How do I see myself? A systematic review of identities in pro- environmental behaviour research

Encouraging pro-environmental behaviour (PEB) is an environmental and societal concern. Encouraging PEB focussing on how consumers see themselves (their identity) has blossomed. However, a theoretical assessment of this research is missing. Three main identity theories seem to best explain the research, specifically, and two-fold, identity, and social identity theory (SIT), collectively known as the unified identity theory (UIT), and place identity theory (PIT). As these theories overlap more than differ in their understanding of identity, we argue that combining these theories is needed to avoid redundancies in identity theorizing, provide a universal approach to identity in terms of the processes and outcomes, and explain the PEB research most succinctly. Therefore, we understand identity similarly between the theories and offer a universal identity theory approach based on the theoretical definitions and assumptions. Finally, we demonstrate how the theory can be used to explain the research. Next, research was identified by conducting a systematic review using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, where 62 studies were relevant. Multiple identities relevant for a given PEB are assumed and evidenced: 99. Identities are assumed to be either individually-, group-, and/or place-focused, drawing on the specific subsets of the universal theory: identity theory, social identity theory, and place identity theory, respectively. Identities are assumed to relate to behaviour, where identity increased PEB with medium effect sizes. Finally, to move the field forward, we provide a theoretical framework of how to test identities in relation to other psychological variables relevant for PEB research.

environmentally harmful products, donating money to environmental charities) are typically regarded as "pro-environmental behaviour" (PEB; e.g., Gilal, Zhang, Gilal, & Gilal, 2019;Nguyen, Lobo, Nguyen, Phan, & Cao, 2016;Rahimah, Khalil, Cheng, Tran, & Panwar, 2018), as PEB is defined as those behaviours that "harms the environment as little as possible or even benefits the environment" (Steg & Vlek, 2009;p. 309). Therefore, the present research will focus on PEB, which includes a whole array of different consumer behaviours. That is, PEBs are actions that minimise the negative impact on (such as preserving and preventing damage to), and/or promoting improvements to, the natural and the built world (Kollmuss & Agyeman, 2002). The built world is the "the human-made space in which people live, work, and recreate on a day-to-day basis" (Roof & Oleru, 2008, p. 24), and PEBs thus span the whole of nature and human-made space. Finally, there is a need for engaging more people in consumer behaviours, which we know are PEBs (Stern, 2000), as well as engaging them in a wider variety of PEBs, which alter the whole of nature and human-made space.

| Theories for understanding PEB research
In recent years, there has been a surge in research on understanding how to encourage PEBs (Chernev & Blair, 2015;Gershoff & Frels, 2015). Specifically, a meta-analysis showed that research in this field was largely based on four dominant theories (Klöckner, 2013), namely, the theory of planned behaviour (Ajzen, 1991), the norm-activation theory (Schwartz, 1992), the value-belief norm theory (Stern, 2000), and habits (Verplanken & Aarts, 1999). However, combining the key constructs of these theories accounted for only 36% of variance explained in a variety of PEBs (Klöckner, 2013), which implies other important factors might be missing in these theories. One factor that seems to have been overlooked in these main theories is identity.

| Identity in PEB research: Theoretical perspectives
Over the last decade, a new stream of research emerged that investigated how identity influences PEBs (Chernev & Blair, 2015;Gershoff & Frels, 2015). How people see themselves, referring to identity (Pronin, 2008), is difficult to define because the definition seems to depend on which theoretical framework is used. In PEB research, identities seem to be best explained within the framework of either identity theory (Stryker & Burke, 2000), social identity theory (Tajfel & Turner, 1979), whereby an integration of both approaches has been argued (Stets & Burke, 2000, formerly known as the "Unified Identity Theory"), and place identity theory (PIT; Proshansky, Fabian, & Kaminoff, 1983). As these theories overlap more than differ in their understanding of identity, we argue that combining these theories is needed to avoid redundancies in identity theorizing, provide a universal approach to identity in terms of the processes and outcomes, and explain the PEB research most succinctly. Therefore, from this point forward, we understand identities in terms of this universal identity approach. According to our universal identity approach, we first define identity on three levels. Level 1 is where we define an individually focused identity as how people label, describe, and recognize one's self individually or personally (Stryker & Burke, 2000). We will refer to this Level 1 as an identity explained by PEB-identity theory (see Table 1, Row 1). Level 2 is where we define a group-focused identity as how people label, describe, and recognize one's self as part of a group (Tajfel & Turner, 1979). We will refer to this Level 2 as an identity explained by PEB-social identity theory (see Table 1, Row 2).
Level 3 is where we define a place-focused identity as how people label, describe, and recognize one's self as part of a place (Proshansky et al., 1983). We will refer to this Level 3 as an identity explained by PEB-place identity theory (see Table 1, Row 3).

| Identity in PEB research: Empirical evidence
Identity has been associated with a wide range of human behaviours (Akerlof & Kranton, 2010), including PEBs (Khare, 2015a(Khare, , 2015b. Research investigating the associations between identity and PEB started in 1992 (Sparks & Shepherd, 1992). Subsequently, increasingly more research has been conducted on this topic (Dagher & Itani, 2014;Graham-Rowe, Jessop,& Sparks, 2015;Hamerman, Rudell,& Martins, 2018). However, a theoretical assessment of this research is missing. Therefore, we use our universal identity approach to explain the PEB research. As a first step in this direction, we conduct a systematic review using the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The guides enable us to identify the relevant studies for this theoretical assessment. By adhering to these guides, researchers can replicate our work and update our review when researchers publish new findings. Furthermore, from this review, we are able to ascertain what assumptions of our identity approach are evidenced or not in the research allowing for future research to see and build on the existing knowledge and gaps. Finally, to move the field forward theoretically, practically, and methodologically, we provide a theoretical framework of how to test identities in relation to other psychological variables relevant for PEB research (Figure 1).
Before we expand on the research aims and conduct the review, we provide a foundation for our universal identity approach by first highlighting the main theories that initially have informed our research, namely, identity theory (Stryker & Burke, 2000), social identity theory (Tajfel & Turner, 1979), whereby an integration of both approaches informed our research specifically (Stets & Burke, 2000), and place identity theory (Proshansky et al., 1983). These theories can be used as conceptual frameworks to best explain identity-PEB research because despite the difficulties in ascertaining which theories seem to be relevant for this context, when references to theory were clearly given, these frameworks most frequently were used, for example, identity theory (Murtagh, Gatersleben, & Uzzell, 2012), social identity theory (Fielding, Terry, Masser, & Hogg, 2008), and place identity theory (Halpenny, 2010). Therefore, these offer the most relevant key insights to identity-behaviour interactions. 1. D1a. PEB-Identity theory Level one is where we define an individually-focused identity as follows: How people label, describe and recognize one's self individually/personally (Stryker & Burke, 2000) which we refer to as an identity explained by PEB-identity theory.
D1b. PEB-Social identity theory Level two is where we define a group-focused identity as follows: How people label, describe and recognize one's self as part of a group (group-focused identity) with which the group has been labelled, described, and recognized (Tajfel & Turner, 1979), which we refer to as an identity explained by PEB-social identity theory. ✓

3.
D1c. PEB-Place identity theory Level three is where we define a place-focused identity as follows: is how people label, describe and recognize one's self as part of a place (place-focused) with which the place has been labelled, described, and recognized (Proshansky et al., 1983), which we refer to as an identity explained by PEB-place identity theory.

5.
A1b. PEB-Social identity theory Any one person can have plural group-focused identities. ✓

6.
A1c. PEB-Place identity theory Any one person can have plural place-focused identities. ✓

7.
A2a. PEB-Identity theory The many identities individually-focused are considered important for any type of behaviour, individually-, group-, or place-focused PEB, irrespective of whether the group/place with said identity is present.
A2b. PEB-Social identity theory The many identities group-focused are considered important for any type of behaviour, individually-, group-, or place-focused PEB, irrespective of whether the group/place with said identity is present.
A2c. PEB-Place identity theory The many identities place-focused are considered important for any type of behaviour, individually-, group-, or place-focused PEB, irrespective of whether the group/place with said identity is present.

10.
A3a. PEB-identity theory These individually-focused identities can be made salient depending on the context a person is in, which in turns promotes particular types of PEBs, individually-, group-, or place-focused.

11.
A3b. PEB-Social identity theory These group-focused identities can be made salient depending on the context a person is in, which in turns promotes particular types of PEBs, individually-, group-, or place-focused. ?

12.
A3c. PEB-Place Identity theory These place-focused identities can be made salient depending on the context a person is in, which in turns promotes particular types of PEBs, individually-, group-, or place-focused.
? 13. A4a. PEB-Identity theory Consumers hold these identities in mind in a specific order that can change, and is referred to as the hierarchy of salience. That is, there is a hierarchy of ✓ (Continues)

| Theoretical foundations of identity
Identity theory suggests identity and behaviour are linked by their shared meaning (Stryker & Burke, 2000). That is, when an identity is in the foreground (salient), this identity will more likely predict behaviour when the meaning of the behaviour corresponds to the meaning of the individual identity. However, from a social identity perspective, if a person identifies with a particular group (the group identity is salient), then he/she internalises the norms of that group and is more likely to act in accordance with those norms (for further discussion, see Stryker & Burke, 2000;Tajfel & Turner, 1979). Hence, an integration of both perspectives implies that when an identity is salient in a situation, the outcomes for persons include common or shared group outcomes (social identity theory perspective) and reciprocal role outcomes (identity theory perspective; for further discussion regarding an integration of both approaches. see Stets & Burke, 2000). Place T A B L E 1 (Continued)
A4b. PEB-Social Identity theory Consumers hold these identities in mind in a specific order that can change, and is referred to as the hierarchy of salience. That is, there is a hierarchy of salience of these many identities group-focused.

15.
A4c. PEB-Place Identity theory Consumers hold these identities in mind in a specific order that can change, and is referred to as the hierarchy of salience. That is, there is a hierarchy of salience of these many identities place-focused. ?

16.
A5a. PEB-Identity theory Consumers can have competing identities which means before any behaviour occurs, the many identities individually-focused need to be similar in behavioural expectations for behaviour to occur.

17.
A5b. PEB-Social Identity theory Consumers can have competing identities which means before any behaviour occurs, the many identities group-focused need to be similar in behavioural expectations for behaviour to occur.

18.
A5c. PEB-Place Identity theory Consumers can have competing identities which means before any behaviour occurs, the many identities place-focused (PEB place identity theory) need to be similar in behavioural expectations for behaviour to occur. ?
F I G U R E 1 The identity-comprehensive action determination model, including key direct predictors of PEB from Klöckner's meta-analysis (2013). The model has been extended by including the assumptions assessed in the identity-review (Table 1) Notes. PEB, pro-environmental behaviour. *The three levels of identity are presented in the bold boxes; Most important predictors of PEB as found in Klöckner's meta-analysis in the three non-bold boxes identity theory can be interpreted in the same way as identity theory and social identity theory. That is, the link between identity and behaviour is its shared meaning. However, these meanings are in accordance with the place (place identity) rather than the reciprocal individual roles corresponding to behaviour (identity theory) or the norms of the group corresponding to the common or shared group outcomes (social identity theory). Place identity refers to the part of people's personal identity that is based on the physical and symbolic features of places (connections and history) where people live (Proshansky, 1978;Proshansky et al., 1983).
There is a key similarity and difference between the three identity theories. With regard to the similarity, identity theory (Stryker & Burke, 2000), social identity theory (Stryker & Burke, 2000;Tajfel & Turner, 1979), and place identity theory (Proshansky et al., 1983) all suggest that when an identity is salient, this identity will more likely predict behaviour when the meaning of the behaviour corresponds to the meaning of the identity, which then leads to common, shared, and reciprocal outcomes. The key difference between the three theories, however, is the type of identities: individually focused identity as per identity theory (Stryker & Burke, 2000), group-focused identity as per social identity theory (Stryker & Burke, 2000;Tajfel & Turner, 1979), or place-focused identity as per place identity theory (Proshansky et al., 1983) and the corresponding, common, shared, and reciprocal outcomes/behaviours: individually-, group-, or place-focused, respectively.
Finally, although the identity theories (identity theory, social identity theory, and place identity theory) can be applied in multiple contexts, here, we assume them differently based on the current applications of identity in PEB research. That is, how identity is assessed in our context seems to be different to how identities are assumed. Therefore, how we assume identities in this context needs updating. To avoid confusing on how we interpret the assumptions of identity based on our context findings, with the original assumptions of the three theories of identity, we label our theories differently, namely, PEB levels of identity, specifically, PEB-identity, -social identity, and -place identity theory. These PEB levels of identity incorpo-
Empirical evidence has identified a wide range of individually-, group, place-focused identities related to different types of PEBs (Gatersleben et al., 2017;Murtagh et al., 2012). However, as different identities can relate to many different behaviours, there is confusion as to which identity does or does not associate with which PEBs.
Indeed, our initial scan of the literature reveals that the different identities are not always associated with behaviour (Matsuba et al., 2012).
Furthermore, as individuals can have plural identities that can relate to a variety of behaviours, there seems to be a trend towards assessing identity types in relation to many behaviours in any given study (Dunlap & McCright, 2008). However, the large number of identity-behaviour interactions studied, without an overview, makes it difficult to know which identities significantly relate to which behaviours. Hence, we aim to provide a comprehensive list of how each identity relates to different types of PEB with respect to significance (Aim 3).
Besides significance, some studies report positive associations (Terry, Hogg, & White, 1999;van der Werff, Steg, & Keizer, 2013), whereas other studies report negative associations (Cook, Kerr, & Moore, 2002;Murtagh et al., 2012). However, without an overview, it is difficult to know which identities positively or negatively associate with PEB. Knowing if identity increases/decreases behaviours will inform researchers as to which identities to utilise with which behaviours depending if they wish to increase/decrease the PEB. Therefore, we will also provide a comprehensive list of how each identity relates to each PEB with respect to direction (Aim 4).  (Hogg, Terry, & White, 1995), the continued critical comparison of the key theories and assumptions within a field, as well as testing and updating them, may help to improve our understanding of identity in PEB (Fritsche, Barth, Jugert, Masson, & Reese, 2017). Therefore, Aim 5 will be to provide a comprehensive list of how each identity relates to each PEB with respect to strength, and, Aim 6 will be to assess if the PEB is either an individual and/or group and/or place-focused PEB relating to an individually-, and/or group-, and/or place-focused identity.

| Systematic review
In the present paper, we aim to test the set of assumptions as proposed by our updated/integrated identity theory (PEB levels of identity, Table 1). Specifically, we test the possibility of multiple identities (Assumptions 1a-c; what we can assume about the concept of identity for understanding PEB. Furthermore, different assumptions, not directly related to identity theories, are used in many papers. That is, identity is often an "add on" construct to other theories that are not identity related (Murtagh et al., 2012). For example, a measure of identity is often an "add on" to the theory of planned behaviour (Fielding, McDonald, & Louis, 2008;Fielding, Terry, et al., 2008;Nigbur, Lyons, & Uzzell, 2010;Shaw, Shiu, & Clarke, 2000;Sparks & Shepherd, 1992;Terry et al., 1999;White, Smith, Terry, Greenslade, & McKimmie, 2009;Yazdanpanah & Forouzani, 2015). In our systematic review, we do not view identity as an add on but specifically zoom in on it.
Inspired by the notion of identity and social identity theory integration (Stets & Burke, 2000), we wondered if we could create a parsimonious theory of identity that included all three theories (identity theory, social identity theory, and place identity theory), assumptions for understanding identity and PEB relationships. That is, how can we make the understanding and future testing of identity more consistent, comparable, and clear? Indeed, the present paper introduces such a theory (PEB levels of identity, Table 1), including assumptions to help researchers understand the use of identity and PEB relationships theoretically.
Our integrated theory of identity helps researchers to (a) understand the construct of identity in its own right, (b As of yet, there is little descriptive clarity on how different identity types relate to various types of PEB. That is, theoretically and conceptually, the concept of identity is too scattered for a metaanalysis to be conducted; a systematic description of the field seems to be more appropriate where the literature stands. A systematic review allows us to provide such a systematic description (Moher, Liberati, Tetzlaff, Altman, The PRISMA Group, 2009). Therefore, we will conduct a systematic review to achieve more descriptive clarity. A systematic review will develop our scholarly understanding of how specific types of identities and behaviours can be classified (i.e., we found that it can be divided into three levels: individually-, group-, and/or place-focused), whether these three specific identities relate to specific PEBs (i.e., to individually-, group-, and/or place-focused PEBs), along with identifying the relationships between the three different identity types and the three PEB types. In other words, the present systematic review revealed the classification of identity types and PEBs, the gaps in the extent to which these relate to each other, and the importance of specific identity types for specific PEB types, as well as several surprising findings, such as inconsistencies within one and the same study, several methodological issues, and new ideas for future research. Therefore, we aim to provide a preliminary understanding and empirical evidence of the three new PEB levels of identity (i.e., PEB-identity theory, PEB-social identity theory, and PEBplace identity theory). We examine the five (implicit) assumptions in the current literature, systematically and coherently in one article, so that these three PEB levels of identity (our "new" theory) can be assessed.
Specifically, we provide a comprehensive list of each identity that is empirically tested in a PEB context (Aim 1). Also, for each of the three PEB levels of identity, we provide a comprehensive list of which theory links to which identity in question (Aim 2).
Furthermore, we provide a comprehensive list of how each identity relates to each PEB with respect to significance, direction, and strength of these relationships (Aims 3-5). In addition, we assess if the PEB is either an individual and/or group and/or place-focused PEB relating to an individually-, group-, and/or place-focused identity (Aim 6). That is, we find holes, gaps, and inconsistencies in the use of identity for understanding PEB. Addressing these aims will help us to clarify how to use the construct of identity

| METHOD
We used the PRISMA method (Moher, Liberati, Tetzlaff, Altman, The PRISMA Group, 2009). The method has been widely used for systematic reviews, especially in the medical sciences (Drubbel et al., 2014;Holden, Haywood, Potia, Gee, & McLean, 2014). PRISMA offers a concise and replicable standard for conducting and reporting systematic reviews by advocating several reproducible steps (Higgins & Green, editors, 2011). We outlined the steps below. Therefore, this review can be replicated and updated, which means our first contribution is by introducing these guidelines to this field which future research can utilise and build on.

| Protocol
A peer-reviewed protocol was necessary a priori to the systematic review. This protocol was preregistered on the Open Science Framework (osf.io/bk6xe).

| Eligibility criteria
To address the aims, we used three criteria for a study's inclusion.
First, the studies needed to test both identity and PEB. The PEB could include self-reported intentions to engage in PEB, self-reported past PEB, or observable measures of PEB. Second, the studies needed to use a design that allowed for the effects between identity and behaviour to be measured, compared, and obtained. Consequently, we only included quantitative research. Third, we focused on primary studies that were published in peer-reviewed academic journals, excluding, for example, reviews, conference proceedings, books, unpublished dissertations, and working papers. We used this criterion to ensure quality and consistency between the studies in terms of methodology and data reporting (Field, 2005).

| Search strategy
Using a modified checklist from the Cochrane Collaboration of Systematic Reviews, search terms and keywords were identified (Open Science Framework, osf.io/bk6xe; and available on request). We chose these search terms to maximise the identification of relevant articles. Hence, we started with many search results, but only a small proportion of these were relevant for the final inclusion.

| Study selection process
In five steps, we identified the total number of relevant studies: 54 articles, comprising of 62 studies, ranging from 1992 to 2015.
1. The total number of records identified via electronic database searching was 5,681.
4. We reviewed the titles and abstracts of the remaining records based on the three eligibility criteria. Two researchers independently assessed 90% of the records, whereby less than 100% of records being checked is acceptable . The researchers disagreed about the inclusion of 16% of the records.
5. Full text articles were reviewed based on the three eligibility criteria, and 50% went through the double-reviewing process, which is acceptable for reviews . The researchers disagreed about the inclusion of articles (9%). After the final step, 300 articles were excluded, leaving a total number of 54 articles comprising of 62 studies.

| Data Extraction Process and Analytical Procedures
We extracted (when possible) information on identity types, PEB type, and the relationships between each identity and each PEB. For the latter, we extracted statistics, such as regression, beta, intercept, odds ratio, split sample multinomial probits, multinomial probits, and Cohen's d. Furthermore, we obtained whether the identity-behaviour associations were significant, positive/negative, the direction of the relationship, and the effect size (small/medium/large; Cohen, 1992).
When effect sizes were not provided, we calculated them when possible (Pearson, 1895). Finally, we extracted the study design used. This extraction process was in line with other research and the recommendations by Cochrane Handbook for Systematic Reviews, upon which PRISMA is based (Bossuyt, Davenport, Deeks, Hyde, Leeflang, & Scholten, 2008). Fifty percent of all analyses went through the double-reviewing process. The two researchers agreed on the results of all analyses.
Finally, we systematically interpreted which identity type and which PEB type they were aligned to. Therefore, identities and PEBs were categorized as either individually-, group-, or place-focused identities in line with PEB-identity theory, -social identity theory, orplace identity theory, respectively.

| Assumptions 1a-c: Any one person can have plural individually-, group-, place-focused identities (Aims 1 and 2)
In the 62 studies, 99 identities were studied in relation to PEB (Aim 1; Table 2). Of these 62 studies, 58% studied one identity, 20% two-, 10% three-, 2% four-, 5% five-, 3% seven-, and 2% studied eight identities. Thus, studies have reported a multitude of identities for understanding PEB. However, most research focused on the contribution of one identity in any given study. Furthermore, identities were categorized as either individually-focused, group-, or place-focused identities in line with PEB-identity theory, PEB-social identity theory, or PEB-place identity theory, respectively (Aim 2; Table 2). Individually focused identities were predominantly studied (n = 59) compared with group-focused identities (n = 34), or placefocused identities (n = 6). In addition, people can have multiple specific identities such as multiple individual (Kiesling & Manning, 2010;Matsuba et al., 2012;Murtagh et al., 2012), group (Bartels & Hoogendam, 2011;Murtagh et al., 2012) and place identities (Halpenny, 2010;Swim et al., 2014 Nigbur et al., 2010;Terry et al., 1999) and a mix of individually-and place-focused identities (Swim, Zawadzki, Cundiff, & Lord, 2014;Tam, 2013). Finally, we did not find participants having a mix of group-and place-focused identities nor individually-, group-, and place-focused identities because we did not find these combinations studied. Therefore, a key conclusion of our work is that combinations of multiple identities are less well understood and need future exploration.
That is, theoretically, it is established that people can, for example, belong to different social groups and therefore have different group identities (Tajfel & Turner, 1979). However, as the review revealed, people's belongingness to different social groups at one time was rarely reported and studied. Therefore, our findings reveal that what we know about identity, theoretically, is not always mapped onto what we as scholars research and, subsequently, what we advise consumers and others to do in order to further understand and encourage PEB. This insight from our systematic review is important, because it shows that there is currently a "blind spot" in our research as a field, and by showing, this we can now take steps to remedy this in future research.

| Assumptions 2a-c: The many identities are considered important for any type of behaviour, individually-, group-, or place-focused PEB (Aims 3-6)
At the study level, and with regard to significance, 72% of the studies found associations between identity and PEB. Fewer, 53% did not find identity-PEB associations. Finally, 17% did not report all the identity-PEB associations (Aims 3-5; we propose that there was a behaviour-observed behaviour gap rather than just and intention-behaviour gap present. In addition, we found both supporting and non-supporting results in the same studies for the same identities relating to like-minded PEBs, such as for "behaviour generic self-identity: pro-environmental" (Whitmarsh & O'Neill, 2010), "environmental movement identity: sympathetic identity" (Dunlap & McCright, 2008), "environmental movement identity: unsympathetic identity" (Dunlap & McCright, 2008), and "Fair Trade consumer identity" (Andorfer & Liebe, 2013). Finally, reporting was not always fully transparent, and associations were not always reported. Therefore, we recommend what to report in future studies to be more transparent (see section "Methodological Contributions"). However, significant relationships were most frequently evidenced (Table 3, Column 8) and mostly correlational (Table 4).

I + I
Environmental identity (Tam,

ID+PEB Type
Environmentally conscious identity (Bhattacharjee et al., 2014 Self-identity with food produced using genetic engineering (Cook et al., 2002 Study 1) Anti-PEB-related
Furthermore, individually focused PEBs were mostly studied (n = 260) and in relation to individually focused identities (n = 166; 64%). Far fewer group-focused PEBs were studied (n = 35), and these were often assessed in relation to group-focused identities (n = 28; 80%). Finally, few place-focused PEBs were assessed and in relation to group-focused identities (n = 9) or place-focused identities (n = 10). First, identity types (individual/group/place) linked to PEB levels of identity (PEB-identity/social identity/-place identity theory) were often matched to the same type of PEB (individual/group/place). However, the review revealed that this matching was not always the case.
Therefore, this review found support for the second assumptions of PEB levels of identity (Table 1, Rows 7-9). However, this evidence was mostly for place-focused identities/PEB-place identity theory.
Individually focused identities, specifically, "self-identity with food produced using genetic engineering" (Cook et al., 2002) and "self-reported number of journeys using public transport for work"- Murtagh et al., 2012). Therefore, this review finds support for the second assumptions of PEB levels of identity (Table 1, Row   7), specifically the aspect that this individually focused identity is most relevant for discouraging individually focused behaviours.
Also, the findings support the assertion that individually focused identities are best for encouraging, rather than discouraging, PEB when compared with group-and place-focused identity types.

| Aim 6: Which group-focused identity best encourages and discourages PEB?
Group-focused identity types, "social identity: group identification" (White et al., 2009), "social identity: group membership of an environmental group" , and "identity similarity with typical recyclers" (Mannetti, Pierro, & Livi, 2004), are most relevant for encouraging individually focused behaviour ("intent to recycle in the social identity*group norm sample"- White et al., 2009; "intent to recycle"- Mannetti et al., 2004) and group-focused behaviour ("intent to do environmental activism in next 6 months"- . Therefore, this review finds further support for the second assumption of PEB levels of identity (Table 1, Row 8). Specifically, the aspect, a group-focused identity, is most relevant for encouraging individuallyand group-focused behaviours.
Group-focused identity types, "social identity: group identificatio"n (Terry et al., 1999) and "group identification rural" , are most relevant for discouraging individually focused behaviour ("intent to recycle in the next fortnight: Time 1 and for the social identity*perceived behavioural control sample"-Terry et al., 1999) and place-focused behaviour ("intent for sustainable agricultural practice"- . Therefore, this review finds further support for the second assumption of PEB levels of identity ( Table 1, Row 8). Specifically, the aspect, a group-focused identity, is most relevant for discouraging individually-and placefocused behaviours.

| Aim 6: Which place-focused identity best encourages and discourages PEB?
Place-focused identity types, "place identity-affect to Point Pelee National Park" (Halpenny, 2010) and "Environmental identity: identification with nature" (Swim et al., 2014), are most relevant for encouraging individually focused behaviour ("intent for general PEB" and "intent for Point Pelee National Park specific PEB"- Halpenny, 2010) and place-focused behaviour ("intent to increase amount pay for township open space"- Swim et al., 2014). Therefore, this review finds further support for the second assumption of PEB levels of identity (Table 1, Row 9). Specifically, the aspect, a place-focused identity, is most relevant for encouraging individually-and place-focused behaviours. Finally, our results support the notion that place-focused identity types are best for increasing PEB, rather than decreasing it. Yet it should be noted that research is scarce on this topic, and more studies could assess how place-identity types decrease PEB and cause anti-PEB.

| Aim 6: Summary of which identities best encourage and discourage PEBs
This review found support for the Assumptions 2a-c of PEB levels of identity (Table 1, Rows 7-9): • An individually focused identity best encourages group-and individually focused PEBs than place-focused PEBs and discourages individually focused PEBs the best versus all the other PEBs. Lastly, individually focused identity best encourages rather than discourages PEB (Table 1, Row 7).
• Place-focused identity best encourages individually-and placefocused PEBs and only encourages PEBs rather than discouraging them ( Table 1, Row 9). However, this finding is preliminary, because our systematic review reveals that research is scarce on how place-identity types discourage PEB or can cause anti-PEB.

| DISCUSSION
We highlighted how to more clearly understand theories of identity in PEB research and hereby the concept of identity in general. For the first time, we proposed an integrated theory of identity that includes the notion of PEB levels of identity, specifically, PEB-identity theory, PEB-social identity theory, and PEB-place identity theory. We provided the definitions, assumptions, and evidence for such assumptions. Furthermore, these theories were devised based on PEB literature only, because we sought to ascertain understanding of this context. However, the theory that we proposed seems not constrained to the PEB context because of the generalizability of the assumptions included in it. Future research can use and build on our assumptions that have been deducted based on the PEB literature and apply in other contexts such as moral/pro-social areas (to which PEB is a subset). Finally, after completing the review, some findings emerged that could not be explained by the assumptions included in our integrated theory. Therefore, we provide additional assumptions, which we document throughout the discussion, to explain the additional findings. Before discussing these findings, we summarise what we revealed.
Specifically, we tested the possibility of multiple identities (Assumptions 1a-c), and our empirical evidence corroborated these assumptions. We also revealed that most research focused on individually focused identity types. Additionally, we tested if identities related to any PEB type (Assumptions 2a-c) and empirical evidence corroborated these assumptions, where we revealed that most research found individually focused identities positively relating to individually-and group-focused behaviours the most. Moreover, from explanations of the nonsignificant findings, we will reveal a further set of assumptions. Specifically, how these identities can be made salient (Assumptions 3a-c), whereby Assumption 3a is mostly supported.
That is, we found that individually focused identities can be made salient depending on the context a person is in and that this can subsequently promote particular types of PEBs. Furthermore, we reveal how these identities can be held in mind in a specific order known as the hierarchy of salience (Assumptions 4a-c). Specifically, we only find evidence for the hierarchy of salience of these individually focused and group-focused identities (4a/A4b, respectively). Finally, we reveal how consumers can have competing identities, meaning identities need to be similar in behavioural expectations for behaviour to occur (Assumptions 5a-c). Specifically, we only found evidence of competing identities for these individually focused and group-focused identities (5a/A5b, respectively). Below the assumptions and evidence are further discussed.

| Assumptions 1a-c: The possibility of multiple identities (Aims 1 and 2)
A trend towards introducing new identity types is evidenced. That is, the large number of identities studied showed a clear need for more consolidation in this field. We propose that our document is useful to avoid duplication as we can see what identities/theories have been studied. For example, if a researcher is interested in studying a certain identity type, Tables 1 and 2  Halpenny, 2010) identity types, were reflected in these 99 different identities and may explain why we did not always find significant results.
We did not always find significant results. We provide three reasons for this finding. The first reason is because each person has more than one identity present in any given situation (individually-/group-/ place-focused; Halpenny, 2010;Murtagh et al., 2012;Stryker & Burke, 2000). Second is the hierarchy of salience, whereby these identities are not equally ordered in mind (Costa-Pinto et al., 2014Eagly & Chaiken, 1993;Murtagh et al., 2012;Stryker & Burke, 2000).
Third is competing identities, whereby these identities have different foci and might not be aligned with each other (Eagly & Chaiken, 1993;Stryker & Burke, 2000). Therefore, the same/similar set of behaviours expected may be different and consequently conflict. This notion is explicitly assumed by identity theory to explain why behaviour occurs (Eagly & Chaiken, 1993;Stryker & Burke, 2000). This assumption needs to be adopted for this context to aid understanding for why significant results did not always prevail. To guide this discussion, we will detail further our reasons why there were not always significant results in the discussion after outlining the hierarchy of salience and competing identities.
In the PEB context, each person has more than one identity present in any given situation. If we assume all the identities documented in the review are different from each other, and can be adopted by any one person, 99 identities can be held in mind, which are relevant for PEB. These identities are prioritised, and more salient, depending on how they are cued. Depending on which identities are salient will depend on which set of expectations and behaviours are prioritised. As these identities can be differently prioritised depending on the cue, these salient identities are transient and ordered in mind, and not all are given equal importance when deciding how to see oneself (PEB-identity theory), in relation to groups social (PEB-social identity theory) or in relation to places (PEB-place identity theory). Furthermore, the salient identities are assumed to be similar because the set of behaviours associated with the identities need to be congruent for behaviour to occur (Eagly & Chaiken, 1993). What we mean by "similar" is that the set of expectations and behaviours linked to the identities are not conflicting (Eagly & Chaiken, 1993). If these identities/behaviours were conflicting, behaviour stops until there is more external and/or internal support to override and select identities/behaviours which are congruent (Eagly & Chaiken, 1993). Therefore, PEB may have not occurred because these multiple identities were present, ordered in a specific way, and were competing and conflicting leading to a lack of PEB. In contrast, PEB may have occurred because these multiple identities were present, ordered in a specific way, and were not competing and conflicting leading to PEB. However, the assumption of multiple identities being relevant in a situation, along with competing/conflicting identities, was scant.

| Assumptions 2a-c: Multiple identity relationships with PEB types (Aims 3-6)
A key implication of documenting identity-PEB associations is that anyone interested in this topic can now find an appropriate identity type for their specific PEB (Table 3). Furthermore, identity-PEB associations were not always significant, which did not always make sense which we now discuss. A possible reason for this occurrence can be seen in how PEBs were studied. For example, some studies make statements in relation to identity and "observed PEB" while measuring past PEB only (Murtagh et al., 2012). It seems plausible that past PEB can make the current identity salient that leads to future PEB (van der Werff et al., 2014a). However, the past, present, and future of identity and PEB were not always clearly defined/measured. This occurrence may contribute to the differences in significance between and within studies. Therefore, future research would benefit from understanding the past, present, and future of identity and PEB, to see how longitudinally it can change.
Furthermore, we observed individually-and group-focused identities significantly associating with PEB more than place-focused identities. A potential explanation might be that PEB from a place-identity perspective depends on whether environmental concerns are consistent with the place's character (Kneafsey, 1998;Kudryavtsev, Krasny, & Stedman, 2012). This consistency is likely to act as a moderator for the effects of place-identity on PEB, thereby suppressing the direct effect of place-identity on PEB (Biddle, Bank, & Slavings, 1987;Stedman, 2002).
Furthermore, identities were often significantly related to PEB regardless of the type of PEB measure (intention or behaviour). That is, identities seem to be equally influential for intentions and behaviours, suggesting that there is no obvious intention-behaviour gap (Kollmuss & Agyeman, 2002). A reason for this finding might be that certain psychological biases, including social desirability and conformity, may have equally affected intention and behavioural measures (Kondo, Saito, Deguchi, Hirayama, & Acar, 2010). Social desirability is where individuals carry out a behaviour based on what they think other individuals expect of them. As PEB is socially desirable and the right thing to do (Allen et al., 2018), people may report they do the behaviour to look good (Kondo et al., 2010). A similar line of reasoning can be proposed for the issue of conformity, referring to responses based on what respondents think others plan to do and do (Haney, Banks, & Zimbardo, 1973;Kondo et al., 2010). As carrying out PEB is considered a behaviour that is commonly planned (intention) and carried out by other people (behaviour), people may be motivated to conform in both settings, again to look good. However, no intentionbehaviour gap could also mean that it therefore does not matter whether we measure intentions rather than behaviour when understanding the contributions of identity for PEB. That is, we perhaps reveal that critiques on the prevalence of intentions compared with behavioural measures (e.g., Kormos & Gifford, 2014;Wood et al., 2015) are less problematic than sometimes thought. However, future research would benefit from collectively assessing these biases, such as checking for social desirability and conformity (Kondo et al., 2010) alongside self-reported intentions, behaviours and observed behaviours collectively, to examine if these biases exist in both these intention-behaviour settings, as well as if this intention-behaviour gap does not exist. Furthermore, social desirability and conformity are related and occur in conjunction with identity.
Social desirability is linked with either an individually-, group-, and place-focused identity, as research found that when a person has never challenged their identity, these people act socially desirable (Orlofsky, Marcia, & Lesser, 1973;Phillips, 2009 (Marcia, 1966;Phillips, 2009 This finding implies that the individually-and place-focused identity are less likely to be acted upon when the group-focused identity is strong and is being challenged by others, whereas the opposite might happen if the group-focused identity is weak and not being challenged by others (Costa-Pinto et al., 2014. Therefore, group conformity indicators can be used to decide how best to encourage people to carry out PEB and via which identity, individually-, group-, and/or place-focused identity, depending on their likelihood to conform to their group-focused identity norms and expectations or their own individually-and place-focused norms and expectations. Finally, some studies find positive relationships, and others find negative relationships between identity and behaviour. These relationships depend to a large extent on the specific identity studied. To clarify which specific type of identity was studied and which specific behaviour was studied, identities and behaviours (identity/behaviour) can be further categorised as follows (Table 3; Columns 2 and 4; identity theme, and PEB theme), respectively: • PEB-related-The identity/behaviour were actively related to the PEB definition.
• Anti-PEB-related-The identity/behaviour were actively NOT related to the PEB definition.
• Neutral to PEB-related-The identity/behaviour were neither related nor unrelated to the PEB definition • Anti-and PEB-related-The identity/behaviour were both actively unrelated, and related to the PEB definition.
Including this categorisation enables us to explore in more detail (inconsistencies in) the negative and positive relationships between identity and behaviour. For example, although we would expect that a "place identity of island high in environmental protection" (Hernández, Martín, Ruiz, & Hidalgo, 2010) would positively relate to PEB-related actions and negatively to anti-PEB-related actions, the relationship between identity and anti-PEB-related actions were not negative. Furthermore, surprisingly "self-identity with food produced using genetic engineering" sometimes positively and sometimes negatively related to an anti-PEB-related action. However, our systematic review also revealed that anti-PEB-related actions were rarely studied, so (in)congruency of anti-PEB-related identity and anti-PEB-related behaviour needs further assessment because we only see a few case examples (Table 3, Column 4 in bold/bold and italic text). Therefore, the qualitative and detailed literature inspection of this systematic review revealed some surprising/unexpected relationships and pointed towards new areas for future research. Additionally, our review revealed many nonsignificant findings, which might be best explained by extending our theory with additional assumptions, specifically, the many identities can be made salient (Assumptions 3a-c).

| Assumptions 3a-c: Identities can be made salient
Another observation was that there were nonsignificant results, which may be because of the context a person is in, for example, as a person might have had more relevant social cues which was evidenced to some extent in some research (e.g., Costa-Pinto et al., 2014, and this provides preliminary evidence that researchers could make different identities salient by changing the context. Before describing these experimental study results, we first want to address an additional underlying assumption that seems relevant to reveal based on this preliminary evidence: PEB levels of identity, specifically, PEBidentity, -social identity, and -place identity theory, assume that individually-, group-, and place-focused identities can be made salient depending on the context a person is in, which in turns promotes particular types of PEBs: Individually-, group-, or place-focused (Table 1, Rows [10][11][12]. Based on past experimental results in the field, we reveal that our theory can be even further extended with additional assumptions, specifically, the hierarchy of salience of identities (Assumptions 4a-c).

| Assumptions 4a-c: The hierarchy of salience of identities
Now, we describe experimental results in relation to our assumptions.
Personal identity, which we refer to as an individually focused identity, and social identity, which we refer to as a group-focused identity, were present and measured via manipulation checks (supporting That is, we reveal that PEB theories of identity need to consider that consumers might hold these identities in mind in a specific order, although that order can change. Based on our findings (

| Assumptions 5a-c: Competing identities
Our research indicates that our theory needs to consider that consumers can have competing identities. This means that before any behaviour occurs, the different individually-, group-, and placefocused identities need to be similar enough and/or be aligned enough for behaviour to occur (  Kudryavtsev et al., 2012). That is, the presence of the group or place may function as a cue, or "make salient", the relevant identity. However, the "boosting effect" of a physically present group or place might be less effective when people identify strongly with a group or place (Dunlap & McCright, 2008;Swim et al., 2014).
One possible reason for this is that merely thinking about the presence of a group or place may function as a cue, or "make salient", the group or place identity and lead to the corresponding behaviours. Therefore, the extent to which people identify with the group/place will determine which cues are needed, and how effective these cues will be for making salient the group/place identity.
However, more research needs to be conducted to test this notion of competing identities for place-focused identities especially as empirical research is largely missing.

| Further support for hierarchy of salience (Assumptions 4a-c) and competing identities (Assumptions 5a-c)
Another observation was that some results did not always make sense, for example, when the PEB-focused identity increased, the similar behaviour, and a PEB focused behaviour, decreased (Murtagh et al., 2012). More in-depth inspection reveals this competing/conflicting proposition but in behavioural focus, which links back to identity. This study assessed "transport-related identity importance: cyclist" along with other identities, such as "transport related identity importance: public transport user." The former identity was negatively associated with "number of journeys using public transport: work." We assumed that in this study a procycling identity was more important than the pro-public transport identity. The set of behaviours expected that are pro-cycling rather than pro-public transportation were prioritised. Therefore, the identities/behaviours conflicted/competed with pro-cycling identity overriding and leading to cycling and not public transport behaviour (one cannot take public transport and cycle to work). Consequently, our systematic review reveals that it is important to carefully choose the identity type when aiming to encourage or discourage PEB. Yet, this finding is preliminary as research into the ordering effects of identities, or how one identity/behaviour can be overridden is largely missing. However, we propose that these two additional main assumptions are further relevant based on this limited evidence-consumers hold these identities in mind in a specific order that can change and is referred to as the hierarchy of salience (Assumptions 4a-c), and consumers have competing identities meaning, the many identities need to be similar in behavioural expectations for behaviour to occur (Assumptions 5a-c).
The assumptions of competing and conflicting identities (Assumptions 4a-c) and hierarchy of salience (Assumptions 5a-c) are consistent with goal-framing theory (GFT; Lindenberg & Steg, 2007). GFT suggests that three goals (hedonic, gain, and normative) are competing/conflicting and that there is a hierarchy of salience of goals. That is, the hedonic goal needs the least external support (like individually focused identities). The normative goal needs the most external support, with the gain goal in the middle (like placeand group-focused identities, respectively). Aligned with the assumptions of GFT; making salient (cueing) collectively the place-, group-, and individually focused identities are recommended and can be achieved using cues that are place (place-focused identity), people (group-focused identity), and self (individually focused identity) focused. Specifically, the cue needs to focus on pro-PEB intention and impact, tailored to the person in question and provided in the form of environmental labelling and/or feedback, similarly to GFT recommended approaches (Lindenberg & Steg, 2007).
To this end, we now introduce our final methodological contributions of this review.

| METHODOLOGICAL CONTRIBUTIONS
To contribute to the methodological development of this field, we introduced the PRISMA method ), a well-known and widely accepted systematic review method from the medical sciences (Drubbel et al., 2014;Holden et al., 2014). By methodically applying the reproducible steps found in the PRISMA guidelines, we achieved two benefits. First, these guidelines can be used to ensure that future reviews will be conducted in a robust and replicable manner. Second, as the steps are easily replicable, future research can build and expand upon this review, which is necessary as the field of identity, and PEB is blossoming. Overall, we suggest that our use of the PRISMA methodology will help to streamline future research reviews within this field, as well as facilitate comparisons with other fields.
Furthermore, methodologically in terms of the research we identified, we noticed identity-PEB associations are often established using correlational designs (Table 4). Making identities salient, for example by cueing, could help to strengthen the identity-behaviour relationship. However, our review shows that making identities salient does sometimes, but not always, result in promoting PEB. For example, making salient environmental self-identity by asking people to note their past PEBs positively affected the number of times people chose a pro-environmental product (e.g., van der Werff et al., 2014avan der Werff et al., , 2014b but did not do so in another study (e.g., van der Werff et al., 2013).
Future research could focus more strongly on experimental studies in which identity is made salient (for example, via cueing) to see under what conditions saliency affects behaviour.
By conducting this systematic review, we also reveal that only eight papers study their selected identity more than once (Costa-Pinto et al., 2014;Nigbur et al., 2010;Owen, Videras, & Wu, 2010;Tam, 2013;van der Werff et al., 2013, van der Werff et al., 2014a, 2014b. Furthermore, as indicated by Table 2, many identities are not explicitly defined. That is, research articles tend to create new identity constructs that are rarely replicated. Furthermore, we are not able to observe whether identities that are labelled similarly actually measure the same construct. As a meta-analysis relies on averaging the effect sizes (weighted by its standard error) from many studies for a single construct (i.e., a single identity type). However, only a single effect size for each identity type is often reported, making this weighted averaging for different constructs not possible at this point.
Yet, what we can do is report the observed effect sizes. By doing so, our systematic review can thus still provide insight into the current state of the literature.
Furthermore, as reporting was not always fully transparent, we recommend what to report in future studies and to do so more transparently. These recommendations emerged because of how identity in PEB research has previously been reported and conducted upon. Specifically, research would benefit from reporting on the following: 1. Explicitly stating all identity types tested (individually-, group-, and/or place-identity) AND state how they are PEB-related, anti-PEB-related, and/or neutral to PEB-related.
2. Explicitly stating the corresponding theory/definition/assumptions of identity in PEB research.
3. Explicitly stating all PEB types tested (individually-, group-, and/or place-PEBs) AND state how they are PEB-related, anti-PEBrelated, and/or neutral to PEB-related. 4. Explicitly stating exactly all the identity and PEB relationships, in terms of direction (positive and/or negative AND identity predicting behaviour and/or vice versa, AND strength, AND significance level).

Explicitly stating the measurements of identity (self-reported
and/or observed).
6. Explicitly stating the measurements of PEBs (intention, and/or behaviour AND self-reported and/or observed).
7. Explicitly stating the study design such the sample (students, and/or non-students AND whether the sample was representative of population in question).
8. Whether research controlled for demographics of participants which could affect the effect estimate.
9. Whether the randomisation of participants to experimental conditions and for recruitment of correlational research was conducted.
10. Whether the inclusion of other controls such as social desirability and conformity biases were assessed and reporting on the effects of these.
11. Reporting precisely whether the research was experimental, quasi-experimental, correlational and/or longitudinal in design.
12. Explicitly stating whether pre-/post-identity checks for baseline identity was assessed to see if the before and after results of pure identity testing affect identity in relation to PEB.
Finally, identity as a separate variable has been criticized, mainly because the identities are considered to overlap with the theory of planned behaviour variables, for example (Fishbein & Ajzen, 2010).
Although identity is similar to variables found in the theory of planned behaviour both theoretically (Fielding & Hornsey, 2016) and empirically (Rise, Sheeran, & Hukkelberg, 2010), regardless, identities are important uniquely (Murtagh et al., 2012). Furthermore, pragmatically, a large growth in the field of identity-PEB associations, alongside the theory of planned behaviour exists, suggesting that identity provides a unique contribution (Rise et al., 2010). Therefore, we shed light on the identities theoretically to allow future work to accurately and comprehensively assess this potential overlap of identity and the theory of planned behaviour, for example. Although the latter part was outside the remint of this article, we show how future research can assess this potential overlap with dominant variables successfully for predicting PEB. For example, identities should be considered in relation to other concepts incorporated in the theory of planned behaviour when explaining PEB as previous research has used the tenets of the theory of planned behaviour to predict PEB attitudes and behaviours (Fielding, McDonald, & Louis, 2008;Fielding, Terry, et al., 2008;Nigbur et al., 2010;Rise et al., 2010;Shaw et al., 2000;Sparks & Shepherd, 1992;Terry et al., 1999;White et al., 2009;Yazdanpanah & Forouzani, 2015). To move the field forward theoretically, practically, and methodologically, we provide a theoretical framework on how to test identities in relation to other psychological variables relevant for PEB research ( Figure 1).

| CONCLUSION
We set out to encourage PEB because it is an environmental and societal concern. We proposed a developing field for addressing this concern was to focus on how consumers see themselves (their identity).
However, a theoretical assessment of the empirical evidence on this topic was missing. Therefore, we ascertained how identity in PEB research can be theorised, namely, under the umbrella term PEB levels of identity, specifically, PEB-identity theory, PEB-social identity theory, and PEB-place identity theory. By conducting a systematic review, using the PRISMA guidelines, we assessed identity in PEB research theoretically. Based on these theorisations, we identified from the literature the extent to which we found evidence for the assumptions associated with the theories (Table 1). In 62 studies, one main assumption evidenced was consumers can have plural individually-/group-/place-identities relevant in this context (Table 2).
Furthermore, we evidenced the assumption that identity and PEB are associated (Table 3), where designs were mostly correlational (Table 4). To move the field forward theoretically, practically, and methodologically, we provide a theoretical framework of how to test identities in relation to other psychological variables relevant for PEB research ( Figure 1). Finally, we introduce the PRISMA guidelines to this field. The guides mean this review can be easily built upon. Furthermore, future reviews using such guides can be easily replicated and updated.