Imported or locally made products? Building customers’ resilience to negative word of mouth (NWOM) about locally made products: Mediation of message quality in Tanzania

Abstract The purpose of this study is to determine the effect of word of mouth (WOM) attributes on the resilience of customers to negative word of mouth (NWOM) about locally made products when mediated by message quality. The study’s findings contribute to a variety of areas. To begin, available empirical evidence has primarily examined WOM from a source perspective, with an emphasis on the message’s design and delivery to receivers. Thus, it is necessary to study WOM from the perspective of the receiver, as the decision to accept or reject products is highly dependent on the perspective of the receiver (user). Further, available literature has not sufficiently connected the attributes of WOM with resilience to NWOM when mediated by message quality, especially of locally made products. This study attempts to cover that gap in knowledge. The study used a cross-sectional design involving 201 customers. Structural Equation Modelling (SEM) revealed that WOM attributes such as expertise, homophily, and trustworthiness between the sender and the receiver positively and significantly influence the message quality, and that message quality influences the resilience of NWOM about locally made products. Furthermore, message quality was found to partially mediate the relationship between WOM attributes and resilience to NWOM. The findings have both theoretical and practical implications.


PUBLIC INTEREST STATEMENT
The economic growth of any country is contingent on the expansion of domestic businesses. There are currently numerous initiatives that aim to improve the performance of businesses. However, in the majority of developing nations, customers still prefer imported goods to those made locally. According to reports, when customers do not effectively support locally made products through positive recommendations, they may engage in negative word of mouth (NWOM) against locally made products, thereby promoting imports. Since Word of Mouth (WOM) is regarded as the most effective form of marketing communication. It can be an effective method for increasing customer resistance to NWOM about locally made products. This study investigates the mediating effect of message quality on the WOM attributes and resilience to NWOM.

Introduction
Seeing that Tanzania is progressing towards a middle-level economy, all small and medium enterprises (SMEs) have become essential sub-sectors. As a base for private sector-led growth with the high potential of helping the country achieve such a transformation, SMEs in Tanzania, like in other parts of the developing world, have for a long time been used as a tool for poverty reduction and as a potential strategy for economic development (Talib et al., 2014;Kira & He, 2012;Matare & Sreedhara, 2020). Globalization's new ideologies have anchored healthy business environments for promoting private sector development and re-engineered economic growth and efficiency. SMEs are seen as a mechanism for the inclusion of marginalized groups in mainstreaming the economy and as an equity mechanism for economic empowerment in developing economies (Gamba, 2019). Specifically, SMEs play a role in generating employment, creating competition, promoting innovation, and generating financial wealth (Arasa & K'Obonyo, 2012;Mazzarol, 2015).
Moreover, apart from these contributions, SMEs in developing countries, specifically in Tanzania, have experienced many challenges which jeopardize their growth. Most SMEs have potentially failed to utilize available resources because of informal practices. The challenges documented in the previous studies that affect SMEs include capital constraints, insufficient business training, and an anti-entrepreneurial culture (Mashenene & Rumanyika, 2014), lack of business training, corruption and lack of access to finance (Tillmar, 2016). Other challenges are changing business environments, inborn individual attributes, competitive activities, location, inadequate human resources and management skills. Also, SMEs face inadequate access to market information, technology, poor linkages from support services, and an unfavorable regulatory environment and policies (Matare & Sreedhara, 2020;Nkonoki & Ericsson, 2010;Ssendi, 2013).
Following the recognition of the importance of SMEs' commercialization in economic development, governments in developing countries and other development partners have implemented a number of policies and frameworks aimed at improving the business environment, such as the creation of a SME development policy that focuses on supporting and promoting the business sector through programs such as sustainable industrial development policy and funding for entrepreneurship development. For example, the government of Tanzania established the National Strategy for Growth and Poverty Reduction (NSGPR I & II), which serves as a mediumterm strategy to eradicate poverty in the country by 2025. Promoting and supporting informal business sectors is the primary target of the strategy (Kira & He, 2012).
However, apart from these initiatives, there is still a concern that global rivalry is rising, foreign enterprises are expanding into new international markets, and local markets are no more appealing as a source of business potential. Local SMEs that had never considered overseas competitors suddenly faced international competition. Therefore, local SMEs are underperforming due to a lack of significant changes in product quality, manufacturing procedures, and product marketing compared to the marketing of imported goods (Kumburu & Kessy, 2021). Various studies have been carried out to determine the factors that impact consumers' decisions to purchase locally produced or imported items. In particular, the interest of policymakers is to promote the idea that purchasing things made within a country is beneficial to the country. This is because, it reduces a country's reliance on imported goods. In the wake of this call, the country gains in various ways, including increased demand for local SMEs while simultaneously cutting unemployment and stimulating economic activity, thus increasing the country's gross domestic product overall (Ayob & Hussain, 2016).
At large, customers' dissatisfaction with locally made products is primarily attributed to their perception of their poor quality, based on personal experience. Many enterprises in developing nations do not generate high-quality items that meet or exceed customer expectations. Based on the customer's impression, quality is determined by the customer's assessment of the overall excellence or superiority of the product under consideration (Suchánek et al., 2017;Zeithaml, 1988). The appraisal of a single purchase once it has been made and the whole experience following the purchase and use of the product over time are important factors that influence consumer satisfaction (Suchánek et al., 2017). Additionally, purchasing behavior can be utilized to assess whether to acquire locally produced or imported products based on consumer satisfaction. This is influenced by the physical characteristics of a product, such as its durability, quality, and performance, and non-physical characteristics, such as its price. Producing high-quality and innovative products, on the other hand, is regarded as a more advanced skill in industrialized countries than in emerging economies. Moreover, customers are satisfied with products from developed countries, especially when seeking quality and innovative features (Ayob & Hussain, 2016). This statement concludes that many customers are not satisfied with locally made products. Customer dissatisfaction has raised another concern about negative word of mouth (NWOM) (Reynolds & Harris, 2009), especially for locally made products. Customers who have had poor durability experiences, low product quality, and poor performance due to their purchases have become agents for spreading NWOM, among others. When customers are unsatisfied, the spread is twice as great as when they are satisfied. There is some indication that business actors may drastically underestimate the potential impact of NWOM, and this sort of customer retribution is becoming increasingly widespread (He & Harris, 2014). In this way, NWOM has aggravated and alleviated customer dissatisfaction based on attributes from worse off or others aggravating dissatisfaction, product evaluation process, and experience. The greater the extent to which the NWOM is used against locally produced goods, the greater the likelihood of consumer xenocentrism. Consumer xenocentrism is characterized by people intentionally paying more for imported goods than they would otherwise pay for domestically produced goods (Mahmoud et al., 2021).
To put it another way, various studies have considered increasing consumer ethnocentrism as the most important means of increasing positive attitudes and beliefs held by consumers about the appropriateness (and even morality) of purchasing locally made products as opposed to imported products, with the expectation that high ethnocentric consumers will refuse to purchase imported products and may even chastise others for doing so (Sharma, 2015). Consumer ethnocentrism may be a vital first step toward a better understanding of how customers compare locally made products to imported products and the mechanism by which they acquire resiliency to imports. Highly ethnocentric consumers tend to favour domestic products over foreign goods (Luque-Martínez et al., 2000). This goes with the fact that local business actors are currently thinking internationally to survive locally. Furthermore, the advancement of technology in product development and skill development has resulted in increased quality to match imported products, which was not the case only a few decades ago. As a result, SMEs can now produce some of the things that were previously available only in developed countries. Given the improvement in the quality of products produced by local enterprises compared to prior years, it is reasonable to conclude that there is a beneficial relationship between enterprises and the senders of information. However, contrary to the known positive relationship between enterprises and the sender of the message, the exact mechanism through which the WOM spreads from the sender to the recipient of the message remains a lingering question in the literature (Amani, 2022). Therefore, for word-of-mouth communications to have theoretical strength, a range of criteria should be considered, including the perceived usefulness of WOM communications to their receivers (Amani, 2022; Sweeney et al., 2014). In addition, it is critical to pay attention to the recipient side of the WOM message since the perceived usefulness of the WOM message indicates the degree to which recipients believe the WOM message will be advantageous in making decisions. So, it follows that receivers can benefit through increasing their resilience against negative WOM, resulting in the cycle of message transmission becoming more effective. As a result, depending on how people receive and perceive the information, WOM can be transformed into either positive WOM or negative WOM.
When negative or erroneous messages are allowed to propagate, they can perpetuate negative public perceptions and, as a result, poor product acceptance. Although much is known about the factors that influence how recipients respond to WOM messages, there is still much more to learn about how they can resist the NWOM. Additionally, the success of WOM should be evaluated by examining the resilience of the receivers to NWOM. Developing a high resistance to negative information is widely regarded as a critical component of successfully competing with foreign competitors in the market. In addition, resilience in the face of bad information has been identified as an additional role for behavior (Elbedweihy et al., 2016) that corresponds to the extent to which customers do not allow negative information about a company to diminish their general view of the company (Augusto et al., 2019;Eisingerich et al., 2011).
Thus, the study makes two significant contributions to the literature. First and foremost, there has been no investigation into the association between WOM attributes and resilience to NWOM with locally made vs. imported products. Second, the study contributes by illuminating the critical role played by the differential expertise between senders and receivers, the perceptual homophily between senders and receivers, the trustworthiness of the WOM message source, and the WOM message quality from the sender on the resilience to the NWOM. As it was noted, individuals exposed to certain positive WOM characteristics are more resilient to negative information than those who have not been exposed. Also, as stated earlier, the effectiveness of the communication, on the other hand, is best captured from the receiver's perspective. Thus, for locally made products to be purchased, the customers must be resilient to the NWOM about locally made products. Therefore, in the absence of resilience to NWOM, hesitancy, defined as the refusal or delay to use locally made products despite the availability of quality, may negatively impact the intention to purchase locally made products, increasing the likelihood of continuing to use imported products. It is the goal of this study to find out how word of mouth (WOM) attributes affect the customers' ability to deal with negative word of mouth (NWOM) about locally made products in Tanzania when the message quality is also taken into account.

Theoretical foundation
The foundation of this study is based on the prospect theory. Prospect theory postulates that individuals make decisions under conditions of uncertainty. It believes that individuals accept gains and losses based on the risks associated with a particular situation (Kahneman & Tversky, 1979). Therefore, individuals should theoretically value the benefits of outcomes by comparing them to the benefits of other alternative options. It means that, as the risk of loss increases, people tend to be more lenient toward behaviours; yet, when the risk of gain increases, people always choose the right option. In this study, individuals' decisions to purchase locally made products over foreign products is considered a difficult decision when presented with several options to choose from (in this case, choosing either to purchase locally made products or imported products). Customers always show a strong preference for the option that offers the most certainty. Given that recipients rely on the message sent by senders, the quality of the message and the relationship between senders and recipients are the most important factors to consider. As a result, to encourage the recipients of the message to prefer locally made products over imported products, the message from senders must be of sufficient quality to engender consumer ethnocentrism, causing customers to demonstrate a negative attitude toward purchasing or supporting imported goods. This implies that the higher the quality of the communication sent between the sender and the receiver, the greater the likelihood of achieving reconciliation concerning NWOM. This means that people who are ethnocentric will have a hard time with foreign goods and prefer to buy things made in their own country.
Furthermore, according to the theoretical premises, customers are subjected to various hurdles due to the NWOM message spread against locally made products. As a result, to accept and purchase locally made products, they must be resilient to NWOM. It also implies that the recipient of the message must measure the advantages of purchasing locally made products against the risks of not purchasing them or purchasing imported products. It is thought that this theory will help people understand why and how they can be resilient in the face of negative WOM messages.

WOM in the context of Tanzania
Culture is defined as shared values that are strongly linked to products, ideas, and events (Lam et al., 2009). Individuals from different cultures manage information in different ways under normal circumstances. Therefore, customers from different communities have different cultural values, and as a result, they may receive and transmit WOM in different ways. Although Tanzania is home to more than 120 tribes, each with its own language, traditions and customs, communication is facilitated by the country's use of the Swahili language, which is almost universally spoken. This has increased the communication's effectiveness. However, WOM has encountered some cultural barriers regarding the acceptance of promoting domestic products over imported products. First, historically, the majority of products in Tanzania are imported. Thus, customers are more involved with imported products than with locally made products (Kumburu & Kessy, 2021). As a result, positive customer referrals frequently favour imported products.
Second, information about products, particularly from media outlets, was frequently endorsed to promote imported goods. This tendency helped customers form a favorable opinion of imported goods. Nevertheless, globalization, on the other hand, has altered the landscape considerably. Due to technological advancements and increased government efforts to foster ethnocentrism among customers, SMEs are now capable of producing goods that they could not previously produce. This is very important because customers who know and like the product's quality are more likely to spread good news about it.
Additionally, globalization has resulted in the modernization of traditional modes of communication. For instance, social media platforms provide customers with sufficient product information, enabling them to make rational purchasing decisions (Paul et al., 2020). Currently, Tanzania has witnessed a high reliance on information from social media. Customers are now exposed and empowered with enough information about products across different countries. This has increased the ability of sources to share reliable information with other customers. Hence, a customer's preference for personal recommendations from friends over advertisements has led to a rise in expertise, trust, and homophily. This is because customers place more value on close friendships than on company promotions.

The differential expertise between senders and receivers
Expertise refers to the favorable personality traits of the WOM source. Many prior studies have demonstrated that source credibility positively impacts overall message quality (Bansal & Voyer, 2000;Tien et al., 2019;Wathen & Burkell, 2002). In this regard, receiving messages from experts rather than non-experts may strongly impact the WOM message quality. Therefore, it is accepted that senders of WOM messages with high levels of expertise are more likely to persuade other customers. In addition, the recipient of the message will be more likely to trust the message because their communications appear to be more trustworthy. Additionally, the sincerity and expertise of the sender are frequently positively correlated with changes in the attitude of the recipients in every communication situation (Baber et al., 2016;Tien et al., 2019). Thus, if the recipients feel that the particular sender has the expertise and can provide an accurate message, they will be less inclined to seek information from those whose level of expertise is deemed inadequate (Bansal & Voyer, 2000). Apart from that, some studies have reported negative and positive results regarding expertise. A study by (Wangenheim & Bayón, 2004) has noted a strong positive impact, while (Martin & Lueg, 2013) revealed a negative influence of expertise on WOM. Generally, regardless of whether the recipient of the WOM message has a real or virtual relationship with the sender, the fact that an expert individual posted the message with a high degree of believability will lead to the high message quality and a favorable judgment of the information's value (Cheung et al., 2008). Hence, it is hypothesized that H 1 : The differential expertise between senders and receivers significantly influences the quality of WOM messages.

The perceptual homophily between senders and receivers
Homophily, often known as similarity, refers to how the source and receiver are perceived to have similar features, values, preferences, demographics, and lifestyle. Positive links between the sender and the recipient are commonly acknowledged to impact the effectiveness of WOM. In particular, WOM seekers are more inclined to speak with homophilous sources due to the mentioned commonalities. However, the results of previous studies have been mixed when it comes to the role of homophily in WOM message outcomes. For example, a study by Le et al. (2018) discovered that homophily had a positive effect on message quality, but (Reichelt et al., 2014) discovered a negative relationship between homophily and practical function. Similarly, Chu and Kim (2011) discovered that homophily is negatively connected to obtaining WOM opinions. These contradictory findings suggest that homophily is a complicated factor influenced by specific homophily sources. Thus, it is reasonable to hypothesize that H 2 : The perceptual homophily between senders and receivers significantly influences the WOM message quality.

The trustworthiness of the WOM message source
Communication from the sender that the receiver may believe is referred to as trustworthiness. One of the most fundamental characteristics of communication is trustworthiness. It exemplifies the standard upon which all positive relationships are based. The literature demonstrates that the degree to which receivers place their trust in WOM sources significantly impacts the effectiveness or influence of WOM messages (Sweeney et al., 2014;Wangenheim & Bayón, 2004). Further, the source's trustworthiness has been associated with credibility, reliability, and transferability of the message quality (Le et al., 2018;Reichelt et al., 2014). Aside from that, WOM can foster trust, giving a significant competitive advantage in a market where consumer trust in televisions and traditional advertising has dwindled significantly (Sweeney et al., 2008). It can be hypothesized that H 3 : The trustworthiness of the WOM message source significantly influences the WOM message quality.

WOM message quality and resilience against NWOM
In the context of message quality, the depth of the message and the power of advocacy are both considered. Content elements, such as the language used and the amount to which a message tells a story or contains vast amounts of information, are considered rich in a message. In contrast, the message's advocacy strength relates to the efficacy with which it is delivered. The quality of the message delivered, its cognitive value, and the depth of the argument are all factors that influence how well the audience receives it. The quality of the message is closely related to how customers react to it, and as a result, it is a vital factor in determining how they will respond (Sweeney et al., 2008). Resilience to negative information has generally been seen as an extra-role behavior that corresponds to how recipients of the WOM message do not allow unfavorable information about the product to harm their overall perceptions. When the message quality lands to recipients, a stronger relationship between sender and recipient can be built around the product's identity, increasing behaviors for resilience against NWOM (Augusto et al., 2019). Therefore, it can be hypothesized that H 4 : The WOM message quality from the sender significantly influences the resilience of negative WOM messages.

Mediating effect of WOM message quality
It is a well-known fact that when recipients receive WOM messages, they process the message based on the characteristics of the source and the quality of the content contained within the message (Le et al., 2018). According to the theory of prospects, they receive a message and make a reasonable choice before engaging in a particular behavior, as previously stated. In general, the quality of a WOM message is influenced by the sources of the message. In this way, depending on the quality of the message, recipients can make decisions based on their evaluation of the characteristics of the sender. Their evaluation is based on their trustworthiness, source expertise, and homophily with the sender. Nonetheless, the effects of these dimensions on the resilience of the message receivers, through the mediating variable of message quality, have not extensively been studied before, particularly in the context of locally made versus imported products. Customer service in the modern world is defined by a constant search for additional information about various products. As the world is filled with imperfect information, consumers have developed a heightened awareness and sensitivity to any piece of information that assists them in distinguishing between these products to satisfy their needs better. The more information consumers have, the better their choices are (Atika et al., 2016;Elsharnouby et al., 2021). In general, perceived information quality positively affects customers' willingness to fight against any bad signals they may receive while selecting a particular product. However, the quality of the message received is critical. Customers' perceptions of the quality of information, such as its accuracy, relevance, and timeliness, are influenced by the information's quality (Atika et al., 2016).
Individuals are resilient to negative information, which means that preventing negative information from affecting their overall view of a product should be a top priority (Eisingerich et al., 2011). As a result, it is critical to understand the processes that contribute to negative information and the mechanisms that assist customers in resisting negative information (Torres & Augusto, 2018). Additionally, role behavior can be demonstrated in the resilience to unfavorable information when consumers support products (Zarei et al., 2020). Due to committed customers who have a stronger propensity to remain robust in the face of unfavorable information, brands can prevent negative effects by remaining resilient in the face of negative information (Torres & Augusto, 2018). Thus, given that the attributes of WOM messages enhance the message's quality and that receiving quality messages can significantly assist customers in becoming resilient to NWOM, it can be hypothesized that H 5 : WOM message quality significantly mediates the relationship between differential expertise and the resilience of NWOM messages.
H 6 : WOM message quality significantly mediates the relationship between the perceptual homophily and the resilience of NWOM messages.
H 7 : WOM message quality significantly mediates the relationship between the trustworthiness of the WOM message source and the resilience of NWOM messages.

Study design and sampling procedures
The research was carried out in Dodoma, Tanzania's capital. The study's major goal was to look into the mechanism of increasing customers' resilience to NWOM on locally made products via message quality mediation. Customers in three primary categories of SMEs in Tanzania were surveyed to test the study model: furniture, fashion, and craftsmanship. The SMEs were chosen for the study because they face threatening competition from imported products and because it was simple to contact their customers. In addition, the study collected data from 201 customers of the selected SMEs in Tanzania using a cross-sectional survey research approach. The study used a structured self-administered questionnaire to obtain data from sampled customers using the convenience sampling technique. However, to avoid selection biases, the study looked at how representative customers were and used diversification to strengthen the convenience samples.

Measurement items
A literature review informed the development of the items. However, items were modified to fit with the nature of the study and methodological approaches. The data collection involved the following itemized constructs: The first construct was differential expertise between senders and receivers (EXP). This construct is defined as the WOM source of experience and knowledge connected to opinion leaders. This meant that WOM recipients placed a higher value on experience than broad knowledge from the public. Five items (EXP1-EXP5) were adopted and modifiedfrom (Le et al., 2018;Martin & Lueg, 2013;Ohanian, 1990;Wangenheim & Bayón, 2004). The second construct was the trustworthiness of the WOM message source (TRS). It is defined as the tendency of the source to pass trusted information to the receiver. Five items (TRS1-TRS5) were adopted and modified from (Le et al., 2018;Ohanian, 1990). Finally, the perceptual homophily between senders and receivers (HOM) was defined as the degree to which sources and receivers were perceived to have similar attributes. Five items (HOM1-HOM5) were adopted and modified from (Le et al., 2018;Sweeney et al., 2008). Also, the WOM message quality from the sender (MES) was defined as the clarity of the message in terms of the degree to which the communication content is explicit. Eight items (MES1-MES8) were adopted and modified from (Le et al., 2018;Sweeney et al., 2012). Resilience to the negative WOM (RES) was defined as the extent to which the recipient of the WOM message does not allow negative information. Therefore, the ability to reject imported products and opt for locally made products depends on the resilience of NWOM. Three items (RES1-RES3) were adopted and modified from (Augusto et al., 2019;Elbedweihy et al., 2016). Finally, a 5-point Likert scale was used to capture information from respondents based on the study items.

Statistical analysis
A descriptive analysis was performed to define the distribution of the characteristics of the convenience sample. WOM attributes such as expertise, trustworthiness, and homophily were defined as the independent variables, while resilience to NWOM was identified as the dependent variable. In addition, message quality was set to be a mediating variable. Confirmatory factor analysis and path estimation were applied using Structural Equation Modeling (SEM). SEM was used because it is the most precise method of estimating the indirect effects of specified exogenous factors on endogenous variables (Hair et al., 2010). Gerbing and Anderson (1988) also argued that the most powerful and highly recommended multivariate data analysis tool for estimating variables' direct and indirect effects in multivariate structures is SEM. In addition, the SOBEL test was used to test the strength of message quality (mediator) on the relationship between WOM attributes and the resilience of NWOM messages. In the process, the criteria used were developed by (Baron & Kenny, 1986). Baron and Kenny suggested that full mediation is achieved when the relationship between independent and dependent variables is no longer significant after controlling the effect of the mediator variable. In contrast, partial mediation is realized when the relationship is significant but substantially reduced.

Common method variance
Additionally, due to the prevalence of self-reported information and the presence of common procedure bias in a study, it may pose a threat to data trustworthiness (Malhotra et al., 2006). It might have a detrimental effect on the study's conclusions (Conway & Lance, 2010). As a result, Harman's single factor test was employed to determine the proportion of variance explained by a single factor. The result indicates that a single factor accounts for less than 50% of the variance, at 45.174%. Hence, one can conclude that the data is devoid of common methodological bias.

Results
This section presents the empirical results and the estimations.

Descriptive analysis
This part focuses on the demographics of the people who were surveyed and how the central tendency was calculated. Table 1 shows that 30 (14.9%) of the customers who participated in this study were between the ages of 18 and 38, 68 (33.8%) were between the ages of 39 and 59, and 103 (51.3%) were over the age of 60. As a result, most customers have an average age of greater than 39, and as a result, they have been exposed to a high volume of positive WOM messages. In addition, the findings reveal that 73 (36.3%) and 128 (63.7%) of the customers had completed secondary and college education, respectively. In other words, they may be able to withstand poor public opinion and hence be resilient to NWOM. Meanwhile, government employees were the vast majority, with 114 (56.8%), 53 (26.3%) working for private companies, and 34 (16.9%) being self-employed. This indicates that the majority of customers are employed in the public sector. This could indicate that most customers have an ethnocentric attitude toward foreign items and that the people with whom they interact have a high level of expertise, homophily, trustworthiness, and message quality.

Measurement of central tendencies
All items addressed to 201 customers were included in this part, and the mean and standard deviation for all items are presented here. Table 2 shows that the mean and standard deviation for all measurement items are 4.9754 and 0.9574, respectively, with the highest mean and standard deviation. Furthermore, the lowest mean and standard deviation are 2.3255 and 0.1644,

Confirmatory factor analysis (CFA)
The obtained thresholds are all within the acceptable range, indicating that the anticipated model perfectly suits the data. The recommended and actual values are indicated in Table 3.

Measurement model assessment
Based on Table 4, the Cronbach's Alpha Coefficient for all constructs were more than 0.7, which indicates that the constructs examined in the study are internally consistent and trustworthy (Pallant, 2000;Tabachnick & Fidell, 2012). Furthermore, all items had Composite Reliability (CR) values greater than 0.7, indicating that the instruments were trustworthy. On the other hand, Average Variance Extracted (AVE) for all constructs were greater than 0.5, indicating a strong convergent validity. Additionally, the Maximum Shared Variance (MSV) was less than the AVE, showing that discriminant validity was achieved (Table 4; Fornell & Larcker, 1981).
Also, as indicated in Table 5, the square root of AVE was higher than the inner correlations between variables which is another proof for claiming discriminant validity (Fornell & Larcker, 1981).

Hypothesis testing
This study has 7 hypotheses. As shown in Table 6, EXP has a favorable and significant impact on MES (β = 0.567; p = 0.004). When EXP increases by 1 unit, it increases MES by 56.7%. Hence, the hypothesis H 1 was supported. Apart from that, according to the regression analysis results, the relationship between HOM and the MES was positive and statistically significant (β = 0.463; p = 0.001), implying that increasing HOM by 1 unit increases MES by 46.4%. Thus, hypothesis H 2 : was supported.
On the other hand, the study revealed a positive and statistically significant link between TRS and MES (β = 0.733; p < 0.001). This means that increasing 1 unit of TRS will increase MES by 73.3%. Consequently, hypothesis H 3 was supported. Finally, the relationship between MES and RES was positive and significant (β = 0.641; p = 0. 009). This further suggests that increasing 1 unit of MES will mean increasing 64.1% of RES. So, hypothesis H 4 : was supported. Generally, all attributes of WOM were found to be determinants of the MES and that MES was revealed to be a determinant of RES.

Testing the mediation effect of message quality
Specifically, the SOBEL test was used to establish how strong the mediator variable was. The SOBEL conditions were based on the criteria proposed by (Baron & Kenny, 1986). Generally, there were 3 models that were used to present the mediation effect of message quality. These models resulted from divisions made on WOM attributes such as expertise, homophily, and trustworthiness. The first criterion requires that an independent variable have a significant relationship with the dependent variable. In this case, all 3 WOM attributes had a significant relationship with resilience; expertise (ß = 0.9347; p < 0.001), homophily (ß = 0.7548; p < 0.001) and trustworthiness (ß = 0.6576; p < 0.001). Hence, the first criterion was fulfilled (Tables 7-9).  The second criterion requires the model to have a significant relationship between the independent and mediator variables. In this case, all 3 WOM attributes were found to have a significant relationship with message quality; expertise (ß = 0.7530; p < 0.001), homophily (ß = 0.4431; p < 0.001) and trustworthiness (ß = 0.5374; p < 0.001). This means the second criterion was fulfilled (Tables 7-9). The third criterion requires the mediator to significantly influence the dependent variable. The findings indicate that message quality had a significant relationship with resilience across all three models; (ß = 0. 0.3332; p < 0.001), (ß = 0.2212; p < 0.001) and (ß = 0.3364; p < 0.001) respectively (Tables 7-9).
The fourth criterion requires the significant levels in the relationship between the independent and dependent variables to decrease to have a partial mediation or to be insignificant to have a full mediation when the mediator variable is included in the model. In this case, the results indicate that message quality has yielded a decreased path coefficient result on all 3 WOM attributes: expertise (ß = 0.0535; p < 0.05), homophily (ß = 0.0543; p < 0.05) and trustworthiness (ß = 0.0023; p < 0.05). This implies that the fourth criterion was fulfilled (Tables 7-9). This further means that hypotheses H 5, H 6 and H 7 were supported.

Discussion of the findings
The main objective of this study was to understand the customers' resilience to negative word of mouth (NWOM) on locally made products. In so doing, the study was engaged in finding out the influence of WOM attributes on resilience on NWOM, especially when mediating these attributes with message quality. It is believed that the more effective WOM attributes are, the more the message quality will be developed between the sender of the message and the receiver of the message. Hence, the higher the possibility that message quality will help customers make informed decisions, especially when resisting the NWOM about locally made products. As a result of the research, it was hypothesized that using message quality as a mediator could be an alternate method of generating resistance to opting for foreign products among the customers, thus increasing the development of the local SMEs.
Apart from that, understanding the link between WOM attributes, message quality, and resilience to NWOM is important in advancing current knowledge on improving customer ethnocentrism and that SMEs might use to entice customers with negative attitudes towards foreign services to promote the country's economy. The study findings are also prompted by existing literature suggesting that the quality of the message, which has been seen for a long time as a mechanism to improve customer decision-making, is still an effective factor in preventing customers from using foreign products and opting for locally made products. The study proposed that examining the relationship between WOM attributes and resilience on NWOM can be done practically by incorporating other mediator variables such as message quality.
Specifically, grounded in prospect theory, the findings on the influence of expertise on message quality show that when the expertise of the sender of the message is taken into account, the receiver of the information tends to look at the experience, knowledge, qualifications, skills, and if the source provides a great deal of the information, For example, suppose customers believe that these items prevail in the discussion between them. In that case, they always saw the sources of information as competent enough to provide correct information, to the point that they saw no requirement to double-check it. As a result, they can trust the information and make well-informed decisions. Also, a range of characteristics, such as qualifications, training, education, career, and experience, can define expertise from the sources. Furthermore, the findings confirmed that when a source delivers expert information, it is assumed that the source has a deeper understanding of the topic than the average person, to the point where individuals may trust their views.
Although expertise is frequently connected with the power of making decisions, it may also be thought that experts can influence others due to the unique knowledge status they have gained through experience and professional conduct. These findings match (Martin & Lueg, 2013;Wangenheim & Bayón, 2004), who posted that sources with specific training skills are more likely to be deemed higher expertise. In recognition of their enviable position, WOM seekers are more likely to seek out and trust the recommendations of such specialists. Trusting expertise is what will make most of the customers go against some foreign products for locally made products.
Aside from that, findings on homophily confirm that the homophily between the sources and the receivers of communication is one of the reasons for the customers' belief in the message quality. The most likely sources of resisting NWOM against locally made products through WOM transmission are close friends and family members. These findings also showed that message quality is likely to improve when connected individuals are comparable in certain features such as social position, belief systems, common language, levels of education, and other such attributes. That is, when receivers  spend free time with the sender or have a similar outlook on life as the sender, they are more likely to believe that the information is trustworthy and can be used to make risky decisions, such as choosing locally made products over foreign products, which are thought to be of higher quality and durability.
Also, if receivers of the message have a common interest with the sender of the message, or if they have similar likes and dislikes with the sender, they will always develop a close social relationship, which can help them create a strong bond of trust. The findings concur with (Wangenheim & Bayón, 2004), who posted that Strong links between the sender and the recipient are widely considered to impact WOM effectiveness. The proximity or intimacy between a sender and a recipient defines the strength of the homophily of their link or tie.
Also, results indicate that the source's trustworthiness determines the message's quality. While making decisions like opting for locally made products over foreign products is considered a difficult decision due to the NWOM spread for a long time against the locally made product, customers need a reliable source of information. Customers will find it challenging to receive a message if they believe the source is dishonest. This also proves that a trustworthy person is courteous, fearless, and aware of others' feelings, making it easier to believe the quality of their message. This is similar to (Le et al., 2018), who suggested that all friendships are built on trust. As a friend, trust is very important because it helps and drives the message's quality, and it also helps the receivers of the message have more faith in the information in the message. Concerning the quality of the message, findings suggest that if the information is informative, clear, and reliable, customers can resist negative information and opt to buy locally made products instead of imported ones. The findings also suggest that introducing message quality will improve the ability to resist the NWOM if the message is specific, elaborative, explicitly, intense, and reinforcing.
This finding further provides empirical evidence that when a source presents a positive WOM for customers who are yet to switch to foreign products, the message quality can create a good link between WOM attributes and resilience against NWOM. Hence, it will drive more customers to continue using locally made products. This situation can be explained by the fact that existing customers relied heavily on the message quality before switching to other products. Further, message quality can influence how the customer reacts to the sender's message. So, that means that messages that have the right attributes are the best predictors of a customer's resistance to NWOM. Furthermore, it means that a message with qualities can lead to a reasonable evaluation. Thus, customers can employ quality messages to avoid NWOM that is most dynamic and driven by strong emotions like anger, impatience, and annoyance. In turn, good message quality encourages people to align their ideas and behaviors and reject information that contradicts their commitment to locally made products. Customers' having connections to sources with high expertise, homophily, and trustworthiness has systematically improved message quality, resulting in resilience to any negative attitudes that may prevent them from buying and using locally made products. This study's findings are congruent with those of (Le et al., 2018), who found that message quality has a positive and substantial effect on positive WOM.

Conclusion
Following the above discussion, it can be inferred that expertise, homophily, and trustworthiness are important for improving the quality of messages for buying and using locally made products and resisting NWOM. Furthermore, because the quality of messages determines customers' ability to reject unfavorable public ideas, it remains critical to customers' resilience to NWOM. The findings add to the existing research on WOM, particularly by focusing on the customer's resilience against NWOM on locally made products.

Practical implications
The practical implications of this study are that SMEs should make sure that they design proper communication strategies that will make senders of the information have clear and proper basis of information so that the senders can take a significant interest in translating their existing experience and knowledge to receivers by connecting the ideas of expertise, homophily, and trustworthiness into the formation of strong positive WOM about locally made products. However, it is advised that senders should provide accurate information instead of promoting poor-quality products made by local SMEs. Promotion of this kind will not help local SMEs; instead, it will increase the chances of making poor products. Furthermore, the culture of knowledge sharing within society should be reinforced. This will help customers have updated technical information before making risk decisions. Finally, the acquired message quality must be effectively utilized in order to develop resilience to negative WOM on locally made products; increasing the number of customers who opt to buy locally made products will reduce the likelihood of using foreign products because buyers of locally made products will act as the senders of positive WOM to others. Finally, the outcomes of this study will assist policymakers in developing appropriate policies and support programs to improve Tanzania's resilience to NWOM towards locally made products. This will benefit SMEs by increasing their tax bases, creating jobs, and contributing to the development of new indigenous technologies.

Theoretical implications
Within the context of theoretical implication, the study deploys prospect theory to study the effects of WOM attributes on message quality, the influence of message quality on resilience to NWOM on locally made products, and the mediating role of message quality on the relationship between message attributes and resilience to NWOM. Based on the prospect theory, which postulates that individuals make decisions under uncertainty by believing that individuals accept gains and losses based on the risks associated with a particular situation, the theory helps understand how customers can choose between foreign and locally made products. Therefore, the study extends theoretical understanding by successfully applying the theory into the WOM discipline, in which the concepts related to the attributes of WOM were successfully applied. Also, by involving a new concept of message quality as a mediator variable, the study confirms that several factors could help researchers and other practitioners make informed decisions when faced with uncertainty. Theoretically, the study extends the prospect theory by theorizing that customers can make an informed decision by being resilient to NWOM on locally made products by using attributes of WOM to influence message quality.

Limitations and areas for further studies
Since this study used a cross-sectional research design combined with a quantitative research approach, future researchers should use a longitudinal strategy to compare any variations in expertise, homophily, trustworthiness, and message quality regarding resilience to negative WOM on locally made products. Alternatively, qualitative investigations can be carried out to augment the quantitative findings, since methodological triangulation may better help to understand the mediating effects of expertise, homophily, and trustworthiness on message quality and message quality on resilience. Finally, this study concentrated on customers from only 3 SMEs: furniture, craftsmanship, and fashion. Different studies may sample more SMEs to increase the generalization ability of the findings. In addition, convenience sampling may result in biases. Hence, whenever the sampling frame is accessible, future investigations may employ simple random sampling.