Environmental citizenship behavior and sustainability apps: an empirical investigation

Purpose – This study aims to investigate the moderating effect of sustainability app on environmental citizenshipbehavioronthebasis of norm-activation model. Design/methodology/approach – A questionnaire survey, which comprises ﬁ ve variables (i.e. awareness of consequences, ascription of responsibility, personal norms, environmental citizenship behavior in a private sphere and environmental citizenship behavior in a public sphere) measured through 16 items, was conducted in the USA by using Amazon Mechanical Turk. With 549 valid respondents ’ answers in hand, the collected data were analyzed applying a multi-group structural equation modelling technique with IBM SPSS AMOS 23 softwareprogram. Findings – The results revealed that there is a positive and signi ﬁ cant relationship between awareness of consequences, ascription of responsibility, personal norms and environmental citizenship behavior in both private and public sphere. Furthermore, this study attested that sustainability apps utilization has a moderating effect onthepredictorsof environmental citizenshipbehaviors. Originality/value – Past studies have seldom examined the contribution of mobile apps to environmental sustainability.Thispaperenrichestheextantacademicliteraturein the ﬁ eldoftechnologyforbehaviorchange,and bears signi ﬁ cant implications on how sustainability apps can be adopted by governments, policymakers, organizationsandteachereducatorstoengagepeopleandstimulateenvironmentalcitizenshipbehaviors.


Introduction
Sustainability, defined as "the development that meets the needs of the present without compromising the ability of future generations to meet their own needs" perceived as manipulative, such advertising campaign that makes individuals feel fearful, obligated or guilty.In fact, empirical studies demonstrate that alarmistic and fear messages are often not effective in creating a behavior change (de Vries, 2020).
In search of innovative and interactive ways to inspire and motivate citizens to be more sustainable in their everyday life, the use of mobile applications (apps) has gained the attention of scholars and practitioners in recent years (Brauer et al., 2016;Nghiem and Carrasco, 2016;Ouariachi et al., 2020).The relationship between technologies and human actors can give birth to new processes, values, and social rules (Troisi et al., 2018;Troisi et al., 2021).Information and communication technologies (ICTs) also contributes to the developments and advances in the practice of teaching and learning worldwide (Visvizi et al., 2018a).Specifically, the utilization of sustainability apps presents huge potential to enhance awareness about sustainability (Yanamandra and Ramesh, 2019) and induce change in behavior without adopting hard paternalism forms, that is, actions that interfere with individuals' liberty or autonomy (Diefenbach et al., 2016).
An analysis of extant literature revealed that researches on sustainability apps have prevalently focused on their classification (Brauer et al., 2016), design characteristics (Mulcahy et al., 2020), adoption intention (Aguiar-Castillo et al., 2018;Whittaker et al., 2021), and educational function (Abner and Baytar, 2019;Montiel et al., 2017).Simultaneously, there is a paucity of empirical evidence on how such technologies might intervene in the process leading to environmental citizenship behavior.

Environmental citizenship behavior
For that reason, the purpose of the current study consists of filling this knowledge gap by conducting a multi-group analysis to evaluate the moderating effect of the use of sustainability app on environmental citizenship behavior.
The remainder of the paper is organized as follows.In Section 2, the background is presented and the research hypotheses are developed.The measurement instrument and data collection procedure are reported in Section 3. Statistical analysis methods and results are introduced in Section 4. Subsequently, in Sections 5, we report a discussion of the findings and implications for theory and practice.In Section 6, conclusions, research limitations and future research directions are presented.

Norm activation model
According to Smederevac-Lalic et al. (2020), environmental citizenship behaviors have a normdriven nature.Therefore, we adopted the norm-activation model developed by Schwartz (1977) as our basic model to explore environmental citizenship behavior.Furthermore, we evaluated the moderating effect of a specific contextual variable, namely, sustainability app usage, on the relationships between the constructs that characterize the norm-activation model.
The norm-activation model is grounded on the assumption that an individual is disposed to sacrifice her/his own self-interest for the collective benefits of others.As highlighted by Schwartz (1977), this form of altruistic behaviors is influenced by personal norms, awareness of consequences and ascription of responsibility.
In the environmental psychology literature, personal norms are defined as "moral obligations to perform or refrain from specific actions" (Schwartz and Howard, 1981, p. 191).Personal norms are cognitive structures varying from subject to subject constructed on the basis of inputs from the external situation and internal reasons in agreement with values, beliefs, conceptions of right and wrong, good or bad (Schwartz, 1977;Thøgersen, 2006).
Awareness of consequences refers to the individual's disposition to become aware of the potential consequences of her/his behavior for the welfare of others or for other things (Schwartz, 1977).If an individual is aware of the consequences that her/his actions may have on others, then norms guiding how she/he should or should not behave are activated and feelings of moral obligation are induced (Kaiser and Shimoda, 1999).For example, if a person is aware of the effects of plastic pollution on marine life, he/she may be likely to engage in recycling of plastic waste.
Ascription of responsibility concerns with the individual tendency to see "the self as responsible for events initially" (Schwartz, 1977, p. 230).People's actions can have positive and negative consequences on the other people, the other species and the environmental wellbeing.Therefore, as highlighted by Stern et al. (1999, p. 83), ascription of responsibility is "the belief or denial that one's own actions have contributed to or could alleviate those consequences".For example, an individual tends to view energy saving in the workplace as her/his own responsibility rather than of her/his organization.

The moderating role of sustainability apps
ICTs are generally a viable way to influence individuals' behavior (Brauer et al., 2016).Example from the health, education and public service domain show that the utilization of apps can serve as successful interventions (Carlo Bertot et al., 2012;Eid et al., 2020;Hirsh-Pasek et al., 2015;Oni et al., 2016;Pai and Alathur, 2019;Reddick and Zheng, 2017).Apps are dedicated software applications that run on portable devices such as smartphones and tablets (Gokgoz et al., 2021).App stores like those from Apple and Google offer the possibility to download different types and categories of apps, for example, social networking apps, online shopping apps, gaming apps, apps for food lovers, mobile wallet apps, health apps and education apps (Mehra et al., 2020).Recently, apps have also proven to be helpful achieving sustainability-related goals (Brauer et al., 2016).Some examples concerning the domain and functions of the most popular sustainability apps are illustrated in Table 1.
Since we are dealing with a young product category, there is little research on sustainability apps.Most studies in the field mainly focused on how technology could be designed to actively influence and change human behavior.One design principle is the "aesthetic of friction" (Laschke et al., 2015), that is, break up the routine to inspire reflection (Diefenbach et al., 2016).For example, a data-driven app, such as a carbon footprint calculator, could make us think about how much our lifestyle is unsustainable and thus render the choice between taking the car or the bike more deliberate.In their study, Brauer et al. (2016) highlighted that sustainability apps could be implemented with one or more of the following functions: educate, gamify, informate, transformate and collaborate.

Environmental citizenship behavior
According to Georges et al. (2015), the main persuasive techniques used by app developers to help people to live more sustainably are eco-feedback, reminder, reward, self-monitoring tool, suggestion and trigger.Several studies (Douglas and Brauer, 2021;Johnson et al., 2017;Mulcahy et al., 2020) also suggest that gamification helps creating a condition favorable for the pro-environmental education of the individual.Apps help reducing the ignorance barrier and contribute to the habit formation.Specifically, gamified apps try to deal with the problem of personal motivation by adopting extrinsic motivation such as points, levels, discounts and badge.
Based on the discussion above, which prevalently focuses on how apps can assist behavior change, we hypothesize that sustainability apps can help strengthen the relationship between personal norms, environmental awareness and responsibility, as well as motivate environmental citizenship behavior in both private and public settings.Specifically, we derived the following hypotheses: H1.The correlation between awareness of consequences and ascription of responsibility is stronger for individuals who use sustainability apps.
H2.The correlation between ascription of responsibility and personal norms is stronger for individuals who use sustainability apps.
H3.The correlation between personal norms and environmental citizenship behavior in private-sphere is stronger for individuals who use sustainability apps.
H4.The correlation between personal norms and environmental citizenship behavior in public-sphere is stronger for individuals who use sustainability apps.
The conceptual model underlying this study is presented in Figure 2.

Measurements and instrument development
This research adopted a self-administered questionnaire survey technique.The questionnaire (Table 2) was designed to explore five different variables and comprised 16 items measured on a seven-point Likert scale (1 = Strongly disagree to 7 = Strongly agree).
The questionnaire started with a filter question to identify two different subsets of survey respondents, namely, individuals who use sustainability apps and individuals who do not use sustainability apps.Individuals who answered to use sustainability apps had also to specify the name of the app and the frequency of usage before to jump to the common section containing a battery of questions concerning the measurement of the environmental citizenship behavior.Questions regarding respondents' demographic information were included at the end of the survey.
The items measuring awareness of consequences, ascription of responsibility and personal norms were adapted from Onwezen et al. (2013).The environmental citizenship behavior was measured on two dimensions, namely, the private-sphere and the publicsphere.The items were generated by taking in consideration the examples of environmental citizenship behavior provided by extant literature, such as Hadjichambis and Paraskeva-Hadjichambi (2020).

Data collection
Amazon Mechanical-Turk (MTurk) was used to recruit participants for our research (Shank, 2016).The survey was lunched June 16, 2021.In the panel option, we specified the target of our online survey.In addition, we selected 550 as number of MTurk workers.In a few days, 550 questionnaires from respondents living in the USA were collected.One questionnaire was excluded from the analysis because the name of the app was not inserted.Therefore, the final dataset comprised 549 valid answers.As presented in Table 3, 53.6% of the participants are female.Most participants are between 18-24 (50.1%) and 25-34 years old (27.3%).Participants who received higher education accounted for 81.3%.Of the 549 respondents, the 48% of participants use sustainability apps.

Data analysis and results
This research employed IBM SPSS Statistics 25 and IBM SPSS AMOS 23 to perform the data analysis.First, we conducted a confirmative factor analysis (CFA) and inspected the reliability and validity of the measurement model.Second, to evaluate the moderating effect of sustainability app utilization on each relationship between environmental citizenship behavior and its predictors we performed a multi-group analysis via partial least squares structural equations modeling.

Reliability and validity analysis
Prior to assess the reliability and validity of the measurement model, with the help of IBM SPSS Statistics 25 we conducted the descriptive statistics analysis and the normality test.As shown in Table 4, skewness and kurtosis value for each item was below 63 and 610 respectively (Kline, 2011); hence, data were normally distributed.

Environmental citizenship behavior
We also estimated the mean of each items of the measurement model and compared the results between the two different groups.As illustrated in Figure 3, with the exception of few items (i.e.AC_2, ECBPUB_2 and ECBPUB_3), the level of agreement of respondents who use sustainability apps is higher than that of respondents who do not use sustainability apps.
The measurement model was conducted using confirmatory factor analysis (CFA).IBM SPSS AMOS 23 was used to perform the CFA.The results of the CFA revealed a good fit.Specifically, the Root mean square error of approximation (RMSEA) = 0.048, the Bentler's comparative fit index (CFI) = 0.977, the Tucker-Lewis index (TLI) = 0.970 and the normed fit index (NFI) = 0.959.In addition, the chi-squared test denoted good model fit too (CMIN/df = 210.675/94= 2.241, p < 0.001), the value, in fact, was less than 3.0 (Hair et al., 2010).To examine the reliability of the measurement model, we used Cronbach's alpha values and composite reliability (CR) values respectively.As depicted in Table 5, Cronbach's a value ranged from 0.749 to 0.938; hence met the cut-off value of !0.70 (Hair et al., 2010).CR ranged from 0.756 to 0.939 that meets the suggested criterion of !0.60 (Hair et al., 2010).Therefore, the results suggest that the reliability is acceptable.The validity of the measurement model was estimated by examining both convergent validity and discriminant validity (Hair et al., 2010).Average variance extracted (AVE) was used to measure convergent validity.As shown in the Table 5, all the AVE of each construct, which ranged from 0.548 to 0.837, also met the suggested criterion of !0.50 (Hair et al., 2010).Thus, the convergent validity is acceptable.Finally, to test the discriminant validity, we compared the square root of the average variance extracted ( ffiffiffiffiffiffiffiffiffiffi AVE p ) with the correlations among the five constructs.The ffiffiffiffiffiffiffiffiffiffi AVE p of each construct was higher than the off-diagonal Environmental citizenship behavior correlation values.Therefore, according to Fornell-Larcker criterion analysis, discriminant validity was supported (Fornell and Larcker, 1981).

Multi-group analysis
A multi-group analysis using IBM SPSS AMOS 23 was conducted to examine the moderating effect of sustainability app utilization.Research participants were divided in two groups, those who use sustainability app in their everyday life (Group 1; n = 261), and those who do not use sustainability apps (Group 2; n = 288).Following Byrne (2004), to determine whether the structural model of the two groups are statistically different from each other, we performed a multi-group invariance testing.The p-value of the chi-square difference test between the unconstrained (X 2 = 417.725,df = 80, p < 0.001) and the constrained (X 2 = 449.018,df = 65, p < 0.001) subset models is significant (DX 2 = 31.293,Ddf = 15, p = 0.008).This means that the hypothesis of invariant factor variances must be rejected; hence, the model differs across the groups.
In both models, the relationships between variables are all positive and statistically significant.However, the model regarding the individuals who use sustainability apps presents larger standardized path coefficients (b ) and coefficient of determination (R 2 ).H1, H2, H3 and H4 are supported.Therefore, sustainability apps utilization exerts a moderating effect on the relationship between awareness of consequences, ascription of responsibility, personal norms and environmental citizenship behavior in both private and public sphere.The detailed results are depicted in Table 6, and Figure 4 respectively.
By adopting a multi-group approach, we simultaneously reproduced the path differences between two groups and reported results in either situation.The main strength of this technique regards the fact that it can be easily executed with software; hence, it is less time consuming than qualitative approach.On the contrary, this quantitative technique focuses on theory testing rather than on theory generation.Furthermore, it might lead to reductionist explanations.Table 7 summarizes the strengths and weakness of the multigroup approach adopted in this research.Second, to the best of our knowledge, this is one of the first study investigating the role of sustainability apps in the relationship between environmental citizenship behavior and some of its predictors, namely, personal norms, ascription of responsibility and awareness of consequences.Third, through the multi-group analysis, it was found prominent differences between individuals who use sustainability apps and those who do not use sustainability apps.Specifically, sustainability apps utilization strengthen the relationship between ascription of responsibility and personal norm.For example, an app such as AWorld uses stories and information about sustainability to encourage the members to take pro-environmental actions and build a shared sense of purpose.Furthermore, individuals who use sustainability apps are more prone to translate their personal norm into environmental citizen behavior in both private and public setting.Therefore, this study supports the hypothesis regarding the moderating effect of sustainability apps.
Fourth, this research indicated that individuals tend to show their environmental citizenship especially through actions associated with their private-sphere, such as reducing  household energy consumption, recycling and opt for eco-friendly products.One possible reason for this finding may be due to the particular model of environmental education.
According to Chawla and Cushing (2007), environmental education focuses principally on the private-sphere environmentalism rather than preparing students for public actions, such as act, protest, lobbing and participation in environmental movements.
The urgency to reimagine and recreate a non-formal as well as formal environmental education for children, youth and older people has been highlighted in a recent paper published by Reid et al. (2021).Specifically, the authors invoke an environmental education grounded on critical thinking, close to science and that enables individuals to identify fake information and ideologies that underestimate the relationship between economic growth and environmental impact.
Another explanation may lie in the fact that sustainability apps, except sporadic cases (Buycott app), are designed to achieve personal sustainability goals.However, as stated by Dobson (2007), private actions have also public implications.For example, our choice to live sustainably allows that others may live well.Furthermore, our actions can inspire other individuals and motivate them to change behavior.
The current study offers to governments and policymakers some practical implications for transforming society through the adoption of ICTs such as sustainability apps.These types of technologies can be used to engage those people who would like to adopt a more sustainable lifestyle but they need a final push.The objective of sustainability app is to trigger behavioral change by means of specific alerting features and gamified scenarios that may be found not to be necessarily cost-effective.
Central governments and cities could develop mobile apps to achieve specific sustainability goals and furnish to citizens personalized services.Furthermore, sustainability apps can be integrated into smart city ecosystems (Kashef et al., 2021;Lytras et al., 2019;Visvizi et al., 2018b;Lytras and Visvizi, 2018;Troisi et al., 2022) to improve pro-environmental activities such as recycling, reduce food waste, donating clothes and sharing stuffs.
As knowledge, skills development and environmental awareness are a prerequisite for environmental citizenship behavior, public and private educational institutions, including universities, can adopt sustainability apps to foster innovative environmental educational approaches that emphasize reflection on direct, concrete experience.Furthermore, sustainability apps favorite value co-creation in educational context (Loia et al., 2016), as well as human connections and a networked learning (Lytras et al., 2018).

Conclusions
In this paper, we assessed the potentials of sustainability apps to contribute to environmental citizenship behaviors.Hence, the study provides insights about how the adoption of this specific technology, which takes advantage of the ubiquitous of smartphones and other portable devices, may strengthen the relationship between environmental citizenship behavior and its predictors based on the norm-activation model.
This study has some limitations.First, we used MTurk to test the hypothetical model.This crowdsourcing marketplace has a small population.Therefore, this might compromise data quality due to the potential expositions of the worker to previous similar surveys (Chandler et al., 2019).Future research could consider alternative sources for the target audience (e.g.emails, and face-to-face interview).
Second, this study collected data from the USA.Future research should collect and compare data from different countries to enhance the generalizability.
A third limitation regards the SEM analysis.We estimated only the direct effects of awareness of consequences and ascription of responsibility.Additional research could Environmental citizenship behavior consider personal norms as a mediator (De Groot and Steg, 2009), and examine the indirect effects of awareness of consequences and ascription of responsibility on environmental citizenship behavior in both private and public sphere.Moreover, future research could consider introducing in the conceptual model a control variable, such as usage frequency or app category.
Like most research, we derived our conceptual model from the extant literature.The norm-activation model showed consistent findings.However, a plethora of potential constructs, such as values and social norms, may be considered within subsequent work.
Figure 1.Examples of environmental citizenship actions in a for-quadrant representation Figure 2. Conceptual model Figure 4. Results of the multigroup structural equation modeling

Table 1 .
The app connects neighbors with each other and local shops so that surplus food can be shared