1 Introduction

In a scene of the Hollywood movie “Her”, Theodore, the main character, purchases an operating system (OS) with artificial intelligence that is designed to adapt and evolve. When Theodore runs the OS, a male voice asks three personal questions: (a) Are you social or anti-social? (b) Would you like a male or female voice? And (c) how would you describe your relationship with your mother? Theodore answers the questions and the OS restarts and it is now customized according to those answers. This scene of the movie exemplifies the aim of this paper, to explore how users’ expressions of identity can be used for information systems’ personalization.

Information technologies (IT) have affected the social structures where our individual and social lives are embedded [1]. For example, individuals’ constant mobility and the lack of division between home and work are supported by technology developments [2]. Prior research analyzed the effects of the relation between IT and individuals’ identity. In this regard, Carter and Grover [3] reviewed and classified those studies according to their focus on: (a) IT as a determinant of individuals’ identity development, where for example Stein et al. [4] analyzed the role of IT artifacts in professionals’ identity construction; (b) IT as a medium for individuals’ self-expression, where for example Walther [5] analyzed how individuals utilize computer-mediated-communications to manage others’ impressions of themselves; and (c) IT as a consequence of individuals’ identities, where for example Lee et al. [6] suggested that work-related IT is easily adopted when aligned with the worker’s self-perception.

Despite these results, and to the best of our knowledge, no study has analyzed the way in which an individual’s identity can be used for IT customization. This concept, also known as personalization, refers to “the ability to provide content and services that are tailored to individuals based on knowledge about their preferences and behaviors” (p. 84) [7 as cited in 8]. The underlying assumption of customization is that harmonizing the IT attributes with users’ attitudes and values will lead to greater users’ satisfaction.

In order to address this gap, in this work-in-progress paper we explore (1) how a social psychology variable like identity can be used to customize a particular information system (IS), and (2) the extent to which individuals are willing to adopt such tailored IS in the face of privacy concerns and the possibility of using such IS in different contexts of their life (e.g. at work, at home).

2 Theoretical Background and Research Model

2.1 Social Identity Theory

One of the most important conceptualizations about the relation between individuals and society was developed by James in his book, ‘Principles of Psychology’ [9]. In the book, the author explains that the sense of self emerges in the interaction between how I see myself (‘I’ self), and how others see me (‘me’ self) in a particular social context. This conceptualization places an individual’s sense of self as a constant cognitive process that emerges in her/his social interactions [10].

One of the theoretical perspectives that has analyzed the dynamics among the society, the individual, and the self is called Symbolic Interactionism [11]. From this perspective, in every social network individuals have different positions known as roles (e.g. parent), and these positions have different expectations about how they should be played (e.g. taking care of children) [12]. In individuals’ recurring interactions with their social networks (e.g. family), they recognize themselves as occupants of these positions and choose to play them accordingly [13]. Through the role-playing, individuals identify and internalize these positions, developing different identities around them. Therefore, individuals have many identities as they play different societal roles, and identities are constantly constructed, revised, and changed across individuals’ different social interactions [14].

From a Symbolic Interactionism perspective, these individuals’ behaviors are associated with their social identifications. According to Stryker [15], the individual self is structured by different identities that are organized in a salience hierarchy. Identity salience is defined as “the probability that an identity will be invoked across a variety of situations” (p. 286) [12]. Hence, in a specific situation individuals’ behaviour is the result of the salience of an identity associated with the role they perform in the social structure. For example, an individual might have a parent and worker identities. In the specific situation of buying a car, if the individual has her/his parent identity as first in the identity hierarchy salience, she/he will prefer the car’s features associated with safety (e.g., air bags) to the car’s features associated with his worker identity (e.g., power, speed).

In sum, from a Symbolic Interactionism perspective, social structures are made of interconnected positions referred to as roles. Each role is linked to activities (e.g., practices) and resources (e.g., products), which in turn become symbols that convey meanings through which individuals interact with their social network. In the individuals’ interaction with society, their interpretation and enactment of these roles generate a self that is structured by different identities. With time, individuals’ patterned social interaction maintains and facilitates the development of new social structures [12].

This theoretical framework is suitable for this study since the ubiquitous nature of IT allows individuals to perform different roles (e.g., co-worker, parent, friend) while interacting with technology [3]. In addition, the salience of a particular individual’s identity may be utilized as a means to customize the IT with which that individual interacts. In this way, the customized IT would become one of the resources the individual utilizes to express her/his salient identity.

2.2 Research Model and Hypotheses

The proposed research model is shown in Fig. 1. The constructs and hypotheses included in the model are described below.

Fig. 1.
figure 1

Research model

Full size image

User Information.

Users’ demographic information has been utilized for personalization, web search, and targeted advertising purposes [16, 17]. In particular, demographics-based recommendation systems have been used as a method for personalization in e-commerce interactions. In those situations, demographic information is explicitly obtained from users and the site may recommend products based on the preferences of users with similar demographics [18]. However, this approach requires that companies are able to collect complete and reliable demographic information and users may be reluctant to provide such information due to inconvenience or privacy concerns [17, 18]. Moreover, this approach assumes that in a particular moment in life, individuals have similar needs and wants [19] and this may not be the case. Therefore, and as suggested by the marketing segmentation literature, this approach might be supplemented by using psychographic and behavioral variables [20].

One such variable could be a user’s identity. An individual’s identity is known to be a motivator of behaviors [3, 21, 22]. In the marketing literature, it has been found that individuals use brands to express and validate their identity [23–25]. Consumers may pay more attention to identity-related stimuli (e.g., a product that appeals to a particular identity), react more positively to advertisement featuring individuals with a desired identity (e.g., a celebrity) or engage in behaviors that are linked to their identity [26]. Moreover, consumers can customize the products they acquire (e.g., selecting a particular ringtone in a mobile phone) to reflect their identity [3]. Considering these findings, it could be expected that users would be willing to adopt an app that is personalized by using their identity information. Users would engage in what is known as verification activities (i.e., behaviors that reinforce individuals’ identities) [27] by using such an app [3]. Bearing in mind the limitations of using a demographic-based customization and the potential of creating a positive association between an individual’s identity and a product (e.g., a personalized app) [26], we propose that users will be more willing to adopt a personalized app that has elicited their identity-related information than one that has elicited their demographic-related information. Therefore, we hypothesize that:

H1: Eliciting identity-related information as the user information will lead to a higher intention to adopt the personalized app than eliciting demographic-related information.

Privacy Concerns.

Information privacy refers to individuals’ ability to control the extent to which their information is acquired and used [28, 29]. Information privacy concerns (herein to be referred to as privacy concerns) refer to “an individual’s subjective view of fairness within the context of information privacy” (p. 337) [30]. Privacy concerns have been found to be a detrimental factor for individual’s willingness to provide personal information to companies and to conduct e-commerce transactions (see for example [31–33]). In the same vein, previous research has found that privacy concerns may negatively influence individuals’ intention to use personalized services online (either through computers or mobile devices) (see for example [8, 34, 35]). In light of the previous results, it is expected that individuals will have privacy concerns related to the collection of their personal information (either demographic- or identity-related) and that those privacy concerns will reduce their intentions to adopt a personalized app that requires providing such personal information. Thus, we hypothesize that:

H2: Privacy concerns are negatively related to the intention to adopt the personalized app.

Usage Across Domains.

As mentioned in the introduction, IT is ubiquitous in people’s social and personal life and as such, it influences their identity development [1]. This has become more salient as IT has become more interconnected (i.e., usable across devices), and people are able to use it across different life situations [3]. At an individual level, people’s enactment of certain roles and identities involves the use technology. For example, a person can use a particular budget application for both her/his work expenses using a desktop computer and her/his house expenses using her/his mobile phone. In addition, at a social level, IT also intertwines with individuals’ social networks. For example, in a single Facebook account, individuals might have work and family friends or groups of friends and may employ different mechanisms to manage their roles (e.g., worker and parent) separately (e.g., restricting content to some viewers) [36]. Considering that individuals receive several benefits from utilizing IT (e.g., mobile devices) in different spheres of their life, such as managing time more effectively, and having flexibility in the performance of tasks [37], it is expected that individuals will be more willing to adopt a personalized app from which they could obtain such benefits. Therefore, we hypothesize that:

H3: The usage across domains is positively related to the intention to adopt a personalized app.

3 Methodology

The hypotheses proposed in the research model will be validated through an experiment with a two-group, between-subject design. The two groups involve a different type of user information collected: identity-related versus demographics-related information used to customize the IS.

3.1 Experimental Procedure

Participants will be adults that (1) can have one of the proposed identities (e.g., worker, parent) as salient and (2) have downloaded and used mobile apps. They will be recruited using a market research firm, after obtaining ethical clearance from the authors’ university. Those participants will be contacted via e-mail and will be provided with a URL, where they will find the experimental treatments and questions to answer.

After obtaining participants’ consent, they will be randomly assigned to one of the two groups of the type of information collected. In the “demographic” group, participants will be asked questions related to their age, gender, occupation, income, and level of education. Those are traditional demographic variables used in marketing segmentation [38]. Participants in the “identity” group will be asked to rank five social identities (i.e., parent, worker, friend, member of a religious group, and student) according to their relative importance in their lives (as per Callero [39]). This ranking of identities is justified considering that identities that are central to an individual’s self have the greatest potential to influence behaviors [3, 40]. Next, participants will be asked questions to make the top-ranked identity salient (see items in Table 1 below). Making an identity salient increases the possibilities of observing the effects of that identity on an individual’s attitudes and behavior [26, 41].

The information elicited from participants will be utilized to personalize the screen that participants will see next. In that screen, participants will see the features the app has to offer. In addition, a sample screen of how the app would look will be shown to participants. Next, they will be asked questions related to their privacy concerns, the extent to which they would use the app in different contexts (e.g., at home, at work), and intentions to adopt the personalized app. Manipulation checks will also be performed after showing participants the app’s screen. Other questions will be collected with the purpose of controlling for their influence on the endogenous construct of the model, such as familiarity with mobile apps, extent of usage of mobile apps, and whether participants have used an app with a similar purpose to that of the app shown during the experiment. Finally, an open-ended question will be asked to probe for reasons participants would decide or not to adopt such a personalized app.

3.2 Measurement Instrument

In order to measure the constructs proposed in the research model, previously validated scales will be adapted to the context of this study. Table 1 below summarizes the scales (with their sources) that will be used in this study.

Table 1. Summary of constructs and sources for their scales
Full size table

3.3 Pilot Study

A pilot study will be conducted to test the personalized screens that participants will see after the demographic- or identity-related information is elicited. In this study, graduate students from the authors’ university will be involved and will be assigned to one of two groups: the “identity” group and the “demographics” group. In each group, participants will see first a non-personalized screen of the app (so they get familiar with it). Next, they will be asked identity- or demographic-related questions as per the experimental procedure described above. Then, participants will see the personalized screen of the app and will be asked to rate the extent to which the screen has been personalized according to the collected information, using a Likert scale. Finally, participants will be given the space (with an open-ended question) to provide their comments or suggestions on how the personalization can be improved.

3.4 Model Validation and Sample Size

The proposed research model will be validated using linear regression analysis. In the model, user’s information will be specified as a dummy variable (i.e., “demographics” condition – 0, “identity” condition – (1). Post-hoc analyses will be run to determine the differences between groups in terms of privacy concerns and possibility to use the personalized app across domains. The required minimum sample size to validate the proposed model can be determined by following Faul et al. [45] requirements to detect a medium size effect, with a power of 0.80 and alpha of 0.05, in between-subjects designs with 2 groups. In this case, the minimum number of participants that would need to be recruited is 64 per group (128 in total). In order to account for potential spoiled responses, a total of 160 participants will be recruited.

4 Potential Contributions and Limitations

From a theoretical point of view, the results will contribute to the advancement of the IS literature in the area of customization by proposing social psychology variables as an input for the customization process. In addition, the study will evaluate the effect of this type of customization on the technology adoption process, as well as the barriers that need to be overcome to drive this adoption. From a practical perspective, the study will provide IS designers with a new variable (i.e., an individual’s identity) to use in the customization of their systems.

This study has some limitations. First, individuals’ identities may not necessarily be constant or coherent. They may change overtime (what has been known as identity projects) or may generate tensions within the individual (due to the pursuit of conflicting goals). Therefore, customization based on identity may need to adapt and evolve according to those changes. Future research may consider this possibility by conducting longitudinal studies and verifying that the customized features keep valid over time for users. Second, this study will test the effect of identity-based customization in one particular app. The antecedents explored here may need to be expanded to consider other types of apps or the use of customized computer software.