The in fl uence of relationship intention on cell phone users ’ attitudes towards complaining and complaint behaviour

3In any service environment, there is a strong possibility that customers’ expectations and the actual service delivery are not in unison. When service failures do occur, customers’ attitudes towards complaining directly infl uence their actual complaint behaviour. As not all customers want to build relationships with service providers, it is imperative that service providers gain a deeper understanding of the behaviour, and specifi cally the complaint behaviour, of those customers who do have relationship intentions. The purpose of this study was to determine the infl uence of relationship intention on customers’ attitudes towards complaining and complaint behaviour following a service failure within the cell phone industry. Convenience sampling was used, and 605 respondents participated in the study. The results indicate that the majority of respondents who participated in this study had a propensity to complain, and that respondents with high relationship intentions are more likely to voice a billing error to their cell phone network provider than respondents with low relationship intentions. Furthermore, relationship intention should be considered as a variable that could infl uence customers’ attitudes towards complaining and complaint behaviour. The results make a valuable theoretical contribution and have managerial implications for service providers in the cell phone


INTRODUCTION
1C ustomers cannot evaluate services before purchase due to their intangibility (Berry 1995: 237;Oliver 1980: 460;Zeithaml, Berry & Parasuraman 1993: 1).Consequently, even service providers who place a high priority on the provision of quality service to customers are prone to service failures (Harrison-Walker 2012: 115;Lacey 2012: 137;Tsarenko & Tojib 2011: 383, 391), when the service does not meet customers' expectations (Harrison-Walker 2012: 115).In the event of a service failure, customers with positive attitudes towards complaining will complain to service providers, thereby enabling service providers to remedy the situation and restore their satisfaction (Tax, Brown & Chandrashekaran 1998: 72), instead of switching to another service provider (Yuksel, Kilinc & Yuksel 2006: 22).
Within the cell phone industry, the continuous new technological developments (Franzak & Pitta 2011: 396), greater variety of services offered, number portability (Seo, Ranganathan & Babad 2008: 182 195), and more cell phone network providers entering the market (ICASA 2012b: 3, 30, 39), have simplified the process of switching to another cell phone network provider when customers are dissatisfied with service provision.In the cell phone industry specifically, customers' switching behaviour is reduced through satisfaction with every service encounter (Stone & Dickey 2002: 480).Despite service providers attempting to continually provide excellent service, however, service failures are unavoidable (Harrison-Walker 2012: 115;Lacey 2012: 137;Tsarenko & Tojib 2011: 383, 391).Service failures result in customer dissatisfaction, which, in turn, could negatively impact customer retention and longterm profitability (Robinson, Neeley & Williamson 2011: 90).A service failure in the cell phone industry could therefore have detrimental effects on cell phone network providers' profitability and can even result in customers switching to another cell phone network provider.
Cell phone network providers do not want customers to switch to competitors, as customer retention results in increased profitability (Jena, Guin & Dash 2011: 23).Enduring relationships between cell phone network providers and customers prevent customers from switching to competitors and, for this reason, cell phone network providers employ relationship marketing strategies to build long-term relationships with customers (Coulter & Ligas 2004: 489;Sheth & Parvatiyar 2002: 4).However, not all customers are receptive to relationship marketing strategies; only some customers have relationship intentions and want to build relationships with service providers (Beetles & Harris 2010: 353, 354;Hess, Story & Danes 2011: 22;Kumar, Bohling & Ladda 2003: 669;Petruzzellis 2010: 625).
Those customers who want to continue their relationships with service providers will provide feedback to service providers when their expectations are not met (Kumar et al. 2003: 670;Lacey 2012: 141;Tsarenko & Tojib 2011: 386).Despite the fact that independent variables such as relationship intention, which is external to the service failure situation, can influence customers' behaviour relating to service failures, this has not been investigated (Holloway, Wang & Beatty 2009: 385).As only certain customers have relationship intentions (Kumar et al. 2003: 670) towards their cell phone network provider, the effect of relationship type on customers' reactions to service failures in terms of their complaint behaviour should be investigated (Hedrick, Beverland & Minahan 2007: 69).The purpose of this study is therefore to determine the influence of relationship intention on customers' attitudes towards complaining and complaint behaviour following a service failure within the cell phone industry.

Literature review
Relationship marketing and relationship intention 1S atisfactory service delivery, together with direct contact between service providers and customers, contributes to the establishment of enduring relationships between service providers and customers (Grönroos 2004: 100;Mason & Simmons 2012: 227).The development of enduring relationships is furthermore the result of invested resources (time, effort and money) in the relationship by both service providers and customers, resulting in higher switching costs (Kinard & Capella 2006: 360) and, thus, prohibiting relationship dissolution (Homburg, Giering & Menon 2003: 44).Long-Tolbert and Gammoh (2012: 397) suggest that where relationships between service providers and customers exist, customers can be considered as active partners of their service provider.However, by focusing relationship marketing efforts on all customers, service providers dissipate their resources (Odekerken-Schröder, De Wulf & Schumacher 2003: 178;Tuominen 2007: 182).To prevent dissipating resources, service providers should consider customers' relationship intentions before any relationship-building efforts are contemplated (Beetles & Harris 2010: 353, 354;Hess et al. 2011: 22;Kumar et al. 2003: 669;Petruzzellis 2010: 625).Targeting customers who are receptive to relationship building (thus, with relationship intentions) should therefore be the first step of relationship marketing strategies (Berry 1995: 239;Liang & Wang 2006: 124).Kumar et al. (2003: 669) define relationship intention as a customer's intention to build a relationship with a particular service provider while purchasing products or services from the service provider.Kumar et al. (2003: 670) propose five constructs to measure customers' relationship intentions, namely, involvement, expectations, fear of relationship loss, forgiveness and feedback.

Involvement
1I nvolvement refers to customers' willingness to engage in relationship activities irrespective of obligation or coercion (Kumar et al. 2003: 670).According to Scott and Vitartas (2008: 54) and Kinard and Capella (2006: 365), customers with stronger feelings of attachment to, and involvement with, their service provider will be more responsive to requests from their service provider for suggestions, and will be more inclined to provide comments about performance or service delivery to their service provider.Highly involved customers will thus not only have contact with their service provider (Scott & Vitartas 2008: 54), but depending on the degree of their expertise in the market, they will give informative and positive feedback to their service provider (Ruiz, Castro & Armario 2007: 1094).This, in turn, generates knowledge for service providers with regard to customer needs and expectations (Mascarenhas, Kesavan & Bernacchi 2004: 486-487) which enables them to remain relevant to customers by knowing and exceeding customers' expectations (Engeseth 2006: 36-37).

Expectations
1C ustomer expectations can be viewed as what customers think service delivery will entail (Kim, Ok & Canter 2012: 60-61).Customer expectations should therefore be managed throughout the service experience (Sharma, Tam & Kim 2012: 531), as expectations act as service delivery standards influencing satisfaction (Berry & Parasuraman 1997: 65;Oliver 1980: 460;Zeithaml et al. 1993: 1).Kumar et al. (2003: 670) argue that high customer expectations portray concern about product or service quality and influence customers' intentions to build relationships with service providers.Expectations are developed based on inherent customer characteristics (Mason & Simmons 2012: 233), non-experiential information from sources that service providers can control (such as advertising), and cannot control (such as wordof-mouth and general media), as well as previous experience with the service provider (Anderson, Fornell & Lehmann 1994: 56).Experience with the service provider and individualised attention to the customer (Parasuraman, Berry & Zeithaml 1991: 41) create a bond between customers and the service provider (Moore, Ratneshwar & Moore 2012: 254), which customers might not want to lose (Kumar et al. 2003: 670).

Fear of relationship loss
1C ustomers form bonds with service providers when they experience repetitive satisfactory service which, in turn, can develop into a lasting commitment towards service providers (Homburg et al. 2003: 44;Liang & Wang 2006: 123;Spake & Megehee 2010: 316, 319-320).Furthermore, the benefits customers receive from long-term relationships with service providers above and beyond the core service, whether it be confidence benefits, social benefits or special treatment benefits, are considered to be relational benefits (Gwinner, Gremler & Bitner 1998: 102, 109-110).Customers make a trade-off between the relational benefits and the costs of maintaining relationships with service providers in deciding between maintaining or ending these relationships (Spake & Megehee 2010: 316).Because customers fear losing their relational benefits and bonds with their current service provider, they will consider the aforementioned benefits and costs to decide whether they will switch from or stay with their current service provider when service failures occur (Chelminski & Coulter 2011: 366).Kumar et al. (2003: 670) therefore argue that bonds between customers and service providers result in customers fearing the loss of their relationships with service providers.

Forgiveness
1F orgiveness provides a relational focus to service failures, because forgiveness releases the negative emotions associated with the failure by instead pursuing actions to restore the relationship between a customer and the service provider (Chung & Beverland 2006: 98;Hedrick et al. 2007: 70;McCullough, Fincham & Tsang 2003: 540;McCullough, Worthington & Rachal 1997: 333;Tsarenko & Tojib 2011: 382).For example, a customer voicing dissatisfaction to the service provider (thus pursuing an action to restore the relationship) instead of switching to another service provider could possibly portray forgiveness.Forgiveness will therefore be used to deal with dissatisfaction (Worthington & Scherer 2004: 402).Kumar et al. (2003: 670) accordingly propose that customers with high relationship intentions will be more inclined to forgive service providers when service failures occur.

Feedback
1C ustomers with higher relationship intentions will voluntarily provide positive or negative feedback to service providers without expecting a return or reward (Kumar et al. 2003: 670).Relationship intention customers are thus good sources from which to obtain feedback, as customers initiate the dialogue without expecting a return or reward.Such feedback furthermore facilitates the dialogue between customers and service providers necessary for creating value (Grönroos 2004: 103).Service providers use positive feedback for identifying strengths that can be further reinforced, and use negative feedback (mostly regarded as complaints) to improve service provision (Berry & Parasuraman 1997: 65;Wirtz, Tambyah & Mattila 2010: 380) and prevent the recurrence of a similar dissatisfactory service (Chelminski & Coulter 2011: 370).
Customers provide feedback to improve the service (Wirtz et al. 2010: 380), and for altruistic reasons (McCullough et al. 1997: 322).Altruism increases when service dimensions relate to public interest.A customer who receives bad-tasting food at a restaurant might switch to another restaurant in future, but a perceived health hazard will result in negative feedback to the service provider to allow for correction in service delivery before other customers are harmed (Hirschman 1980: 434).Most dissatisfied customers do not, however, provide negative feedback to service providers about their dissatisfaction, but would rather choose to take a form of private action (Ha 2004: 200;McCollough, Berry & Yadav 2000: 133).Otherwise satisfied customers, who want to continue the relationship with their service provider, are more likely to deploy negative feedback to the service provider (Lacey 2012: 141;Proença & Rodrigues 2011: 205;Tsarenko & Tojib 2011: 386), because they expect service providers to rectify the situation so that the relationship can be restored to its former state (Tax et al. 1998: 72).
Service failure 1A ll service providers, no matter how consistently good the quality of the service they provide, and despite intentions to always deliver quality service, are subject to service failures (Harrison-Walker 2012: 115;Lacey 2012: 137;Tsarenko & Tojib 2011: 383, 391).A service failure occurs whenever a service outcome does not meet customers' expectations (Harrison-Walker 2012: 115).The intangible nature of services prevents customers from evaluating services prior to purchase (Berry 1995: 237), resulting in subjective evaluations of services against expectations.Whenever service failures occur, customers will experience dissatisfaction (Oliver 1980: 460) which, in turn, can result in complaints, retaliation, negative word-of-mouth, lingering anger, resentment, hostility and even switching to another service provider, all of which result in considerable loss of future value for the service provider (Nikbin, Ismail, Marimuthu & Abu-Jarad 2011: 19;Tsarenko & Tojib 2011: 382).However, customers consider which dimensions of the service are contributing to their feelings of dissatisfaction; to whom the responsibility for the service failure should be attributed; what the service provider can do to rectify the problem; and the extent to which the customer can influence the service provider to take corrective action before forming attitudes towards complaining (Day 1984: 497).
Attitude towards complaining 1At titude towards complaining can be defined as customers' tendencies to voice their dissatisfaction to service providers (Bearden & Mason 1984: 492;Day 1984: 497;Richins 1982: 505).Three dimensions that influence customers' attitudes towards complaining are, firstly, how customers balance their perception of the objective (such as the time and money spent on the complaint process) and psychological costs in relation to service providers' responsiveness to complaints; secondly, customers' individual norms concerning complaining; and, lastly, customers' perceptions of the societal benefits resulting from complaining (Richins 1982: 505).
Moreover, attitudes are customer context-specific: customers decide on the strength and direction of the attitude with regard to a specific aspect of the environment, within a given context (Petruzzellis 2010: 615).Therefore, customers' attitudes towards complaining can be considered in terms of two directions, namely, propensity to complain (or positive attitude towards complaining) and negative attitude towards complaining (Yuksel et al. 2006: 17).Customers with more favourable attitudes towards complaining are more likely to react to their dissatisfaction by expressing their complaint intention to service providers (Bearden & Mason 1984: 495;Richins 1982: 505;Yuksel et al. 2006: 15, 22), as they want to continue supporting the service provider in the future (Chang & Chin 2011: 128).Customers who voice their dissatisfaction to service providers therefore have a propensity to complain (Phau & Sari 2004: 422;Yuksel et al. 2006: 17), while customers with a negative attitude towards complaining will probably switch to another service provider when service failures occur (Yuksel et al. 2006: 22).According to Kumar et al. (2003: 670), customers with higher relationship intentions would also want to continue supporting the service provider and will, therefore, complain to service providers.
As customers' attitudes influence their behaviour (Tung, Shih, Wei & Chen 2012: 998), customer complaint behaviour changes in relation to customers' attitudes towards complaining, in that customers complain more if they believe that their complaints will be taken seriously.Customers' attitudes towards complaining thus influence their actual complaint behaviour (Richins 1982: 505).

Customer complaint behaviour
1C ustomer complaint behaviour encompasses the multiple behavioural and nonbehavioural responses of customers to service failures (Singh 1988: 94).According to Blodgett, Hill and Tax (1997: 187), Hirschman's theory of exit, voice and loyalty (Hirschman 1980: 432, 434-435, 439), the expectancy disconfirmation paradigm (Oliver 1980: 466), the satisfaction/dissatisfaction model (Day 1984: 497) and the attribution theory (Folkes, Koletsky & Graham 1987: 534) can all be considered as the foundations of customer complaint behaviour.However, this study chose the taxonomy of customer complaint behaviour developed by Singh (1988: 104) to operationalise customer complaint behaviour, as shown in Table 1.mcmlxxiii Complaining to an external agency Source: Constructed from Singh (1988: 104) According to Singh's taxonomy, customers can choose to take no action, to take private action or to take public action.Instead of choosing to take no action by not complaining, most dissatisfied customers either spread negative word-of-mouth or voice their dissatisfaction to service providers (Chelminski & Coulter 2011: 361;Tax & Brown 1998: 79).Considering private action, dissatisfied customers take the effort, time and money associated with the establishment of a new relationship with another service provider into consideration before they decide to switch to another service provider (Lam, Shankar, Erramilli & Murthy 2004: 297).Furthermore, negative word-of-mouth presents a problem to service providers in terms of future customer acquisition and future customer spending, as customers view word-of-mouth to be credible, and other customers will thus not support a service provider about whom negative word-of-mouth is spread (Nikbin et al. 2011: 20).
With regard to public action, voicing is considered as the level of comfort customers have to proactively verbalise concerns to service providers in the form of negative feedback, which also aids service providers in understanding customers' expectations (Lacey 2012: 141).Furthermore, through voicing, negative word-ofmouth is prevented and customers get the opportunity to reduce the dissonance caused by the service failure (Chelminski & Coulter 2011: 370).Customers who voice their dissatisfaction expect service providers to restore their satisfaction in order to restore the relationships with these service providers (Tax et al. 1998: 72).Since customers with relationship intentions want to continue their relationships with their service providers (Kumar et al. 2003: 670), relationship intention should influence customers' complaint behaviour following a service failure.

Problem statement and objectives
1C omplaint behaviour following service failures is influenced by customers' attitudes towards complaining (Richins 1982: 505).Furthermore, customers' decisions to continue relationships with service providers, or switch to another service provider, also influence their complaint behaviour (Tax et al. 1998: 72).Customers with relationship intentions would like to continue their relationships with their service providers (Kumar et al. 2003: 669), and relationship intention should therefore influence attitudes towards complaining and customer complaint behaviour.Kumar et al. (2003: 670) hypothesised that customers with relationship intentions care for and are involved with their service providers and have high expectations of and fear losing their relationships with their service providers.Such customers will also forgive service failures and provide feedback to their service providers.Previous research, especially in South Africa, on relationship intention focused mainly on scale development (De Jager 2006: 5;Delport, Steyn & Mostert 2011: 277;Kruger 2010: 5;Kruger & Mostert 2012: 45;Mentz 2007: 2).The influence of relationship intention on customers' attitudes towards complaining and complaint behaviour after a service failure has not been investigated.
The South African cell phone industry was chosen as the context for this study, as service failures do occur (ICASA 2012a: 28) within this highly competitive industry (Morrisson & Huppertz 2010: 250).The Independent Communications Authority of South Africa (ICASA) received a wide range of complaints (4 553 complaints for the 2011-2012 period) in which cell phone users complained about billing (31%), contract terms and conditions (27%), quality of service (21%) and some other complaints (ICASA 2012a: 28).Cell phone network providers are furthermore perceived as being similar, and customers do not perceive a clear differentiation between cell phone network providers (Haenlein & Kaplan 2012: 467).Considering these characteristics of the cell phone industry and the fact that customers can switch to another cell phone network provider if necessary (Seo et al. 2008: 182 195), marketers should attempt to build relationships with customers showing relationship intentions to retain them.
The aim of this study was to determine the influence of relationship intention on cell phone users' attitudes towards complaining and complaint behaviour.The following objectives were formulated for this study: • Determine cell phone users' attitudes towards complaining • Establish the influence of relationship intention on cell phone users' attitudes towards complaining • Identify cell phone users' complaint behaviour following a service failure • Determine the influence of relationship intention on cell phone users' complaint behaviour following a service failure.

Research methodology
Research design and target population 1A descriptive research design in the form of non-probability convenience sampling was used for this study.The target population included Gauteng residents 18 years or older who had used a cell phone network provider for at least three years.

Questionnaire design
1C losed-ended questions were used throughout the questionnaire with an unlabelled five-point Likert scale for all scale items.The questionnaire started with a preamble explaining respondents' rights and the purpose of the study, followed by screening questions to ensure that respondents form part of the target population of the study.To capture all the information necessary for the study, four different sections were used.
Section A captured classification and patronage habit information concerning respondents' cell phone network providers.Section B established relationship intention using the measuring instrument proposed by Kruger and Mostert (2012: 45), utilising Likert scales where 1 = definitely not, and 5 = definitely.
Section C dealt with respondents' attitudes towards complaining and complaint behaviour following a service failure using a scenario.Concerning the scenario of a service failure within the cell phone industry, ICASA (2012a: 28) states that 31% of all complaints regarding cell phone network providers relate to billing.For this reason, it is believed that the target population would consider a service failure scenario pertaining to a billing error as relevant and credible.Respondents were presented with a hypothetical service failure scenario pertaining to a billing error by their cell phone network provider (see Annexure A), and respondents' complaint behaviour was measured according to their reaction to this scenario.Instead of asking respondents to recall an event, which could result in biases due to memory lapse, the use of scenarios in service failure research is becoming common practice ( Attitude towards complaining was measured with scale items adapted from Phau and Sari (2004: 414, 417), Walters (2010: 242, 243) and Yuksel et al. (2006: 16), where 1 = strongly disagree and 5 = strongly agree.Customer complaint behaviour was measured using scale items adapted from Chelminsky and Coulter (2011: 363-364), DeWitt and Brady (2003: 205), Ekiz and Au (2011: 335), Singh (1988: 105), Walters (2010: 240-244) and Yuksel et al. (2006: 16), where 1 = strongly disagree and 5 = strongly agree.
Section D obtained respondents' demographic details such as their gender, age and population group.A pilot study to test the feasibility of the questionnaire and identify any problems that should be rectified before fielding the questionnaire (Zikmund & Babin 2010: 61-62) was done with 27 respondents from the target population.

Data collection and data analysis
1P ersonal in-home interviews were conducted by trained fieldworkers using interviewer-administered questionnaires.This approach is appropriate, as the service failure scenario had to be explained to respondents (Bradley 2007: 128).
The Statistical Package for Social Sciences (Version 21) and the SAS statistical programme (Version 9.3) were used for statistical analysis.This study used a confidence level of 95% and subsequent significance level of 0.05.However, because statistical significance does not indicate the strength of the significance, r-values of Cohen for Pearson product-moment correlation coefficients, and d-values of Cohen for ANOVAs, to determine practical significance by means of effect size (Steyn 1999: 3), were also calculated.Practical significance in terms of r-values is considered to be small at 0.1, medium at 0.3 and large at 0.5 (Cohen 1988: 79-81).Practical significance in terms of d-values is considered to be small at 0.2, medium at 0.5 and large at 0.8 (Cohen 1988: 25-26).Since medium effect sizes have ample practical effect, as differences between respondent groups can already be noticed with the naked eye (Cohen 1988: 20), medium and large effect sizes were regarded as practically significant when interpreting results.All r-values and d-values were rounded off to one decimal.

Respondent profi le
A total of 605 respondents participated in this study.Table 2 presents the frequencies and percentages for the respondent profile.
As indicated in Table 2, slightly more than half the respondents were female (53.7%) and the majority of respondents were in the 30 to 39 years age group (22.3%), or in the 20 to 24 years age group (21%).Furthermore, the majority of respondents were Black African (33.6%) and White (28.3%).

Respondents' cell phone patronage habits
1T able 3 presents the frequencies and percentages for respondents' cell phone patronage habits.From Table 3 it can be observed that 43.0% of respondents used Vodacom, 34.4% used MTN and 16.5% used Cell C as their cell phone network provider, while only 3.3% and 2.8% of respondents used 8-ta and Virgin Mobile respectively.Just over half the respondents had a contract with their cell phone network provider (52.2%), and the majority of respondents spent between R101 and R250 per month on cell phone expenses (36.2%).

Reliability and validity
1Th e reliability of a scale examines whether the same results will be obtained if the study is repeated (Bradley 2007: 64) and, therefore, reliability reflects consistency (Zikmund & Babin 2010: 334).Through examining internal-consistency reliability, the reliability of the test based on the consistency that occurs within the test is examined (Sprinthall 2003: 480).The internal consistency reliability of the relationship intention scale, attitude towards complaining scale, and compliant behaviour scale were assessed through the calculation of Cronbach's alpha coefficient values, where coefficient values of 0.6 and more are considered reliable (Bagozzi 1994: 18;Malhotra 2010: 319).
An advantage when examining Cronbach's alpha coefficient values (indicated in Table 4 for all underlying dimensions of the three scales used in this study) includes that each item is individually assessed for variability.For this reason, items contributing to the overall reliability of the scale and items not contributing to the reliability of the scale can be identified (Zikmund & Babin 2010: 334;Sprinthall 2003: 487).Although two of the underlying dimensions of the customer complaint behaviour scale, namely no action and negative word-of-mouth, had only one item measuring the dimensions, and no Cronbach's alpha coefficient value could be determined for these two dimensions, the impact on the Cronbach's alpha coefficient for the total customer complaint behaviour scale value when removing these two items was considered.The Cronbach's alpha coefficient value increased by only 0.02 when removing the item measuring no action, and increased by 0.01 when removing the item measuring negative word-of-mouth.The two items measuring no action and negative word-of-mouth thus do not negatively impact on, but contribute to, the reliability of the customer complaint behaviour scale.Furthermore, the scale of customer complaint behaviour, with the underlying dimensions used in this study, had been widely used and validated (Chelminsky & Coulter 2011: 363-364;DeWitt & Brady 2003: 205;Ekiz & Au 2011: 335;Singh 1988: 105;Walters 2010: 240-244;Yuksel et al. 2006: 16).For these reasons, the underlying dimensions of no action and negative word-of-mouth were also considered for further analysis to gain more insight from the underlying dimensions of customer complaint behaviour of the respondents who participated in this study.It should be noted that no hypotheses testing was done, as the results cannot be generalised beyond the context of this study.Furthermore, considering the aforementioned decision, and after determining the Cronbach's alpha coefficient values for the underlying dimensions of all three scales where multiple items were used (indicated in Table 4), it can be concluded that the measurement scales, namely relationship intention with the five underlying dimensions, attitude towards complaining with the two underlying dimensions, and customer complaint behaviour with the five underlying dimensions, are reliable for the purposes of this study.
Table 4 presents the Cronbach's alpha coefficient values for the relationship intention scale, the attitude towards complaining scale, and complaint behaviour scale, as well as the underlying dimensions of these scales as determined through confirmatory factor analyses when examining the construct validity.Validity can be defined as the extent to which a test truthfully represents a concept (Zikmund & Babin 2010: 335).Whether the measure does in fact measure what it is supposed to measure is determined through construct validity (Zikmund & Babin 2010: 337).Confirmatory factor analyses were performed to determine the underlying dimensions of constructs and to determine construct validity (Bagozzi 1994: 342-344).The scales of relationship intention, attitude towards complaining, and complaint behaviour are considered valid, since the measure of sampling adequacy (MSA) was above 0.60, and more than 50% of the variance was explained by the underlying dimensions of each scale.By examining the eigenvalue for each scale, the underlying dimensions were uncovered and labelled, as indicated in Table 4.
Classifying respondents according to their relationship intentions 1A mean score was calculated for respondents' overall relationship intentions.Considering that the midpoint of the Likert scale used (where 1=definitely not, and 5=definitely) was 3, mean scores indicate that respondents participating in this study showed a tendency to higher relationship intentions towards their cell phone network provider (mean=3.60).In order to determine the influence of respondents' relationship intention levels on their attitudes towards complaining and their complaint behaviour, respondents were categorised into three almost equally sized groups, using the 33.3 and 66.6 percentiles as cut points on their overall mean scores for relationship intention.Table 5 presents the frequencies, means and standard deviations (SD) for the three relationship intention groups determined from the aforementioned categorisation.The number of respondents per group differed due to the fact that ties occurred in the continuous data.From Table 5, it can be deduced that 200 respondents were categorised as having low relationship intentions (mean=2.90),208 respondents as having moderate relationship intentions (mean=3.62),and 197 respondents as having high relationship intentions (mean=4.29).

Attitude towards complaining
1M ean scores were calculated for the underlying dimensions of attitude towards complaining.Respondents were subsequently categorised according to their overall mean scores for propensity to complain and negative attitude towards complaining.Respondents with a higher mean score for propensity to complain than negative attitude towards complaining were grouped into the propensity to complain group, while respondents with a higher mean score for negative attitude towards complaining than propensity to complain were grouped into the negative attitude towards complaining group.Respondents with equal mean scores for both propensity to complain and negative attitude towards complaining were removed from the analysis (10 respondents).Table 6 presents the descriptive statistics related to respondents' attitudes towards complaining.Reverse scoring was used for negative attitude towards complaining, which entails that a low score indicates that respondents have low negative attitudes towards complaining, and a high score indicates that respondents have high negative attitudes towards complaining.From Table 6, it is evident that approximately 90% of respondents were grouped in the propensity to complain group, while the remainder were grouped in the negative attitude towards complaining group.Mean scores indicate that respondents in the propensity to complain group had a higher mean score (mean=4.29)for propensity to complain than negative attitude towards complaining (mean=2.35).Furthermore, respondents in the negative attitude towards complaining group had a higher mean score for negative attitude towards complaining (mean=3.91)than propensity to complain (mean=3.10).It can therefore be concluded that the majority of respondents who participated in this study had a propensity to complain.

Relationship intention and attitude towards complaining
1Th e relationship between respondents' relationship intentions and their attitudes towards complaining was determined using Pearson product-moment correlation coefficients.Table 7 presents the r-value of the Pearson product-moment correlation coefficients between respondents' relationship intentions and propensity to complain, as well as negative attitude towards complaining.7 indicates that there was a significant large positive correlation between respondents' relationship intentions and their propensity to complain (r=0.5),where propensity to complain increases as relationship intention increases.It can therefore be concluded that an increase in respondents' relationship intentions is concurrent with their propensities to complain.
From Table 7 it can also be observed that a significant negative correlation exists between respondents' relationship intentions and negative attitude towards complaining.The strength of the correlation is small (r=-0.1),however, and therefore not of practical significance.It can therefore be concluded that there is no relationship between respondents' relationship intentions and negative attitude towards complaining.

Customer complaint behaviour
1T he mean scores were calculated for the underlying dimensions of customer complaint behaviour.Table 8 presents the mean scores and standard deviations (SD) for no action, negative word-of-mouth, switching, voicing and complaining to an external agency.Reverse scoring was used for no action, which entails that a low score indicates respondents would take action, and a high score indicates that respondents would take no action after a billing error by their cell phone network provider.As indicated in Table 8, respondents are likely to take action after a billing error by their cell phone network provider (mean=1.67).Table 8 furthermore indicates that respondents are likely to spread negative word-of-mouth about their cell phone network provider following a billing error (mean=4.23),and are prone to voicing their dissatisfaction about a billing error to their cell phone network provider (mean=3.71).Respondents are, however, less likely to complain about a billing error by their cell phone network provider to an external agency (mean=2.83),and least likely to switch to another cell phone network provider after a billing error (mean=2.77).

Relationship intention and customer complaint behaviour
1Th e relationship between respondents' relationship intentions and their complaint behaviour after a service failure by their cell phone network provider was investigated using Pearson product-moment correlation coefficients.Table 9 presents the r-value of the Pearson product-moment correlation coefficients between respondents' relationship intentions and no action, negative word-of-mouth, switching, voicing, as well as complaining to an external agency.9 indicates no significant correlation between relationship intention and no action.From Table 9, significant positive medium correlations between respondents' relationship intention and negative word-of-mouth (r=0.3), as well as voicing (r=0.3), are furthermore evident.It can therefore be concluded that as respondents' relationship intentions increase, so will their negative word-of-mouth and voicing after a billing error by their cell phone network provider.A significant positive correlation between relationship intention and complaining to an external agency is also evident in Table 9.The strength of the correlation is small (r=0.2),however, and therefore not of practical significance.It can therefore be concluded that there is no relationship between respondents' relationship intentions and complaining to an external agency after a billing error by their cell phone network provider.Furthermore, Table 9 indicates a significant negative correlation between relationship intention and switching, albeit small (r=-0.1),and therefore not of practical significance.It can therefore be concluded that there is no relationship between respondents' relationship intentions and switching after a billing error by their cell phone network provider.
To determine whether respondents with different levels of relationship intention differ with regard to their complaint behaviour following a service failure, analyses of variance (ANOVAs) were performed to determine whether differences between the means of dimensions existed.Table 10 shows the descriptive statistics for each complaint behaviour dimension as well as Tukey's comparison (statistical significance at the 0.05 level) and d-values (effect sizes) when comparing the means of the underlying dimensions of complaint behaviour, namely, no action, negative word-of-mouth, switching, voicing and complaining to an external agency, for the different relationship intention levels.From Table 10, it is evident that statistically significant differences exist between respondents with low and moderate relationship intentions, and between respondents with low and high relationship intentions for all the underlying complaint behaviour dimensions.
Table 10 indicates a medium effect size (d=0.7)between respondents with high relationship intentions and respondents with low relationship intentions regarding no action.A medium effect size (d=0.5)between respondents with moderate relationship intentions and respondents with low relationship intentions is also evident for the no action response.The mean scores indicate that respondents with low relationship intentions (mean=2.11)are more likely not to take action than respondents with high relationship intentions (mean=1.49),as well as respondents with moderate relationship intentions (mean=1.65).It can therefore be concluded that respondents with low relationship intentions are more likely not to take action after a billing error by their cell phone network provider than respondents with moderate and high relationship intentions.
Regarding negative word-of-mouth, Table 10 shows a medium effect size (d=0.7)between respondents with high relationship intentions and respondents with low relationship intentions.A medium effect size (d=0.5)between respondents with moderate relationship intentions and respondents with low relationship intentions for negative word-of-mouth is also evident.The mean scores indicate that respondents with high relationship intentions (mean=4.49)as well as respondents with moderate relationship intentions (mean=4.33)are more prone to negative word-of-mouth than respondents with low relationship intentions (mean=3.89).It can therefore be concluded that respondents with high and moderate relationship intentions are more likely to tell their friends or family about a billing error by their cell phone network provider than respondents with low relationship intentions.
Although statistically significant differences between respondents with high relationship intentions and respondents with low relationship intentions, as well as respondents with moderate relationship intentions and respondents with low relationship intentions concerning switching were found, the effect sizes are small (d=0.3 and d=0.2, respectively), and therefore not of practical significance.When the differences between respondent groups are not practically significant, the differences should not be considered for marketing strategies, as the effect of these differences on customer switching is very small.
Table 10 also indicates a medium effect size (d=0.5)between respondents with high relationship intentions and respondents with low relationship intentions with regard to voicing.Mean scores indicate that respondents with high relationship intentions (mean=3.95) are more likely to voice their dissatisfaction than respondents with low relationship intentions (mean=3.38).Although a statistically significant difference was found between respondents with moderate relationship intentions and respondents with low relationship intentions concerning voicing, the effect size is small (d=0.4), and therefore not of practical significance.It can therefore be concluded that respondents with high relationship intentions are more likely to voice a billing error to their cell phone network provider than respondents with low relationship intentions.
Although statistically significant differences were found between respondents with high relationship intentions and respondents with low relationship intentions, as well as respondents with moderate relationship intentions and respondents with low relationship intentions with regard to complaining to an external agency, the effect sizes are small (d=0.4 and d=0.2, respectively) and therefore not of practical significance.When the differences between respondent groups are not practically significant, the differences should not be considered for marketing strategies, as the effect of these differences on complaining to an external agency is very small.

Discussion
1C ustomers' attitudes towards complaining influence their complaint behaviour (Richins 1982: 505).The latter is also influenced by customers' decisions to continue relationships with their service provider (Tax et al. 1998: 72).As customers with relationship intentions want to continue their relationships with their service provider (Kumar et al. 2003: 669), relationship intention should influence customers' attitude towards complaining as well as their complaint behaviour.This study examined the influence of relationship intention on customers' attitudes towards complaining and compliant behaviour following a service failure scenario by their current cell phone network provider.
Results indicated that respondents had a tendency for higher relationship intentions towards their cell phone network provider, and that the majority of respondents participating in this study had the propensity to complain.One of the main findings of this study was that, as respondents' relationship intentions increase, so do their propensities to complain.This implies that those respondents with higher relationship intentions also have higher propensities to complain than respondents with lower relationship intentions.These findings support the premise of Kumar et al. (2003: 670) that higher relationship intention results in more concern for, and involvement with, the service provider through, amongst other actions, complaints to service providers to communicate customers' expectations.
Considering customer complaint behaviour, the results of this study suggest that respondents will take action after a billing error by their cell phone network provider.Respondents with high and moderate relationship intentions are more prone to take action after a billing error by their cell phone network provider than respondents with low relationship intentions.While customers with higher relationship intentions might not object to paying a premium price for their service (Kumar et al. 2003: 672-673), respondents will not let a billing error by their cell phone network provider go unnoticed.
Respondents will spread negative word-of-mouth after a billing error by their cell phone network provider, which will increase as relationship intention increases.Respondents with high and moderate relationship intentions will spread more negative word-of-mouth after a billing error by their cell phone network provider than respondents with low relationship intentions.In contrast to suggestions by Kumar et al. (2003: 673) that customers with relationship intentions will spread positive wordof-mouth, the findings of this study indicate that cell phone users will also spread negative word-of-mouth after a billing error by their cell phone network provider.It is important to note, however, that this study did not measure positive word-of-mouth.As suggested by previous authors (Hirschman 1980: 434;McCullough et al. 1997: 322), altruism might be a factor, as respondents with higher relationship intentions might want to caution other customers against experiencing similar service failures.It is recommended that cell phone network providers assure customers with higher relationship intentions that a similar service failure will not occur again in an effort to keep negative word-of-mouth to a minimum.
Respondents are also prone to voice their dissatisfaction to their cell phone network provider, which will again increase as relationship intention increases.Findings support the supposition by Kumar et al. (2003: 670) that customers with high relationship intentions care about the service provider and provide feedback to improve the service (Wirtz et al. 2010: 380) of their service provider, as respondents with high relationship intentions are more prone to voice their dissatisfaction to their cell phone network provider than respondents with low relationship intentions.In line with the arguments of Kumar et al. (2003: 670), Lacey (2012: 141), Tax et al. (1998: 72) and Tsarenko and Tojib (2011: 386), it is concluded that customers who voice their dissatisfaction have the intention to continue their relationship with their service provider.It is recommended that cell phone network providers use customer voicing to identify customers with higher relationship intentions for relationshipbuilding purposes.
The findings furthermore suggest that respondents who participated in this study are unlikely to complain to an external agency and the least likely to switch to another cell phone network provider after a billing error.No relationships were found between respondents' relationship intentions and complaining to an external agency or switching to another cell phone network provider in the event of a service failure.

Theoretical and practical implications
1Th e findings of this study make theoretical contributions to broadening the field of attitude towards complaining and customer complaint behaviour research.The findings of this study also have practical implications that can be used to improve marketing strategies relating to customer complaint behaviour and relationship marketing.

Theoretical implications
1A lthough it is acknowledged that customers wanting to continue their relationships with service providers are the customers most likely to voice their dissatisfaction to service providers (Lacey 2012: 141;Proença & Rodrigues 2011: 205;Tax et al. 1998: 72), and that customers with a propensity to complain are most likely to voice their dissatisfaction to service providers (Bearden & Mason 1984: 495;Richins 1982: 505;Yuksel et al. 2006: 15, 22), this study specifically identifies relationship intention as an influence on customers' propensity to complain and complaint behaviour (no action, negative word-of-mouth and voicing).Relationship intention should therefore be considered as an important influence on cell phone users' propensity to complain, as well as their response of no action, negative word-of-mouth and voicing customer complaint behaviour.Furthermore, relationship intention should be considered as a variable that could influence customers' attitudes towards complaining and complaint behaviour, which would warrant investigating other influences as well, such as the influence that relationship intention has on satisfaction, loyalty and retention following service recovery.

Practical implications
1T he propensity to complain and voicing of respondents with high relationship intentions enable cell phone network providers to restore these customers' satisfaction, improve service and prevent a similar re-occurrence of the service failure, resulting in a sustainable competitive advantage within the cell phone industry.As recommended by Hedrick et al. (2007: 70), cell phone network providers can utilise complaint behaviour to identify customers with high relationship intentions, as those customers are open to relationship-building strategies.
Cell phone network providers, in particular, might thus not only have higher yields from the increased profitability of the lifetime value of customers with high relationship intentions (Kumar et al. 2003: 673), but also from the constant efforts of customers with higher relationship intentions to maintain relationships with their current cell phone network provider.Cell phone network providers should focus on customers with higher relationship intentions for relationship-building purposes, as these customers will lower the cost of maintaining relationships and will provide the highest return on cell phone network providers' relationship-marketing investments.
Limitations and directions for future research 1A lthough convenience sampling can be used for the theory-testing objectives used in this study (Calder, Phillips & Tybout 1981: 197 199 204;Mittal 1995: 666), the methodology applied in this study, where only one service failure scenario was used, inhibits the generalisability of the findings, as another scenario or real service failure could result in different responses from respondents.Replicating this study across multiple industries, and more South African provinces, using probability sampling with multiple service failure scenarios, or real-life service failures, would address these identified limitations.The relationship between attitude towards complaining and customer complaint behaviour was furthermore not taken into consideration.Moreover, only negative word-of-mouth and not positive word-of-mouth was measured, which limited the conclusions and recommendations that can be drawn from this study.Furthermore, as both no action and negative word-of-mouth were measured with only one item respectively, as done by Singh (1988: 105) and Singh (1990: 7) who developed the taxonomy of customer complaint behaviour used to operationalise customer complaint behaviour in this study, future research could measure no action and negative word-of-mouth with multiple items.
As reasons for customers' complaints are related to the severity of service failure (Chelminski & Coulter 2011: 361, 366;Tax & Brown 1998: 79), future research could add the perception of the severity of service failure to the constructs used in this study to provide a more comprehensive picture.The influence of reasons why customers choose not to complain, such as customers thinking that it might not help to complain (Komunda & Osarenkhoe 2012: 95;Lacey 2012: 141), should be examined along with the constructs of this study.Furthermore, there is an affective component to cell phone users' behaviour after service failures (Hedrick et al. 2007: 69) which, along with coping differences (Duhachek 2005: 52), could be explored in future research.
Annexure A: Service failure scenario 1P lease read the following scenario and imagine you are experiencing the situation: After signing a contract with your cell phone network provider for 150 free minutes to any cell phone number during office hours, you receive your bill and see that you have in fact been charged for all the calls you made during office hours and not just for the calls exceeding the 150-minute frame.

Table 1 :
Taxonomy of customer complaint behaviour

mcmlxiiiC ustomer complaint behaviour mcmlxivT axonomy mcmlxv Dimension(s) within each taxonomy
mcmlxviN o action mcmlxviiT he customer forgets about the service failure and does nothing mcmlxviiiP rivate action mcmlxixN egative word-of-mouth mcmlxxS witching to another service provider mcmlxxi Public action mcmlxxiiV oicing (feedback on dissatisfactory service to the service provider)

Table 5 :
Relationship intention groups mmxcviii Relationship

Table 6 :
Attitude towards complaining categories

Table 7 :
Relationship intention and attitude towards complaining

Table 8 :
Mean scores for the underlying dimensions of customer complaint behaviour mmclU nderlying

Table 9 :
Relationship intention and customer complaint behaviour mmclxviiiC orrelation

Table 10 :
Levels of relationship intention and complaint behaviour