Keywords

1 Introduction

During the 2020 COVID-19 pandemic, millions around the globe have found themselves to a large extent confined to their homes for weeks on end. Social distancing measures and lockdown regulations have suspended or limited a wide variety of activities that involve in-person interactions [1]. Yet, the need to help others has prevailed, and a host of online initiatives have launched via social media [2].

In Poland, numerous groups offering community help emerged. One of the fastest-growing support groups in Poland, ‘Visible Hand’ (Widzialna Ręka), amassed over 90,000 members during the first weeks of its existence [3].

1.1 Online Interaction

Despite online interaction’s mirroring of ‘real-life’ behaviour, it retains a number of distinct characteristics – many of which have become subjects of research (for a review of the current state of knowledge, see [4]). Studies have demonstrated that online engagement varies, depending on the characteristics of online platforms [5, 6]. Another factor that plays a role in how people experience technology mediated interactions is the form these interactions take. A recent study on copresence and well-being during COVID-19 lockdown showed that copresence experienced by respondents was positively correlated with time spent on video calls (as opposed to audio-only calls) [7]. Importantly, personal disposition affects decisions to engage (or not) in online behaviours; individuals differ in their propensity to interact with others through the internet [8]. Other traits and possible factors identified as those that moderate human-technology engagement and adoption are psychological resilience, optimism, innovativeness, self-efficacy, habit, social influence, risk-taking [9], and individuals’ perception of their security online [10, 11]. Considering the aforementioned research, it is transparent that multiple factors come into play in decisions on whether to engage with others online.

1.2 Help Mediated by Technology

A subset of online social interactions that is less widely researched is altruistic behaviours. The body of literature on the differentiation between online and offline helping behaviour, and on how technology mediates online prosociality, remains limited.

Several studies examining prosocial behaviour online suggest that helping others online is influenced by altruism and reciprocity [12,13,14,15]. Moreover, literature indicates that online and offline prosocial behaviour are positively related [12, 15]. Wright [14] demonstrated that patterns of cyber prosocial and antisocial behaviour imitate their real-life counterparts.

Some personality traits that supported the helping behaviour exhibited by its subjects differed depending on, for instance, their relation to the person who received the help. The participants’ openness to experience contributed to their tendency to help only when strangers were the recipients [16]. Moreover, extraversion contributed to all helping behaviours, while agreeableness only enhanced their tendencies to help friends [16, 17].

Also, situational factors can affect altruism. Public acts of help bring promises of reciprocity, and attract so-called egoists, who are primarily interested in the personal benefits that result from helping. In contrast, altruists more often decide to act prosocially, regardless of personal gain [18].

In the case of helping behaviours, social media enables feelings of engagement with no more than a click, without subsequent action that would realistically fulfil others’ needs (see clicktivism; [19, 20]). Although online donations reach wider networks, they do not necessarily lead to more frequent (or of higher monetary value) donations [21].

Considering the above, we suspect that specific characteristics exist that influence online interactions, such as willingness to interact with others and perception of technology.

The Technology Adoption Propensity (TAP) index and the General Online Social Interaction Propensity (GOSIP), scales rooted in the psychology of individual differences, can assist in predicting prosocial behaviour.

TAP aims to identify the personal variables that moderate the use of technology [11]. Its authors identified four subscales on which human-technology interaction relies: two are considered contributors (Optimism and Proficiency); and two are inhibitors (Vulnerability and Dependence). Their purpose is to indicate whether subjects perceive technology as enabling them to meet their goals more conveniently (Optimism); whether subjects consider themselves able to swiftly master new technologies (Proficiency); how strongly subjects believe that using technology exposes them to being exploited (Vulnerability); and how likely subjects are to become unhealthily dependent on technology (Dependence).

GOSIP scale was developed to understand how personality influences involvement in online conversations [8]. As with willingness-to-communicate, an individual trait responsible for the tendency to initiate and engage in face-to-face interactions [22], the authors of the GOSIP scale attempted to measure individuals’ tendency to be involved in online communication [8]. The measure aims to encompass several aspects of online communication: their enjoyment of online communication; the social needs covered by online interactions; and the role of individuals in group dynamics [8]. These factors contribute to a general disposition towards online engagement that can mediate human-to-human online communication.

1.3 Current Study

We were unable to identify any studies that link technology adoption and propensity for online social interaction to prosocial behaviour; therefore, our hypotheses are driven by other effects described in the literature.

Since openness to experience and extraversion have previously been linked to prosocial behaviour, we expected that propensity for online social interaction would demonstrate similar effects, in addition to significant connections with higher frequency and higher engagement in helping activities (Hypothesis 1a for online and 1b for offline behaviours). We hypothesised that the TAP subscales would connect significantly with individuals’ prosocial behaviour. We expected that TAP Proficiency and Optimism would correlate positively (Hypotheses 2a and 2b, respectively), and TAP Vulnerability negatively with participants’ willingness to participate in helping activities (Hypothesis 2c).

Online and offline activities must be differentiated. We predicted that both TAP and GOSIP would exhibit stronger correlations with online than offline behaviours (Hypothesis 3a for TAP and 3b for GOSIP), since both constructs describe individual tendencies pertaining to technology and internet use. Finally, we hypothesised that TAP and GOSIP were independent predictors of individuals’ helping behaviours (Hypothesis 4a for TAP and 4b for GOSIP), since social interaction and propensity for technology are clearly distinct in the context of the definitions [8, 11].

2 Method

2.1 Participants and Procedure

The study was conducted during the first Polish lockdown at the onset of the COVID-19 pandemic (March-May 2020). Participants were recruited via social media: invitations were published in Facebook groups comprising individuals requesting and offering community help, in addition to groups for university students. Any adult volunteer could participate. Informed consent was obtained from all participants. Aims of the study were not masked. Study was implemented in accordance with the 1964 Helsinki Declaration and its later amendments. Study was approved by the ethics committee of the National Information Processing Institute, Warsaw, Poland. Study was fully anonymous; participants’ personal data was not collected. A preliminary study’ report was available via email. The participants, totalling 234, completed questionnaires. One observation was excluded due to missing data. With consideration for the aims of the study, we analysed only the data from the participants who had declared at least one prosocial activity. Thirty-seven participants (21 women and 16 men, with a mean age of 31.4) declared no such activity; therefore, only 196 observations (162 women, 29 men, and 5 who declared ‘other’, with a mean age of 34) were further analysed.

2.2 Measures

Three independent sub-scales of the TAP index [11] were used to investigate how the characteristics they describe (Optimism, Proficiency, and Vulnerability) related to the frequency and extent to which subjects engaged in helping behaviours, both online and offline. The Dependency subscale was excluded since it was judged irrelevant to our hypotheses. A Polish language translation, courtesy of J. Kowalski, can be found in Appendix 1. In the original study, the internal consistency coefficients varied between 0.73 and 0.87 [11].

A Polish language translation of the GOSIP scale [8] was produced for the purpose of the study. It was generated based on four independent translations and evaluated by four independent experts. The final translation can be found in the Appendix 1. The internal consistency coefficient reported by the authors was 0.92 [8].

Engagement in helping behaviours was assessed using a set of questions devised for the purpose of this study (see Appendix 2). Subjects were asked to state whether they participated in various helping activities during the ongoing lockdown. The activities on which the questionnaire enquired were divided into two groups (online and offline), and included financial or psychological support, and organizing social events online. Responses were given on a scale describing the frequency of occurrence of each behaviour (from ‘never’ to ‘daily’). Consequently, we were able to discern two dependent variables: the number of activities (the number of declared types of activity engaged in); and the frequency of helping (the accumulated frequencies of the declared behaviours). An additional variable, a number of activities x frequencies was created as a simple accumulation of the two previous variables, which reflects both factors.

The reliability coefficients (Cronbach’s Alpha) for TAP and GOSIP are presented in Table 1.

3 Results

Table 1 presents the correlations, means, standard deviations, and internal consistency coefficients for each of the variables used in the study.

Statistical analyses were conducted using IBM SPSS Statistics 26 software. Hypothesis 1a was confirmed: propensity for online social interaction demonstrated significant positive correlation with the declaration of online behaviours (from 0.22 to 0.30; p < 0.01 for the correlations with declarations of online behaviours – see Table 1). However, the correlation with offline behaviours was not confirmed (Hypothesis 1b; 0.5 to 0.10; p > .05). The FDR correction for multiple comparisons revealed significant results [23]. TAP, however, demonstrated significant effects only on the Vulnerability subscale (Hypothesis 2c), which correlated significantly and negatively with all categories of helping (from −0.27 to −0.17; p < 0.05 for all the correlations with online and offline helping declarations – see Table 1). All the correlations with TAP Vulnerability remained significant (p < 0.05) following the FDR correction for multiple comparisons [23]. However, Cronbach's Alpha was moderately low for TAP Vulnerability compared to other subscales (see Table 1; see more in the Limitations section).

The correlation with TAP Optimism (Hypothesis 2a) and TAP Proficiency (Hypothesis 2b) did not correlate significantly with helping; thus, only Hypothesis 2c was confirmed.

Table 1. Correlation coefficients, means, standard deviations and internal consistency coefficients.
Full size table

To test whether TAP and GOSIP exhibited stronger relations with online than offline behaviours (H3a and H3b), we used the Fisher R-to-Z-transformation. For TAP (H3a), our analyses suggested that it was not more strongly connected with the type of declared helping activities, (Z = 0.126, p = 0.45 for online/offline frequency of occurrence; Z = 1.28, p = 0.10 for number of activities; and Z = −0.515, p = 0.303 for number x frequency). For GOSIP (H3b), the relationships with online frequency, number of activities and variable combining frequency, and number of activities were stronger than those declared offline (Z = 2.664 p = 0.004; Z = 2.678 p = 0.004; and Z = 1.998, p = 0.023, respectively).

Finally, to test whether TAP and GOSIP were independent predictors of helping behaviours (Hypotheses 4a and 4b), we conducted a series of regression analyses (see Table 2) with frequency and number of declared helping activities as dependent variables and TAP Vulnerability and GOSIP as predictors. The analyses revealed that TAP Vulnerability (H4a) and GOSIP (H4b) independently explained the variance of helping behaviours. Moreover, we observed that TAP Vulnerability alone was significant in explaining offline behaviour, whereas, during analysis of online activities, both TAP Vulnerability and GOSIP were significant predictors. GOSIP, however, had higher incremental R in all models that analysed online activities, adding more explained variance to each model when entered to the analyses after TAP Vulnerability. In summary, Hypotheses 4a and 4b were confirmed extent: GOSIP and TAP proved to be independent predictors of online helping – although only the TAP Vulnerability subscale was considered.

Table 2. Regression analyses with type of help as dependent variables with TAP Vulnerability subscale and GOSIP as predictors.
Full size table

4 Discussion

This study investigated the relationships of technology adoption and propensity for general online social interaction with online and offline helping activities during the first Polish COVID-19 lockdown. The results demonstrated support for Hypothesis 1a: propensity for social interaction related significantly to higher frequency and to higher numbers of declared helping activities, both online and offline. In TAP, only the Vulnerability subscale correlated significantly with the dependent variables (Hypothesis 2c). The results returned small negative correlations with both the number of activities and the number of frequencies. In propensity for general online social interaction, Hypothesis 3b was also confirmed: the correlations were significantly higher for online than for offline behaviours. In TAP (H3a), the difference was insignificant: no stronger correlation for online activities was observed. For online activities, TAP and GOSIP were significant and independent predictors of offering help. TAP also proved a significant predictor for offline helping offers (Hypotheses 4a and 4b).

It is noteworthy that the majority of items in the questionnaire required high degrees of interaction with others. It was unsurprising, therefore, that the respondents’ results on the GOSIP scale correlated positively with the frequency and the number of types of helping behaviour they offered. Predictably, this relationship was stronger for online activities, since GOSIP was designed to measure individuals’ propensity to engage specifically in online interactions. Yet, it would be reckless to conclude that propensity to interact with others is a straightforward predictor of more general altruistic behaviours. As other study reports, the relationship between extraversion and altruism might not be linear, but rather U-shaped, meaning when playing the dictator game, individuals with very high or very low extraversion scores (−2SD) were more likely to give a higher percentage of the endowment to the other player than those who scored moderately low (−1SD) [17]. With this knowledge, charity groups or services would benefit from enabling various ways of engaging with their causes. The altruistic potential of highly extroverted individuals can easily be harnessed in direct forms of assistance, while introverts would more likely engage in other forms – specifically, volunteer work that can be accomplished alone, without the need to interact with or approach others.

The negative correlation between Vulnerability and the frequency of engagement in helping activities, both online and offline, compels us to enquire why this technology-related scale applies in a similar way to such behaviour, regardless of whether it demands the use of technology. An individual’s score on the Vulnerability scale should reflect to what extent that person believes that using technology would increase their chances of falling prey to malicious schemes, or of becoming a victim of financial fraud or identity theft. These fears might be related to anxious personality disposition [24, 25], which, in turn, might inhibit individuals’ proneness to empathetic reactions. There is evidence that experimentally evoked anxiety decreased the strength of empathy responses [27, 28]. Our results align with these findings, as we failed to prove the technology-specific aspect of TAP Vulnerability subscale. Anxiety might limit both online and offline altruistic behaviours similarly. It may inhibit one’s initiative to act or react due to threatening interpretations of the reality, or concentration on one’s own safety. Lam, Chiang and Parasuraman found that feelings of insecurity around technology use were rare for services deemed low-risk – even among individuals who tended to deem technology unsafe [28]. The designers of successful online platforms understand that the fear of being exploited via technology can be mitigated by a variety of social cues [29]. Users’ perception of the extent to which they are capable of applying safeguarding strategies helps to predict whether they will decide to engage in helping behaviours or not.

Our findings demonstrate that although GOSIP is a significant predictor of online community assistance, its accuracy fails to extend offline. We conclude that GOSIP measures a technology-specific type of engagement in social interactions. This might also suggest that the social aspect of helping behaviours varies between the online and offline worlds. We may conclude that those who are socially engaged online could also be effective online helpers. TAP and GOSIP both independently predicted engagement in helping behaviours. Such results indicate that separate processes are responsible for the effects. The absence of significant correlation between TAP Vulnerability and GOSIP appears to confirm this interpretation. It is probable that individuals are differently motivated to help; thus, when facilitating helping behaviours, it is imperative that this diversity is accommodated. The independence of TAP and GOSIP in this study also offers evidence for differentiation between the two constructs, which to our knowledge, have not previously been tested. The lack of correlation between TAP Vulnerability and GOSSIP may appear surprising, since in the eyes of many a tendency to interact with others should be negatively connected with one’s internet mistrust. However, in our opinion independence of GOSSIP and TAP Vulnerability simply adds to perceived accuracy of both constructs: GOSSIP reflects individual attitudes toward social online interactions and TAP towards technology.

4.1 Limitations and Future Directions

This study was correlational in nature and the participants were recruited on social media platforms. Conducting a similar study with the use of a more representative sample could serve to eliminate this limitation. An experimental approach would allow us to explore ways of facilitating helping behaviours by proposing different types of engagement that are better suited to individuals – including those who possess lower extraversion or need for security. Future studies could investigate whether other variables describing individual traits, such as extraversion or neuroticism, act as predictors of online altruism. Such research could potentially mediate the effects reported in this study.

Moreover, the progression of the COVID-19 pandemic could entail limited offline help due to the public anxiety around such exceptional circumstances. The number of suggested online activities, therefore, was likely inflated, when compared to the ‘normal’, pandemic-free context; thus, a future study with consideration for helping methods prior to the emergence of COVID-19 could prove useful. Consideration for the perceived pandemic threat could serve as a relevant mediator to such a study.

This study has also elucidated issues on the content validity of applied questionnaires. A study on the external validity of TAP would be compelling in terms of exploring usability –particularly that of the TAP Vulnerability subscale – which is not cyber-specific and reaches beyond the personality traits on which this study focused. Additionally, the Cronbach’s alpha coefficient of our results attained by TAP was relatively low; it is noteworthy, however, that the authors of TAP reported similarly low alpha coefficients, which probably also results from the length of the scale, i.e., TAP Vulnerability subscale consists of only 3 items [11]. Last but not least, a longer research exploring the role of TAP and GOSIP in online behaviours could involve all the TAP subscales including the Dependence subscale. Although this study in no way addressed social media and internet addiction, it is worth adding this aspect into future research considerations. It is possible that helping others online may be one of the behaviors undertaken to satisfy the need to represent oneself on social media, or some way to rationalize the large number of hours spent online.

4.2 General Conclusions

This study occupies a gap in the literature by offering new insights into the role of human attitudes towards online interactions and technology in helping behaviours. Understanding what may limit some individuals’ propensity to participate—in addition to which features might remedy their concerns—should assist in enhancing engagement in social initiatives. Online platforms carry unique potential in enabling ways to engage in altruistic behaviours for those who avoid social interactions. Facilitating methods of participation that meet varying individual needs may result in the inclusion of volunteers who are less socially engaged and are more anxious in their daily lives.