Relationship marketing and the not-for-profit sector

Relationship marketing has been one of the major paradigms in the marketing literature over the last ten years. In this time it has also had a major impact on marketing activities in the for-profit sector, being credited with increased customer cooperation, increased purchases and decreased customer defection (Morgan and Hunt 1994; Gummesson 1999). Traditional marketing has emphasized the importance of acquiring new customers (Gummesson 1997). Relationship marketing, however, has put a more overt emphasis on the importance of developing long-term supportive relationships with existing customers and posits that energy and resources are better spent on this group (Gronroos 1997) than on attempting to attract new customers. Several theorists have suggested that relationship-marketing activities would be particularly suited to the non-profit sector (Sargeant 2001a; Burnett 1998).

Add to shortlist Cite

Relationship marketing and the not-for-profit sector

Introduction

This seems particularly relevant as non-profit organizations (NPOs) operate in a highly competitive environment (Sargeant 2001a). Sargeant argues that competition from the increasing number of NPOs, combined with a decreasing funder pool (Pharoah and Tanner 1998) has led many NPOs to rely increasingly on a small number of key funders (NCVO 1999). To survive and thrive in this competitive environment, theorists such as Burnett (1998), Sagawa (2001) and Sabo (2002) propose that NPOs should concentrate on maintaining and developing relationships with existing funders.

Applying relationship marketing to funder relationships

While relationship-marketing concepts have been applied to many different stakeholders such as employees, suppliers and communities, most studies focus on relationship marketing in the context of the customer. This is because customers are the group that drive the cashflow and business success most directly (Zadrozny 2006; Gronroos 1997). In the not-for-profit context, funders are often seen as the group that drive cashflow most directly (Burligame 2001). It is for this reason that funder–organization relationships will be the key focus of this chapter. But what are the similarities and differences between funder and customer relationships and how can relationship-marketing concepts be applied to the not-for-profit sector?

Rothschild (1979) argues that funder relationships with NPOs are fundamentally different from orthodox customer–organization relationships. He believes that individuals in customer relationships are usually the direct consumers of the organization’s services. Funders, however, are not the direct consumers of an NPO’s services, nor do they normally have direct experience or involvement in service delivery. Consequently, this is why theorists have argued that funder trust in NPOs is critical (Sargeant and Lee 2001). Even when there is a direct customer transactional relationship with an NPO, Garbarino and Johnson (1999) found that trust is more important in generating long-term loyalty than the benefits received in the exchange itself.

Trust is central to relationships with organizations (Kramer and Tyler 1996). The development of trust is a key activity in relationship marketing generally (Gummesson 1997) and in the NPO sector (Thomas et al. 2002). Trust has been associated with many positive organizational outcomes, not specifically related to NPOs. These include, for example: buffering against poor economic conditions (e.g. Taylor 1996); long-term competitive advantage (e.g. Silinpaa and Wheeler 1998); long-term financial success and shareholder value (e.g. Collins and Porras 1998); innovative solutions to organizational challenges (e.g. Kay 1993) and organizational advantage from intellectual capital (e.g. Nahapiet and Goshal 1998).

The organizational outcomes are the consequence of the various supportive behaviours. These behaviours, in turn, are anteceded by trust. These include, for example, long-term commitment and loyalty (e.g. Bagozzi 2000); lower levels of coercion and opportunism by stakeholders towards their organizations (e.g. Cummings and Bromiley 1996); cooperation and positive functional conflict between stakeholders and their organizations (e.g. Morgan and Hunt 1994); flexibility (e.g. Young-Ybarra and Wiersma 1999) and creativity (e.g. Nahapiet and Goshal 1998). Commitment is perhaps the most important of these supportive behaviours as it drives many of the other behavioural outcomes (Morgan and Hunt 1994). Funder commitment is important to NPOs because this group often provides the NPO with its main source of income. But what type of relationship will generate this commitment? While trust is still likely to be a key driver of commitment, the key question is how NPOs cause funders to trust and be committed to them when there is no direct consumer transaction.

The drivers of trust and commitment

Figure 2.1 presents the Morgan and Hunt (1994) model of relationship marketing. According to Morgan and Hunt (1994), trust and commitment are at the heart of any successful relationship with customers. They believe that commitment depends on four variables: relationship benefits, relationship termination costs, shared values and trust. Trust itself is dependent upon three variables: shared values, communication and opportunistic behaviour.

The view that commitment is dependent on the above antecedents has a long history, based on a number of theories. The prediction that relationship benefits and termination costs drive commitment has its origins in exchange theory (e.g. Blau 1964; Chadwick-Jones 1976). The shared values–commitment link, on the other hand, can be seen to be derived from the social–psychological theory of attraction based on similarity (e.g. Berscheid 1985). The link between trust and commitment is based on theories of long-term exchange (Perlman and Duck 1987).

Figure 2.1 The Morgan and Hunt model of relationship marketing

The view that trust is developed from the history of a relationship, through shared values, communication and opportunistic behaviour, can also be supported from other sources deriving from the interpersonal relationship literature (for reviews of these theories, see Duck 1997; Duck and Ickes 2000; Hinde 1997; Seligman 1997). In essence, all these antecedents of trust have the same theoretical foundations and are based on socio-cognitive learning theories (e.g. Heider, 1958, 1980), which assume that individuals are rational and learn from the past. Individuals interact, experience and observe the actions of a relationship partner (e.g. how they have communicated, whether they have kept commitments or been honest, etc.) and use these perceptions to develop a view of how the partner will act in the future (i.e. their trust in the organization).

Material and non-material benefits

Morgan and Hunt (1994) see the benefits from customer relationships in terms of product profitability, customer satisfaction and product performance. Clearly, these dimensions are less directly applicable in NPO–funder relationships. A more in-depth look at exchange theory provides insights into the selection of more appropriate benefits in this context. Rempel et al. (1985), for example, categorized the benefits in relationships into three types: ‘extrinsic’, ‘instrumental’ and ‘intrinsic’. Of these, they found that extrinsic and intrinsic were the most important and were thus chosen as sub-constructs of benefits in this study. Extrinsic benefits are the material benefits that are exchanged in a relationship. Intrinsic benefits, on the other hand, are the benefits inherent in the relationship. They are more intangible than extrinsic benefits and are usually tacit.

A series of exploratory qualitative interviews was carried out with organizational funders to investigate the nature of the benefits they received in their relationship with the NPO. The general nature of intrinsic and extrinsic benefits was explained to funders, who were asked to identify examples under each category to verify the dichotomy and to develop appropriate measures (Strauss and Corbin 1990). In the process, the terms were renamed ‘material’ and ‘non-material’ benefits. These material benefits included, for example, funders learning from the NPO and receiving positive publicity from their relationship with it (Polonsky and Macdonald 2000). Non-material benefits, on the other hand, included the belief that the NPO was making efficient use of its donated funds and that it was having a positive impact on the people for whom these funds were intended. This is similar to the concept of operational competence, identified by Mayer et al. (1995) and Sirdeshmukh et al. (2002).

Given that Rothschild (1979) emphasized the non-transactional nature of funder relationships with NPOs, supported by other theorists who have argued that funder trust in NPOs is critical in these relationships, it is likely that non-material benefits will be closely associated with trust. In fact, we advance this further, suggesting that trust is the key antecedent of non-material benefits. In the absence of direct consumption, we suggest that trust will be one of the major factors used to assess the non-material benefits provided by NPOs.

Communication

Communication, as operationalized by Morgan and Hunt (1994), has three sub-constructs. These are the frequency, relevance and timeliness of communication from the organization to the customer. However, a review of the literature on NPO fundraising suggested that this was a rather limited view of communication. Burnett (1992), for example, has recognized the importance of dealing with funders individually, recognizing their motivation for giving and their expectations of the NPO. Sargeant (2001b) found that when funders were asked about the communication process, their loyalty increased. Thus, we can infer from these sources that communication must be a two-way process involving listening to as well as informing funders. Listening would include seeking funders’ opinions about the NPO and identifying their needs and motivations in the relationship.

In the interviews, funders often made judgements about the NPO on the basis of contacts they had experienced with the NPO’s staff. Staff interactions can therefore be considered as an important element in NPO communication with funders. This statement is consistent with the points made by Brennan and Brady (1999) who suggest that NPO marketing relationships will be strengthened by the behaviour of employees in a variety of roles. Reichheld (1993) found that employee behaviour had a significant impact on customer loyalty. Building on this, Flood et al. (2000) found that good communication by employees had a positive impact on the financial performance of organizations.

Garbarino and Johnson (1999), in reviewing theories of relationship marketing, include both transactional and relational exchanges. Transactional exchanges refer to encounters in which customers relate to any number of interchangeable employees for strictly transactional ends. In relational exchanges, on the other hand, customers interact with key individuals for both transactional and relational reasons (Gronroos 1997). Peltier et al. (2002) endorse the importance of this distinction in the NPO sector. Moreover, Sirdeshmukh et al. (2002) identified front-line employee behaviour as a key driver of trust in relationship marketing. In designing the questionnaire, therefore, a number of questions were included that addressed funder perceptions of the NPO staff.

The following employee behaviours were mentioned in the interviews by funders as important in shaping their positive perceptions of the NPO: a knowledgeable and professional approach, friendly service and employees who appeared passionate about their work. This may serve as a proxy for the impact of the NPO on its clients. We may therefore hypothesize that communication is a key antecedent of non-material benefits, given the close association of both concepts with trust. In the absence of direct consumption, we suggest that funders will have to rely on both communication and trust when judging the non-material benefits provided by an NPO.

The outcomes of trust and commitment

The Morgan and Hunt model (see Figure 2.1) also included the outcomes of commitment of trust. They posit five outcomes. In the NPO context, these outcomes would be:

acquiescence: the degree to which a funder accepts or adheres to the NPO’s requests;
propensity to leave: the likelihood that the funder will leave the NPO relationship in the (reasonably) near future;
cooperation: the funder sharing information with the NPO and working with it to achieve mutual goals;
functional conflict: the extent to which a funder seeks to resolve disputes amicably and in a way that can enhance understanding in the relationship, or negatively, i.e. dysfunctionally;
decision-making uncertainty: the extent to which the funder perceives it has enough information to make key decisions, can predict the consequences of these decisions and has confidence in those decisions.

It is likely that some of these variables will be strongly linked to trust, such as the extent to which a funder is prepared to cooperate and treat conflict in a constructive way. MacMillan et al. (2004) provide other outcomes, that are suggested as generic themes that can be customized and adapted to different organizations and stakeholder groups. These are now contextualized to the funder–NPO context as a way of illustrating other outcomes in funder–NPO relationships:

retention: the extent to which a funder intends to stay with an organization;
extension: the extent to which funders are likely to extend their relationships with an NPO. For example, funding different and new projects outside those they already fund, or supporting the NPO in a different role (e.g. as a volunteer);
advocacy: the extent to which a funder is likely to recommend and defend an NPO; Ŷ cooperation: the extent to which a funder will both actively and reactively cooperate with the NPO, by sharing information, solving conflicts in a functional way and seeking mutual benefit in the relationship;
subversion: the extent to which a funder will actively seek to harm the NPO, by, for example, using coercive power or seeking the support of others to harm the NPO.

Following more recent works such as MacMillan et al. (2004) and Money and Hillenbrand (2006) it seems reasonable that outcome variables can be tailored to specific circumstances. For example, advocacy – the extent to which a funder will recommend and defend the NPO to others – may be of particular importance in certain circumstances, and so on. Measurement of negative outcomes, such as subversion – the extent to which funders may seek to harm the NPO – may also be relevant. It is thus suggested that the exact nature of outcomes should be defined in a process which draws upon perceived organizational need and the current circumstances in a relationship. As the work of Money and Hillenbrand (2006) suggests, it is envisaged that the outcome variables should be customized on each occasion.

The model

The Morgan and Hunt model was amended to include both material and non-material benefits as shown in Figure 2.2, with trust being an antecedent of non-material benefits.

Case study: applying the model to organizational funders of an NPO in South Africa

Many NPOs rely on donations from both individuals and organizations (such as corporations, trusts and governments). While both sources of funding are important, the average value of each organizational funder is often suggested to be higher than that of each individual (see Burlingame 2001; Sargeant and Stephenson 1997). Therefore, NPOs would benefit from focusing on strengthening their relationships with organizational funders. It is for this reason that organizational funders were chosen as the focus of the case study.

A crime-prevention NPO based in South Africa was chosen as the setting for the research. The majority of the NPO’s funders are organizations, including many leading trusts and companies in South Africa. Of these organizational funders, approximately 50 per cent also funded other NPOs while 50 per cent only funded the NPO in question. The entire sample was drawn from one sector in one country. The benefit of this was that the whole sample was subject to the same business environment and public policy context, thus minimizing extraneous sources of variation. Furthermore, the study set out to test the applicability of relationship-marketing approaches to the NPO sector while not seeking to develop a general model of NPO funding. A study based on one NPO and its funders was therefore appropriate. For reasons of simplicity and focus, the case study

Figure 2.2 A model of funder–NPO relationships

excludes consideration of the outcome variables, which are left as suggestions for further research later in the chapter. At this stage, we concentrate only on the drivers of trust and commitment.

Exploratory fieldwork was done in two ways. First, a focus group was held with the board of the NPO to identify their perceptions of the relationship dimensions with organizational funders. The Morgan and Hunt factors were used as prompts in this discussion. Second, a series of in-depth, onsite interviews was conducted with fifteen senior managers of the funding companies. In addition to exploring the specific benefits of the relationships as noted above, the study also aimed to gain a holistic understanding of the relationship and assess the validity of the main concepts used in the Morgan and Hunt model. Respondents were encouraged to suggest any changes that were necessary to fit the NPO context. From these discussions, it became clear that some of the Morgan and Hunt concepts were appropriate while others required modification. A draft questionnaire was designed using the measures explained below and sent to a different sample of senior managers in funding organizations. After receiving their feedback and making the relevant changes, a final draft of the questionnaire was developed.

Data collection

The questionnaire was administered by post to senior managers (responsible for funding decisions) of all 120 funding organizations, i.e. the total population of organizational funders for this particular NPO. A total of 41 responded, representing a response rate of just over 33 per cent. This response rate is in line with typical response rates where respondents are top managers or organizational representatives (Baruch 1999).

Of the 41 respondents, 21 (51 per cent) were commercial organizations, 6 (15 per cent) were government departments, 8 (20 per cent) were either trusts, foundations or development organizations and 6 (15 per cent) organizations did not indicate what sector they were from. Participants were asked to indicate the extent to which they agreed or disagreed with a set of statements concerned with their commitment to the NPO and this was done with reference to a 7-point Likert-type scale. Point 1 on this scale indicated strong agreement with a statement, point 7 strong disagreement and point 4 neither agreement nor disagreement.

Measures

The constructs in the Morgan and Hunt model (see Figure 2.1), were used as the basis for finding appropriate measures for our initial model shown in Figure 2.2. The multi-item scales used by Morgan and Hunt to measure their constructs were adapted by adding or removing items/questions so as to make the domains of the constructs relevant to this study. New questions were developed through expert focus groups and by drawing upon the literature. The measures that are shown in Figure 2.2 are described in more detail in Table 2.1.

Preliminary testing of the measures

Some preliminary analysis was carried out to test the reliability and validity of the multi-item scales. The methods used were those recommended by Churchill (1979).

First, each scale was subjected to a Cronbach Alpha reliability test. The Cronbach Alphas ranged from 0.74 to 0.89, and thus were all in excess of the generally accepted required level of 0.70 (Foreman et al. 1998).

Second, a principal component factor analysis was performed on each of the multi-item scales. The scales factored as expected with loadings for items being in excess of 0.5. This demonstrated convergent validity.

Finally, composite equally weighted scales were created to obtain measures of the constructs in the model. The correlations between these scales were calculated and are shown in Table 2.2.

NB: Some of the N figures are less than the sample size of 41, revealing that there are a small number of missing values. While all the measures exceeded the minimum of 30 observations needed for partial least squares (PLS) analysis (Wixom and Watson 2001), the missing values were replaced with the mean, as suggested by Chin and Newstead (1999) when using PLS to perform the structural modelling exercise.

All but one of the correlations is significant at the 1 per cent level. The exception is the correlation between termination costs and commitment. This link is therefore removed from the initial model, leaving shared values, non-material and material benefits as the only antecedents of commitment. It should be noted that both the qualitative and quantitative research suggested the level of termination costs to be very low in this relationship. This is likely to have contributed to the lack of a significant link between termination costs and commitment.

Table 2.1 Measures used in the case study

Commitment	Communication	Benefits	Shared values	Trust	Non-opportunistic behaviour	Termination costs
The conceptualization of commitment is that of Morgan and Hunt. It is described in terms of statements relating to the participant’s intention to stay in the relationship with the NPO, and willingness to put in an effort to maintain the relationship with the NPO.	The conceptualization of communication is based on a subset of the informing measures provided by Morgan and Hunt. To these were added the listening aspects suggested by Burnett (1992) and the NPO staff behaviour aspects suggested in the inter view stage of the research. It is thus described in terms of statements relating to providinginformation to (i.e. informing) funders (characterized by the frequency, relevance and timeliness of communication), listening to funders (i.e. seeking information about funders’ needs and motivations), staff interactions (i.e. with staff who are responsive, knowledgeable and passionate about the NPO).	As no appropriate instrument for the measurement of these benefits is available in the literature, an instrument was specifically developed for this study. This was derived from Rempel et al.(1985). A distinction is made between material - the tangible benefits explicit in the relationship (e.g. receiving positive publicity from their relationship with the NPO, learning skills from the NPO) andnon-material benefits: the benefits inherent to the relationship (e.g. believing that funds were making a positive impact on thetargeted clients and that funds were used efficiently). These sub-scales are intended to function as independent measures.	Following Morganand Hunt (1994), shared values are described in terms of statements relating to the participants’ perceptions that an NPO has similar values to themselves. An instrument developed by Swasy(1979) embraces the conceptualization as described above in a context-free set of statements. It was, therefore, adapted for use in this study. The adaptation involved changing the wording of individual statements in the Swasy(1979) instrument to capture individuals’ perceptions of the extent to which they share the same values as the NPO.	The Morgan and Hunt conceptualization of trust, having been developed for a commercial customer-company relationship, was seen to be tootransactional and thus not adopted. Instead, the measure of trust was based on aconceptualization of Rempel et al. (1985), in terms of the following sub-constructs: reliability, the extent towhich a funder expects the NPO to be consistent and predictable in keeping its commitments; dependability, the extent to which funders expect the NPO will tell them the truth, act in their interests, and not take advantage of them in the future, and faith, the extent towhich funders expect the NPO to act in their interests in an unpredictable and uncertain world.	Morgan and Hunt (1994) use a construct labelled opportunistic behaviour. We use a construct labelled non-opportunistic behaviour. This is done to simplify the model for the reader, because with this labelling all the links in the model can be hypothesized to be positive. The items used tomeasure non-opportunistic behaviour are therefore represented by statements relating to the participants’ perceptions that an NPO has not taken advantage of them in the past. An instrument developed by Cummings and Bromiley(1996) embraces such a conceptualization and has, therefore, been adapted andadopted for use in this study. This, however, involved changing the wording of the individual Cummings and Bromiley statements to capture funders’ perceptions of the extent to which they perceived they have, or have not been, treated opportunistically.	The conceptualization of termination costs is an extension of that used by Morgan and Hunt to incorporate their suggestions for further research and is described in terms of statements relating to thenancial costs offiending a relationship (i.e. switching andopportunity costs, dissolution expenses) and the non-nancialficosts of ending a relationship (e.g. loss of reputation, aggravation). Questions were added to obtain a measure incorporating both dimensions.

Table 2.2 Correlation matrix of the construct measures

		Termination costs	Material benefits	Non-material benefits	Shared values	Communication	Non-opportunistic behaviour	Trust	Commitment
Termination costs	Pearson	1.00	.211	.021	.085	.353*	–.059	–.003	.159
	Sig.(1-tailed)	.	.098	.450	.305	.016	.367	.494	.171
	N	39	39	39	38	37	35	36	38
Material benefits	Pearson	.211	1.00	.695**	.419**	.528**	.374*	.411**	.440**
	Sig.(1-tailed)	.098	.	.000.0	.004	.000.0	.011	.005	.002
	N	39	41	41	40	39	37	38	40
Non-material benefits	Pearson	.021	.695**	1.00	.507**	.539**	.470**	.598**	.549**
	Sig. (1-tailed)	.450	.000	.	.000	.000	.002	.000	.000
	N	39	41	41	40	39	37	38	40
Shared values	Pearson	.085	.419**	.507**	1.00	.361*	.260	.630**	.465**
	Sig. (1-tailed)	.305	.004	.000	.	.012	.060	.000	.001
	N	38	40	40	40	39	37	38	39
Communication	Pearson	.353*	.528**	.539**	.361*	1.00	.363**	.582**	.395**
	Sig. (1-tailed)	.016	.000	.000	.012	.	.014	.000	.007
	N	37	39	39	39	39	37	38	38
Non-opportunistic behaviour	Pearson	–.059	.374*	.470**	.260	.363**	1.00	.648**	.378*
	Sig. (1-tailed)	.367	.011	.002	.0	.014	.	.000	.012
	N	35	37	37	37	37	38	36	36
Trust	Pearson	–.003	.411**	.598**	.630*	.582**	.648**	1.00	.436**
	Sig. (1-tailed)	.494	.005	.000	.000	.000	.000	.	.003
	N	36	38	38	38	38	36	38	37
Commitment	Pearson	.159	.440**	.549**	.465**	.395**	.378*	.436**	1.00
	Sig. (1-tailed)	.171	.002	.000	.001	.007	.012	.003	.
	N	38	40	40	39	38	36	37	40

Notes:

* Correlation is significant at the 0.05 level

** Correlation is significant at the 0.01 level

The amended NPO–funder research model is shown in Figure 2.3 below. This model is now formally tested through the application of structural equation modelling.

Data analysis

The data were analysed in two separate, but sequentially related, stages of analysis by means of structural equation modelling, using PLS. These stages were:

Stage 1: Testing the measurement model by performing a validity and reliability analysis on each of the measures of the research model.

Stage 2: Testing the structural model by estimating the paths (links) between the variables in the model, determining their significance as well as the predictive ability of the model.

Structural equation modelling: partial least squares

Partial least squares (PLS), a structural equation modelling technique, was used in this study to perform the analysis for the NPO–funder model depicted in Figure 2.3. PLS is the preferred approach when one or more of the following is present: the model includes formative constructs; the sample size is relatively small; assumptions of normality are not satisfied (Chin and Newstead 1999). The software used was PLS-Graph Version 3.0.

A PLS application requires a minimum sample size which is ten times the greater of: (a) the number of items comprising the most formative construct; or (b) the number of independent constructs directly influencing a dependent construct (see Wixom and Watson 2001:28). In this study, none of the constructs are formative – all are reflective. A discussion on reflective and formative indicators can be found in Chin (1998: ix).

Figure 2.3 The research model of NPO–funder relationships

Trust is the dependent construct, with the three most independent antecedents influencing it being shared values, communication and opportunistic behaviour. This application of PLS therefore requires a minimum sample size of 30 which is exceeded by our sample size of 41. The outputs from the PLS software are used first to test the measurement model and then to test the fit and performance of the structural model. The results for the two stages of analysis now follow.

Stage 1: The measurement model: reliability and validity of the measures

The measures used for the constructs are assessed in terms of:

Reliability by means of the Fornell and Larker internal consistency index (Fornell and Larker, 1981). These are not part of the PLS output and have to be computed from the loadings and average residuals reported in the PLS output. Consult Hair et al. (1998:612) for the formula. A measure is considered reliable if the reliability index is at least 0.70 (Nunally 1978). The reliability of the measures is reported in Table 2.3.

Table 2.3 Reliability and convergent validity (AVE) report

	Mean	Standard deviation	Loading
Shared values (Fornell and Larcker reliability = 0.84; AVE = 0.64)
1 In general, their opinions and values are a lot like ours.	4.83	1.43	0.73
2 We like and respect their values.	5.71	0.89	0.82
3 We share a very similar set of values (e.g. in terms of their beliefs about the way staff be treated …).	5.29	1.10	0.84
Trust (Fornell and Larcker reliability = 0.87; AVE = 0.53)
1 The NPO are very unpredictable. I never know how they are going to act from one day to the next.	5.63	1.13	0.55
2 I can never be sure what the NPO are going to surprise us with next.	5.74	0.99	0.80
3 I am confident that the NPO will be thoroughly dependable, especially when it comes to things that are important to my organization.	5.55	1.03	0.89
4 In my opinion, the NPO will be reliable in the future.	5.69	0.92	0.75
5 Though times may change and the future is uncertain, I know that the NPO will always be willing to offer my organization the support it may need (e.g. even if we had not funded them recently).	5.26	0.85	0.62
6 The NPO would not let us down, even if they found themselves in an unforeseen situation (e.g. competition from other funders, changes in government policy).	5.10	0.94	0.69
Non-material benefits (Fornell and Larcker reliability = 0.86; AVE = 0.54)
1 The chance to support programmes that will have a good long-term impact on society.	6.05	0.80	0.80
2 Innovative and cutting-edge solutions to the problems it faces (e.g. crime-related issues and opportunities in South Africa).	5.51	1.12	0.81
3 The knowledge that our support is used in an ethical way (e.g. good accounting systems, staff who do not waste money).	5.59	1.26	0.74
4 The opportunity to target our support to issues that we care about.	5.56	1.03	0.58
5 Being able to see that our support has an impact on the people for whom it is really intended.	5.41	1.26	0.73

	Mean	Standard deviation	Loading

Commitment (Fornell and Lacrker reliability = 0.94; AVE = 0.85)
1 The relationship my organization has with … is something we intend to maintain in the long term (e.g. over the next two years, possibly beyond).	5.59	1.43	0.90
2 The relationship my organization has with … is something that we will put a lot of effort into maintaining in the future.	5.30	1.40	0.96
3 The relationship my organization has with … is something we are very committed to.	5.62	1.08	0.89
4 The relationship my organization has with … is very important to us.	5.34	1.46	0.93
Non-opportunistic behaviour (Fornell and Larcker reliability = 0.88; AVE = 0.71)
1 In the past the NPO have succeeded partly because they have taken advantage of my organization.	5.51	1.43	0.80
2 I feel that the NPO generally try to get the upper hand when they deal with us.	5.67	1.11	0.87
3 The NPO have taken advantage of my organization when it was in weak position.	6.23	0.83	0.86
Communication (Fornell and Larcker reliability = 0.88; AVE = 0.56)
1 Staff who talk with passion and experience about the work of…	5.63	0.89	0.78
2 The opportunity to work with knowledgeable, professional and approachable staff.	5.0	1.22	0.79
3 The NPO keep me informed about new developments that are relevant to us.	5.08	1.47	0.61
4 The NPO provide frequent communication about issues that are important to us.	4.83	1.52	0.74
5 Even when things don’t go quite according to plan, the NPO do their best to listen to us (e.g. my own and my organization’s ideas, concerns and suggestions).	5.28	1.24	0.72
6 Whatever the circumstances, the NPO usually take notice of the suggestions I make about my organization’s work with them.	4.97	0.95	0.80
Material benefits (Fornell and Larcker reliability = 0.87; AVE = 0.70)
1 Opportunities to be involved with projects that tie in with our company’s vision (funding something that in some way matches our strengths, builds our reputation).	5.57	0.99	0.89
2 An efficient use of our support (whatever form this takes).	5.66	1.39	0.79
3 Being able to support programmes that will ultimately also benefit my organization and its staff (e.g. creating a society that is good for our business and our people and their families).	5.39	1.18	0.83

Convergent validity by assessing the average variance extracted (AVE). This is standard output from PLS. Measures with AVE 0.50 or higher are said to exhibit convergent validity (Chin 1998). The AVEs reported in Table 2.3 all exceed 0.50, confirming that all measures demonstrate satisfactory convergent validity.

Discriminant validity is established from the latent variable correlation matrix. This matrix has the square root of AVE for the measures on the diagonal and the correlations among the measures as the off-diagonal elements. The matrix has to be constructed from the PLS output. Discriminant validity is determined by looking down the columns and across the rows. Should the diagonal elements be larger than offdiagonal elements, discriminant validity is deemed satisfactory. Discriminant validity is demonstrated as these conditions are satisfied (see Table 2.4). Table 2.4 The latent variable correlation matrix: discriminant

Table 2.4 The latent variable correlation matrix: discriminant validity

	Shared values	Trust	Non-material benefits	Commitment	Non-opportunistic behaviour	Communication	Material benefits
Shared values	0.797
Trust	0.623	0.726
Non-material benefits	0.519	0.680	0.737
Commitment	0.535	0.469	0.684	0.921
Non-opportunistic behaviour	0.335	0.668	0.474	0.359	0.844
Communication	0.437	0.614	0.603	0.524	0.438	0.741
Material benefits	0.413	0.403	0.652	0.507	0.431	0.605	0.836

The structural model is now tested.

Stage 2: Testing the structural model: path coefficients and predictive ability

The key PLS outputs for this analysis are:

R² values, reported in Figure 2.4, which provide an indication of the predictive ability of the independent variables. Trust, non-material benefits and commitment with R² values of 0.68, 0.52 and 0.52 respectively are considered to provide adequate evidence of the predictive ability of the model.
Path coefficients and their statistical significance, reported in Figure 2.4. PLS, being a distribution-free technique, uses the bootstrapping resampling technique to determine the significance of the paths. Two hundred resamples were taken in performing the bootstrap. The results reveal that all the links in the model, except material benefits to commitment, are significant at the 5 per cent level.

The findings, reported in (1) and (2) above, indicate that a relationship-marketing approach can be applied to a funder relationship with an NPO. This provides much needed empirical support to previous theoretical arguments and endorses the assertions of Burnett (1998) and Brennan and Brady (1998) that relationship marketing can be applied to the NPO sector.

Non-material benefits is the main driver of funder commitment, since it has the highest path coefficient of 0.51. Material benefits, however, is not identified as being a significant link and is eliminated from the model. It should be noted that there is a high correlation between material and non-material benefits (see Table 2.1). Thus, if non-material benefits are removed from the analysis, the impact of material benefits on commitment approaches significance. Receiving a certain level of material benefits is, therefore, important in influencing funders’ commitment to an NPO. Nevertheless, in this case, differences in non-material benefits are the main predictor of continued giving.

It is noted that trust is developed directly by the behaviours of the NPO towards its funders. This is through communication and non-opportunistic behaviour. To develop commitment, it is essential to provide non-material benefits. Trust, in turn, is the key driver of non-material benefits. Clearly, therefore, NPOs must engage in behaviours that will develop the trust of their organizational funders.

Figure 2.4 The fitted model of NPO–funder relationships based on the results