Analysis of text factors impacting donation behavior in public welfare crowdfunding projects

Abstract

BACKGROUND:

The textual description of a public service crowdfunding project is an important factor influencing the audience’s donation behavior, but the existing studies on the textual characteristics of the project are rather scattered.

OBJECTIVE:

This paper attempts to systematically sort out the characteristics of project texts along the lines of linguistic and non-linguistic factors, clarifying the relationship between the characteristics of project texts, project sources, and social donation behavior.

METHODS:

Based on Aristotle’s persuasion theory, language factors are measured from three dimensions of appeal to personality, appeal to logic, and appeal to emotion, while other text features unrelated to persuasive language are classified as non-language factors. When discussing the influence path of linguistic and non-linguistic factors on donation behavior, this paper controls the project type to test the moderating role played by the identity characteristics of crowdfunding initiators.

RESULTS:

The results show that the use of personality-based language (third-person words), logic-based language (money words and quantitative words), and emotion-based language (tone of voice and negative emotions) all have a significant positive effect on the audience’s donation behavior, while the use of second-person words in personality-based language has a negative effect on donation behavior; the identity of the project initiator (project origin) plays a complex and diverse moderating role in the influence of project text features on donation behavior.

CONCLUSION:

There are obvious differences in the description of different text strategies adopted by the project initiator.

Keywords

Crowdfunding donation behavior persuasion theory textual analysis

Li Wei is a Professor at School of Economics and Management, Chongqing University of Posts and Telecommunications.

Yang Dongshan is a graduate student at School of Economics and Management, Chongqing University of Posts and Telecommunications.

Yuxin Sun is an undergraduate student at School of Economics and Management, Chongqing University of Posts and Telecommunications.

1 Introduction

Since 2005, the Ministry of Civil Affairs has been organizing the “China Charity Award”, which has played an important role in promoting the spirit of charity, spreading the culture of charity, and inspiring the community to participate in charity. In 2015, the Ministry of Civil Affairs and the Ministry of Human Resources and Social Security jointly issued the “Guidance on Establishing and Improving the Charity Recognition and Reward System” to guide and encourage social forces to actively carry out charitable activities to help the needy groups around them. In 2016, the Charity Law came into force, providing legal protection for public charity in China. This shows the importance the country attaches to it and the need for public welfare to attract more people to participate in it. The development of technology has allowed the internet to play an important role in social governance and building people’s livelihoods. The Internet has done a good job of bridging the digital divide, building bonds of trust with transparent public welfare, and advancing the development of charity with rational public welfare. In 2011, a few university students launched the first crowdfunding project in China, the “Smoke-free Ride”. Subsequently, various crowdfunding platforms have emerged and public welfare platforms have grown rapidly. By April 2021, the total amount of money raised on Tencent’s public welfare platform exceeded 11.7 billion yuan, helping more than 90,000 projects in total.

At this stage, most crowdfunding projects are released through crowdfunding platforms. The numerous release platforms inevitably bring about the problem of overload in the number of projects released and similarity in the way content is presented. For example, Tencent Public Welfare can have more than 5,000 projects in fundraising in a single day, and with other crowdfunding platforms releasing project information, facing such a large number of crowdfunding platforms poses a great difficulty for the community. A project usually includes text and images to describe the person’s situation, but due to the limitations of the platform, it is not possible to mark the key points described in the text, and the excessive and unfocused text descriptions put a lot of pressure on platform users to read, which in turn affects their donation behavior. How to systematically attract the public to crowdfunding through project text descriptions is an issue that needs to be studied by platform operators.

The traditional crowdfunding research mostly focused on the psychology of pro-social behavior of platform donors, the rise of online public welfare platforms, and crowdfunding research focused on project text description [1, 2], but the existing research suffers from the deficiencies of theoretical is diverse perspectives and relatively scattered research variables and indicators. This paper attempts to use language and non-language as logical threads to systematically sort out textual description features, reclassify mainstream variables and indicators more clearly, and improve the discernment of research findings. Specifically, the paper proposes to classify the textual features of crowdfunding projects into linguistic and non-linguistic factors, where the linguistic factors, which are diverse and complex in structure, are sorted out and measured using Aristotle’s theoretical framework of persuasion in three dimensions: appeal to personality, appeal to logic and appeal to emotion, while other textual features that are not related to persuasive language are classified as non-linguistic factors. In exploring the pathways through which linguistic and non-linguistic factors influence public giving behavior, the identity of the crowdfunding initiator is introduced and its moderating role in each pathway is examined, with the aim of discovering whether there should be differences in the textual description strategies appropriate for different types of project initiators.

2 Literature review

In traditional crowdfunding campaigns financial constraints, reputation, and reciprocity are the main factors influencing donation behavior. In the face of current crowdfunding projects on online platforms, scholars have analyzed and come up with a variety of influencing factors. In this paper, the following three aspects will be addressed.

In terms of the users of crowdfunding platforms, the psychology of donors can be divided into two directions egoism and altruism, with the former being motivated by the desire to reap the rewards of their actions and the latter being motivated by an intrinsic desire to help others, regardless of personal costs and benefits, and the joy of giving being a positive, self-directed, emotional response experienced through helping others, referred to as a “warm light” [3]. “ When comparing altruism (PSM) with public service motivation, piatak found that altruism had no effect on prosocial behavior after controlling PSM [4], while Jiao et al. gave the opposite conclusion [5], a plausible explanation for this may be that the charitable proj1cts studied are different and have different characteristics. Altruism can thus be found to be an important driver for donors to undertake giving activities. In terms of trust between donors and charity platforms, Kim et al. Believe that past behaviors have had a great impact on the willingness to donate, and the experience of individual past donation behaviors has lowered the entry threshold, which is conducive to donation [6]. Donors with more donation experience have higher loyalty to the public welfare platform and are more likely to have donation behavior. On the contrary, those with less donation experience have lower loyalty to the public welfare platform and lower willingness to donate. Yang et al. suggest that users show loyalty to specific social platforms and have a willingness to continue using them out of an emotional commitment to social media [7]. The relationship between the act of giving and the happiness gained from giving has been demonstrated by scholars, with results showing that giving increases a person’s happiness, and that people with a high level of happiness are more likely to develop empathy and are more likely to give [8]. Financial constraints can be understood as the personal financial pressure felt by donors when they develop a psychology of giving. Financial constraints experienced by potential donors limit their ability to donate [9]. In the current context of charity crowdfunding, people who are financially stressed are more inclined to re-share on social networks because it is a way to bring some support to the projects they care about without spendingmoney [10].

In terms of the textual characteristics of crowdfunding projects on platforms, many scholars have investigated whether the number of words can have a positive impact. More comments on a platform can reflect more useful information and motivate platform users to make decisions [11]. However, the text length is too long and the efficiency of information transmission is low, which will make it difficult for users to make choices and is not conducive to taking action [12]. As for the overall presentation of words, Yangzhijing believes that the presentation of information will have a certain impact on the effect of fund-raising. Taking language style as an independent variable, she designs an experiment by combining frame effect with linguistic style. The research shows that the combination of illustrative and positive frame can better stimulate people’s willingness to donate [13]. Linguistic styles were categorized into three types: descriptive, narrative, and pictorial. Pictorial is the use of pictures to convey information in a way that is visually expressive. In crowdfunding projects crowdfunders often place pictures in a prominent position in the hope that the pictures will convey information about the crowdfunding to the donor so that they can receive donations [14]; Nisbett’s research on the psychology of pictures as a way to provoke donors to give has shown that the vividness of pictures is more likely to provoke empathy in donors than the straightforwardness of words. The emotions of the people presented in the pictures can have a different emotional impact on the donor [15]. Narrative texts, on the other hand, focus on the description of the situation, with images, vividness, and insight being their main characteristics. Kazoleas notes that narrative texts are more vivid, attract more attention and interest, and create a more concrete picture in the mind of the donor [16]; Taylor et al. also suggest that donors are more likely to recognize the main messages in a case description, which can influence their behavior [17]. In terms of emotional transmission, emotional texts influence users’ consumption behavior; Majumdar’s findings suggest that emotional factors in crowdfunding projects can increase the success rate of fundraising on crowdfunding platforms [1]; Yang Zhijing’s research concluded that if the message framed in response is positive, this message will show positive results after the donation and have a positive impact on subsequent donors. Conversely, negative comments reflect a negative mindset after donation [13]; Ritzenhein conducted another in-depth study and concluded that emotional expressions are used more frequently than logical expressions in crowdfunding projects. However, compared to logical expressions, emotional expressions have a limited effect on crowdfunding results [18]; studies have been conducted on successful crowdfunding projects and found that successful crowdfunding projects use varying degrees of logicalexpressions in their language [19]; Wang Wei has also shown again that projects with a logical language style on crowdfunding platforms are more likely to be successful in raising funds [2].

3 Theoretical foundations and research hypotheses

3.1 Theoretical basis

In this paper, we classify the factors that influence the donation behavior of crowdfunding platform users into linguistic and non-linguistic factors. The cognitive information that donors obtain through the project text and the emotions they feel in the project

context is considered linguistic factors, while non-linguistic factors are features that are not related to the attractiveness of the project text information. Non-verbal factors include the amount of the goal, the length of the text, and the number of images.

Aristotle’s rhetorical theory states that the three means of persuasion –“appeal to personality, appeal to logic, and appeal to emotion” –are effective in changing the attitudes and behavior of the public. Ethos, the most effective means of persuasion, appeals to the credibility of another person’s speech. The speaker gains the trust and support of the audience by demonstrating his or her abilities, honors, experience, etc. In the case of crowdfunding projects, personality indicates the need for the initiator of the project to demonstrate his or her moral and intellectual level to the donor and lies in gaining the trust of the donor. Logic (Logos), appeals to human reasoning. The speaker demonstrates the logical rigor the presentation of the message by means of logical reasoning. The initiator of a crowdfunding project uses real examples such as relevant texts and objective data to prove his point and make his claim clear to the donor. Emotion (Pathos) appeals to the audience’s feelings. The speaker mobilizes the emotions of the audience during the speech, eliciting feelings of sympathy, compassion, etc. from the public. This paper, therefore, classifies linguistic factors into three categories, appeal to personality, appeal to logic, and appeal to emotion, based on Aristotle’s theory of persuasion. The research model for this paper is as follows:

Fig. 1

Research model.

3.2 Research hypotheses

3.2.1 Linguistic factor facet

The language factor, i.e. the information contained in the text content of a crowdfunding project and the way the information is described and characterized, such as the vividness, authenticity, and emotional expression of the text, is the primary channel for helpers to understand the needs of the project, generate recognition and form a willingness to donate Different language factor characteristics will bring readers different perceptions of trust and directly determine their autonomous donation behaviour.

The impact of appeal to personality language on donor behavior

According to Aristotle’s theory of persuasion, personality, as the most powerful means of persuasion, lays the foundation for all communication, and speakers should highlight their attributes and positive self-image to gain persuasion by conveying credibility. More indicators have been proposed by scholars for the measurement of appeal to personality, such as credibility (first-person words) [20], assertiveness (second-person words) [20], and credibility claims (the presence of a female message) [1], among which the widely recognized measure is mainly that of personal pronouns. In this paper, we also use the third and second person to measure the linguistic factors of the resort to personality category.

Durand’s study found that fundraising campaigns described in the third person presented more objective information about the recipient and therefore raised relatively more money [21]. Newman et al. found that liar communication was often characterized by the use of fewer third-person pronouns [22], while more truthful requests made greater use of the third person thereby gaining more trust and leading to an increased chance of donation [21]. By comparing experimental and control groups, Fergusson found that writing in the third person facilitated the expression of emotion and helped people understand the event better than writing in the first person [23]. In the context of charity crowdfunding, the use of third-person pronouns not only reduces the emotional stress of the recipient, but also provides a realistic description of the recipient’s situation, eliciting understanding and sympathy from the viewer, which in turn generates donations. Therefore, the following hypothesis is proposed. H1a: Third-person word rate has a positive effect on donation behavior.

Simmons et al. found that the use of second-person pronouns can be detrimental to kinship emotional expressions and the relationship between words, leading to more criticism and blame and not being able to persuade others well [24]. In comedy, second-person pronouns are often used by people of high status [25], and the more second-person pronouns are used, the more authoritative the dictatorship becomes [26]; however, in the context of educational crowdfunding, the use of second-person pronouns is not appropriate for the presentation of the recipient’s situation and is not conducive to persuading the donor, which can easily have a negative impact on giving behavior. Therefore, the following hypothesis is proposed. H1b: Second-person word rate has a negative effect on donation behavior.

The influence of appeal to logic language on donation behavior

Aristotle’s theory of persuasion states that a speaker can persuade an audience by appealing to human reason through a true description of the information situation, ensuring the clarity of his or her logic. More metrics have been proposed by scholars for the measurement of appeal to logic, such as argument structuredness (number of conjunctions) [20], the average number of words per sentence (second person words) [20], appeal to reason (money-related evidence) [1], and cause-and-effect words [2], among which the main widely recognized measures are logic-related words and money words, but the quantitative language makes it easier for the reader to understand the expression of the text. Therefore, this paper uses quantitative language and money words to measure the linguistic factor of recourse to logic.

The objective data in the text is not influenced by emotions, opinions or personal feelings and is quantifiable and measurable language. The availability of quantitative data on finance and money indirectly demonstrates the credibility, reliability and accuracy of the text and increases supporters’ confidence in the achievement of their goals [27]. Objective data on products has been shown in research on advertising and marketing to influence consumer attitudes [28]. The objective data contained in the text helps the donor’s understanding of the message of the project, which in turn increases the persuasive effect of the project text. In charitable giving, donors come from a wide range of industries and are more likely to be attracted to donation projects that contain specific data, as this gives the donor a deeper understanding of the recipient’s situation. Therefore, the following hypothesis is proposed. H2a: Quantitative language word rate has a positive effect on the number of donations.

The use of money-related terms has a significant impact on giving behavior. when Pamala et al. studied the impact of money attitudes and charitable giving, they found that giving increased when the cost of giving decreased [29]; narratives of potential money can show true information about recipients, increasing credibility and attracting more funding [1]; and Zhang et al. combined persuasion theory to analyse medical crowdfunding from different perspectives, and found that high-frequency monetary evidence facilitates the success of crowdfunding for healthcare [30]. In the context of philanthropic crowdfunding, monetary vocabulary can demonstrate relevant background information about the recipient, influence rational claims, and thus attract donations. Therefore, the following hypothesis is proposed.

H2b: Money word rate has a positive effect on donation behavior.

The impact of appeal to emotion-based language on donation behavior

According to Aristotle’s theory of persuasion, appeals to emotion influence giving behavior by igniting attitudes and moral values in potential donors, causing them to show empathy and mobilizing their positive emotions. More metrics have been proposed by scholars for the measurement of appeals to emotion, such as promising (positive words) [20], reverse motivation (negative words) [20], and emotional appeals (negative words) [1], among which the widely recognized measures are mainly positive and negative words, but tone words, as an expression of emotion, also contain a wealth of emotion from the author [31]. Therefore, in this paper, both tone words and negative emotional words are used to measure the linguistic factor of resorting to logical categories.

Hynd et al. define a tone word as an aspect of a text that indicates a relationship between the author and the reader, the content, and a possible association between personal tone and reader response [32]. Tone words in texts usually serve to enhance emotional color to better express characters’ feelings and enhance the tone of sentences, and can be divided into several broad categories of tone features: statements, questions, supplications and exclamations. Braber believes that the modal particle in German better identifies the mood of the sentence and contains the rich emotions that the speaker wants to express [31], and Tetlock et al. concluded that the tone of media content can capture aspects of a company that is difficult to quantify beyond its fundamentals and have an impact on its future earnings [33]. In the context of educational crowdfunding, the use of tone of voice can better convey the conveyed sentiment and donors are more likely to be sympathetic. Therefore, the following hypothesis is proposed. H3a: Tone word rate has a positive effect on donation behavior.

The story of a charity project usually drives the viewer’s emotions, and the vivid emotions bring the viewer into the experience of the story’s protagonist. Most charity projects contain negative emotional appeals, all of which express the sadness, pain, anxiety, and anger of the story’s protagonist. The emotional changes that occur in the project, sadness, doubt, anger, and guilt, can all motivate potential donors to give. Majumdar’s study of charitable giving on the RAOP platform found that negative emotional words in charity messages increased the likelihood of giving [1]; Xu et al. found that in the project text of medical crowdfunding, the use of tragic narrative to create a fragile image will arouse the sympathy of donors and lead to donation behavior [34]. In the context of educational crowdfunding, negative emotive words will convey the difficult situation of the protagonist, which in turn will elicit sympathy from the viewer and lead to donation behavior. Therefore, the following hypothesis is proposed. H3b: Negative emotional word rates have a negative facet effect on donation behavior.

3.2.2 Non-verbal factors

Non-verbal factors refer to other means and vehicles of information transmission other than the content of the project text (verbal and textual), in order to investigate the influence of indirect factors on the number of donations. For example, the image information contained in the project, the length of the text, the setting of the crowdfunding amount, etc. In addition to the language factor, the inclusion of images can add a certain vividness and authenticity to the monotonous text descriptions. As an important part of a platform’s crowdfunding project description, the text should generally describe the reason for the fundraising, the purpose, and the extent of the project’s progress. The longer the text description, the more factually accurate and detailed the description can be to give potential donors a true picture of the project, and the greater the value of the project. Longer commentary information provides a detailed and objective narrative of the project information [11], increasing communication with potential supporters and attracting investment in individual projects [35]. Shorter text descriptions can lead to inaccurate information descriptions and discourage donors from giving [36]; too brief text descriptions may indicate that the project initiating organization has not prepared well enough for donors to obtain useful information from it, and they may perceive such donations as meaningless. In the context of charity crowdfunding, longer text items can describe the real situation of the recipients more clearly, reduce the information asymmetry between the recipients and donors, increase the credibility of the project and generate donation behavior. Therefore, the following hypothesis is proposed in this paper. H4: Text length in crowdfunding projects has a negative effect on donation behavior.

In charity crowdfunding, images can be more intuitive, and the combination of necessary text descriptions and images allows the viewer to access the information more quickly and directly than if it were purely text-based. The use of story images is considered to be a more powerful psychological cue to keep the donor’s attention on the issue and the protagonist to avoid unnecessary distraction. Pictures in projects convey vivid messages to donors [14]. Persuasive messages such as high-quality word pictures can influence participants’ investment behavior [1], and Nisbett’s research found that pictures are more likely to inspire donor empathy compared to words [15]. In the context of educational crowdfunding, the use of images will show the difficult plight of the project initiator, causing potential donors to sympathize and thus elicit donations. Therefore, this paper proposes the following hypothesis. H5: The number of images in a crowdfunding project has a positive effect on donation behavior.

Crowdfunding projects attract donors by describing the situation of the recipient, and the fundraising target varies from project to project. The higher the fundraising goal, the more difficult the background of the project initiator and the more help he or she needs, and the more likely it is that potential donors will feel compassionate. By examining the data collected in the context of product crowdfunding, Aleksina et al. found that for the same category of fundraising projects across different platforms, campaigns with smaller total project amounts raised more quickly [37], while the opposite was true for charity crowdfunding. In the context of charity crowdfunding, as potential donors prioritize recipients in more difficult living conditions, they will develop a preference for fundraising projects with higher target numbers to fund their fundraising goals. Therefore, the following hypothesis is proposed in this paper. H6: The target amount of a crowdfunding project has a positive effect on donation behavior.

In the field of sociology, Bromley argues that reputation represents the sum total of opinions about an entity (e.g. individual, group, organisation) [38]. Reputation is considered to be a moderator of other different contexts, such as consumer attitudes and behavioural responses, brand performance, customer satisfaction and loyalty. Building on the research of other scholars, Laird et al. examined the moderating effect of personal reputation on the perceived accountability-tension relationship and showed that a positive personal reputation leads to less perceived job tension and depression at work and higher job satisfaction, while those with a poorer personal reputation experienced the opposite results [39]. In their study of the social dimensions of online shops, Tractinsky et al. found that self-evaluation and others’ evaluations were critical. A high rating by others means that the online shop has a good reputation in the consumer’s mind and can motivate shopping activity [40]. Therefore, it is important for both individuals and organisations to have a good reputation.

In crowdfunding-based industries, both verbal and non-verbal factors in project descriptions have a significant impact on the donor decision-making process. As there is no material reward for investors’ monetary donations in charity crowdfunding, donors may express concerns about the reputation of the grantee and whether their donations are being misused. As a result, donors will evaluate to assess the credibility of a project before donating. Project credibility can be expressed in several ways, and in the context of philanthropic crowdfunding, the reputation of the originator is an important indicator for donors to assess the credibility of a project. Aristotle believed that persuasion is achieved through the personal character of the speaker, and that a person’s education, experience, public role and reputation conferred by publications qualify him to speak on an issue and make it credible to the audience. It is therefore necessary for donors to seek the reputation of the project initiator from the field of giving and to assess the credibility of the donated crowdfunding project in this respect. Thus, the reputation of the project initiator significantly enhances the influence of the persuasion factor on donor behavior. For this reason, the following hypothesis is proposed: H7: Project origin plays a positive moderating role in the relationship between linguistic factors (first person word rate (H7a), second person word rate (H7b), quantitative word rate (H7c), money word rate (H7d), structural word rate (H7e), tone word rate (H7f), and negative emotional word rate (H7g)) on donation behavior. H8: Non-verbal factors (text length (H8a), number of images (H8b), target amount (H8c)) positively moderated the relationship between item source and donation behavior.

4 Data acquisition and analysis

4.1 Data collection

In this paper, the Tencent Public Welfare Platform was chosen as the data source for the following reasons: Firstly, it has the largest number of projects and the most active users, which can ensure the adequacy of data. Secondly, the platform has the greatest national influence, which ensures the authenticity of the projects and the reliability of the data. With the help of Python software, we crawled 14,449 projects about education grants under Tencent Philanthropy from January 1, 2016 to June 30, 2020, and finally obtained 9,220 valid texts by filtering out the missing texts.

4.2 Variables and measurements

Dependent variable. The dependent variable selected for this paper is the audience’s donation behavior, measured by the cumulative number of donations to the crowdfunding project.

Explanatory variables. The explanatory variables were designed in terms of both linguistic and non-linguistic factors of the donation crowdfunding project text, where the linguistic factors include three dimensions of appeal to personality, appeal to logic and appeal to emotion, and the measurement variables used were: appeal to personality linguistic factor measured by third-person word rate and second-person word rate; appeal to logic linguistic factor measured by quantitative words, money words and structural words; appeal to tone words and negative The third-person word rate and second-person word rate were used to measure recourse to personality; quantifiers, money words and structure words were used to measure recourse to logic; and tone words and negative emotions were used to measure recourse to affect. In terms of how the variables are assigned, the paper uses the Dalian University of Technology’s ontology of emotion words and the C-LIWC dictionary. The third-person word rate is the ratio of the number of third person words appearing in the item body to the total number of words in the item body; The second-person word rate is the ratio of the number of second-person words occurring in the item body to the total number of words in the item body; The quantitative word rate is the ratio of the number of numeric occurrences in the body of the item to the total number of words in the body of the item; The money word rate is the ratio of the number of money words occurring in the item body to the total number of words in the item body; The tone word rate is the ratio of the number of auxiliary words occurring in the item body to the total number of words in the item body; The negative emotional word rate is the ratio of the number of negative emotional words occurring in the item body to the total number of words in the item body. For non-verbal factors, this paper uses three measurement variables: item text length, number of pictures and target donation amount, where text length refers to the total number of words in the item body content, number of pictures is the number of pictures contained in the item body, and target is the target amount size set by the item initiator.

Moderating variables. The moderating variable chosen in this paper is the identity characteristics of the originator of the crowdfunding project, and the variable is named project source. Based on the sample information we categorized the identity of the project initiator as a foundation (or non-profit organization) and an individual organization, with values of 1 and 0 respectively.

Control variables. As public service projects include different categories such as education, medical care, and people search, the characteristics of the corresponding donation projects and the influencing factors of the audience giving are likely to be different, so in order to improve the validity of this study, we controlled for the type of project and only chose education donation projects as the research object. The project texts collected for this study are strictly for the education category.

4.3 Descriptive statistics

Table 1 shows the descriptive statistical analysis of the data in this study. There are 9220 data objects, most of the projects on Tencent Public Welfare got donations, but there are also projects that did not get donations, the text description text of the projects is relatively long; most of the projects have less than 10 pictures; some of the donation projects did not set a target amount, most of the projects have a target amount of 100,000 or more.

Table 1
Descriptive statistical analysis

Variable N Average Median Standard deviation Min Max

Number of donations 9220 2440.705 458.500 11286.367 0.000 626859.000

Text length 9220 2073.554 1963.000 692.535 245.000 8234.000

Number of images 9220 8.203 7.000 5.076 0.000 74.000

Target amount 9220 376556.946 150000.000 1339258.942 10000 7676000.000

Third-person word rate 9220 0.006 0.005 0.005 0.000 0.047

Second-person word rate 9220 0.002 0.001 0.002 0.000 0.022

Quantifiable word rate 9220 0.034 0.033 0.009 0.003 0.125

Money word rate 9220 0.035 0.034 0.012 0.000 0.095

Tone word rate 9220 0.035 0.034 0.010 0.005 0.072

Negative emotional word rate 9220 0.007 0.006 0.004 0.000 0.058

Variable	N	Average	Median	Standard deviation	Min	Max
Number of donations	9220	2440.705	458.500	11286.367	0.000	626859.000
Text length	9220	2073.554	1963.000	692.535	245.000	8234.000
Number of images	9220	8.203	7.000	5.076	0.000	74.000
Target amount	9220	376556.946	150000.000	1339258.942	10000	7676000.000
Third-person word rate	9220	0.006	0.005	0.005	0.000	0.047
Second-person word rate	9220	0.002	0.001	0.002	0.000	0.022
Quantifiable word rate	9220	0.034	0.033	0.009	0.003	0.125
Money word rate	9220	0.035	0.034	0.012	0.000	0.095
Tone word rate	9220	0.035	0.034	0.010	0.005	0.072
Negative emotional word rate	9220	0.007	0.006	0.004	0.000	0.058

The results of the correlation coefficient analysis in Table 2 show that most of the variables have a significant effect on donation behaviour, while the correlation coefficients between the variables are below 0.5 and the maximum variance inflation factor VIF between the variables is 1.49, which does not exceed 10, so there is no multicollinearity between the variables in this study.

Table 2

Correlation coefficient table

Variable	1	2	3	4	5	6	7	8	9
Text length	1.000
Number of images	0.381	1.000
Target amount	0.072	0.032	1.000
Third-person word rate	0.029	0.024	–0.022	1.000
Second-person word rate	–0.031	0.025	–0.027	0.050	1.000
Quantifiable word rate	0.030	0.038	0.000	–0.172	0.045	1.000
Money word rate	–0.374	–0.248	0.030	–0.119	0.018	0.114	1.000
Tone word rate	0.059	0.118	–0.042	0.481	0.152	–0.255	–0.256	1.000
Negative emotional word rate	0.051	–0.010	–0.006	0.174	–0.002	–0.100	–0.156	0.200	1.000

4.4 Regression results

As the dependent variable is the number of donations, which is a count variable, an OLS model may lead to heteroskedasticity and thus a less accurate estimate of significance. Common models used for the analysis of count variables are negative binomial regression and Poisson regression. In this study, the variance of the number of donations is much larger than its mean, the dispersion is high and it follows a negative binomial regression, therefore a negative binomial model is appropriate for this study.

As can be seen from Table 3, compared with model 1, the absolute value and AIC value of log likelihood of model 2 are reduced, indicating that the overall goodness of fit of the model is improved. Therefore, we adopt the variable structure and regression results of model 2.First, in terms of language factors, the regression results show that: ding172 appealing to personality language has a significant impact on donation behavior, and there are reverse differences in the direction of influence between different personal pronouns. The increase of the third person word rate can improve the trust, understanding, and sympathy of helpers, and then promote donation behavior (β= 8.206, p < 0.1): However, the increase of the second person word rate will significantly reduce the donation behavior of the helpers to the project (β= –31.601, p < 0.01), which may be due to the fact that the second person conveys more assertive information, and the readers’ perception of such information is more negative or even disgusting. ding173 In terms of appealing to logical language, the project text contains more quantitative data (β= 11.468, p < 0.01) and money-related evidence (β= 14.059, p < 0.01), which will improve the helpers’ sense of authenticity and trust in the donation event, and then the donation behavior will occur, which means that the quantitative information will help the readers to make rational analysis and judgment, and then effectively improve the logical persuasion of the text description. ding174 In terms of appealing to emotional language, the more modal particles the project text contains, the more it can stimulate the emotional and perceived value of the helpers and make them produce donation behavior (β= 7.6948, p < 0.05): The more negative emotional words in the project text, the easier it is to stimulate the helpers’ compassion and compassion, and then produce a positive donation emotional tendency, thus enhancing the helpers’ donation behavior to the project (β= 15.976, p < 0.01). Secondly, in terms of non-verbal factors, the regression results show that the length of the project text, the number of pictures and the target amount will have a significant positive impact on the charitable donation behavior of the helpers.

Table 3
Regression results

Independent variable Model 1 Model 2

Linguistic factors Appeal to personality Third-person word rate 8.206^*

Second-person word rate –31.601^

Appeal to logic Quantifiable word rate 11.468^*

Money word rate 14.059^*

Appeal to emotion Tone word rate 7.6948^*

Negative emotional word rate 15.976^*

Non-verbal factors Text length 0.00025^* 0.00034^*

Number of images 0.0127^* 0.0161^*

Target amount 1.326^* 1.291^***

Log likelihood –75491.667 –75407.653

AIC 150993.3 150837.3

		Independent variable	Model 1	Model 2
Linguistic factors	Appeal to personality	Third-person word rate		8.206^*
		Second-person word rate		–31.601^**
	Appeal to logic	Quantifiable word rate		11.468^***
		Money word rate		14.059^***
	Appeal to emotion	Tone word rate		7.6948^***
		Negative emotional word rate		15.976^***
Non-verbal factors		Text length	0.00025^***	0.00034^***
		Number of images	0.0127^***	0.0161^***
		Target amount	1.326^***	1.291^***
Log likelihood		–75491.667	–75407.653
AIC		150993.3	150837.3

Note: ^*p < 0.1, ^**p < 0.05, ^***p < 0.01.

4.5 Moderating effect test

Add the interactive items from the project source on the basis of model 2, and the results are shown in Table 4, from which it can be concluded that item source strengthens the effects of the target amount (β= 0.57909, p < 0.01), third-person word rate (β= 27.01713, p < 0.01), second-person word rate (β= 274.7052, p < 0.01), modal particle rate (β= 27.25272, p < 0.01) and negative emotion word rate (β= 119.1711, p < 0.01) on donation behavior, but weakens the effects of the length of the text (β= –.0002547, p < 0.05) and money word rate (β= –23.73952, p < 0.05) on donation trips. However, the project source has no significant moderating effect on the influence path between the number of pictures, quantitative word rate and donation behavior. The above results show that project sources play a complex and diverse regulatory role in the process of the influence of linguistic and non-verbal factors on donation behavior, which means that there are obvious differences in the text description strategies suitable for project initiators with different identity characteristics when writing the text of public welfare projects.

Table 4
Analysis of moderating effects

Interaction term Significance Conclusion

Item Source^* Third-person word rate 27.01713^*** H7a holds

Item source^*Second-person word rate 274.7052^*** H7b holds

Item source^Quantitative word rate Not significant H7c does not hold

Item source^Money word rate –23.73952^*** H7d does not hold

Item source^*Tone word rate 27.25272^*** H7e holds

Item source^*Negative emotional word rate 20.83325^** H7f holds

Item source^*Text length –0.0002547^*** H8a does not hold

Item source ^* Number of images Not significant H8b does not hold

Item source ^* Target amount 0.57909^*** H8c holds

Interaction term	Significance	Conclusion
Item Source^* Third-person word rate	27.01713^***	H7a holds
Item source^*Second-person word rate	274.7052^***	H7b holds
Item source^*Quantitative word rate	Not significant	H7c does not hold
Item source^*Money word rate	–23.73952^***	H7d does not hold
Item source^*Tone word rate	27.25272^***	H7e holds
Item source^*Negative emotional word rate	20.83325^**	H7f holds
Item source^*Text length	–0.0002547^***	H8a does not hold
Item source ^* Number of images	Not significant	H8b does not hold
Item source ^* Target amount	0.57909^***	H8c holds

5 Conclusion and discussion

5.1 Conclusion

In this paper, we used textual analysis to study the influence of donation texts on audiences’ donation behavior in terms of both linguistic and non-linguistic factors. For the linguistic factors, which are diverse and complex, we used Aristotle’s model of persuasion theory to sort out and measure the variables in three dimensions: appeal to personality, appeal to logic and appeal to emotion.The theoretical model was tested with the help of texts obtained from Tencent’s crowdfunding platform on education donation projects.The main theoretical research results can be summarized into two aspects. First of all, in terms of the influence path of text features (linguistic and non linguistic factors) on donation behavior, the empirical test results show that appealing to personality language (third person words), appealing to logic language (money words and quantitative words) and appealing to emotional language (mood and negative emotion) have a significant positive impact on the donation behavior of the public.The second-person words in personality language have a negative effect on donation behavior;Among the nonverbal factors, the length of text, the number of pictures and the target amount all have a positive impact on the donation behavior, but the influence coefficients are relatively small. Secondly, by examining the moderating effect of project sources on the impact path, it is found that project sources have a positive moderating effect on the target amount, third person usage rate, second person usage rate, tone and negative emotion rate, while money words and text length have a negative moderating effect. This means that when designing the text content and writing style of crowdfunding projects, charitable organizations or public welfare platforms need to combine their own identity characteristics to choose more effective description strategies.

5.2 Management implication

Based on the above theoretical research findings, aiming at improving the donation behavior of the masses, this paper puts forward the following suggestions on the project text description strategy of the public welfare crowdfunding platform. Firstly, the initiator of the project can enhance the audience’s persuasive awareness of the text description from the following aspects: first, the third-person pronouns are used more in the text description and the second-person pronouns are reduced. The third-person pronouns can more truly reflect the real situation of the helpers. At the same time, the helpers can more truly feel the current plight of the helpers, while the second-person pronouns will make people feel unreal, Not conducive to donation; Secondly, the situation described in the project description of the public welfare platform should be as detailed as possible, and more pictures, objective data or money data should be used to increase the authenticity of the current dilemma. Appropriately increasing the amount of help for the project can better reflect the urgency of the current situation; Finally, add more mood particles to the text emotional description of the platform project. At the same time, negative emotional words can stimulate the inner sympathy and empathy of the helpers. Second, the project initiator selects an effective project text description style according to its own identity characteristics (differences in credibility). Among them, foundations and non-profit organizations with high social credibility can adopt more linguistic factors such as the third-person, linguistic auxiliary words and negative affective words, and appropriately increase non-linguistic factors such as the target amount, because the main effect is positive and the adjustment effect is positive; At the same time, try to avoid language factors such as second-person words (the main effect is negative and the adjustment effect is positive) and money words (the main effect is positive and the adjustment effect is negative) and non-language factors such as text length (the main effect is positive and the adjustment effect is negative); Individual organizations with low social credibility should strengthen the use of language factors such as the third-person words, language auxiliary words and negative emotion words, and reduce the use of second-person, money words and other factors, in order to improve the public’s attention to the project, persuasive perception and stimulate donation behavior.

5.3 Limitation and future expectation

The limitation of this paper is that it only reveals the direct path of influence of verbal and non-verbal factors on donation behavior, while the influence process of verbal and non-verbal factors as influencing factors stimulating donors is more complex. This paper lacks research on the mediating factors in the influence process, while the research on moderating factors only considers the moderating effect of the source of the item and lacks research on other moderating variables. In subsequent research, we will add financial ties as a moderating variable and motivation (internal and external motivation) as a mediating variable in the process of influence of textual characteristics on giving behavior, to enrich the research on the influence of textual characteristics on giving behavior.

Footnotes

Acknowledgments

We would like to thank Miss. Li for her guidance on this paper, and also the Chongqing Institute of Social Sciences Planning for its financial support of this paper.

Author contributions

CONCEPTION: Wei Li

METHODOLOGY: Wei Li

DATA COLLECTION: Dongshan Yang

INTERPRETATION OR ANALYSIS OF DATA: Dongshan Yang and Yuxin Sun

PREPARATION OF THE MANUSCRIPT: Dongshan Yang and Yuxin Sun

REVISION FOR IMPORTANT INTELLECTUAL CONTENT: Wei Li

SUPERVISION: Wei Li

References

Majumdar

, Bose

, My words for your pizza: An analysis of persuasive narratives in online crowdfunding, Information & Management 2018;55(6):781–94.

Wang

, Chen

, Zhu

, Wang

, The success rate of crowdfunding and the persuasiveness of language style - An Empirical Study Based on KickStarter, Management World 2016;(05):81–98.

Bekkers

, Wiepking

, A literature review of empirical studies of philanthropy: Eight mechanisms that drive charitable giving, Nonprofit and Voluntary Sector Quarterly 2011;40(5):924–73.

Piatak

, Holt

, Prosocial behaviors: A matter of altruism or public service motivation? Journal of Public Administration Research and Theory 2020;30(3):504–518.

Jiao

, Qian

, Liu

, et al., Why do people support online crowdfunding charities? A case study from China, Frontiers in Psychology 2021;12:130.

Kim

, Han

, The application of the theory of planned behavior to identify determinants of donation intention: Towards the comparative examination of positive and negative reputations of nonprofit organizations CEO, Sustainability 2020;12(21):9134.

Yang

, Hu

, Kpandika

, et al., Effects of social attachment on social media continuous usage intention: The mediating role of affective commitment, Human Systems Management 2021;40(4):619–31.

Lai

, Yang

, Mao

, et al., When do good deeds lead to good feelings? Eudaimonic orientation moderates the happiness benefits of prosocial behavior, International Journal of Environmental Research and Public Health 2020;17(11):4053.

Müller

, Lindenmeier

, Exploring the role of charitable ethnocentrism and donation motives in international giving: Empirical evidence from Germany. Journal of Philanthropy and Marketing. 2021:e1729.

10.

, He

, Song

, et al., Factors impacting donors’ intention to donate to charitable crowd-funding projects in China: A UTAUT-based model, Information, Communication & Society 2018;21(3):404–15.

11.

Tariq

, Tanveer

, Organic consumption through ICT: A moderated mediation model of consumer attitude and perceived irritation25J], Human Systems Management 2021;40(2):223–34.

12.

Liu

Research on the relationship between information quality of crowdfunding project plan and financing performance. Nanjing University; 2021.

13.

Yang

, The effect of framing effect and linguistic style on the effectiveness of microprofits fundraisings, Psychological Technology and Applications 2015;8:17–22.

14.

George

, Deborah

, The scarecrow and the tin man: The vicissitudes of human sympathy and caring, Review of General Psychology 2007;11(2):112–26.

15.

Nisbett

, Lee

Inference: Strategies and Shortcomings of Social Judgement. Upper Saddle River: Prentice Hall; 1980.

16.

Kazoleas

, A comparison of the persuasive effectiveness of qualitative versus quantitative evidence: A test of explanatory hypotheses, Communication Quarterly 1993;41(1):40–50.

17.

Taylor

, Thompson

, Stalking the elusive” vividness” effect, Psychological Review 1982;89(2):155.

18.

Ritzenhein

, Content analysis of fundraising letters, New Directions for Philanthropic Fundraising 1998;(22):23–36.

19.

Tirdatov

, Web-based crowd funding: Rhetoric of success, Technical Communication 2014;61(1):3–24.

20.

Qitoto , Liu

, Wang

, Zhou

, A study on the persuasive effect of the linguistic style of knowledge payment product descriptions - the moderating role of knowledge producers’ reputation, Nankai Management Review 2020;23(05):159–70.

21.

Durand

, Peters

, Eltorai

AEM

, et al., Medical crowdfunding for organ transplantation, Clinical Transplantation 2018;32(6):e13267.

22.

Newman

, Pennebaker

, Berry

, et al., Lying words: Predicting deception from linguistic styles, Personality and Social Psychology Bulletin 2003;29(5):665–75.

23.

Fergusson

Writing about traumatic events using the third person pronoun: Psychological and health effects. University of Waterloo; 1994.

24.

Simmons

, Chambless

, Gordon

, How do hostile and emotionally overinvolved relatives view relationships? What relatives’ pronoun use tells us, Family Process 2008;47(3):405–19.

25.

Murphy

, Archer

, Demmen

, Mapping the links between gender, status and genre in Shakespeare’s plays, Language and Literature 2020;29(3):223–45.

26.

Davidson

, Financial markets and authoritative proximity in personal finance magazines, Public Understanding of Science 2014;23(6):734–49.

27.

Larrimore

, Jiang

, Larrimore

, et al., Peer to peer lending: The relationship between language features, trustworthiness, and persuasion success, Journal of Applied Communication Research 2011;39(1):19–37.

28.

Chang

CLH

, Wu

, Social network service based on ABC theory, Human Systems Management 2021;40(4):535–47.

29.

Wiepking

, Breeze

, Feeling poor, acting stingy: The effect of money perceptions on charitable giving, International Journal of Nonprofit and Voluntary Sector Marketing 2012;17(1):13–24.

30.

Zhang

, Xue

, Li

, Liu

, Effect of textual features on the success of medical crowdfunding: Model development and econometric analysis from the tencent charity platform, Journal of Medical Internet Research.e 2021;23(6):22395.

31.

Braber

, Emotional and emotive language: Modal particles and tags in unified Berlin, Journal of Pragmatics 2006;38(9):1487–503.

32.

Hynd

, Chase

, The relation between text type, tone, and written response, Journal of Reading Behavior 1991;23(3):281–306.

33.

Tetlock

, Saar-Tsechansky

, Macskassy

, More than words: Quantifying language to measure firms’ fundamentals, The Journal of Finance 2008;63(3):1437–67.

34.

, Wang

, Kindhearted People, Please Save My Family: Narrative strategies for new media medical crowdfunding, Health Commun 2020;35(13):1605–13.

35.

Belleflamme

, Lambert

, Schwienbacher

Crowdfunding: An industrial organization perspective. In Workshop Digital Business Models: Understanding Strategies’. Paris; 2010, pp. 25-26.

36.

Ahlers

GKC

, Cumming

, Günther

, Schweizer

, Signaling in equity crowdfunding, Entrepreneurship Theory and Practice 2015;39(4):955–80.

37.

Aleksina

, Akulenka

, Lublóy

. Success factors of crowdfunding campaigns in medical research: Perceptions and reality, Drug Discovery Today 2019;24(7):1413–20.

38.

Bromley

Reputation, image and impression management. John Wiley & Sons; 1993.

39.

Laird

, Perryman

, Hochwarter

, Ferris

, Zinko

, The moderating effects of personal reputation on accountability-strain relationships, Journal of Occupational Health Psychology 2009;14(1):70.

40.

Tractinsky

, Srinivasan Rao

, Incorporating social dimensions in Web-store design, Human Systems Management 2001;20(2):105–21.