. Journal of Management, 37(1), 39–67. https://doi.org/10.1177/0149206310388419
Connelly, B. L., Certo, S. T., Ireland, R. D., & Reutzel, C. R. (2011)。訊號理論：回顧與評估。管理學期刊，37(1)，39–67。https://doi.org/10.1177/0149206310388419

Damman, O. C., De Jong, A., Hibbard, J. H., & Timmermans, D. R. M. (2016). Making comparative performance information more comprehensible: an experimental evaluation of the impact of formats on consumer understanding. BMJ Quality & Safety, 25(11), 860–869. https://doi.org/10.1136/BMJQS-2015-004120
Damman, O. C., De Jong, A., Hibbard, J. H., & Timmermans, D. R. M. (2016)。〈使比較績效資訊更易理解：格式對消費者理解影響的實驗評估〉。《BMJ 品質與安全》，25(11)，860–869。https://doi.org/10.1136/BMJQS-2015-004120

Dudgeon, P. (2016). A Comparative Investigation of Confidence Intervals for IndependentVariables in Linear Regression. Multivariate Behavioral Research, 51, 139–153. https://doi.org/10.1080/00273171.2015.1121372
Dudgeon, P. (2016)。線性迴歸中自變數信賴區間的比較研究。《多變量行為研究》，51，139–153。https://doi.org/10.1080/00273171.2015.1121372

Duggan, J. T., Carbery, R., McDonnell, A., & Sherman, U. (2023). Algorithmic HRM control in the gig economy: The app‐worker perspective. Human Resource Management. https://doi.org/10.1002/hrm.22168
Duggan, J. T., Carbery, R., McDonnell, A., & Sherman, U. (2023)。零工經濟中的演算法人力資源管理控制：應用程式工作者的觀點。《人力資源管理》。https://doi.org/10.1002/hrm.22168

Dumont, G. (2024). Fulfilling the process promise in new venture creation research: The ethnography/accelerator approach. Strategic Entrepreneurship Journal. https://doi.org/10.1002/sej.1494
Dumont, G. (2024)。在新創事業創建研究中實現流程承諾：民族誌／加速器方法。《策略創業期刊》。https://doi.org/10.1002/sej.1494

Edeling, A., & Wies, S. (2024). Embracing entrepreneurship in the creator economy: The rise of creatrepreneurs. International Journal of Research in Marketing. https://doi.org/10.1016/j.ijresmar.2024.07.003
Edeling, A., & Wies, S. (2024)。擁抱創作者經濟中的創業精神：創作者企業家的崛起。《國際行銷研究期刊》。https://doi.org/10.1016/j.ijresmar.2024.07.003

Elmalech, A. (2023). Empowering the middle management: Incorporating data summarisation and visualisation techniques in management information systems. Journal of Information Science, 016555152311766. https://doi.org/10.1177/01655515231176649
Elmalech, A. (2023)。賦能中層管理：在管理資訊系統中整合資料摘要與視覺化技術。《資訊科學期刊》，016555152311766。https://doi.org/10.1177/01655515231176649

Eslami, M., Vaccaro, K., Lee, M. K., Elazari Bar On, A., Gilbert, E., & Karahalios, K. (2019). User Attitudes towards Algorithmic Opacity and Transparency in Online Reviewing Platforms. Human Factors in Computing Systems, 494. https://doi.org/10.1145/3290605.3300724
Eslami, M., Vaccaro, K., Lee, M. K., Elazari Bar On, A., Gilbert, E., & Karahalios, K. (2019)。使用者對線上評論平台中演算法不透明性與透明性的態度。《人因計算系統會議論文集》，494。https://doi.org/10.1145/3290605.3300724

Glatt, Z. (2023). The intimacy triple bind: Structural inequalities and relational labour in the influencer industry. European Journal of Cultural Studies. https://doi.org/10.1177/13675494231194156
Glatt, Z. (2023)。親密三重綁架：影響者產業中的結構性不平等與關係勞動。《歐洲文化研究期刊》。https://doi.org/10.1177/13675494231194156

Guiñez-Cabrera, N. and Aqueveque, C. (2021), “Entrepreneurial influencers and influential entrepreneurs: two sides of the same coin”, International Journal of Entrepreneurial Behaviour & Research. https://doi.org/10.1108/IJEBR-08-2020-0563
Guiñez-Cabrera, N. 與 Aqueveque, C. (2021)。「創業型影響者與具影響力的企業家：同一枚硬幣的兩面」，《國際創業行為與研究期刊》。https://doi.org/10.1108/IJEBR-08-2020-0563

Guo, S., Xu, K., Zhao, R., Gotz, D., Zha, H., & Cao, N. (2018). EventThread: Visual Summarization and Stage Analysis of Event Sequence Data. IEEE Transactions on Visualization and Computer Graphics, 24(1), 56–65. https://doi.org/10.1109/TVCG.2017.2745320
Guo, S., Xu, K., Zhao, R., Gotz, D., Zha, H., & Cao, N. (2018)。EventThread：事件序列資料的視覺摘要與階段分析。《IEEE 視覺化與電腦圖學匯刊》，24(1)，56–65。https://doi.org/10.1109/TVCG.2017.2745320

Hunter, S. T., Blocker, L. D., Gutworth, M. B., & Allen, J. (2022). Why we Support some Original Ideas but Reject Others: An Application of Signaling Theory. Journal of Creative Behavior, 57(2), 199–220. https://doi.org/10.1002/jocb.570
Hunter, S. T., Blocker, L. D., Gutworth, M. B., 與 Allen, J.（2022）。我們為何支持某些原創想法卻拒絕其他想法：訊號理論的應用。《創意行為期刊》，57(2)，199–220。https://doi.org/10.1002/jocb.570

Jago, A. S., & Carroll, G. R. (2023). Who Made This? Algorithms and Authorship Credit. Personality and Social Psychology Bulletin, 1461672221149815. https://doi.org/10.1177/01461672221149815
Jago, A. S., 與 Carroll, G. R.（2023）。這是誰做的？演算法與作者權利。《人格與社會心理學通訊》，1461672221149815。https://doi.org/10.1177/01461672221149815

Janson, M. P., & Dickhäuser, O. (2023). More Fit, Less Load? Cognitive Processes of Feedback Framed toward Own Preferences. Journal of Experimental Education. https://doi.org/10.1080/00220973.2023.2287453
Janson, M. P., & Dickhäuser, O. (2023)。更合適，負擔更輕？針對自身偏好框架化的回饋之認知過程。《實驗教育期刊》。https://doi.org/10.1080/00220973.2023.2287453

Jeitschko, T. D., & Tremblay, M. J. (2020). Platform competition with endogenous homing. International Economic Review, 61(3), 1281–1305. https://doi.org/10.1111/IERE.12457
Jeitschko, T. D., & Tremblay, M. J. (2020)。具有內生性歸屬的平臺競爭。《國際經濟評論》，61(3)，1281–1305。https://doi.org/10.1111/IERE.12457

Kahneman, D., & Tversky, A. (1984). Choices, values, and frames. American Psychologist, 39(4), 341–350. https://doi.org/10.1037/0003-066X.39.4.341
Kahneman, D., & Tversky, A. (1984)。選擇、價值與框架。《美國心理學家》，39(4)，341–350。https://doi.org/10.1037/0003-066X.39.4.341

Kerrigan, S., Grushka, K., Chand, A., Street, K., Shadbolt, J., & Lawry, M. (2022). Creative industries careers: shifting aspirations and pathways from high school to university—a NSW case study. Australian Educational Researcher. https://doi.org/10.1007/s13384-022-00574-9
Kerrigan, S., Grushka, K., Chand, A., Street, K., Shadbolt, J., & Lawry, M. (2022)。創意產業職涯：從高中到大學的志向與路徑轉變—新南威爾斯州案例研究。《澳洲教育研究者》。https://doi.org/10.1007/s13384-022-00574-9

Key-DeLyria, S. E., Bodner, T., & Altmann, L. J. P. (2019). Rapid Serial Visual Presentation Interacts with Ambiguity During Sentence Comprehension. Journal of Psycholinguistic Research, 48(3), 665–682. https://doi.org/10.1007/S10936-018-09624-9
Key-DeLyria, S. E., Bodner, T., & Altmann, L. J. P. (2019)。快速連續視覺呈現與句子理解過程中的模糊性交互作用。《心理語言學研究期刊》，48(3)，665–682。https://doi.org/10.1007/S10936-018-09624-9

Kotturi, Y., Yu, J., Khadpe, P., Gatz, E., Zheng, H., Fox, S., & Kulkarni, C. (2024). Peerdea: Co-Designing a Peer Support Platform with Creative Entrepreneurs. Proceedings of the ACM on Human-Computer Interaction, 8(CSCW1), 1–24. https://doi.org/10.1145/3637384
Kotturi, Y., Yu, J., Khadpe, P., Gatz, E., Zheng, H., Fox, S., & Kulkarni, C. (2024)。Peerdea：與創意創業者共同設計的同儕支持平台。《ACM 人機互動論文集》，8(CSCW1), 1–24。https://doi.org/10.1145/3637384

Kruikemeier, S., Boerman, S. C., & Bol, N. (2021). How Algorithmic Systems Changed Communication in a Digital Society. Media and Communication, 9(4), 116–119. https://doi.org/10.17645/MAC.V9I4.5005
Kruikemeier, S., Boerman, S. C., & Bol, N. (2021)。演算法系統如何改變數位社會中的溝通。《媒體與傳播》，9(4), 116–119。https://doi.org/10.17645/MAC.V9I4.5005

Lee, C., & Jin, K. (2024). Toward the Dynamic Relationship Between AI Transparency and Trust in AI: A Case Study on ChatGPT. International Journal of Human-Computer Interaction, 1–18. https://doi.org/10.1080/10447318.2024.2405266
Lee, C., & Jin, K. (2024)。邁向 AI 透明度與對 AI 信任之動態關係：以 ChatGPT 為案例研究。《國際人機互動期刊》，1–18。https://doi.org/10.1080/10447318.2024.2405266

Lee, S., Chang, Y., Lee, O.-K. D., Ryu, S., & Yin, Q. (2024). Exploring online social platform affordances for digital creators: a multi-method approach using qualitative and configurational analysis. Industrial Management and Data Systems, 124(4), 1501–1530. https://doi.org/10.1108/imds-12-2023-0951
Lee, S., Chang, Y., Lee, O.-K. D., Ryu, S., & Yin, Q. (2024)。探索數位創作者的線上社交平台可供性：採用質性與組態分析的多方法研究。《產業管理與資料系統》，124(4)，1501–1530。https://doi.org/10.1108/imds-12-2023-0951

Lisi, M., Mongillo, G., Mongillo, G., Milne, G., Dekker, T. M., & Gorea, A. (2021). Discrete confidence levels revealed by sequential decisions. Nature Human Behaviour, 5(2), 273–280. https://doi.org/10.1038/S41562-020-00953-1
Lisi, M., Mongillo, G., Mongillo, G., Milne, G., Dekker, T. M., & Gorea, A. (2021)。透過序列決策揭示的離散信心水準。《自然人類行為》，5(2)，273–280。https://doi.org/10.1038/S41562-020-00953-1

Ma, R., Gui, X., & Kou, Y. (2023). Multi-Platform Content Creation: The Configuration of Creator Ecology through Platform Prioritization, Content Synchronization, and Audience Management. International Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3544548.3581106
Ma, R., Gui, X., & Kou, Y. (2023)。多平台內容創作：透過平台優先排序、內容同步與觀眾管理構建創作者生態系統。國際人因計算系統會議。https://doi.org/10.1145/3544548.3581106

Maric, S., Schuessler, E. S., & Thäter, L. (2024). Neither Employment nor Self‐Employment: Avenues Towards Functional Platform‐Mediated Work Systems. British Journal of Industrial Relations. https://doi.org/10.1111/bjir.12855
Maric, S., Schuessler, E. S., & Thäter, L. (2024)。非雇傭亦非自僱：邁向功能性平台媒介工作系統的途徑。《英國產業關係期刊》。https://doi.org/10.1111/bjir.12855

Meng, Z. (2024). Cross-platform sensitivity and algorithmic adaptability: How transnational creators navigate algorithms across Chinese and US-based platforms. New Media & Society. https://doi.org/10.1177/14614448241307578
孟哲（2024）。跨平台敏感性與演算法適應性：跨國創作者如何在中國與美國平台間駕馭演算法。《新媒體與社會》。https://doi.org/10.1177/14614448241307578

Metoyer, R., Zhi, Q., Janczuk, B., & Scheirer, W. J. (2018). Coupling Story to Visualization: Using Textual Analysis as a Bridge Between Data and Interpretation. Intelligent User Interfaces, 503–507. https://doi.org/10.1145/3172944.3173007
Metoyer, R., Zhi, Q., Janczuk, B., & Scheirer, W. J.（2018）。將故事與視覺化結合：利用文本分析作為數據與詮釋之間的橋樑。《智慧使用者介面》，503–507。https://doi.org/10.1145/3172944.3173007

Moffett, J. W., Folse, J. A. G., & Palmatier, R. W. (2021). A theory of multiformat communication: mechanisms, dynamics, and strategies. Journal of the Academy of Marketing Science, 49(3), 1–21. https://doi.org/10.1007/S11747-020-00750-2
Moffett, J. W., Folse, J. A. G., & Palmatier, R. W.（2021）。多格式溝通理論：機制、動態與策略。《行銷科學學會期刊》，49(3), 1–21。https://doi.org/10.1007/S11747-020-00750-2

Möhlmannn, M., Salge, C., & Marabelli, M. (2023). Algorithm Sensemaking: How Platform Workers Make Sense of Algorithmic Management. Journal of the Association for Information Systems, 24(1), 8. https://doi.org/10.17705/1jais.00774
Möhlmannn, M., Salge, C., & Marabelli, M. (2023)。演算法意義建構：平台工作者如何理解演算法管理。《資訊系統學會期刊》，24(1)，8。https://doi.org/10.17705/1jais.00774

Murphy, D. H., & Knowlton, B. J. (2022). Framing effects in value-directed remembering. Memory & Cognition, 50(6), 1350–1361. https://doi.org/10.3758/s13421-022-01317-y
Murphy, D. H., & Knowlton, B. J. (2022)。價值導向記憶中的框架效應。《記憶與認知》，50(6)，1350–1361。https://doi.org/10.3758/s13421-022-01317-y

Nefedov, N. N. (2022). Explaining Why the Computer Says No: Algorithmic Transparency Affects the Perceived Trustworthiness of Automated Decision‐Making. Public Administration Review, 83(2), 241–262. https://doi.org/10.1111/puar.13483
Nefedov, N. N. (2022)。解釋為何電腦會說不：演算法透明度影響自動化決策的信任感知。《公共行政評論》，83(2)，241–262。https://doi.org/10.1111/puar.13483

Ostinelli, M., Bonezzi, A., & Lisjak, M. (2024). Unintended effects of algorithmic transparency: The mere prospect of an explanation can foster the illusion of understanding how an algorithm works. Journal of Consumer Psychology. https://doi.org/10.1002/jcpy.1416
Ostinelli, M., Bonezzi, A., & Lisjak, M. (2024)。演算法透明度的非預期效應：僅有解釋的可能性即可促進對演算法運作方式的理解錯覺。《消費者心理學期刊》。https://doi.org/10.1002/jcpy.1416

Pavleska, T., & Jerman Blažič, B. (2017). User bias in online trust systems: aligning the system designers’ intentions with the users’ expectations. Behaviour & Information Technology, 36(4), 404–421. https://doi.org/10.1080/0144929X.2016.1239761
Pavleska, T., & Jerman Blažič, B. (2017)。線上信任系統中的使用者偏誤：使系統設計者的意圖與使用者期望一致。《行為與資訊技術》，36(4)，404–421。https://doi.org/10.1080/0144929X.2016.1239761

Polk, E., & Diver, S. (2020). Situating the Scientist: Creating Inclusive Science Communication Through Equity Framing and Environmental Justice. Frontiers in Communication, 5. https://doi.org/10.3389/FCOMM.2020.00006
Polk, E., & Diver, S. (2020)。定位科學家：透過公平框架與環境正義創造包容性的科學傳播。《傳播前沿》，5。https://doi.org/10.3389/FCOMM.2020.00006

Ren, Q. (2024). Navigating the Creator Economy: An Analysis of Content Promotion and View Allocation Policies on Digital Content Platforms. Marketing Science. https://doi.org/10.1287/mksc.2023.0573
任強（2024）。《駕馭創作者經濟：數位內容平台上的內容推廣與觀看分配政策分析》。行銷科學。https://doi.org/10.1287/mksc.2023.0573

Rospitasari, M., Pratiwi, N., & Zebua, H. (2023). Creator Culture: An Introduction to Global Social Media Entertainment. Social & Cultural Geography, 24, 1281–1282. https://doi.org/10.1080/14649365.2023.2173394
羅斯皮塔薩里，M.、普拉蒂維，N.、& 澤布亞，H.（2023）。《創作者文化：全球社群媒體娛樂導論》。社會與文化地理學，24，1281–1282。https://doi.org/10.1080/14649365.2023.2173394

Salamon, E. (2025). Peripheral creator labor: Navigating regional marginalization and resistance in social media entertainment. New Media & Society. https://doi.org/10.1177/14614448241308520
薩拉蒙，E.（2025）。《邊緣創作者勞動：社群媒體娛樂中區域邊緣化與抗爭的導航》。新媒體與社會。https://doi.org/10.1177/14614448241308520

Sherratt, T. N., & O’Neill, E. (2023). Signal detection models as contextual bandits. Royal Society Open Science, 10(6). https://doi.org/10.1098/rsos.230157
Sherratt, T. N., & O’Neill, E. (2023)。信號偵測模型作為情境強盜。皇家學會開放科學，10(6)。https://doi.org/10.1098/rsos.230157

Shin, H., Kim, D. Y., & Choi, S. W. (2024). Developing scale for affective response: anxiety, anger, depression. European Psychiatry, 67(S1), S527. https://doi.org/10.1192/j.eurpsy.2024.1093
Shin, H., Kim, D. Y., & Choi, S. W. (2024)。情感反應量表的發展：焦慮、憤怒、抑鬱。歐洲精神病學，67(S1)，S527。https://doi.org/10.1192/j.eurpsy.2024.1093

Slovic, P. (1987), “Perception of risk”, Science, Vol. 236 No. 4799, pp. 280-285. https://doi.org/10.1126/SCIENCE.3563507
Slovic, P. (1987)，“風險感知”，科學，236 卷第 4799 期，頁 280-285。https://doi.org/10.1126/SCIENCE.3563507

Song, H. J. (2023). On the Strength of Trust: The Moderating Role of Certainty in Judgments of Authorities. Communication Research, 009365022211501. https://doi.org/10.1177/00936502221150144
Song, H. J. (2023)。信任的強度：確定性在權威判斷中的調節作用。傳播研究，009365022211501。https://doi.org/10.1177/00936502221150144

Spence, M. (1973), Job market signaling, Quarterly Journal of Economics, Vol. 87 No. 3, pp. 355-374. https://doi.org/10.2307/1882010
Spence, M. (1973)，職場訊號，<i>經濟季刊</i>，第 87 卷第 3 期，頁 355-374。https://doi.org/10.2307/1882010

Stegeman, H. M., Are, C., & Poell, T. (2024). Strategic Invisibility: How Creators Manage the Risks and Constraints of Online Hyper(In)Visibility. Social Media and Society. https://doi.org/10.1177/20563051241244674
Stegeman, H. M., Are, C., & Poell, T. (2024)。策略性隱形：創作者如何管理線上超（不）可見性的風險與限制。<i>社群媒體與社會</i>。https://doi.org/10.1177/20563051241244674

Verwiebe, R., Buder, C., Weissmann, S., Osorio-Krauter, C., & Philipp, A. (2024). The algorithm is like a mercurial god”: Exploring content creators’ perception of algorithmic agency on YouTube. New Media & Society. https://doi.org/10.1177/14614448241307931
Verwiebe, R., Buder, C., Weissmann, S., Osorio-Krauter, C., & Philipp, A. (2024)。演算法如同善變的神祇：探討內容創作者對 YouTube 演算法代理性的感知。《新媒體與社會》。https://doi.org/10.1177/14614448241307931

Yang, Y., & Zhang, Y. (2024). Design of Human-Machine Integration System to Meet Diverse Interactive Tasks. International Journal of Human-Computer Interaction. https://doi.org/10.1080/10447318.2024.2307689
楊洋、張宇（2024）。滿足多樣互動任務的人機整合系統設計。《國際人機互動期刊》。https://doi.org/10.1080/10447318.2024.2307689

Ye, D., Liu, M. J., Luo, J., & Yannopoulou, N. (2022). How to Achieve Swift Resilience: the Role of Digital Innovation Enabled Mindfulness. Information Systems Frontiers, 1–23. https://doi.org/10.1007/s10796-021-10225-6
葉東、劉明傑、羅傑、Yannopoulou, N.（2022）。如何實現快速韌性：數位創新促成的正念角色。《資訊系統前沿》，1–23。https://doi.org/10.1007/s10796-021-10225-6

Ye, Z. Q., Huang, Q., & Krijnen, T. (2024). Douyin’s playful platform governance: Platform’s self-regulation and content creators’ participatory surveillance. International Journal of Cultural Studies. https://doi.org/10.1177/13678779241247065
葉志強、黃琦、Krijnen, T.（2024）。抖音的遊戲化平台治理：平台自我規範與內容創作者的參與式監督。《國際文化研究期刊》。https://doi.org/10.1177/13678779241247065

Figure 1 Interaction Effect of Framing and Platform Dependency on Perceived Risk
圖 1 框架與平台依賴性對風險感知的交互作用

Figure 2: Transparency × Dependency on Perceived Risk
圖 2：透明度 × 依賴性對風險感知的影響

Figure 3 Cognitive Load and Decision Confidence by Feedback Modality
圖 3 認知負荷與決策信心依反饋模式之比較

Figure 3：「Cognitive Load and Decision Confidence by Feedback Modality」
圖 3：「認知負荷與決策信心依反饋模式」

Table 1 Means (SD) for Perceived Risk and Entrepreneurial Intention by Framing and Platform Dependency (study 1)
表 1 按框架與平台依賴性分組的感知風險與創業意向平均值（標準差）（研究 1）

Table 2 Means (SD) for Platform Trust, Perceived Risk, and Entrepreneurial Intention by Transparency and Platform Dependency
表 2 依據透明度與平台依賴性之平台信任、感知風險與創業意圖之平均值（標準差）

Table 3 Means (SD) for Decision Confidence, Cognitive Load, Perceived Risk, and Entrepreneurial Intention by Feedback Modality
表 3 不同回饋模式下決策信心、認知負荷、感知風險與創業意圖之平均值（標準差）

Table 4. Hypothesis Testing Summary
表 4. 假設檢驗摘要

Appendix 1. Scenario Manipulation and Survey Items (Study 1)
附錄 1. 情境操作與調查項目（研究 1）

Framing Manipulation (Simulated Notification)
框架操作（模擬通知）

Negative Framing Version:
負面框架版本：
“Your recent performance has significantly declined. Engagement rates dropped by 42% this week. If this trend continues, your visibility and income may be at risk. Please take immediate action to avoid further losses.”
「您近期的表現顯著下滑。本週互動率下降了 42%。若此趨勢持續，您的曝光率與收入可能面臨風險。請立即採取行動以避免進一步損失。」

Positive Framing Version:
正向框架版本：
“Your engagement metrics show a temporary fluctuation. Short-term dips are common and offer an opportunity to recalibrate. With slight adjustments, your content can quickly regain upward momentum.”
「您的互動指標顯示出短暫波動。短期下滑屬常見現象，並提供了重新調整的機會。透過些微調整，您的內容能迅速恢復上升動能。」

Sample Survey Items
樣本調查題項

Perceived Risk (α = .86)
感知風險（α = .86）

"I feel uncertain about the future of my platform career."
「我對我的平台職涯的未來感到不確定。」

"This feedback makes me feel my career is at risk."
「這些回饋讓我覺得我的職涯有風險。」

Entrepreneurial Intention (α = .81)
創業意圖（α = .81）

"I intend to continue investing in content creation."
「我打算繼續投資於內容創作。」

"I will persist in growing my presence on this platform."
「我將持續在此平台上擴大我的影響力。」

Appendix 2. Transparency Manipulation and Trust Scale Items (Study 2)
附錄 2. 透明度操控與信任量表項目（研究 2）

High Transparency Version
高透明度版本

"The algorithm reduced visibility due to declining engagement among your target audience. This change is based on shifts in watch time and like-to-view ratios. Your predicted performance for next week is 18% lower unless adjustments are made."
演算法因目標觀眾參與度下降而降低了能見度。此變化基於觀看時間及按讚與觀看比率的變動。若未進行調整，預測下週表現將下降 18%。

Low Transparency Version
低透明度版本

"Your content visibility has changed. Please continue monitoring your dashboard for performance updates."
您的內容能見度已變化。請持續監控您的儀表板以獲取表現更新。

Trust in Platform Scale (α = .89)
平台規模信任度 (α = .89)
(adapted from McKnight et al., 2002)
（改編自 McKnight 等人，2002）

“I trust the platform to treat creators fairly.”
「我相信該平台會公平對待創作者。」

“The platform provides useful and understandable feedback.”
「該平台提供有用且易於理解的回饋。」

“I believe the platform cares about creators’ long-term success.”
「我相信該平台關心創作者的長期成功。」

Appendix 3. Feedback Modality Materials (Study 3)
附錄 3. 反饋模式材料（研究 3）

Numeric Version
數字版本

Table showing 7-day engagement, view count, and click-through rate.
表格顯示 7 天的參與度、觀看次數及點擊率。

Raw values only, e.g., “Engagement: 5.3%, CTR: 2.8%, Views: 4,230.”
僅原始數值，例如：「參與度：5.3%，點擊率：2.8%，觀看次數：4,230。」

Visual Version
視覺版本

Line chart of engagement trend
互動趨勢折線圖

Bar chart of daily views
每日觀看次數長條圖

Color-coded alerts (green: stable, red: decline)
色碼警示（綠色：穩定，紅色：下降）

Narrative Version
敘事版本

“Your engagement slightly decreased this week. Compared to last week’s peak, your views declined by 15%. This is a normal fluctuation. Consider uploading earlier in the day for higher reach.”
「您的互動本週略有下降。與上週高峰相比，您的觀看次數減少了 15%。這是正常的波動。建議您考慮在較早的時間上傳，以獲得更高的觸及率。」