The relationship between information hoarding and selective exposure: the role of information overload, identity bubble reinforcement, and intolerance of uncertainty

Abstract

Background

While related studies have examined the influencing factors of information hoarding, relatively little research focused on its potential consequences (e.g., selective exposure). Based on the information ecosystem theory, this study investigated the relationships between information hoarding, information overload, identity bubble reinforcement, intolerance of uncertainty, and selective exposure.

Methods

Using a cross-sectional survey design, we tested the proposed model with a final sample of 717 valid responses.

Results

The findings showed that information hoarding was positively associated with selective exposure, that information overload and identity bubble reinforcement mediated this relationship, respectively. In addition, intolerance of uncertainty acted as a moderator between information hoarding and selective exposure, as well as between information overload and selective exposure.

Conclusion

These findings not only enhance our understanding of how information hoarding relates to selective exposure, but also provide recommendations for individuals and platform managers to mitigate selective exposure.

Introduction

The development of digital devices and cloud storage services has provided convenience for individuals in preserving information. According to recent statistics from AAG IT, the number of global users of personal cloud storage services (e.g., Dropbox, Google Drive, and iCloud) has grown remarkably from 1.1 billion in 2014 to an estimated 2.3 billion in 2024 [1]. With just a few clicks, individuals can effortlessly save resources on digital devices or cloud platforms, offering tremendous convenience for both work and personal life. Unfortunately, more and more people are getting lost and losing control when it comes to acquiring or saving digital information resources. They worry about missing out on digital information resources, accumulate an excessive number of similar resources, and are hesitant to delete unnecessary ones. Researchers refer to this phenomenon of continuously saving, accumulating clutter and have difficulty deleting digital information resources as information hoarding [2, 3], which is considered a maladaptive information use behavior in the information age.

What cannot be ignored is that in the information age, changes in information storage and use behaviors can trigger chain reactions that affect individual information processing. Previous studies have found the “Google effect” of individual information processing, that is, individuals remember more about the address where the content is stored than the stored content [4]. This indicates that the information storage and use behaviors based on digital devices have changed individual information processing. However, previous research pays less attention to the impact of these information storage and use behaviors on individual information processing, especially information hoarding on information selection and processing. Selective exposure, a type of information processing, refers to the tendency of individuals to select information that aligns with their own views during the information selection process [5]. Selective exposure can lead to incorrect judgments and the belief in misinformation [6], but also aggravate the phenomena of information cocooning and group polarization [7]. It is crucial to explore the influencing factors of selective exposure.

According to information ecology theory, information, the information environment, and information subject form a dynamic ecosystem in which these elements interact [8, 9] and collectively influence individual information processing. Therefore, this study adopts the framework of information ecology theory to examine the relationship between information hoarding and selective exposure. Specifically, we identify information overload (information), identity bubble reinforcement (information environment), and intolerance of uncertainty (information subject) as key factors that are associated with information processing.

Information hoarding and selective exposure

Information hoarding can be seen as a contemporary manifestation of traditional hoarding in the digital age. Research has demonstrated a significant association between hoarding behaviors and digital hoarding [10]. Similar to traditional hoarding behavior, individuals who engage in information hoarding may perceive their accumulated information resources as an extension of themselves [2]. More importantly, these information resources may reinforce their self-identity [11], making them more likely to focus on information that aligns with their hoarded content [16]. In other words, individuals who hoard digital information may be more inclined to engage in selective exposure to content consistent with their existing digital resources. Second, numerous studies have shown that digital hoarding causes anxiety and stress [3, 12, 13]. These emotional states may interfere with their cognitive processing and decision making [14], potentially resulting in selective exposure. In addition, more direct evidence showed that hoarding behavior is related to impaired sustained attention and impaired decision-making [15, 16]. Information hoarding may produce cognitive bias, as it shares essential characteristics with traditional hoarding. Therefore, information hoarding may be associated with selective exposure.

The mediating role of information overload

Based on the information ecosystem theory [8, 9], information serves as the fundamental component, directly shaping individuals’ information selection and processing strategies. It is worth noting that while information holds great significance for individuals, more information is not necessarily better. However, an excessive amount of information may overwhelm individuals, exceeding their capacity to effectively process it. This phenomenon is known as information overload [17]. Most previous studies on the causes of information overload have focused on information perspectives, such as: information diversity, information complexity, and information pushing [17]. However, information overload is a subjective perception of individuals, which is not only related to the information itself but may also be associated with individual maladaptive information usage behavior (e.g., information hoarding). According to cognitive load theory [18], information overload increases cognitive burden, making it difficult for individuals to filter and evaluate information effectively. As a result, they may resort to less demanding strategies, such as selective exposure [19], to reduce the complexity of information processing. Therefore, drawing upon cognitive load theory, we conceptualize information overload as a key factor in the information ecosystem theory, examining its role in the relationship between information hoarding and selective exposure.

Information hoarding may contribute to information overload. Driven by psychological motives such as fear of missing out and attachment anxiety [2, 20], individuals who hoard information are likely to spend long periods online, continuously acquiring and saving information. This behavior increases the likelihood of perceiving information overload. More unfortunately, due to the desensitization effect [21], the benefits of information hoarding in alleviating fear of missing out and providing attachment security gradually diminish. As a result, individuals may accumulate even more digital information to maintain psychological satisfaction. This results in greater perceived information overload. Secondly, due to the endowment effect [22], individuals with information hoarding will hate loss and struggle to delete information. This psychological tendency depletes cognitive resources, contributing to the perception of information overload. In addition, existing studies have shown that individuals who hoard information often experience symptoms of information overload, such as stress and confusion [12], providing more direct evidence for the relationship between information hoarding and information overload.

Faced with overloaded information, individuals process information in two main ways. Information avoidance and discontinuing social networking sites are strategies for coping with information overload [19, 23]. Another approach is information filtering, where individuals selectively read and process information based on their interests, needs, and cognitive bias [25, 30]. This strategy reflects selective exposure, that is, information overload may lead to selective exposure. Notably, driven by the basic psychological needs of individuals with information hoarding [2, 20], selective exposure, rather than information avoidance, may be the best way to deal with information overload. First, as a feeling of stress [24], information overload can induce negative emotions, such as fatigue and anxiety [17]. Negative emotions will naturally have adverse effects on the cognitive process [25], which may lead to selective exposure. Moreover, to mitigate these negative experiences, individuals tend to passively filter out irrelevant information and retain relevant information, closely linked to selective exposure [19, 23]. Direct evidence has showed that information overload is associated with selective exposure [19]. Therefore, Information overload may mediate between information hoarding and selective exposure.

The mediating role of identity bubble reinforcement

Apart from information overload, the information environment also plays a crucial role in shaping individuals’ information processing [8, 9], especially in today’s predominantly social information landscape. Social media platforms offer key avenues for individuals to satisfy communication needs and exchange information with others or groups. In this context, Keipi et al. proposed the concept of identity bubble reinforcement, which refers to individuals being more actively involved in and identifying with a group of like-minded individuals’ activities in the online world [26, 27]. Previous studies have largely regarded identity bubble reinforcement as a key factor influencing individual online behaviors, such as social networking site addiction and cyberbullying [28, 29]. However, there has been limited research exploring the relationship between maladaptive information use, information processing, and identity bubble reinforcement. Drawing on motivated reasoning theory [30], individuals tend to seek and interpret information in a way that aligns with their pre-existing beliefs and group identities, often disregarding conflicting viewpoints. Identity bubble reinforcement amplifies this tendency by creating a psychologically safe and confirmatory informational environment. This study conceptualizes identity bubble reinforcement as an environmental factor within the information ecosystem framework, and integrates insights from motivated reasoning theory to examines its role in the relationship between information hoarding and selective exposure.

Information hoarding may be associated with identity bubble reinforcement. Individuals with information hoarding not only accumulate instrumental information but also emotional social resources, such as idol and academic information [2, 31]. The hoarding of emotion-based information enhances self-definition and categorization [10, 11], increasing the likelihood of connecting with specific groups and strengthening identity bubbles. Second, when individuals discover that a particular information source regularly publishes the information they require, they tend to hoard it by following or subscribing [10], which reinforces the identity bubble. Meanwhile, platform algorithms and personalized recommendations push content from individuals or groups who post relevant information [32]. Furthermore, based on social homophily [33], individuals who hoard relevant information are more likely to engage in communication and interactions with others who also hoard similar information, further reinforcing identity bubbles.

Identity bubble reinforcement may be a key factor in selective exposure. Deindividuation is a significant manifestation of identity bubble reinforcement [26, 34]. Rather than being a loss of individual self-concept, deindividuation involves a shift towards group self-definition [35]. This shift leads individuals to exhibit selective information filtering based on group identity [36]. Specifically, they tend to exclude and reject content that falls outside their identity group [27] while displaying more acceptance and absorption of information within their group. Secondly, according to social identity theory [37], individuals may adopt behaviors to integrate themselves into the group in order to maintain its identity [38]. These behaviors may include actively reading and sharing information that aligns with the group’s viewpoint, refuting information that contradicts the group’s perspective [27], thus generating selective exposure. Therefore, identity bubble reinforcement may mediate between information hoarding and selective exposure.

The moderating effect of intolerance of uncertainty

Within the framework of information ecosystem theory, individual characteristics also influence information selection and processing [8, 9]. Previous studies have found that critical thinking [39], knowledge [40], and personality traits [41] are moderating factors in information processing. However, previous studies did not closely correlate information subject with information and the information environment, and overlooked the specific characteristics of information and information environment– uncertainty [42]. Uncertainty represents one of the primary informational challenges individuals face, and individuals exhibit significant differences in their tolerance of uncertainty. Therefore, intolerance of uncertainty should be conceptualized as a key factor within the information subject of the ecosystem, warranting further investigation of its role in the interactions among information, the environment, and selective exposure.

Intolerance of uncertainty refers to an individual’s inclination to react to uncertain situations in terms of their cognitive, emotional, or behavioral responses [43]. It is considered a personality trait that varies among individuals [44]. According to previous studies, intolerance of uncertainty is an important boundary or moderating factor in examining the relationship between variables, such as enhancing the relationship between online health information search and health anxiety [45], and the relationship between information search and COVID-19 fear [48], weakening the relationship between compassion and moral disengagement [46], and the relationship between moral harm and self-harm [47]. Therefore, intolerance of uncertainty may strengthen or weaken the moderating role depending on the specific variable relationships.

Researchers have identified two behavioral trends in individuals with intolerance of uncertainty: approach and avoidance [48]. One type of behavioral tendency is approach, in which individuals exhibit behaviors aimed at approaching a goal or solving a problem in order to obtain the highest degree of certainty. We hypothesized that intolerance of uncertainty mitigates the relationship between information overload and selective exposure. Information overload is not only characterized by a large amount of information, but also by the accompanying information contradictions. Individuals with high intolerance of uncertainty will have more anxiety when faced with contradictory or inconsistent information [49]. To alleviate anxiety and uncertainty, individuals are more likely to process and understand information. They may also search for additional relevant information to relieve uncertainty [42], which may lower the likelihood of selective exposure. In contrast, in the face of overloaded information, individuals with low intolerance of uncertainty have less anxiety or forced certainty [50]. This leads to less processing and understanding of information, resulting in a higher likelihood of selective exposure.

However, approach behavior sometimes does not reduce anxiety, but instead leads to more negative and uncertain emotions [42], particularly when such behavior threatens self-identity. In this study, we hypothesized that intolerance of uncertainty would enhance the relationship between information hoarding and selective exposure, and the relationship between identity bubble reinforcement and selective exposure. This reduces the possibility of more information processing by individuals with high intolerance of uncertainty, and thus more selective exposure. Information hoarding and identity bubble reinforcement are more strongly associated with self [11] or group self [26, 34]. According to cognitive dissonance theory, individuals actively avoid information that contradicts their self or group identity, while accepting information that aligns with it [51]. Hence, individuals with a higher intolerance for uncertainty are more likely to avoid information that is inconsistent with their self or group identity in order to alleviate the discomfort caused by uncertainty. On the other hand, individuals with a lower intolerance for uncertainty are more inclined to accept and process information that is inconsistent with their self or group identity, which may result in lower selective exposure.

The present study

Drawing on the information ecosystem theory [8, 9], this paper aims to investigate the potential relationship between information hoarding and individual selective exposure. Specifically, we will examine the mediating role of information overload, identity bubble reinforcement, and the moderating role of intolerance of uncertainty (Fig. 1). The findings of this study will contribute to a deeper understanding of maladaptive information use, addressing gaps in research on its effects on information processing. These insights will offer new perspectives for improving information management strategies and optimizing platform designs to reduce selective exposure. In summary, we propose the following hypothesis:

Fig. 1.

Hypothetical model

Hypothesis 1

Information hoarding will be positively associated with selective exposure.

Hypothesis 2

Information overload will mediate between information hoarding and selective exposure.

Hypothesis 3

Identity bubble reinforcement will mediate between information hoarding and selective exposure.

Hypothesis

a: Intolerance of uncertainty will act as a moderator between information hoarding and selective exposure such that this relation will be weaker for low intolerance of uncertainty than for high intolerance of uncertainty.

Hypothesis

b: Intolerance of uncertainty will act as a moderator between information overload and selective exposure such that this relation will be stronger for low intolerance of uncertainty than for high intolerance of uncertainty.

Hypothesis

c: Intolerance of uncertainty will act as a moderator between identity bubble reinforcement and selective exposure such that this relation will be weaker for low intolerance of uncertainty than for high intolerance of uncertainty.

Materials and methods

Participants

This study employed convenience sampling and utilized questionnaires to recruit participants from the Chinese college student population through an online recruitment advertisement. All participants were required to provide informed consent prior to participation, and those who declined were automatically excluded from the study. A total of 769 participants were initially recruited, with 717 valid responses retained after excluding invalid questionnaires, including those with response times shorter than 1-minute, inconsistent answers, or incorrect responses to attention check questions. Of the valid participants, 158 were male (22.04%) and 559 were female (77.96%); 198 were freshmen (27.62%), 229 were sophomores (31.94%), 202 were juniors (28.17%), 30 were seniors (4.18%), and 58 were graduate students (8.09%). On average, participants reported using social networking sites for 4.61 h per day (SD = 2.83) and reading information on social networking sites for 2.56 h per day (SD = 2.29).

Procedure

The research design was approved by the institutional ethics committee of the first author’s university. Participants were recruited through online advertising, and questionnaires were completed through an online platform. After obtaining informed consent, participants anonymously completed the questionnaire based on their actual situation and received a certain amount of remuneration.

Measures

Demographic characteristics

Participants reported demographic characteristics, including gender, grade, and social networking site (SNS) usage. Gender and grade were assessed using multiple-choice questions. SNS usage was measured with two items: “On average, how many hours per day do you spend using social networking sites?” and “On average, how many hours per day do you spend consuming information on social networking sites?” Participants provided their responses in hours.

Informaiton hoarding scale

The present study utilized a modified version of the cyber hoarding behavior scale developed by Bai and colleagues [2]. This modification involved changing the original scale formulation from online resources to online information (e.g., “I saved online information that I think will be useful in the future”). The information hoarding scale consisted of 12 questions that were divided into three dimensions: continuously saving, accumulating clutter, and difficulty in deleting. Participants were asked to rate each item using a 5-point Likert scale, ranging from 1 (not at all) to 5 (completely). Higher scores on the scale indicated more severe information hoarding. In this study, the internal consistency of the scale was good, with Cronbach’s α values of 0.922, 0.941, 0.940, and 0.954 for continuously saving, accumulating clutter, difficulty in deleting, and total, respectively. McDonald’s ω values were 0.923, 0.945, 0.941, and 0.967. The AVE values for the three dimensions were 0.749, 0.803, and 0.798, respectively. Additionally, CFA results demonstrated a good model fit (CFI = 0.972, TLI = 0.963, RMSEA = 0.082, SRMR = 0.029), indicating good construct validity.

Selective exposure scale

The selective exposure scale developed by Peng was adapted by replacing health information with information in the original scale (e.g., I prefer to select information that is consistent with my own views when I searching for online information) [52]. The scale consists of 5 questions and is scored on a 5-point Likert scale, ranging from 1 (not at all) to 5 (completely). The higher the score, the more serious the selective exposure. The reliability analysis showed excellent internal consistency (Cronbach’s α = 0.956, McDonald’s ω = 0.957). The AVE value was 0.815, and CFA results supported a well-fitting model (CFI = 0.994, TLI = 0.988, RMSEA = 0.079, SRMR = 0.009).

Information overload scale

The information overload scale developed by Jensen et al. was used to measure individual senses of subjective information overload (e.g., there is so much information on the Internet now that I don’t know which one to pay attention to) [53]. The scale consists of 7 questions and is scored on a 5-point Likert scale, ranging from 1 (never) to 5 (always). The higher the score, the higher the level of information overload perceived by the individual. The Cronbach’s α for this scale was 0.940, and McDonald’s ω was 0.926. The AVE value was 0.680. CFA results showed an adequate model fit (CFI = 0.957, TLI = 0.928, RMSEA = 0.146, SRMR = 0.032).

Identity bubble reinforcement scale

The identity bubble reinforcement scale developed by Kaakinen et al. was used (e.g., In social media, I belong to one or more groups and I will abide by the rules or responsibilities of this group) [27]. The scale consists of nine questions and is divided into three subscales: social identity, homogeneity, and information bias. Participants were asked to rate each item using a 10-point Likert scale, ranging from 1 (not a description of me at all) to 10 (a complete description of me). The higher the score, the higher the individual’s level of identity bubble reinforcement. In this study, the Cronbach’s α values for social identity, homogeneity, information bias, and total scale were 0.779, 0.903, 0.821, and 0.898, respectively. McDonald’s ω values were 0.813, 0.906, 0.824, and 0.903. The AVE values for the three dimensions were 0.571, 0.759, and 0.608. CFA results showed a moderately acceptable model fit (CFI = 0.904, TLI = 0.856, RMSEA = 0.148, SRMR = 0.081).

Intolerance of uncertainty scale

The study used the intolerance of uncertainty scale developed by Buhr and Dugas [59] and revised by Huang et al. [54]. The scale includes three dimensions of uncertainty impeding action and having negative self-reference, avoidance of uncertainty, and uncertainty causing stress and anxiety, with a total of 11 items. The scale is scored on a 5-point Likert scale ranging from 1 (not at all) to 5 (completely). The higher the score, the less the individual can tolerate uncertainty. In this study, the internal consistency of the scale was high, with Cronbach’s α values of 0.909, 0.885, 0.897, and 0.964 for each dimension and total scale, respectively. McDonald’s ω values were 0.934, 0.905, 0.923, and 0.974. The AVE values for the three dimensions were 0.715, 0.736, and 0.726. CFA results indicated an acceptable model fit (CFI = 0.927, TLI = 0.917, RMSEA = 0.131, SRMR = 0.042).

Results

Common method bias test

To mitigate potential common method bias [55], we implemented several procedural remedies during questionnaire design and administration, including anonymous responses and attention checks. Additionally, to assess the potential influence of common method bias, we conducted Harman’s single-factor test and the unmeasured latent method factor approach. The results of Harman’s single-factor test indicated that seven factors had eigenvalues greater than 1, with the first factor accounting for 41.952% of the variance, which is below the recommended threshold of 50%, suggesting that common method bias is unlikely to be a major concern. Furthermore, we employed the unmeasured latent method factor approach by comparing a baseline model (without a common method factor) and a common method factor model (including a common method factor). The results showed that the model fit indices of the baseline model were CFI = 0.837, TLI = 0.827, IFI = 0.837, RMSEA = 0.089, SRMR = 0.054, while those of the method-factor model were CFI = 0.836, TLI = 0.827, IFI = 0.836, RMSEA = 0.089, SRMR = 0.058. The inclusion of the common method factor did not lead to a substantial improvement in model fit, indicating that serious common method bias was not present in this study.

Descriptive and correlational analysis

Table 1 shows the results of descriptive statistics and correlation analysis for each variable. All variables were significantly and positively associated.

Table 1.

Descriptive statistics and correlation analysis of variable

	M (SD)	1	2	3	4	5
1 Information hoarding	3.05 (0.79)	1
2 Selective exposure	3.35 (0.92)	0.442***	1
3 Information overload	3.24 (0.83)	0.721***	0.488***	1
4 Identity bubble reinforcement	6.00 (1.59)	0.238***	0.217***	0.196***	1
5 Intolerance of uncertainty	3.07 (0.91)	0.622***	0.478***	0.699***	0.111**	1

Notes: ^***p < 0.001

Testing for the proposed moderated mediation model

The study used the Process 3.0 macro in SPSS 24.0 for the moderated mediation model test. The model 15 was selected for the test, and the results were obtained using the bias-corrected nonparametric percentile Bootstrap method with 5000 replicate samples as shown in Table 2. Information hoarding had a significant positive effect on information overload (B = 0.749, p < 0.001), identity bubble reinforcement (B = 0.477, p < 0.001), and selective exposure (B = 0.119, p < 0.05), and information overload had a significant positive effect on selective exposure (B = 0.166, p < 0.01), and identity bubble reinforcement had a significant positive effect on selective exposure (B = 0.077, p < 0.001). This result suggests that information overload and identity bubble reinforcement can mediate between information hoarding and selective exposure, and hypothesis 2c and 3c holds. A significant positive effect of the interaction between information hoarding and intolerance of uncertainty on selective exposure (B = 0.177, p < 0.001), a significant negative effect of the interaction between information overload and intolerance of uncertainty on selective exposure (B = -0.230, p < 0.001), and the interaction between identity bubble reinforcement and intolerance of uncertainty had no significant effect on selective exposure (B = -0.020, p = 0.229).

Table 2.

Testing the moderated mediation model

Regression Equati		Fitting Index			Significance of Coefficients
Outcome	Predictors	R	R²	F	B	t	LLCI	ULCI
IO	gender	0.730	0.533	162.141***	0.140	2.684**	0.038	0.242
	grade				0.002	0.158	-0.027	0.032
	SNS hours				0.015	1.753	-0.002	0.032
	SNSI hours				-0.034	0.011**	-0.055	-0.013
	IH				0.749	27.427***	0.695	0.803
IBR	gender	0.244	0.059	8.970***	-0.078	-0.551	-0.357	0.200
	grade				0.032	0.782	-0.049	0.113
	SNS hours				0.22	0.950	-0.024	0.069
	SNSI hours				-0.011	-0.363	-0.067	0.046
	IH				0.477	6.418***	0.331	0.623
SE	gender	0.573	0.329	31.396***	-0.054	-0.773	-0.193	0.084
	grade				0.012	0.585	-0.028	0.052
	SNS hours				0.003	0.253	-0.020	0.026
	SNSI hours				-0.005	-0.327	-0.033	0.024
	IH				0.119	2.158*	0.011	0.227
	IO				0.166	2.848**	0.052	0.281
	IBR				0.077	3.967***	0.039	0.115
	IU				0.277	6.087***	0.187	0.366
	IH×IU				0.177	3.866***	0.087	0.267
	IO×IU				-0.230	-5.454***	-0.326	-0.154
	IBR×IU				-0.020	-1.205	-0.053	0.013

Notes: ^*p<0.05; ^**p<0.01; ^***p<0.001; SNS hours = daily usage time of social networking sites; SNSI hours = daily usage time for reading information on social media sites; IH = information hoarding; SE = selective exposure; IO = information overload; IBR = identity bubble reinforcement; IU = intolerance of uncertainty; Gender was coded as 1 = male, 2 = female

To further explore the moderating effect of intolerance of uncertainty, the regression coefficients for the relationships between information hoarding and selective exposure, and between information overload and selective exposure, were selected as values of the moderating variables at M ± 1SD in this study to test whether they were significant. Participants categorized as having high intolerance of uncertainty (i.e., those scoring above M + 1SD) comprised 121 individuals, whereas those categorized as having low intolerance of uncertainty (i.e., those scoring below M-1SD) comprised 120 individuals. As is shown in Fig. 2a, the effect of information hoarding on selective exposure was stronger for individuals with high levels of intolerance of uncertainty (simple slope = 0.279, t = 3.885, p < 0.001) than for individuals with low levels of intolerance of uncertainty (simple slope = -0.042, t = -0.632, p = 0.527). As is shown in Fig. 2b, the effect of information overload on selective exposure was weaker in individuals with high levels of intolerance of uncertainty (simple slope = -0.051, t = -0.655, p = 0.513) than in individuals with low levels of intolerance of uncertainty (simple slope = 0.384, t = 6.085, p < 0.001).

Fig. 2.

The moderating role of intolerance of uncertainty in selective exposure. a illustrates the moderating effect in the relationship between information hoarding and selective exposure, while b presents the moderating effect in the relationship between information overload and selective exposure. The y-axis represents predicted values of selective exposure derived from the regression model. The categories of intolerance of uncertainty are defined as one standard deviation above (high) and below (low) the mean (M ± 1SD)

Discussion

Based on the information ecosystem theory [8, 9], this study examined the relationship between information hoarding and selective exposure, considering the mediating roles of information (information overload) and information environment (identity bubble reinforcement), as well as the moderating role of information subject (intolerance of uncertainty).

Information hoarding and selective exposure

The result showed that information hoarding was significantly positively related to selective exposure, suggesting that hypothesis 1 was supported. This result supports the self-validation theory [56]. Self-validation theory proposes that individuals are motivated to create a self-validating environment, which can lead to subjective selective exposure. Individual hoarding provides a self-validating online information environment, which in part enhances the self-awareness [2, 3] and is associated with selective exposure. Specifically, it could be explained in two ways. For one thing, an individual’s memory of hoarded information affects the selection of future information. It was found that when reading new information, individuals pay more attention to and process information that has prior memory [57]. This essentially reflects the “laziness” of human cognitive processing, where individuals tend to exert the least cognitive effort to process information, such as the availability heuristic [58]. Consequently, this cognitive “laziness” predisposes individuals to process new information resembling the previously hoarded information, thereby reinforcing selective exposure. From the other hand, individuals may view the hoarded information as part of the self, representing their views and attitudes [2, 3]. When individuals read information that is inconsistent with the views of the hoarded information, they may have cognitive dissonance reactions [51] and tend to avoid the possibility of conflicting information, which in turn causes selective exposure.

Information overload as a mediator

Second, the findings indicated that there was also a significant positive relationship between information overload and information hoarding and selective exposure. Further, information overload can mediate between information hoarding and selective exposure, supporting hypothesis 2.

Information hoarding is a typical manifestation of “entropy-increasing distress”, and its continuously saving, accumulating clutter, and difficulty in deletion are all related to information overload [2, 13]. First, continuously saving. Under the background of information explosion, it is impossible for an individual to obtain or store all information [3]. Individuals will always be aware that there is still information worth saving or not yet, which may cause the perception of information overload. Second, accumulating clutter. The psychological feeling of confusion caused by clutter is related to information overload [3]. Even sorting out these cluttered information resources can lead to increased cognitive load and stress [12, 13], resulting in the overall sense of “entropy-increasing distress” and information overload. Thirdly, it is difficult to delete the information saved by individuals [59]. Over time, this can make the information hoarded more cluttered and increase the perception of information overload.

In addition, the results of research show that information overload was significantly positively related to selective exposure. The results provided empirical support for the application of cognitive load theory in cyberpsychology [18]. When individuals are confronted with an overwhelming amount of information on the Internet, they often rely on information filtering to manage this information overload [19, 23]. This leads to selective exposure, where individuals attend to information that aligns with their beliefs and ignore conflicting information [20]. Furthermore, the current mainstream research view is that information overload will induce individuals to develop network avoidance behaviors, such as information avoidance [23] and online social avoidance [60]. However, in the age of Internet information, individuals who completely escape and avoid using the Internet may be socially excluded and cause more mental health problems [61]. Therefore, individuals with information overload do not completely avoid information, but will only focus on information that they need or that is consistent with their personal viewpoints, which in turn leads to active selective exposure.

Identity bubble reinforcement as a mediator

The results of the study showed that there was also a significant positive association between identity bubble reinforcement and information hoarding and selective exposure. Furthermore, identity bubble reinforcement can mediate between information hoarding and selective exposure, and hypothesis 3 holds.

First, information hoarding is associated with identity bubble reinforcement. According to the social identity theory [37], individuals tend to engage in social interactions with groups that are similar to their own identity, and this preference holds true in online social interactions as well. In this study, the ego represented by hoarded information [11] influences individuals’ choice of groups in online interactions. Individuals tend to choose groups that are consistent with the hoarded information and their views [27], and produce identity bubble reinforcement. Meanwhile, in the current social media environment dominated by personalized recommendations, individuals’ behaviors like downloading and bookmarking information are factored into the reference range of information recommendation algorithms [32]. Consequently, individuals are more prone to encountering information and information communities akin to what they’ve previously amassed, thereby facilitating communication with specific groups and reinforcing the formation of identity bubbles.

Identity bubble reinforcement is related to selective exposure, providing empirical evidence for motivated reasoning theory [30]. Previous research has found that individuals with identity bubble reinforcement have a strong need to belong and project the need to belong to certain specific groups [27]. The reinforcement of group affiliation through identity bubbles leads individuals to seek information aligned with their group’s perspective while filtering out contradictory information. The outcome of this process results in a phenomenon known as selective exposure. More importantly, identity bubble reinforcement reflects individual self-categorization [27]. By substituting the group self for the original self, individuals attach a specific group label to the self that might otherwise receive other information [26]. This limits communication with individuals outside the group and closes off alternative information channels, resulting in selective exposure.

Intolerance of uncertainty as a moderator

Finally, intolerance of uncertainty plays an enhanced moderating role between information hoarding and selective exposure, such that this relation was stronger for higher intolerance of uncertainty than for lower intolerance of uncertainty, consistent with Hypothesis 4a. Information hoarding is characterized by continuously saving, accumulating clutter and have difficulty deleting [2, 13], all of which can trigger individuals with high intolerance of uncertainty to pay excessive attention to the hoarded information. The excessive attention to information by individuals with high intolerance of uncertainty may reinforce the memory of the hoarded information [68] and thus subconsciously influence selective exposure. In contrast, individuals with low intolerance of uncertainty are more tolerant of cluttered information or less likely to obsess over whether to delete hoarded information [13], meaning they have less concern about hoarding information and are less prone to deep memory and selective exposure than individuals with high intolerance of uncertainty.

It was found that intolerance of uncertainty plays a weakening moderating role between information overload and selective exposure, consistent with Hypothesis 4b and the results of previous studies [47]. Information overload represents the possibility that individuals are exposed to more information with more divergent perspectives [62]. Individuals with high intolerance of uncertainty are more motivated to find the truth when confronted with conflicting perspectives [42, 50]. At the same time, individuals with high intolerance of uncertainty are more likely to spend longer time reading and thinking about information from a wide range of perspectives, which reduces the risk of selective exposure to information. In contrast, individuals with low intolerance of uncertainty are more likely to tolerate contradictory information or to read information in general [43]. This means that individuals with low intolerance of uncertainty have less motivation to think than individuals with high intolerance of uncertainty, which motivates individuals to use heuristic analysis to process and selective exposure to information [58].

The findings also found that the moderating effect of intolerance of uncertainty between identity bubble reinforcement and selective exposure was not statistically significant and was inconsistent with Hypothesis 4c. This indicated that intolerance of uncertainty did not weaken the effect of identity bubble reinforcement on selective exposure. Unlike information overload, identity bubble reinforcement individuals tend to be exposed to an information environment with more consistent information [27]. Also, unlike hoarded information, the information environment of identity bubble reinforcement individuals is less likely to require sorting and categorizing of information [59]. Individuals experiencing identity bubble reinforcement are in groups with specific viewpoints and preferences, and these groups tend to exchange information that is consistent with the group’s values [27]. In a stable, harmonious and consistent information environment, individuals experience less uncertainty, explore less information from different viewpoints, and have an extremely strong selective exposure. Thus, intolerance of uncertainty is less able to play a role in identity bubble reinforcement and selective exposure.

Implications and limitations

This research has some value in theory and practice. In theory, from the perspective of the information ecosystem theory [8, 9], this study discusses the influence of information subject (intolerance of uncertainty), information (information overload) and information environment (identity bubble strengthening) on selective exposure in the era of digital information. Our findings further validate the core tenet of this theory, which posits that the interaction among the information subject, information, and information environment collectively shapes individuals’ information selection behaviors. Meanwhile, this study enriches the information ecosystem framework by incorporating cognitive load theory [18] and motivated reasoning theory [30], thereby clarifying the psychological mechanisms through which core elements of the ecosystem—information and the information environment—interact with individual information processing strategies such as selective exposure. Secondly, this study addresses the research gap concerning the role of information hoarding in information selection, revealing its potential association with selective exposure. This discovery provides new insights for future research on information hoarding and digital hoarding. Last but not least, we find that whether intolerance of uncertainty can mitigate individual selective exposure should be discussed in the context of specific influencing factors or situations. At the same time, we provide empirical support for the behavioral tendency of approach or avoidance of intolerance of uncertainty [48]. In practice, individuals should be mindful of the influence of information hoarding on selective exposure. They should apply critical thinking to distinguish and selectively store useful information, avoiding compulsive information hoarding that leads to information overload and selective exposure. Additionally, individuals must critically assess the views of specific groups to prevent reinforcing identity bubbles. For digital platform managers, strategies to reduce information overload and selective exposure include offering personalized content management tools to help users prioritize information and reduce cognitive overload. Platforms can also design algorithms that promote exposure to diverse viewpoints, counteracting identity bubble reinforcement and fostering a more balanced information environment.

However, there are still some research limitations and future perspectives of this study that should be noted. First, although this study was conducted based on previous studies and classical theories, its cross-sectional design inherently limits the validity of mediation analysis and causal inference [63]. While our findings suggest an association between information hoarding and selective exposure, future research should employ longitudinal studies or experimental designs to better establish mediation models and causal relationships. For instance, experimental manipulations could involve altering the ideological bias of hoarded information and subsequently measuring individuals’ click preferences when browsing new information or using a dot-probe paradigm to assess attentional biases toward future information [64]. Moreover, longitudinal studies should be conducted to examine whether these effects persist over time. Although our findings demonstrate the statistically significant moderating effect of intolerance of uncertainty, the practical impact of changes in selective exposure remains unclear, and its long-term influence on decision-making warrants further investigation. Second, this study did not specifically explore the impact of different information hoarding. Previous studies have found that the Internet information hoarded by individuals has instrumental value and emotional value [2, 31]. These types of information may have different effects on selective exposure behavior, depending on their personal relevance or sensitivity. Third, this study measured the relevant variables by means of a self-report questionnaire method, which may have some bias, especially social desirability bias and recall bias. Future research could adopt experimental approaches to verify findings; for example, there are many classical paradigms in previous studies on intolerance of uncertainty especially in intolerance of uncertainty [65, 66]. Additionally, the absence of control variables in our analysis makes it unclear whether the observed effects are specific to intolerance of uncertainty or linked to broader factors like anxiety and emotion regulation [67–69]. Future studies should include relevant controls to better isolate its unique effects.

Notably, a potential limitation of this study is the gender imbalance in the sample, with 77.96% of participants being female. This gender bias may have influenced the findings, even though gender was included as a control variable. Our results indicate that gender significantly predicts information overload, with females being more likely than males to perceive higher levels of information overload. This finding is consistent with prior research [70, 71], which found that female tend to experience greater cognitive load when facing excessive information. Furthermore, although gender did not emerge as a significant predictor for other variables, its potential influence cannot be overlooked, particularly in selective exposure and intolerance of uncertainty. For example, prior studies suggest that males are more prone to engage in selective exposure processing compared to females [72]. Meanwhile, female typically exhibit lower tolerance for uncertainty than male [73, 74], which may impact the relationship between information hoarding and selective exposure. Therefore, future research should further explore the role of gender in selective exposure, intolerance of uncertainty, and information overload.

Abbreviations

SNS hours

daily usage time of social networking sites

SNSI hours

daily usage time for reading information on social media sites

IH

information hoarding

SE

selective exposure

IO

information overload

IBR

identity bubble reinforcement

IU

intolerance of uncertainty

Author contributions

Study design: Xuqing Bai, Shuailei Lian; Data collection and analysis: Xuqing Bai, Shuailei Lian, Jingjing Liu; Manuscript writing: Xuqing Bai; Manuscript editing: Xuqing Bai, Shuailei Lian, Xiaojun Sun, Gengfeng Niu; Supervision: Xiaojun Sun.

Funding

This research was supported by the Humanities and Social Sciences General Research Program of the Ministry of Education (Grant 22YJA190009).

Data availability

Data and materials can be obtained by contacting corresponding authors.

Declarations

Ethics approval and consent to participate

All human experiments in this study comply with relevant ethical guidelines and regulations, and have been approved by the Human Research Ethics Review Committee of Central China Normal University. The approval number for this study is CCNU-IRB-202311014b. Meanwhile, informed consent was obtained from all participants for this study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.