Investigating the Effectiveness of Received Offline Word of Mouth: Role of Acquired and Ascribed Homophily

This research examines how similarities in educational and ethnic backgrounds (status homophily), between word of mouth (WOM) participants, influences the effectiveness of received offline WOM on the recipient’s behavioural intentions. The educational similarity between individuals is conceptualized as acquired homophily, whereas ethnic similarity is defined as ascribed homophily. This study employs scenario-based experiments, and the results suggest that positive and negative WOM received from homophilous sources is more effective in influencing the recipient’s purchase and WOM re-transmission intentions than recommendations received from heterophilous sources. This research expands the literature about offline WOM and the multidimensional status-homophily construct by providing insights on the differential influence of positive and negative recommendations received from WOM sources similar or dissimilar to the WOM recipients in terms of their educational and ethnic backgrounds. This is important from both the theoretical and practitioner perspective as the recent focus of scholarly research is on understanding the relationship between online homophily and e-WOM at the expense of offline WOM-homophily relationship.


Introduction
Word of mouth (hereafter, WOM) or informal social talk amongst non-commercial communicators is widely acknowledged as an influential phenomenon affecting an individual's purchase behaviour and thus boosting (in case of positive WOM) or dampening (in case of negative WOM) product sales (Berger, 2014;Bhaiswar, Meenakshi & Chawla, 2021;Eisingerich, Auh & Merlo, 2014;Harrison Walker, 2001;Iyer & Griffin, 2021;Luo, 2007;2009). In addition to its acknowledged influence, WOM is widespread. For instance, an average American consumer engages in 73 brand-related conversations weekly (Engagement Labs, 2021). In this study, the focus is on offline face-to-face WOM because they constitute the bulk of social conversations (Keller & Libai, 2009). In addition, this research follows best practice in WOM research by investigating both dimensions of WOM valence; positive and negative face-to-face WOM (East, Romaniuk & Lomax, 2011).
Face-to-face WOM conversations do not occur in social isolation but typically include a WOM dyad comprising a sender and a receiver who may share social relations amongst themselves such as tie-strength or degree of homophily between individuals which can potentially influence the effectiveness of WOM given or received (Bansal & Voyer, 2000;Brown & Reingen, 1987;Mazzarol, Sweeney & Soutar, 2007;Sweeney, Soutar & Mazzarol, 2008;Trivedi & Teichert, 2021). This research is located within a strand of WOM literature which seeks to identify the factors that may influence the effectiveness of offline WOM given or received within a WOM dyad. These factors have been collectively labelled as characteristics of WOM (East, Hammond & Wright, 2007) and can be classified in four broad categories; (a) inter-personal factors (e.g., tie-strength, homophily, and WOM actively sought or WOM volunteered); (b) personal factors (e.g., WOM participants expertise and experience); (c) message characteristics (e.g., WOM content and strength of WOM delivery) and (d) situational factors (e.g., perceived risk and audience size) (Sweeney et al., 2008).
This study investigates if status homophily or similarity between individuals on sociodemographic attributes (Brown & Reingen, 1987) in a WOM dyad can influence the effectiveness of offline WOM on the recipient's purchase intentions and their intentions to retransmit WOM received within their social networks. This research is important from a theoretical perspective. For instance, the bulk of offline WOM research has focused its research attention on understanding the influence of tie-strength between the WOM participants on the effectiveness of articulated WOM on either the sender (Chawdhary & Dall'Olmo Riley, 2015) or the receiver (Bansal & Voyer, 2000;Brown & Reingen, 1987;Hoye & Lievens, 2007;Mladenovic, Bruni & Kalia, 2021;Nitzan & Libai, 2011) at the expense of understanding the influence of homophily. Furthermore, much of the recent research on WOM-homophily relationship is within the online context (Lin & Xu, 2017;Ladhari, Massa & Skandrani, 2020) neglecting to understand this vital relationship in the offline context. It is important to note that online homophily is conceptually distinct from homophily in the offline context (Brown, Broderick & Lee, 2007) and thus it may be erroneous to generalize the results secured in the online environment to the offline context.
This study addresses this current imbalance in the extant WOM literature by investigating the influence of the under-researched multi-dimensional status homophily construct on the effectiveness of received offline WOM. Primarily our contribution to the WOM literature is twofold: First, the current research investigates a richer conceptualisation of the status homophily construct highlighting the different aspects of the construct wherein status homophily is a multi-dimensional variable with twin dimensions; acquired and ascribed homophily. Investigating the multi-dimensional perspective overcomes the limitation of the existing marketing literature which has primarily understood the influence of status homophily as a unidimensional construct. This study defines ascribed homophily as ethnic similarity between a pair of individuals whilst acquired homophily is similarity between individuals in terms of their education (McPherson, Smith-Lovin & Cook, 2001). Ethnicity and education are two of the biggest strata of any society and thus appropriate attributes to conceptualize the twin dimensions of status homophily (McPherson et al., 2001).
Second, this study addresses the limitations of prior WOM studies which examined offline WOM-homophily relationship but neglected to understand the differential influence of WOM valence. The current research investigates the influence of homophily on the effectiveness of both positive WOM (hereafter, PWOM) and negative WOM (hereafter, NWOM). Importantly, understanding the influence of the multi-dimensional status homophily construct on the effectiveness of received face-to-face PWOM and NWOM might potentially explain the equivocal results about offline WOM-status homophily relationship in the existing marketing literature (Brown & Reingen, 1987;Nitzan & Libai, 2011).
The remainder of this paper is structured as follows: first, the extant literature examining the relationship between WOM and homophily in both the offline and online environment is critically evaluated and discussed, second, the research methodology adopted in this study is described, third the results are reported followed by a discussion of findings and theoretical contributions and last, this research elaborates on the managerial implications, limitations, and directions for future research.

Homophily and Online Homophily
Rooted in the sociology literature, the underpinning logic of homophily or love of the same is to surround oneself with people who are alike or similar on some attributes (Hanks, Line & Yang, 2017;Lazarsfeld & Merton, 1954), or those who are reflective of one's selfimage and have similar life experiences and events (Hanks et al., 2017;McPherson et al., 2001;Nambisan, 2011). The perceived similarity between individuals or homophily as opposed to dissimilar individuals or heterophily (van Esch, Arli, Castner, Talukdar & Northey, 2018) does result in greater levels of social affiliation (Ruef, Aldrich & Carter, 2003). Accordingly, scholars have found that, students who received lower SAT and GPA scores were more likely to choose fellow underperforming students as study partners or friends compared to students with higher SAT and GPA scores (Carrell, Sacerdote & West, 2013).
One of the earliest conceptualizations of homophily was proposed by McCroskey, Richmond and Daly (1975), who viewed homophily as a four-dimensional construct comprising of attitude, background, morality, and appearance similarity between a pair of individuals (Ladhari et al., 2020). In the marketing literature, homophily in the offline context is typically defined as; (a) similarity between individuals on socio-demographic attributes also referred as status homophily (Brown & Reingen, 1987;Lo & Lin, 2017;Rogers & Bhowmick, 1970) and (b) similarity between individuals on values also known as perceptual homophily (Gilly, Graham, Wolfinbarger & Yale, 1998;Lo & Lin, 2017;Wangenheim & Bayon, 2004). Few scholars have treated homophily as a conceptually similar construct to tie-strength (Gatignon & Robertson, 1985;Sweeney, Soutar & Mazzarol, 2014). However, this study conceptualizes homophily as a distinct construct from tie-strength (Brown & Reingen, 1987) wherein tie-strength is a 'relational property that manifests itself in different types of social relations varying in strength ' (Brown & Reingen, 1987, p. 354) and homophily refers to the 'degree to which pairs of individuals are similar in terms of certain attributes such as age, sex, education, and social status' (Brown & Reingen, 1987, p. 354). Alternatively stated, an individual's relation with their father will constitute a strong tie, but the same individual can have a very low degree of homophily with their father on education.
Advent of internet resulted in conceptualization of online homophily which is distinct from the homophily construct in the offline environment Brown et al., 2007).The assessment of homophily in the offline environment is based on cues such as gender, age, social and professional status, and ethnicity, but in an online environment, these cues may be filtered out by the WOM participants and the demographic information might be either missing, camouflaged or even intentionally falsified (Brown et al., 2007). For example, in a crosscultural study, Leohardt, Pezzuti and Namkoong (2020) found that consumers in collectivist cultures discount differences (e.g., differences in lifestyle or personality) between themselves and other social media users, which fosters a sense of similarity with others (i.e., perceived homophily). In fact, in an online environment, homophily is mostly about similarity between; (a) pair of individuals in terms of their shared group interests and group mind-set and (b) between an individual and the website typically not associated with offline homophily (Brown et al., 2007). Although in an online environment individuals do not have face-to-face interactions, they can still make inferences about similarity with review providers. For example by appraising the review content and checking profile information (Filieri, McLeay, Tsui & Lin, 2018;Ismagilova, Slade, Rana & Dwivedi, 2020).
Irrespective of context, multiple theories have supported the influence of homophily in the extant marketing literature notably; social identity theory (Muda & Hamzah, 2021); social comparison theory (Zhang, Liang & Qi, 2021); dual process theory (Filieri, et al., 2018) and source attractiveness theory (Kelman, 1961). The principal premise of these theories is that recipients of information can better identify with sources or senders of information who are similar to themselves thus augmenting the influence of information transmitted to the receiver. For instance, recipients of the information are more likely to be persuaded by a message or information originating from in-group or "like-me" members that are similar on demographics or values rather than information emanating from out-group or "unlike-me" members who may be heterophilous or dissimilar (Brock, 1965;McGarty, Haslam, Hutchinson & Turner, 1994).
In addition to information from similar sources being more persuasive, Nambisan (2011) argues that individuals are more likely to process information received from similar sources more effectively than information received from dissimilar sources. Given that WOM is an exchange of social information between individuals (Berger, 2014), scholars have identified, homophily between individuals as one of the inter-personal factors that can influence the effectiveness of conversations or social information exchanged between individuals (Sweeney et al., 2008). Accordingly, marketing scholars have investigated the influence of homophily in both the online and offline environment on the effectiveness of e-WOM and traditional face-toface WOM and on the subsequent firm-related outcomes such as purchase behaviour and brand attitudes (Lin & Xu, 2017;Nitzan & Libai, 2011;Steffes & Burgee, 2009).

Online Homophily and e-WOM
There is evidence that e-WOM received from sources with similar demographics such as age are more likely to influence the purchase decisions of the recipient than e-WOM received from sources with dissimilar demographics (Steffes & Burgee, 2009). Furthermore, findings show that online homophily significantly influences perceived e-WOM usefulness and credibility, intention to purchase, and e-WOM adoption (Fu, Yan & Feng, 2018;Ismagilova et al., 2020;Zhang et al., 2021). The extant literature also acknowledges contrarian results related to the influence of online homophily wherein e-WOM received from ethnically similar senders was not found to be more effective in positively influencing the recipients brand attitudes and purchase intentions as compared to e-WOM received from ethically distinct e-WOM sources (Lin & Xu, 2017). Recent e-WOM studies found that, recommendations received from friends similar on values stimulates recipient's intention to spread brand-related information. and engage in future e-WOM. However, it is unclear if this effect on the recipient's behavioural intentions is due to friendship (i.e., strong ties) or value homophily (Lo & Lin, 2021). Additionally, Lo and Lin (2017) report that homophily driven customer-to-customer websites which represent values similar to the individual act as proxies for a friend and thus the content available on these websites positively influence the recipients' behavioural intentions. Subsequently, Kim, Kandampully and Bilgihan (2018) found that higher levels of homophily between the consumer and the website led to a more positive attitude towards the website and e-WOM information available on the website.

Offline Homophily and Offline WOM
There is extensive research in the marketing literature that investigates the role of perceptual homophily in influencing the effectiveness of offline WOM. Gilly et al. (1998) found that perceptual homophily between the WOM participants positively influences the recipient's decision. Subsequently, in two separate studies, Wangenheim and Bayon (2004; found that perceptual homophily between the WOM participants would strengthen the influence of PWOM on the recipient's behaviour. However, Wangenheim and Bayon (2004; did not examine their research model under the NWOM condition. This limitation was addressed by Sweeney et al. (2014) who found that perceptual homophily had a significant impact on perceived influence of both PWOM and NWOM messages on the receiver and this effect was stronger for NWOM. In addition, de Bruyn and Lillien (2008) also found that recommendations from WOM sources who are similar to the receiver on values are more effective in influencing the receiver's decision-making process then recommendations from dissimilar WOM sources. Consistent with prior literature, Asada and Ko (2016) found that perceptual homophily moderated the impact of source expertise on WOM influence. Review of the extant marketing literature on offline WOM-perceptual homophily relationship reveals an unequivocal agreement amongst scholars that WOM received from individuals similar on values is more effective than WOM received from individuals who are dissimilar on values on the recipient's behaviour (Asada & Ko, 2016;Gilly et al., 1998;de Bruyn & Lillien, 2008;Sweeney et al., 2014;Wangenheim & Bayon, 2004;. However, scholars have found mixed results pertaining to the influence of status homophily on the effectiveness of offline WOM. These conflicting results in the existing offline WOM-status homophily literature can potentially be due to; (a) failure to acknowledge the multi-dimensional nature of the status homophily construct and (b) overlook the influence of WOM valence. Therefore, the focus of this research is to understand the influence of the multi-dimensional status homophily construct on the effectiveness of both offline PWOM and NWOM on the WOM recipients purchase intentions and WOM retransmission intentions. Review of offline WOM-status homophily relationship is discussed next. Feldman and Spencer (1965) conducted one of the earliest studies, investigating the influence of status homophily between the WOM participants on the effectiveness of recommendations received. They found that couples with children (as opposed to no children) were more likely to be influenced by recommendations given by couples with similar demographics potentially influencing their behavioural intentions than recommendations given by WOM sources with dissimilar demographics. However, de Bruyn and Lillien (2008) in their research found that status homophily between individuals is less effective in influencing recipients' decision-making process then demographic dissimilarity.
de Bruyn and Lillien (2008) contended that when the information required is objective and factual and the product category is for non-personal use (e.g., cat food), then individuals may favour the recommendations of people who are "unlike-me" compared to "like-me" to tap into a broader range of information and experience available within the social networks thus highlighting the importance of source expertise over source homophily in augmenting the persuasiveness of WOM.
Furthermore, findings related to the influence of status homophily on the recipients purchase decisions are equivocal. Brown and Reingen (1987) and Gilly et al. (1998) found no influence of status homophily on the effectiveness of received WOM and subsequently on the recipient's purchase decision, whilst Nitzan and Libai (2011) found that greater the similarity between individuals on status homophily, the more likely it will influence the purchase decisions and intention formation of the WOM recipient. These contrarian results in the existing marketing literature can be due to the unidimensional conceptualization of the status homophily construct. The current study conceptualizes status homophily as a multidimensional construct acknowledging both its acquired and ascribed dimensions.
Understanding the differential influence of acquired and ascribed dimensions of status homophily is important because there is evidence in the sociological literature that acquired and ascribed homophily have a differential effect on the recipient (Alsott, Madnick & Velu, 2014) but the marketing literature is largely silent on the differential influence of acquired and ascribed homophily. The next discussion focuses on hypotheses development.

Hypotheses Development
Underpinned by source-attractiveness theory (Kelman, 1961) and reviewed literature which suggests that perceived similarity between individuals augment the persuasiveness of the information received compared to information received from dissimilar sources (Brock, 1965;Nitzan & Libai, 2011), this research posits that WOM received by the recipient from homophilous sources on both ascribed and acquired traits will be more effective in influencing the receiver's purchase intentions and WOM re-transmission intentions than recommendations received from heterophilous sources.

Purchase Intentions
Purchase intentions is one of the most widely investigated firm-related outcome in the literature that investigates offline WOM-homophily relationship (Brown & Reingen, 1987;Gilly et al., 1998;de Bruyn & Lillien, 2008;Nitzan & Libai, 2011;Sweeney et al., 2014;Wangenheim & Bayon, 2004;. However, much of the research either overlooked the importance of acknowledging the differential influence of PWOM and NWOM and/or neglected to understand the multi-dimensional nature of status-homophily.

Methodology
Based on literature review and hypotheses development the research framework is presented in Figure 1. This research adopted the scenario-based experimental research design, as this research design is widely employed when investigating face-to-face WOM due to difficulty in observing offline WOM as it occurs (Chawdhary & Dall'Olmo Riley, 2015;East, Lomax & Narain, 2001;Garnefeld, Helm & Eggert, 2011;Sukhu & Bilgihan, 2021). In contrast to retrospective surveys, experimental research designs reduce biases stemming from memory lapses and have stronger internal validity (Wien & Olsen, 2014).

H1a
Mobile phone services (hereafter, MPS) are the research context in this study as it is a highly familiar and widely used services category within the United Kingdom; the principal geographic context of this research, with nearly 96% adults owning a mobile device (MobileUK, 2020). The familiarity of the research context amongst participants enhances the realism of the scenarios employed in this research, which can also alleviate concerns about low external and ecological validity associated with scenario-based experiments (Chawdhary & Dall'Olmo Riley, 2015;Gelbrich, 2011). Furthermore, MPS is a widely used research context in WOM research (Chawdhary & Dall'Olmo Riley, 2015;Garnefeld et al., 2011;Garnefeld, Eggert, Helm & Tax, 2013).

Scenario Development
A total of nine scenarios were developed for the current research which included one introductory scenario and eight main scenarios (see appendix A). Main scenarios were developed to describe a PWOM or an NWOM episode emanating from either a WOM source similar to the WOM recipient on acquired/ ascribed attributes or a heterophilous WOM source. Scenarios illustrate a situation whereby a WOM sender who is defined as an existing customer of an MPS provider gives either a positive or a negative recommendation to a WOM recipient who is a prospective customer. Employed main scenarios are almost identical in length in terms of word count, with the difference in the word count between scenarios less than 10%. Identical word count in experimental scenarios is important as past research shows that longer arguments presented in scenarios are perceived as more persuasive by the respondents than shorter ones (Hamilton, Vohs & Mcgill, 2014). In addition, scenarios are gender neutral (Bendapudi & Leone, 2003

Constant Variables
Guided by extant WOM literature, this study identified key variables that can potentially confound the experimental results and therefore held these variables constant at a single level in the scenarios (Aronson, Ellsworth, Carlsmith & Gonzales, 1990). First, two key inter-personal factors; (a) tie-strength and (b) direction of WOM initiative (e.g., WOM sought/WOM Volunteered) and one situational factor; (c) audience size (narrowcasting/broadcasting) are held constant as they can influence the dependent variables (Bansal & Voyer, 2000;Barasch & Berger, 2014;Sweeney et al., 2014). Tie-strength between the WOM participants in the scenarios reflects only weak ties, whereas the direction of the WOM initiation is WOM volunteered by the sender. In addition, audience size in the scenarios is narrowcasting wherein a WOM sender communicates with only one person as opposed to multiple people i.e., broadcasting (Barasch & Berger, 2014). Furthermore, the nature of WOM given in all the scenarios is explicit and not implied (Mazzarol, et al., 2007).

Pre-Tests
Pre-Tests were conducted to test the realism and effectiveness of the scenarios employed in the present study. Student participants primarily based in the United Kingdom were recruited for pre-tests. Prior WOM studies have recruited student sample for pre-tests (Saenger & Thomas, 2021). PWOM and NWOM are discrete variables and therefore were not subjected to any manipulation checks (Garnefeld et al., 2013). Scenarios about homophily (acquired and ascribed) and heterophily were assessed as distinct by the respondents (n=13). Following Liao (2007), scenarios were assessed via a seven-point semantic differential scale for experimental and mundane realism. Overall, the participants (n=10) found the scenarios to be realistic.

Main Study
Two studies were conducted to test the research framework. Study 1 examines the relative effectiveness of received PWOM from similar and dissimilar WOM sources on the recipient's purchase intentions and intentions to re-transmit received PWOM to others. Study 2 investigates the hypotheses under the NWOM condition. To avoid contamination of the main results, respondents of the pre-tests were excluded from participating in the main study (Feldman & Lynch, 1988).

Measures and Validity of Scales
Current research employed adapted version of Bansal and Taylor's (1999) sevenpoint, three-item semantic differential scale to measure purchase intentions. In addition, Zeithaml, Berry and Parasuraman's (1996), seven-point, three-item likelihood scale is adapted to measure the willingness of the recipient to re-transmit received WOM to others.
Cronbach's alpha for both the scales satisfied the benchmark of 0.70 under both PWOM and NWOM conditions (Nunnally & Bernstein, 1994).

Design and Sample Profile
The total sample size for the PWOM study is n=137. The sample size for the acquired dimension is n=71, wherein the sub-sample for acquired homophily is n=38 and acquired heterophily is n=33. Furthermore, the sample size for the ascribed dimension is n=66, wherein the sub-sample for ascribed homophily is n=30 and ascribed heterophily is n=36. A convenience sample was recruited following the procedure adopted by Cheng and Cho (2021) via social media platforms such as Facebook, Instagram, Snapchat, LinkedIn tapping into researcher(s) personal networks and contacting the respondents directly via e-mails and text messages. 63% of the respondents are females in the PWOM sample.
Age of respondents range from 18 to 65 and above. 45% of the respondents are in the age group of 18-24. Respondents were randomly allocated to one of the treatment scenarios and after reading the introduction and the main scenario were requested to answer questions about the dependent variables. Data are analysed using an independent sample t-test. An independent t-test is appropriate as data was collected from independent samples via different conditions of the scenario-based experiments to compare means (Field, 2018). All the formulated hypotheses are directional and therefore results reported are based on a one-tail ttest (Singh, 2000).

Purchase Intentions Acquired Homophily
Results suggest that PWOM received from homophilous sources on acquired attributes will hold a greater effect on purchase intentions of the recipient then PWOM received from heterophilous sources with MAcquired Homophily = 5.35 and MAcquired Heterophily = 3.82. This difference between the two groups is also significant with t(69) = 5.234, p<.05.
Thus, H1a is supported. This represents a large effect size, r = 0.53.

Purchase Intentions Ascribed Homophily
Findings indicate that PWOM received from homophilous sources on ascribed attributes will hold a greater effect on purchase intentions of the recipient then PWOM received from heterophilous sources with MAscribed Homophily = 6.00 and MAscribed Heterophily = 5.45. This difference between the two groups is also significant with t(64) = 2.096, p<.05.
Thus, H3a is supported. This represents a small effect size, r = 0.25.

Retransmission of WOMAcquired Homophily
PWOM received from homophilous sources on acquired attributes is more likely to increase the WOM re-transmission intentions of the recipient compared to when PWOM is received via heterophilous sources with MAcquired Homophily = 5.09 and MAcquired Heterophily = 3.98.
This difference between the two groups is also significant with t(69)= 4.090, p<.05. Thus, H2a is supported. This represents a medium effect size, r = 0.44.

Retransmission of WOMAscribed Homophily
Results suggest that PWOM received from homophilous sources on ascribed attributes is more likely to increase the WOM re-transmission intentions of the recipient compared to when PWOM is received via heterophilous source with MAscribed Homophily = 5.46 and MAscribed Heterophily = 4.96. This difference between the two groups is also significant with t(64)= 1.804, p<.05. Thus, H4a is supported. This represents a small effect size, r = 0.21

Design and Sample Profile
In study 2, the research framework was tested under the NWOM condition. The total sample size for the NWOM study is n=142. The sample size for the acquired dimension is n=67, wherein the sub-sample for acquired homophily is n= 33 and acquired heterophily is n = 34.
Furthermore, the sample size for the ascribed dimension is n=75, wherein the subsample for ascribed homophily is n = 38 and ascribed heterophily is n= 37. Similar to study 1, a convenience sample is recruited. 50% of the respondents in the NWOM sample are females.
Age of respondents in the NWOM sample range from 18 to 65 and above. 37% of the respondents are in the age group of 25-34. All the participants were randomly assigned to one of the four scenarios, after which they read the introductory and the main scenarios and completed a small questionnaire. Data are analysed using an independent sample t-test. All the formulated hypotheses are directional and therefore results reported are based on a onetail t-test (Singh, 2000).

Purchase Intentions Acquired Homophily
Results suggest that NWOM received from homophilous sources on acquired attributes will not hold a greater dampening effect on purchase intentions of the recipient compared to NWOM received from heterophilous sources with MAcquired Homophily = 2.28 and MAcquired Heterophily = 2.31. This difference between the two groups is non-significant with t(65) = -.110, p>.05. Thus, H1b is not supported, even though the results are in the intended direction. This represents a small effect size, r = 0.01.

Purchase Intentions Ascribed Homophily
Findings reveal that NWOM received from homophilous sources on ascribed attributes will hold a greater dampening effect on purchase intentions of the recipient compared to NWOM received from heterophilous sources with MAscribed Homophily = 2.04 and MAscribed Heterophily = 2.58. This difference between the two groups is also significant with t (73) = -1.876, p<.05. Thus, H3b is supported. This represents a small effect size, r = 0.21.

Retransmission of WOMAcquired Homophily
NWOM received from homophilous sources on acquired attributes is more likely to increase the WOM re-transmission intentions of the recipient compared to when NWOM is received via heterophilous source with MAcquired Homophily = 4.87 and MAcquired Heterophily = 3.57.
This difference between the two groups is also significant with t(65)= 4.631, p<.05. Thus, H2b is supported. This represents a medium effect size, r = 0.49.

Retransmission of WOMAscribed Homophily
Results suggest that NWOM received from homophilous sources on ascribed attributes is unlikely to increase the WOM re-transmission intentions of the recipient compared to when NWOM is received from heterophilous sources with MAscribed Homophily = 4.72 and MAscribed Heterophily = 4.44. This difference between the two groups is non-significant with t(73)= 1.231, p>.05. Thus, H4b is not supported even though the results are in the intended direction. This represents a small effect size, r = 0.14.
The results of hypotheses testing are presented in Table 2.  (Singh, 2000) Source: Author's own work

Discussion and Theoretical Contributions
The findings of this study primarily contribute to a sub-set of WOM literature which identifies the factors that may potentially influence the effectiveness of WOM given or received (East et al., 2007;Sweeney et al., 2008). The current research presents novel evidence about the differential influence of the multi-dimensional status homophily construct on the effectiveness of received PWOM and NWOM. background as the receiver was found to be more effective in influencing the receiver to engage in re-transmission of NWOM to others but not their purchase intentions. Results of this study are discussed below.

Effect on Recipient Purchase Intentions
As hypothesised, findings of this study indicate that PWOM received from similar WOM sources on either acquired or ascribed attributes will be more effective in influencing the WOM recipient's purchase intentions than PWOM received from demographically dissimilar WOM sources. Alternatively stated, WOM recipients are more likely to purchase a product or service if they receive a positive recommendation from homophilous sources.
Results pertaining to ascribed homophily (ethnic similarity) are particularly significant as past research in the online homophily and e-WOM context found no differential influence of ethnic homophily (vs. ethnic heterophily) on the recipients purchase intentions (Lin & Xu, 2017). This result also highlights the need to conduct more research in the offline environment and to avoid generalizing results of online homophily to homophily in the offline context as they are conceptually distinct constructs (Brown et al., 2007).

Limitations and Future Research
Limitations of this study should be taken into consideration when evaluating the findings. These limitations can be a guide for future research.

External Validity
Whilst recruitment of a convenience sample is acceptable in WOM research (Sweeney et al., 2014), it might limit the representativeness of the findings which need to be corroborated in future studies by conducting replication studies. Replications for this study can be done with different samples or research contexts (Kirk, 2013;East & Uncles, 2008) for theory development and refinement (Easley, Madden & Dunn, 2000). In addition, experimental research designs suffer from low external and ecological validity, which to some extent has been alleviated in this study by undertaking realism tests and conducting research within a familiar research context (Chawdhary & Dall'Olmo Riley, 2015).

Other Limitations
Current research investigates only one attribute each for both acquired and ascribed dimensions of status homophily and thus future studies can investigate other acquired (e.g., occupation) and ascribed dimensions (e.g., gender). Furthermore, future studies can investigate the potential interaction between acquired and ascribed dimensions in influencing the effectiveness of received WOM. In addition, future research can also investigate if recommendations from dissimilar individuals can result in recipients' repulsion from the brand recommended (Rosenbaum, 1986). Finally, this study does not consider the varying degrees of homophily between WOM participants which can be low, moderate, and high (Nejad, Amini & Babakus, 2015) and can be investigated in future studies.