1. Introduction

Digital music refers to the type of music that is saved and transmitted using advanced digital technology. It can achieve efficient transmission and sharing of music through the Internet and wireless networks. Compared to traditional music, digital music has the characteristics of being easier to download, copy, and play, and while retaining high-quality sound quality, it can also provide a more flexible and autonomous digital experience for music fans [1]. This has enabled digital music to achieve excellent performance in terms of speed, convenience, fashion, and has become one of the popular ways of music dissemination in today’s society. China’s digital music industry enjoys strong development momentum. The 2023 Global Music Report by the International Phonographic Industry reveals that China has ascended to the position of the fifth largest music market in the world. Overall, streaming remains the core distribution method for video and music, driven by the growth of streaming. The streaming market is still leading the rapid development of the entire digital music industry [2]. China, the second largest market in Asia, saw a growth rate of 28.9 percent. It can be seen that China’s digital music market is sizeable and has broad space for development. In the era of mobile Internet, the payment consciousness of Internet digital music streaming media users has increased significantly. The post-90s and post-00s generation has become the dominant group of digital music consumption, and their willingness to pay for quality content services is becoming increasingly important.

Against this backdrop, the following questions naturally arise: What factors influence the willingness of young people, particularly college students, to consume digital music? Do these factors interact with one another? And how can targeted marketing strategies be formulated based on the findings? Unraveling the influencing factors and their interactions regarding the willingness of young people, especially college students, to consume digital music holds not only academic significance in gaining a deeper understanding of this demographic’s consumption behavior but also provides practical guidance for music platforms to devise targeted marketing strategies, enhance user experiences, and foster the healthy development of the music market.

In 2023, the scale of China’s digital music industry will reach 89.345 billion yuan, a year-on-year increase of 5%, with a slight slowdown in growth rate. The number of online music users has reached 726 million, accounting for 67.6% of the total number of netizens [2]. With the gradual improvement of the copyright protection system and the continuous growth of users’ demand for high-qualimusic content, the number and payment ratio of paying users on major digital music platforms have both experienced double-digit growths.

The rise of streaming platforms is reshaping the global entertainment industry at an astonishing speed. Over the past decade, the streaming industry has witnessed explosive growth, giving birth to many leading companies. By mid-2023, China’s online video users had surpassed 1 billion, becoming a significant force in the global video streaming market. Among them, iQiyi, Youku, and Tencent Video have stood out as major players. Meanwhile, Southeast Asia has become the fastest-growing Internet economy. The COVID-19 pandemic restricted people’s movements but also accelerated the development of streaming services, with mobile streaming users in Southeast Asia increasing by 60%. An increasing number of streaming services and content providers have begun to enter the market, such as Netflix, Disney + , and TikTok. The development of digital music and streaming media platforms is closely linked. The rise of digital music provides a rich content foundation for streaming media platforms, while streaming media platforms promote the popularity of digital music through convenient online playback and personalized recommendations. Both have jointly changed users’ music consumption habits, shifting from physical purchases to online listening and subscription services. In addition, streaming media platforms have further promoted the digital transformation and commercial development of the music industry through data analysis and copyright management. The rise of digital music streaming, widespread social media usage, and the increasing prominence of big data and algorithms in music recommendation systems have fundamentally transformed music consumption patterns and motivations [3]. These changes not only impact consumers’ choices and experiences but also reshape the ecosystem for artists, producers, and the entire music industry. In a complex market environment, online videos exert a multifaceted influence on recorded music sales, encouraging users to actively explore music on social platforms [4]. This tendency towards seeking inspiration and social connections has turned music consumption into a social activity, especially with the emergence of short video platforms like TikTok, where users broaden their musical horizons through creation and sharing [5].

Within this context, the digital environment has fueled consumers’ desire for diversity and personalization, leading to a reevaluation of traditional music consumption models and necessitating novel understandings [6]. The rapid evolution of digital technology has garnered widespread attention regarding shifts in music consumption motivations and behaviors. Existing research highlights significant changes in how consumers acquire and consume music, profoundly affecting their motivations and behavior patterns.

In the German market, online videos’ impact on recorded music sales is intricate, influenced not only by platform availability but also by social interaction, individual preferences, and cultural backgrounds [4]. Further exploration of user motivations on digital music platforms reveals that users seek not just personal entertainment but also inspiration and social connections (Nunes and Birdsall, 2022). Consequently, the surge in digital music consumption has fostered an openness to new music among consumers and prompted reflections on traditional consumption patterns, indicating a growing public emphasis on music diversity and innovation in the digital age [6]. An essential topic for the further development of digital music is understanding the factors that influence consumers’ choices in digital music consumption. Notably, personalized needs and convenient experiences significantly impact consumer decisions [7]. Meanwhile, the socio-cultural background plays a crucial role in shaping consumers’ digital music consumption motivations, assigning music diverse meanings across different cultural or social groups [8]. Consumers’ music choices mirror not only their preferences for notes and melodies but also their identity [9]. Therefore, the evolution of music consumption motivations is not solely a direct consequence of digital technology development but also reflects the intricate interplay of social interaction, cultural identity, and personal experience. Among teenagers, using music sharing to construct personal images and seek recognition and support from peers has become a prevalent phenomenon [10].

From the perspective of user structure, college students are the leading group of music consumption although they have less income. China Mobile’s digital music paying users are mainly low-income students and high-income professional elites. College students, representing the younger demographic, have matured alongside the evolution of the Internet and are more likely to have a typical relationship with music. In particular, the post-90s and post-00s are the major users and fans of digital music platforms, and their paying power cannot be underestimated. Moreover, college students represent a critical demographic for digital music platforms due to their unique behavioral and socio-cultural characteristics. First, as “digital natives” who have grown up in the internet era, they exhibit higher adaptability to technological innovations and digital consumption patterns compared to older generations [8]. This makes them early adopters of digital services, providing insights into future market trends. Second, despite their limited disposable income, their consumption behavior is heavily influenced by social dynamics, which aligns with the study’s focus on subjective norms and user participation [9]. Additionally, college students’ music consumption habits are often intertwined with identity construction and cultural expression, making them a strategic group for understanding the interplay between psychological needs and consumption intentions [10]. Finally, their concentrated presence in urban educational hubs like Chengdu offers logistical advantages for targeted data collection while reflecting regional market dynamics[11]. By focusing on this cohort, the study captures both the immediacy of youth-driven digital consumption and its implications for broader industry strategies.

As an important supporting part of the booming development of China’s digital music industry, the construction of regional music industry bases not only provides strong support and guarantee in many key aspects such as hardware facilities, resource integration and sharing, talent cultivation and education, and market expansion, but also firmly builds the development cornerstone of the core links of the digital music industry chain [11]. As mentioned earlier, although previous studies have investigated people’s willingness to consume digital music from various perspectives, these studies lack systematicity and the research subjects are not focused enough. Therefore, based on previous research, this article systematically studies the influencing factors of digital music consumption willingness among college students in Chengdu, China. The theoretical framework for consumer behavior analyzed statistically in this paper encompasses four dimensions: perceived value, behavioral attitude, subjective norm, and user involvement.

1. When making consumption decisions, consumers comprehensively consider factors such as product value, price, and quality. Products with high perceived value are more likely to gain favor among consumers. Existing research has shown that perceived value is a significant factor influencing consumers’ purchase intention (Consumer Perceived Value, Involvement, Trust, Susceptibility to Interpersonal Influence, and Intention to Participate in Online Group Buying).
2. Consumers’ attitudes toward a product influence their purchase intention. Positive behavioral attitudes promote purchase behavior, while negative attitudes inhibit it. Studies have indicated that behavioral attitude is a crucial factor affecting consumers’ purchase intention (A New Model to Predict Consumers’ Willingness to Buy Fair-Trade Products; Consumer Antecedents Towards Green Product Purchase Intentions).
3. Consumers’ purchase behavior is influenced by those around them. When a product is widely recognized by others, consumers are more likely to develop purchase intention. Research has demonstrated that subjective norm is a significant factor influencing consumers’ purchase intention (The Roles of Values and Social Norm on Personal Norms and Pro-Environmentally Friendly Apparel Product Purchasing Behavior: The Mediating Role of Personal Norms).
4. User involvement refers to the effort and engagement invested by users during product usage. Users with high involvement are more likely to develop a sense of identification and loyalty to the product, leading to purchase intention. Studies have shown that user involvement is an important factor affecting consumers’ purchase intention (Consumer Engagement in Social Media Brand Communities: A Literature Review). Perceived value is a well-established construct in consumer behavior research, particularly in the context of digital goods and services. In the context of digital music, perceived value is crucial because college students, who often have limited disposable income, are likely to weigh the benefits of paid music services against their costs. We chose this variable because it directly addresses the cost-benefit analysis that college students are likely to perform when deciding whether to pay for digital music. In the context of digital music consumption, behavioral attitude captures students’ perceptions of the ethical and practical aspects of paying for music. Given the historical prevalence of free music consumption on the internet, understanding how attitudes toward paying for music influence consumption decisions is critical. We included this variable to explore how students’ attitudes toward paid music services impact their willingness to consume. In the context of college students, peer influence and social norms play a significant role in shaping consumption behaviors. We chose this variable because it captures the social dynamics that are likely to influence college students’ decisions to pay for digital music, especially in a collectivist culture like China. In the context of digital music, user participation can include activities such as creating playlists, sharing music, and interacting with other users on the platform. We included this variable because it reflects the active role that users play in shaping their own consumption experiences, which is particularly relevant for college students who are highly engaged with digital platforms. While other variables such as psychological needs and user stickiness are also relevant, they were included as mediating variables in our model rather than independent variables. This decision was based on the theoretical framework of the Technology Acceptance Model (TAM) and the Theory of Planned Behavior (TPB), which emphasize the importance of perceived value, attitudes, and social norms as primary drivers of behavioral intention. Based on the aforementioned theoretical foundations and research findings, this study proposes the following hypotheses:

H1: Perceived value has a significant positive impact on college students’ willingness to consume digital music.

H2: Behavioral attitude has a significant positive impact on college students’ willingness to consume digital music.

H3: Subjective norm has a significant positive impact on college students’ willingness to consume digital music.

H4: User involvement has a significant positive impact on college students’ willingness to consume digital music.

In 2016, the Opinions of the Chengdu Municipal People’s Government on Supporting the Development of the Music Industry “ guideline has been issued by the Chengdu municipal government, which clearly proposes the grand vision of building Chengdu into the music capital of China and even a globally renowned music city. Based on this, a series of targeted and effective related work have gradually been carried out. As of the end of 2021, the number of college students in Chengdu has reached 1.102 million, including 981000 undergraduate students and 120000 graduate students (including 12000 doctoral students). This has made Chengdu the city with the highest number of college students in southwestern China. The study on the influencing factors of digital music consumption willingness among college students in Chengdu will provide new ideas and research basis for theoretical construction in related fields, expand the framework of influencing factors of digital music consumption willingness, and provide empirical research results. At the same time, it will also provide reference for the study of regional differences.

On the basis of in-depth analysis and criticism of existing literature on the motivation and behavioral changes of digital music consumption, as well as the interaction between digital and traditional channels, this article conducts a unique study on the impact of college students’ willingness to pay for digital music consumption. Firstly, in line with the profound impact of digital technology on music consumption behavior emphasized in existing literature [4,5] used descriptive statistical methods to describe in detail the basic characteristics of college students in Chengdu, especially their preferences and habits in digital music consumption, providing an empirical basis for understanding music consumption behavior in specific regional cultural backgrounds. This approach not only enriches the geographical diversity of existing literature, but also enhances the pertinence and practicality of research.

Secondly, based on the summary of relevant research content, this article not only identifies the main factors affecting online digital music payment, but also constructs an innovative research model through questionnaire design, questionnaire survey, and data retrieval. This model not only incorporates core variables such as personalized needs and convenient experiences in digital music consumption [12], but also considers complex factors such as social interaction and cultural background. These factors have been widely discussed in existing literature, but research on their impact on specific payment intentions is still insufficient. Thirdly, the empirical analysis in this article draws unique conclusions based on critical reference to existing literature. In line with the existing literature [6] emphasizing that the increase in digital music consumption promotes consumers’ open attitudes towards new music, this article finds that college students exhibit a high pursuit of diversity and innovation in digital music consumption. However, unlike existing literature that mostly focuses on macro level trend analysis, this article reveals the specific manifestations of this pursuit among college students and the complex motivations behind it through detailed empirical analysis. In addition, this article also explores the potential impact of digital and traditional channel interaction on the willingness of college students to pay for music consumption [13], but focuses more on this specific consumer group, providing new evidence for understanding the role of multi-channel interaction in segmented markets.

In summary, this article conducts an in-depth analysis of the impact of college students’ willingness to pay for digital music consumption through empirical research and innovative research models. This study not only enriches the theoretical framework of digital music consumption behavior, but also provides targeted market strategy recommendations for music industry practitioners, especially in how to attract and cultivate the important consumer group of college students. The unique contribution of this article lies in its consideration of regional cultural backgrounds, analysis of the comprehensive effects of multiple factors, and exploration of the interaction between digital and traditional channels, providing a new perspective for understanding the complexity and diversity of music consumption in the digital age.

2. Methods

2.2. Survey instrument

The survey asked students about their preferences for consuming digital music, and other information like their economic situation as well as demographic information. The full survey information is provided in S1 and S2 Appendix. The author (Yu Shan) is a music teacher at Xihua University, with part-time teaching positions at Sichuan University of Culture and Arts and Geely University, where she delivers the course ‘Music Appreciation’. Teaching four classes of 150 students each semester provides convenient access to collect music-related questionnaires. The survey for this study was conducted in these Music Appreciation classes. The questionnaire is distributed randomly to college students. The questionnaire was distributed in electronic form for a limited time to ensure the quality of the questionnaire filling. The purpose of the study and the method used were explained in detail to the surveyed college students, and after getting the consent, the questionnaires were sent to them, and the questionnaire items and notes for filling were informed with unified guidelines. If there were any questions, the researchers would answer them in time, and the time for completing the questionnaire was controlled within 20–30 minutes. After the questionnaire is completed, the researchers will check the filling, remove unqualified questionnaires in time, and recover them uniformly. The exclusion criteria were: the questionnaire filling time was less than 10 minutes; Missing rate ≥5%; The answers to the questionnaire showed regularity; Inconsistent logic, or there are obvious errors in the questionnaire. If a questionnaire meets any of these criteria, it will be considered as an unqualified questionnaire as shown in Fig 1.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Data collection procedures.

https://doi.org/10.1371/journal.pone.0324168.g001

According to the research needs, 7 factors, such as perceived value, behavioral attitude, subjective norms, user participation, user stickiness, psychological demand and consumption intention, were selected as the variables of this study, and descriptive statistics were carried out on the 7 variables with relevant scales. Likert-type 5-point scoring method was used for the above scales, ranging from 1 “strongly disagree” to 5 “strongly agree”. Based on the basic principle of demographic profile and the survey characteristics, this study formulated 7 items as shown in Table 1, including gender, grade, average monthly consumption during school, source of living expenses, monthly expenses for digital music, living expenses, and spending methods.

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Measurement items for demographics.

https://doi.org/10.1371/journal.pone.0324168.t001

2.3. Data analysis

The data analysis had two main components. First, we designed operational question items for all variables and formed a questionnaire with 42 question items. Second, based on this, the validity analysis (content validity - IOC test and structure validity - EFA) and reliability analysis were carried out for 36 variable data. Additional details are provided below. According to the previous assumptions, the study selected perceived value, behavioral attitude, subjective norms and user participation as independent variables of this study, and measured them with relevant scales. These four items were given in the form of multiple-choice questions. The settings and options of each item are shown in Table 2.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Measuremens of independent variables.

https://doi.org/10.1371/journal.pone.0324168.t002

Consumption intention is selected as the dependent variable of this study to test the consumption intention of college students, and it is given in the form of multiple-choice questions. The items are set in Table 3.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Measurements of dependent variables.

https://doi.org/10.1371/journal.pone.0324168.t003

User stickiness and psychological needs are selected as the mediating variables in this study, which are given in the form of multiple-choice questions. The item Settings are shown in Table 4.

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Measurements of mediating variables.

https://doi.org/10.1371/journal.pone.0324168.t004

After completing the questionnaire design, SPSS and AMOS were used to input, process and analyze the collected data [14]. The following 6 main data analysis programs were adopted in this study. Firstly, SPSS software was used for reliability analysis, validity analysis and exploratory factor analysis (EFA) for the pilot test and sample data quality [15]. Then AMOS model was used for confirmatory factor analysis (CFA) and structural equation model (SEM), and the research hypothesis was tested with bootstrap estimation.

Since the measurement items in this study were in the form of Likert 5 scale, Cronbach’s Alpha was used to measure the intrinsic reliability of the questionnaire. The larger the coefficient in Cronbach’s Alpha measurement, the higher the internal consistency of the questionnaire [16]. According to the researchers, if Cronbach’s Alpha coefficient is higher than 0.8, it indicates high reliability, and if Cronbach’s alpha coefficient is between 0.7 and 0.8, it indicates good reliability, and if the value is between 0.6 and 0.7, it indicates that the reliability is acceptable, and if the value is less than 0.6, it indicates poor reliability. Content validity is an objective evaluation index, which refers to whether the setting of each item is representative and comprehensive, whether it is within the research field, and whether it achieves the research purpose Item-Objective Congruence (IOC) is a procedure used to assess the content validity of a project during the development phase of the test development phase and is limited to the evaluation of one-dimensional items or the measurement of specified items. During this process, the questionnaire is checked by an expert in the field of study. The IOC value is calculated based on the score given by each expert. When the IOC value is greater than 0.5, the measured items in the questionnaire can be accepted, indicating that the content validity of the questionnaire is good.

Convergence validity, also known as aggregation validity or internal consistency, is an important indicator when evaluating the quality of a measurement tool or questionnaire. It primarily measures the degree to which different measurement items (such as different questions in a questionnaire) are related to each other under the same concept or construct. When there is a high degree of correlation between the measurement items under the same construct, the convergence validity is good. The mean extraction variance is also an important index for evaluating convergence validity. Its formula is as follows:

When AVE value is greater than 0.5, it means that the variance of the measured variable explained by its potential variable is greater than the variance explained by its measured error, indicating that the questionnaire has good convergence validity. Discriminant validity (also known as discriminant validity) refers to the degree of decorrelation between a concept and other different concepts, that is, when measuring different variables using different methods, the observed values should be able to distinguish between them. According to existing research [17], in confirmatory factor analysis (CFA), the evaluation method for discriminant validity is to use the square root of AVE and the correlation coefficient between latent variables and other latent variables. If the former is much larger than the latter, then each latent variable shares more variance with its own measurement items than with other measurement items; Further demonstrate that the differences between latent variable measurement items have significant discriminant validity.

This paper carries out exploratory factor analysis (EFA) on the pilot test data, and conducts comprehensive analysis on KMO value, common degree, variance explanation rate value, factor load coefficient value and other indicators to verify the validity level of the data. The data in this study was rotated using the maximum variance rotation method (varimax) in order to find out the corresponding relationship between the factor and the research item [18]. KMO value was used to judge the suitability of information extraction, common degree value was used to exclude unreasonable research items, variance interpretation rate value was used to illustrate the level of information extraction, and factor load coefficient was used to measure the correspondence between factors (dimensions) and items. Combined with the factor load coefficient, confirm whether the correspondence between the factor (dimension) and the research item is consistent with the expectation. If it is consistent, it indicates the validity; otherwise, it needs to be re-adjusted. When the absolute value of factor load coefficient is greater than 0.4 (0.3 standard is also adopted in foreign countries), it indicates that there is a corresponding relationship between the option and the factor.

SEM is a statistical method of analyzing relationships between variables that can be used to build global models and estimate and test causal relationships. It can not only handle multiple dependent variables at the same time, but also allow them and the independent variables to have a certain measurement error, while estimating factor structures and relationships, allowing for a more elastic measurement model, and estimating the degree of fit of the model. Due to the existence of variable relationships in this study and the interpretation of the AMOS results’ View Text ‘, we gradually expanded the initial model by adjusting the SEM correction index (M.I.) to correct the AMOS.

In this paper, AMOS software is used to test the structural equation model, the non-parametric percentile bootstrapping method is used to correct the deviation, and the percentile 95% confidence interval is set in AMOS software (24–28). The confidence interval is the possible range of the whole parameter. A 95% confidence interval is when the overall parameter is within this range with about 95% probability, or when the overall parameter is within this range but has only 95% confidence. If the upper and lower values of the confidence interval have different signs, that is, the range between the upper and lower values contains 0, then the coefficient of the confidence interval is not significant; If both values of the confidence interval are positive, then the range between them does not contain 0, indicating that the coefficient of the confidence interval is significant. Therefore, Bootstrapping estimation method combined with structural equation model (SEM) was selected to verify the research hypothesis. In the previous chapters, a 4-step research design procedure is developed, and the quantitative research is conducted according to it. (1) From the measurement dimension, we designed operational question items for all variables and formed a questionnaire with 42 question items. (2) Based on a combination of purposive sampling and stratified sampling, 431 samples were obtained. (3) The data analysis procedure is summarized in detail. (4) Based on this, the validity analysis (content validity - IOC test and structure validity - EFA) and reliability analysis were carried out for 36 variable data.

3. Results and discussion

In this section we discuss the results of the survey. We begin by describing the respondents to our survey. Then, descriptive statistics and analysis of the data variables are conducted (with indicators such as frequency, percentage, mean, and standard deviation), and the quality of the sample data is evaluated using reliability, validity analysis, and confirmatory factor analysis (CFA) (with analysis indicators such as factor loading, composite reliability (CR), average variance extracted (AVE), and model fit indices). Finally, path analysis and Bias-Corrected Bootstrap procedures are used to test the research hypotheses. The data in the Demographic Profile Questionnaire (Part1) of the respondents were analyzed and expressed in frequency and percentage terms, including gender, age, average monthly consumption, sources of living expenses, monthly expenses for digital music, living expenses, spending methods, etc. Through data analysis, the statistical results are shown in Table 5 There were 221 males (51.28%) and 210 females (48.72%) in the surveyed sample, with a slightly higher proportion of males than females. In grade level, 420 sophomores (97.45 percent) were enrolled, while students from other grades accounted for a relatively small proportion. In terms of monthly consumption, 226 students (52.44 percent) spend between 1,000 and 1,500 yuan, and 134 students (31.09 percent) spend between 1,500 and 2,000 yuan. In terms of the source of living expenses, 318 students (73.78 percent) came from their families, followed by 112 students (25.99 percent) who came partly from their families and partly earned by themselves. In terms of monthly expenses for digital music, the largest number of students were between 1 and 200, with 236 (54.76%), followed by no consumption, with 192 (44.55%). There are 234 people (54.29%) who think that their living expenses are just enough, followed by 146 people (33.87%) who think that they have enough. In terms of the way they spend money, 275 people (63.81 percent) said they plan to spend while they can, followed by 71 people (16.47 percent) who want to save money.

Download:

• PNG
  larger image
• TIFF
  original image

Table 5. Frequency analysis of the sample.

https://doi.org/10.1371/journal.pone.0324168.t005

The demographic profile of the respondents in this study is diverse in terms of study courses and economic backgrounds. However, we acknowledge that the grade distribution is skewed, with 97.45% being second-year students. This imbalance may reduce the representativeness of the entire student population. Future studies could adopt stratified random sampling across all grades to enhance generalizability. Despite this limitation, the data quality was rigorously ensured through validity and reliability tests (Cronbach’s α > 0.8).The observed variables of the study are examined in the second part of the questionnaire. In this paper, descriptive statistical analysis is conducted on all variables from two levels.

The first level is descriptive statistical analysis for each item. Mean and standard deviation are included in the analysis indicators. The statistical results are shown in Table 6. According to the previous assumptions, the study selected perceived value, behavioral attitude, subjective norms and user participation as independent variables of this study, and measured them with relevant scales. Among them, the mean value of perceived value is between 3.657–4.049, which indicates that the perceived value of college students surveyed by the research is above the medium level. The mean value of behavioral attitude questionnaire is between 3.51 and 3.893, which indicates that the behavioral attitude of college students investigated by the research are at the upper middle level. The mean value of the subjective norm questionnaire was between 3.564 and 3.893, indicating that the subjective norm of college students surveyed by the research was at the upper middle level. The mean of user participation was between 3.418 and 3.696, indicating that the user participation of college students surveyed by the research was at the upper middle level.

Download:

• PNG
  larger image
• TIFF
  original image

Table 6. Descriptive analysis of observed variables.

https://doi.org/10.1371/journal.pone.0324168.t006

The dependent variable of this study is consumption intention, and the results show that the average intention to consume of the university students is between 3.55 and 3.645, which indicates that the level of consumption intention of college students is good. User stickiness and psychological needs are selected as the mediating variables. It shows that the overall mean of the questionnaire on user stickiness of college students is 4.048, and the mean of its sub-dimensions is between 3.981–4.176, which indicates that the stickiness of college students is above the medium level. The overall mean of the survey data on psychological needs of college students is 3.941, and the mean of its sub-dimensions is between 3.696–4.035, which indicates that the psychological needs of college students are at the upper middle level.

The second level is an overall descriptive statistical analysis of all the variables, the results of which are presented in Table 7. It is not difficult to see from the results that the average values of perceived value, behavioral attitude, subjective norms, user participation, user stickiness, psychological needs, and consumption intention are 3.858, 3.704, 4.071, 3.581, 4.048, 3.941, and 3.599, respectively. It is obvious that all these values exceed 3, indicating that the measurement results of all observed variables are good.

Download:

• PNG
  larger image
• TIFF
  original image

Table 7. Overall descriptive statistics of all observed variables.

https://doi.org/10.1371/journal.pone.0324168.t007

Based on the theory of correlation analysis, it was found that there is a correlation between UV, PR, CW and PV, BA, SN, UP, respectively, and they were studied. Pearson correlation coefficient can be used to indicate the strength of the correlation relationship. The specific analysis results are shown in Table 8.

Download:

• PNG
  larger image
• TIFF
  original image

Table 8. Correlation analysis.

https://doi.org/10.1371/journal.pone.0324168.t008

The results showed that UV was significantly associated with PV, BA, SN, and UP. The correlation coefficient values between them are 0.313, 0.330, 0.329, and 0.380, respectively. These correlation coefficient values are all greater than 0, indicating a positive correlation between UV and the four terms PV, BA, SN, and UP.

The results showed that PR was significantly associated with PV, BA, SN, and UP. The correlation coefficient values between them are 0.344, 0.401, 0.439, and 0.338, respectively. These correlation coefficient values are all greater than 0, indicating a positive correlation between PR and the four terms PV, BA, SN, and UP.

Similarly, significant differences were observed between CW and PV, BA, SN, and UP. The correlation coefficient values between them are 0.443, 0.460, 0.487, and 0.498, respectively. These correlation coefficient values are all greater than 0, indicating a positive correlation between CW and the four terms PV, BA, SN, and UP.

In addition, the correlation coefficient between CW and UV is 0.453 and shows significance at the 0.01 level, indicating a significant positive correlation between CW and UV. The correlation coefficient between CW and PR is 0.491 and shows a significance level of 0.01, indicating a significant positive correlation between CW and PR.

The validity evaluation of sample data includes two aspects: content validity evaluation and structural validity evaluation. Before conducting exploratory factor analysis (EFA), KMO and Bartlett tests should be performed first.

Upon analyzing the results, it was found that the KMO value of the survey questionnaire was 0.933, which is clearly greater than 0.7. The results also showed that the chi-square value of Bartlett’s sphericity test was 8106.028, the degree of freedom was 630, and the significance was 0.000. The data can be effectively extracted, indicating that there are common structural factors among the variables in the questionnaire, which can be analyzed for factors as shown in Table 9.

Download:

• PNG
  larger image
• TIFF
  original image

Table 9. KMO and Bartlett’s test results.

https://doi.org/10.1371/journal.pone.0324168.t009

Through the reliability and validity analysis of this section, it was found that the evaluation of sample data has high reliability and validity. AMOS software was used to analyze the measurement model composed of independent variables by fixed loading method and maximum likelihood method (ML). The model consists of 7 variables, such as independent variable perceived value, behavioral attitude, subjective norm, user participation, dependent variable consumption intention and intermediate variable user stickiness and psychological demand. Based on the confirmatory factor analysis (CFA) model, analyze independent variables, mediating variables, and dependent variables. It can be seen that this CFA model involves 7 factors and 36 measurement indicators as shown in Fig 2.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Confirmatory factor Analysis diagram.

https://doi.org/10.1371/journal.pone.0324168.g002

According to the T rule of the measurement model identification criteria ()), the CFA model has a total of 36 observation indicators, so the total degree of freedom is, and a total of 79 parameters need to be estimated:36 factor loads, error variances of 36 measurement indicators and correlation coefficients between 7 factors. , which meets the necessary conditions for model identification [19].

On the premise that the data in the model meets the standards, the impact paths between variables are determined based on regression analysis, and the mediating effects of the mediating variables are tested using Bootstrap sampling method.

Table 10 shows that perceived value, behavioral attitude, subjective norms and user participation, as independent variables, all have significant positive effects on user stickability and psychological demand of mediating variables. Specifically, when studying the impact of perceived value on user stickiness, the standardized path coefficient value was 0.147 > 0 and showed a significance level of 0.05 (CR = 2.497, p = 0.013 < 0.05), indicating that perceived value has a significant positive effect on user stickiness. Similarly, when studying the impact of behavioral attitudes on user stickiness, 0.184 > 0, CR = 3.072, p = 0.002 < 0.01; When studying the impact of subjective norms on user stickiness, 0.183 > 0, CR = 2.914, p = 0.004 < 0.01; When studying the impact of user participation on user stickiness, 0.254 > 0, CR = 4.401, p < 0.001。These indicate that behavioral attitudes, subjective norms, and user participation have a significant positive impact on user stickiness.

Download:

• PNG
  larger image
• TIFF
  original image

Table 10. Statistical data of path analysis among variables.

https://doi.org/10.1371/journal.pone.0324168.t010

When studying the impact of perceived value on psychological needs, the standardized path coefficient value is 0.142 > 0, and this path shows a significance level of 0.05 (CR = 2.605, p = 0.009 < 0.05). This indicates that perceived value has a significant positive impact on psychological needs. Similarly, when studying the impact of behavioral attitudes on psychological needs, 0.26 > 0, CR = 4.596, p=<0.001; When studying the impact of subjective norms on psychological needs, 0.304 > 0, CR = 5.056, p=<0.001; When studying the impact of user participation on psychological needs, 0.116 > 0, CR = 2.221, p < 0.001。 These indicate that behavioral attitudes, subjective norms, and user participation have a significant positive impact on psychological needs.

The independent variables (perceived value, behavioral attitude, subjective norms, user participation, and mediating variables such as user stickiness and psychological needs) all have a significant positive impact on the dependent variable (willingness to consume). Specifically, when studying the impact of perceived value on consumption intention, the standardized path coefficient value was 0.173 > 0, and this path showed a significance level of 0.05 (CR = 3.433, p=<0.001), indicating that perceived value has a significant positive impact on consumption intention. Similarly, when studying the impact of behavioral attitudes on consumer willingness, 0.198 > 0, CR = 3.51, p=<0.001; When studying the impact of user participation on consumption intention, 0.164 > 0, CR = 3.016, p = 0.003 < 0.01; When studying the impact of psychological needs on consumer willingness, 0.177 > 0, CR = 3.209, p = 0.001 < 0.011。 These indicate that behavioral attitudes, user participation, and psychological needs have a significant positive impact on consumer willingness.

Table 11 shows that all indexes in the path analysis model constructed by the research reach the standard level, indicating that the model has a good fit.

Download:

• PNG
  larger image
• TIFF
  original image

Table 11. Model fitting index.

https://doi.org/10.1371/journal.pone.0324168.t011

In order to study the mediation effect, based on the suggestion of Preacher and Hayes, 5000 times bootstrap repeated sampling method is adopted to deal with the situation that the data in this study do not conform to the normal distribution. And select 95% confidence interval of percentile and 95% confidence interval of Bias-Corrected to test the significance of path coefficients. Therefore, we examined the Critical Ratio (C.R.), i.e., lower bound and upper bound, and two-tailed significance to test whether the effect between potential variables is significant.

The results of Bootstrapping estimates are shown in Table 12.

Download:

• PNG
  larger image
• TIFF
  original image

Table 12. Results of Bootstrapping analysis.

https://doi.org/10.1371/journal.pone.0324168.t012

The results in Table 4.16 show that in the path of “PV=>UV=>CW”, the indirect effect of “UV” has 95%CI confidence interval (0.004, 0.069), the confidence interval does not include 0, and the significance P = 0.019 < 0.05, indicating that the intermediary effect is statistically significant and the intermediary effect exists. The effect size (non-standard) was 0.026, and the proportion of intermediate effect size was 10.88%.

In the path of “BA=>UV=>CW”, the indirect effect of “BA” has 95%CI confidence interval (0.007, 0.084), the confidence interval does not include 0, and the significance P = 0.006 < 0.05, indicating that the intermediary effect is statistically significant and the intermediary effect exists. The effect size (non-standard) was 0.034, and the proportion of mediating effect size was 11.26%. In the path of “SN=>UV=>CW”, 95%CI confidence interval (0.006, 0.096) for indirect effect of “SN”, confidence interval excluding 0, significance P = 0.013 < 0.05, indicating that the intermediary effect is statistically significant and the intermediary effect exists. The effect size (non-standard) was 0.035, and the proportion of intermediate effect size was 10.57%. In the path of “UP=>UV=>CW”, the indirect effect of “UP” was 95%CI confidence interval (0.012, 0.083), the confidence interval did not include 0, and the significance P = 0.002 < 0.05, indicating that the intermediary effect was statistically significant and the intermediary effect existed. The effect size (non-standard) was 0.038, and the proportion of mediating effect size was 14.29%.

In the path of “PV=>PR=>CW”, the indirect effect of “PV” has 95%CI confidence interval (0.005, 0.072), the confidence interval does not include 0, and the significance P = 0.011 < 0.05, indicating that the intermediary effect is statistically significant and the intermediary effect exists. The effect value (non-standard) was 0.027, and the proportion of the intermediary effect size was 11.30%.

In the path of “BA=>PR=>CW”, the indirect effect of “BA” has 95%CI confidence interval (0.016, 0.112), the confidence interval does not include 0, and the significance P = 0.004 < 0.05, indicating that the intermediary effect is statistically significant and the intermediary effect exists. The effect size (non-standard) was 0.051, and the proportion of mediating effect size was 16.89%. In the path of “SN=>PR=>CW”, the indirect effect of “BA” had 95%CI confidence interval (0.016, 0.112), the confidence interval did not include 0, and the significance P = 0.004 < 0.05, indicating that the intermediary effect was statistically significant and the intermediary effect existed. The effect size (non-standard) was 0.051, and the proportion of mediating effect size was 16.89%.

In the path of “UP=>PR=>CW”, the indirect effect of “UP” was 95%CI confidence interval (0.001, 0.051), the confidence interval did not include 0, and the significance P = 0.038 < 0.05, indicating that the intermediary effect was statistically significant and the intermediary effect existed. The effect size (non-standard) was 0.019, and the proportion of mediating effect size was 7.14%.

Total effect size of PV-CW =0.186+0.026+0.027=0.239

Total effect size for BA-CW =0.217+0.034+0.051=0.302

Total effect size for SN-CW =0.233+0.035+0.063=0.331

Total effect size for UP-CW =0.209+0.038+0.019=0.266

Therefore, overall, subjective norms have the greatest influence on consumption intention, while perceived value has the least influence on consumption intention. Based on the results of bootstrapping estimates, we show the results of the conceptual framework in Fig 3.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Test results for conceptual framework annotations.

https://doi.org/10.1371/journal.pone.0324168.g003

Through empirical analysis of structural equation modeling (SEM) with bootstrap estimation, the test results of four research hypotheses were obtained, summarized in Table 13. Table 13 shows that all four hypotheses are supported, which indicates that our validation analysis is scientifically sound. In addition, compared with regression coefficient, subjective norms have the greatest influence on consumption intention, while perceived value has the least influence on consumption intention.

Download:

• PNG
  larger image
• TIFF
  original image

Table 13. Test results of four research hypotheses.

https://doi.org/10.1371/journal.pone.0324168.t013

To gain a more comprehensive understanding of the implications of the findings in this study, we compared them with relevant international research. Existing research [20,21] has explored the impact of online music videos on recorded music sales, finding that music videos can effectively boost record sales. This echoes our study’s results, where factors such as user engagement and behavioral attitudes have significant effects on college students’ willingness to consume digital music. Despite differences in research subjects and market environments, these commonalities suggest that digital music platforms should focus on enhancing user experience and a sense of participation when formulating marketing strategies. Furthermore, when investigating the role of popular music in cultural identity formation in the streaming era, the importance of music as a tool for shaping cultural identity was emphasized [22–26]. This perspective is also reflected in our study, as college students may be influenced by subjective norms (such as recommendations from friends and family) when choosing to consume digital music. Therefore, digital music platforms can leverage word-of-mouth marketing through channels such as social media to expand their user base [27–30]. Based on the findings of this study, specific marketing strategies can be formulated for digital music platforms and the music industry.

User experience enhancement: By optimizing user interfaces, providing high-quality music content, and offering personalized recommendation systems, the perceived value of digital music platforms among users can be improved. This, in turn, helps to increase user retention and, consequently, boost consumption intentions [31–35]. User engagement encouragement: Online music events and user creation contests can be organized to encourage active participation in the construction and interaction of digital music platforms. This not only increases user involvement but also attracts more new users through word-of-mouth promotion. Social media marketing utilization: Social media platforms (such as Weibo and WeChat) can be leveraged for word-of-mouth marketing, expanding brand influence through user shares, comments, and other interactions. Additionally, influencers, musicians, and other figures can be invited for collaborative promotions to attract more targeted users [36–39]. Differentiated marketing strategy formulation: Differentiated marketing strategies should be developed for college student groups based on factors such as gender, grade, and consumption level. For example, high-end membership services can be offered to students with higher consumption levels, while discounted packages or free trials can be provided to those with lower consumption levels [40].

4. Conclusion and limitation

Supporting information