The prosperity of online medical communities depends on the sustained participation of physicians, especially their knowledge contribution to the community. To encourage physicians’ contributions, online medical communities have implemented a content-charging model to provide economic rewards for physicians’ online knowledge activities. The success of platforms with similar revenue models, such as Quora and Zhihu Live, demonstrates that economic incentives enhance users’ performance in paid knowledge activities [1].

Providing incentives for physicians’ knowledge contributions in online medical communities is crucial for both physicians and patients. For physicians, offering channels to monetize their knowledge enhances their earnings from online contributions, increasing their willingness to share and efficiency in sharing knowledge. For patients, the availability of paid knowledge supplements their access to information, providing higher-quality and more specialized medical knowledge. Figure 1 illustrates the paid and free science popularization pages on the Haodf platform in China.

Incentive mechanisms are one of the most effective measures to encourage active participation in knowledge contribution within online communities. They hold significant value for the thriving development of online health communities. Effective incentive mechanisms can significantly enhance user engagement. Common online community incentives include economic incentives (such as monetary rewards and point systems) and social incentives (such as online reputation). This study focused on economic incentives, specifically manifested as patients paying fees to access paid articles published by physicians. The fees generated from this process directly translate into economic incentives for physicians. Thus, the core focus of this study was on the impact of economic incentives in the context of online health communities. Currently, paid knowledge and free knowledge coexist in online medical communities, and previous studies have found that economic incentives directly affect users’ knowledge contributions [2,3] and sustained participation [4]. However, it is also crucial to understand how economic incentives influence unpaid contributions to knowledge activities, which may affect overall community performance [5]. On the one hand, free content, which has no payment threshold and accumulates significantly, can reach a wider range of users and maintain user engagement on the platform. On the other hand, medical knowledge products are of public interest, and free medical knowledge provides a barrier-free channel for users to obtain medical information, effectively protecting the basic interests of patients. Therefore, if economic incentives result in physicians contributing less to free content, it would be detrimental to the long-term development of the community. Conversely, if economic incentives encourage physicians to further contribute to free content, it would significantly enhance physician-patient interactions within the community. Previous studies have confirmed that, in generalized knowledge communities, economic incentives have a spillover effect on nonincentive activities, boosting users’ overall participation in knowledge communities [5].

Although many scholars have conducted research in this field, several key issues are still unresolved. First, the findings from generalized knowledge communities may not apply to medical contexts. In online medical communities, physicians face significant pressures from offline diagnoses and clinical duties, limiting their time and energy for online knowledge services. Thus, the conclusions drawn from generalized knowledge communities may not be applicable. Second, there is no consensus on the spillover effects of economic incentives. Existing studies have not yet reached a unified conclusion on whether economic incentives have spillover effects on free service performance and the specific direction of the impact. The answer varies across different research contexts. In general, the impacts of participating in paid knowledge activities on contributions to free knowledge activities can be summarized as being both positive and negative. Positive impacts include (1) reputation establishment (engaging in free knowledge activities without thresholds can enhance reputation in the community and attract patients) [6], (2) community contribution (reciprocity motivates physicians to contribute to free knowledge activities due to the benefits received from paid activities) [7], and (3) exposure opportunity (the creation threshold of free knowledge is relatively low, and by sharing free knowledge, physicians can quickly obtain more exposure opportunities) [8]. Negative impacts include (1) the substitution effect (participation in paid knowledge activities leads to physicians’ inability to allocate sufficient time and energy for free knowledge activities) [9] and (2) the crowding-out effect (charging for knowledge activities weakens physicians’ intrinsic motivation to contribute freely, thus reducing the prosocial quality of physicians’ participation in free knowledge contribution) [10,11].

How does physicians’ participation in paid knowledge activities affect their contribution to free knowledge activities? For platforms, it is crucial to develop knowledge-paying models without harming the interests of the vast number of free knowledge-sharing groups and balance physicians’ free and paid knowledge contributions. This study examined the experience of a Chinese leading medical platform, Haodf Online. To clarify the differences in physicians’ free content contribution before and after economic incentives, this study used a quasi-experimental design that combines the propensity score matching (PSM) technique with the multi-period difference in differences (DID) analysis, and investigated on the spillover effect of paid knowledge services on physicians’ free knowledge services. The propensity score matching (PSM) method was used to construct comparable groups of paid and nonpaid physicians to solve the sample selection bias problem. The multi-period difference in differences (DID) method was used to compare the changes in physicians’ free content contribution before and after economic incentives. Its advantage lies in eliminating individual and time trends and accounting for policy stimuli across multiple time points. This research will contribute to the application of economic incentives in online medical communities.

Motivation theory is a fundamental framework for analyzing user behavior patterns in online communities [28-30], categorizing user motivations as intrinsic and extrinsic [31]. Intrinsic motivation stems from the inherent satisfaction or enjoyment derived from engaging in specific behaviors, serving as a core driving force that prompts users to participate spontaneously. In contrast, extrinsic motivation is associated with external rewards or incentives that prompt users to undertake particular actions due to external stimuli. Existing research identifies 2 possible pivotal effects. First, overemphasizing or primarily relying on extrinsic incentives can trigger a “crowding-out effect,” in which extrinsic incentives undermine or even replace intrinsic motivation. Second, extrinsic incentives can trigger a “spillover effect” if they aim to enhance individuals’ perceived competence without compromising their sense of autonomy. That is to say, when extrinsic incentives are regarded as acknowledgments of individual effort and contribution, they are more likely to bolster intrinsic motivation rather than diminish it [29,31].

This study focused on online health knowledge contribution, distinguishing between free and paid health knowledge contributions. In this case, physicians who participate in free knowledge contributions are primarily driven by intrinsic motivation, which arises from their passion for knowledge sharing, professional achievement, and social responsibility. Conversely, physicians’ involvement in paid knowledge contributions is largely influenced by extrinsic motivation, namely, economic incentives. Notably, given the typically strong prosocial orientation of physicians, their behavior in sharing knowledge often stems from a robust intrinsic motivation. Thus, we believe that the rewards from paid contributions not only elevate their sense of achievement and self-confidence but also further enhance their intrinsic motivation, prompting them to engage more actively in unpaid knowledge contributions. Moreover, receiving economic incentives may prompt physicians to devote additional time and effort to free knowledge activities to avoid being perceived as solely motivated by economic incentives and establish a positive reputation [23]. Consequently, we hypothesize that, on online medical platforms, physicians’ engagement in paid knowledge activities will produce a spillover effect on their free knowledge activities rather than a crowding-out effect. This effect is manifested through a significant increase in the quantity of free knowledge contributions, notable improvements in their quality, and an expansion of content diversity. To evaluate this, this study developed measurement indexes for content contributions from 3 dimensions: quantity, quality, and diversity. Quantity directly measures the frequency of physicians’ participation in knowledge services, serving as the most fundamental evaluation metric. It reflects the physicians’ enthusiasm and level of engagement in knowledge contribution. Quality pertains to the professionalism and accuracy of the physicians’ knowledge service content, with high-quality services meeting readers’ medical needs. Diversity assesses the richness of the physicians’ knowledge services, evaluating whether they provide a wide range of content in various formats to address different issues, emphasizing variety and comprehensiveness.

The quantity of physicians’ free article contributions directly reflects the intensity of physicians’ online knowledge contributions, which is crucial for patients to evaluate physicians’ online performance. On the basis of this, this study proposed that participation in paid knowledge activities has a positive spillover effect on the quantity of free content contributions made by physicians (hypothesis 1).

As important as quantity, the quality of free articles is also vital for the sustainable development of online medical platforms. By writing high-quality articles and showcasing their profound medical knowledge and rich clinical experience, physicians can establish a professional image and win the trust and respect of patients. High-quality content can attract more user attention and participation, accumulate a good reputation and reputation for the platform, and lay a solid foundation for future sustainable development.

Whether physicians will improve the quality of free content when they receive economic incentives has not yet been conclusive. Some studies have found that small incentives have a negative impact on contribution quality [32], such as reducing the length of comments [33] and conveying one-sided emotions [33,34], which is due to the weakening of intrinsic motivation. However, from the perspective of reputation building, improving the quality of free content helps physicians establish a reputation for quality content services in the community, thus reducing patients’ uncertainty before purchasing content, which will have a positive impact on improving physicians’ paid services. On the basis of this, this study proposed that participation in paid knowledge activities has a positive spillover effect on the quality of free content contributions by physicians (hypothesis 2).

Diversity is the distribution of differences in certain attributes among members [35]. Diversity is important for community performance, such as user diversity promoting community learning [36] and improving content collaboration efficiency and outcomes [35]. Question-and-answer communities that encourage participants to answer more new questions have a good performance in answering questions, creating long-term value and improving difficulty resolution.

Diversity in online medical communities can be divided into media diversity and content diversity. Media diversity refers to the diversity of media adopted by physicians to spread medical knowledge. High media diversity indicates that physicians have adopted more ways to participate in knowledge contribution. Richer media are more likely to attract consumers’ attention and affect their evaluation of content [37]. The existence of economic incentives may encourage physicians to adopt more diverse means to present content to patients.

Content diversity refers to the differences in vocabulary involved in the creation of medical knowledge by physicians. These differences in vocabulary can affect the content richness of the physicians’ articles. When the content diversity of an article is high, the article may have richer and more comprehensive explanations [38]. Research has found that economic incentives increase the content diversity of online community discussions [22]. On the basis of this, this study proposed that participation in paid knowledge activities has a positive spillover effect on the diversity of free content contributions by physicians (hypothesis 3).

The research model is shown in Figure 2.

This study used the health knowledge column on the Haodf Online platform [39] as the data source. Haodf is one of the largest online medical websites in China, where many physicians provide diagnosis and treatment consultation services.

The paid medical knowledge service of the Haodf Online platform was introduced in 2016. The observation period for this study spanned 2016 to 2021. Due to the limited participation of physicians in knowledge activities in the short term, short intervals cannot effectively capture physicians’ knowledge activities and trends. Therefore, this study examined the impact of economic incentives on physicians’ knowledge activities over a relatively long period. Our time frame is divided into 6-month intervals [6], with the first interval covering data from January 1, 2016, to May 31, 2016, designated as spell 1. Each phase was set to 6 months to allow for a more accurate capture and analysis of the trends in physicians’ participation in online knowledge activities. Finally, the entire observation window is divided into 12 spells.

First, we comprehensively and systematically collected all the articles shown in the “Popular Science and Knowledge” section of the Haodf Online platform, totaling 63,151 articles. Subsequently, the authors of these articles, namely, the physicians, were the focus of our study. These physicians come from various departments covering the major medical specialties on the platform. After data cleaning, which involved removing physicians who had left the community and whose basic information could no longer be accessed, a total of 14,459 physicians were retained. Of these 14,459 physicians, 2070 (14.32%) participated in paid knowledge activities from 2016 to 2021, which constitutes the focal sample of this study. Using PSM with a 1:3 nearest neighbor matching technique, we matched physicians who participated in paid contributions with those who did not on a period-by-period basis, resulting in a comparison sample of 7453 physicians. Within this group of 7453 physicians, 2070 (27.77%) attempted to publish paid articles (although not necessarily in every period), whereas 5383 (72.23%) only published free articles. After conducting PSM on a period-by-period basis over the 6-year observation window, a total of 15,820 physician samples were obtained, including 6617 (41.83%) physician samples who participated in paid knowledge activities and 9203 (58.17%) physician samples who participated in free knowledge activities. The sample in this study is representative of the overall physician population in the Haodf Online medical community.

This study used a quasi-experimental design that combines PSM with DID to study the impact of physicians’ participation in paid knowledge activities on their performance in free knowledge activities.

This study did not require ethics approval as the authors only used publicly available content on the Haodf website and these data did not involve patient information.

In this study, we used the one-to-three nearest neighbor matching method with replacement for PSM [63]. In addition, in view of the time distribution of the articles published by physicians in each half-year span from 2016 to 2021, the year-by-year matching method (each spell as a year, representing half a year in our case) was adopted to find the matching control group for the treatment group in each spell. Taking the first period (ie, the first half of 2016) as an example, the predicted probability value of each physician’s engagement in free knowledge activities was first calculated in combination with the observable matching variables, and then 3 control group physicians who were not involved in paid knowledge activities were found for each physician participating in paid knowledge activities (the treatment group). After matching, physicians who were not successfully matched were excluded, resulting in 336 physicians in the treatment group for the first period. This method was consistently applied to subsequent periods, as shown in Table 2. After 12 periods, we obtained a total of 15,820 samples.

Following PSM, we conducted a balance test using the 2-tailed t test method, and the results are presented in Table 3. The test results in Table 3 show that the absolute values of the bias of all matching variables were basically <5% after matching, indicating that the matching variables and matching methods selected in this study were reasonable. In addition, the t statistics after matching were not significant, suggesting no significant differences between the treatment and control groups after matching. Therefore, the samples obtained after matching ensure the randomness of sample processing and the reliability of the estimated results in this study. After PSM, the kernel density plots of the treatment group and the control group are closer to each other (as shown in Figure 3), indicating that the matching process successfully reduced the differences in covariates between the two groups, achieving better balance.

bV(C): variance in control group.

After PSM, we conducted descriptive statistical analyses of the data (Table 4). The descriptive statistics revealed that the participation rate of physicians in paid knowledge services was 30.3% (4793/15,820) across all physician-period combinations. The median of 0 (IQR 0.000-1.000) indicates that the overall sample had a low participation rate in paid knowledge services. Furthermore, the data distribution of physicians within the sample shows significant variation in the dimensions of quantity, quality, and diversity of knowledge services, indicating potential areas for further improvement.

On the basis of the PSM estimation, this study identified a new control group with similar characteristics to those of the treatment group. The treatment and control groups encompassed a total of 7453 physicians and 15,820 samples. For these matched samples, we constructed dependent variables from the dimensions of quantity, quality, and diversity. The multi-period DID method was used to examine the impact of physicians’ engagement in paid knowledge activities on their participation in free knowledge activities. The detailed estimation results are shown in Tables 5 and 6. Table 5 illustrates the pure treatment effects on the dimensions of quantity, quality, and diversity of various dependent variables derived from a multi-period DID analysis after eliminating confounding factors. To validate the robustness of these findings, Table 6 incorporates additional regression analyses with control variables. The results of Table 6 are largely consistent with those of Table 5, thereby strengthening the reliability and validity of the treatment effects.

In the quantity dimension, model 10 shows that physicians’ participation in paid knowledge activities led to a significant 473.1% increase in the number of free articles (P<.001). This substantial increase aligns with the findings of previous research [1,6]. The consumer’s knowledge payment behavior helps to promote the contribution of free knowledge.

According to Table 6, in the dimension of quality, the participation in paid knowledge activities significantly promoted physicians’ performance in terms of content length, frequency of functional words, and frequency of causal words. The result of model 11 demonstrates that physicians’ participation in paid knowledge activities had a positive spillover effect on the length of free articles (108%; P=.009), which is inconsistent with the conclusion that economic incentives do not affect the quality of free services in previous study [6,22]. This may be due to physicians’ desire to enhance the online reputation of their free content and establish a good online reputation, thereby attracting more traffic to their paid services. The result of model 12 indicates that physicians’ participation in paid knowledge activities had a negative effect on the consistency between titles and content in free articles, with a reduction of 115.5% (P<.001). This is inconsistent with our initial hypothesis. Further investigation on examples of lower similarity scores is shown in Table S1 in Multimedia Appendix 1. It indicates that physicians often use eye-catching expressions and summaries in titles to engage readers, which may lead to less alignment between titles and content. We further conducted supplementary analyses (Table S2 in Multimedia Appendix 1). First, a t test comparing the consistency between titles and content in free versus paid knowledge after physicians engaged in paid knowledge activities was conducted. The results revealed that, within 12 months of initiating paid activities, the consistency between titles and content in free knowledge services remained significantly higher than in paid knowledge services. This may be because paid knowledge products require more attention-grabbing and creative titles to enhance click-through and conversion rates, thereby reducing consistency between titles and content. In the short term, physicians might also adjust the titles of their free knowledge services, resulting in a decrease in title-content consistency for free services after adopting the paid service, although consistency remained higher than for paid services. Second, over the 12 months following the initiation of the paid service, consistency between titles and content in both free and paid knowledge services gradually improved. This trend may be attributed to physicians accumulating more experience in providing knowledge services over time, enabling them to better balance creativity and content, thereby progressively enhancing the consistency between titles and content. The result of model 13 shows that physicians’ participation in paid knowledge activities had a positive effect on the frequency of functional words in free articles, which increased by 0.6% (P=.001). Physicians adopt different language styles in the expression of free content, which may be a differentiation strategy to distinguish paid content from free content. Model 14 indicates that there was no significant effect on the frequency of health-related words in free articles. Similarly, the result of model 15 shows no significant effect on the frequency of biological words. The changes in health word and biological word frequency before and after receiving the economic incentive were not significant, which shows that physicians did not pursue the output of professional content, possibly to maintain the universal positioning of free content. The result of model 16 reveals that physicians’ participation in paid knowledge activities positively impacted the frequency of causal words in free articles, which increased by 0.2% (P<.001). Physicians increased the frequency of causal words, which may be to narrate medical content with greater logic.

In the dimension of diversity, participation in paid knowledge activities had a significant impact on the content information entropy of free knowledge services. The result of model 8 indicates that physicians’ participation in paid knowledge activities had a positive impact on the media diversity of free articles, but the significance level was relatively low (P=.02). However, model 17 shows that physicians’ participation in paid knowledge activities had no significant impact on the media diversity of free articles. This discrepancy may arise from path dependency in physicians’ choice of media as they tended to continue using previously used media, leading to minimal variation in media use before and after engaging in paid knowledge activities. The result of model 18 indicates that physicians’ participation in paid knowledge activities had a positive effect on the content information entropy of free articles, which increased by 6.6% (P=.006). This finding aligns with the conclusion by Wang et al [6]. Economic incentives encourage physicians to produce more diverse and rich content to the community, contributing to its vitality and development.

Overall, the results demonstrate that the physicians’ participation in paid knowledge activities significantly enhanced their engagement in free knowledge activities. In terms of quantity, participation in paid knowledge activities substantially increased the number of free articles published by physicians. In terms of quality, paid knowledge participation had a significant positive impact on the length, function word frequency, and causal word frequency of free articles while negatively affecting the consistency between title and content. It had no significant impact on the frequency of health-related and biological words. In terms of diversity, paid knowledge participation significantly improved the content information entropy of free articles but did not notably affect media diversity.

The research question addressed in this study was the impact of economic incentives on physicians’ free knowledge contribution in online medical communities. To elucidate this issue, this study used a quasi-experimental method combining PSM with multi-period DID to verify the spillover effect of knowledge payment activities on free knowledge contributions across dimensions of quantity, quality, and diversity. The sample selection bias problem was controlled through PSM. The findings revealed that economic incentives exerted spillover effects on physicians’ free knowledge contributions, affecting all 3 dimensions. Specifically, economic incentives had a positive spillover effect on the quantity, length, and frequency of functional words; frequency of causal words; and content information entropy of free articles while negatively impacting the consistency between titles and content of free articles.

Theoretically, this study enriches understanding of economic incentives within the realm of online health care and explores physicians’ online knowledge contributions from a multidimensional perspective. The contributions of this study are as follows. First, this study comprehensively explored the impact of economic incentives on physicians’ free online medical knowledge contributions, enriching the literature on digital health care and knowledge contribution. Previous studies have mostly focused on nonspecific fields or broad knowledge-sharing platforms [5], with few investigating the motivational mechanisms for knowledge contribution in the specific context of health care. By introducing motivation theory, this study validated the spillover effect of paid knowledge contributions on free knowledge contributions by physicians, providing a theoretical basis and decision support for the sustainable and orderly development of paid knowledge services in online medical communities. Second, this study constructed a comprehensive, 3D knowledge evaluation framework. Previous research has primarily focused on the quantity and quality of knowledge [31,34]. This study further introduced the important dimension of diversity. In terms of measurement methods, earlier research has largely relied on structured data. In contrast, this study used text-mining techniques to extract information related to knowledge quality, such as the consistency between titles and content, function words, health-related vocabulary, physiological process terms, causal words, and entropy values related to knowledge diversity. Third, this study used a year-by-year PSM and a multi-period DID method to scientifically evaluate the impact of economic incentive models on the platform. By constructing a counterfactual framework, the PSM method was used to introduce physicians who did not participate in paid knowledge contributions as a control group, scientifically assessing the impact of participation in paid knowledge contributions on physicians’ free knowledge activities. Simultaneously, the year-by-year DID reduced the selection bias of physicians who published paid content at different time points and, in combination with text-mining methods, explored the impact of economic incentives on physicians’ online knowledge participation on the platform.

This study also has certain practical implications. First, for the platform, providing economic incentives to physicians can promote the overall development of the platform’s knowledge content ecosystem. The introduction of economic incentives can be a feasible business model for knowledge platforms to stimulate users’ contributions in paid and related unpaid activities. Currently, some free medical knowledge–sharing platforms, such as Dingxiangyuan and Kuaishou Health, have introduced paid knowledge through paid courses or short videos. This study holds significant reference value and strategic guidance for these platforms. In addition, this study also provides guidance for nonpaid medical knowledge platforms such as Chunyu Doctor and Baidu Health, offering valuable insights and strategic direction for introducing brand-new paid knowledge models. For paid knowledge platforms in other countries that operate on a membership basis or support one-time payments, such as medical-related communities on Patreon and medical-related modules on Substack, the research framework and related conclusions of this study can also provide valuable insights and references. Second, regarding physicians, they need to maintain their attention and investment in free knowledge activities and effectively manage them as a traffic entrance for paid users. Third, for patients, economic incentives have spillover effects on their free experience, helping them enjoy more, better, and more diverse knowledge articles for free.

There are still some limitations to this study, which can be studied in depth in the future. First, the observation period was relatively short. This study only chose half a year as an observation period for the release of paid articles, primarily focusing on the medium- to long-term economic incentive effects. In the future, the observation period can be extended to study the long-term changes in spillover effects and further explore the impact of economic incentives on physicians’ long-term professional development and knowledge-sharing behaviors. Second, in addition to economic incentives, the influence of noneconomic incentives (such as reputation) on physicians’ online knowledge service behaviors is a significant topic for future research. In the future, further in-depth exploration can be conducted on the noneconomic incentive measures of the platform and their impact on physicians’ online knowledge service behaviors. Third, in terms of the presentation format of knowledge content, only the textual form of products was studied. In the future, we can use speech transcription–related tools to extract information from videos, speech, and other forms of products and fully consider all types of articles. Fourth, this study was based on an online medical platform in China. Given that cultural backgrounds would profoundly influence individuals’ values, motivational systems, and decision-making processes, the conclusions drawn from Chinese platforms may offer only limited applicability, and there are some limitations in applying these findings to other countries and regions. Future research should fully consider the influence of cultural backgrounds and conduct more in-depth studies in this regard.