center23002457459410012100center818008745855Lawrence Guy Supervised by Dr


center23002457459410012100center818008745855Lawrence Guy
Supervised by Dr. Norman Chiliya
941009200Lawrence Guy
Supervised by Dr. Norman Chiliya
center300003207385The Influence of Social Media Marketing on a NPO’s Brand Image: A Research Thesis
9410036300The Influence of Social Media Marketing on a NPO’s Brand Image: A Research Thesis

Table of Contents: TOC o “1-3” h z u Chapter 1 : Introduction to the Study PAGEREF _Toc523863695 h 21.1Background/Overview of the Study: PAGEREF _Toc523863696 h 21.2Purpose of the Study: PAGEREF _Toc523863697 h 51.3Research Questions: PAGEREF _Toc523863698 h 61.4Significance and Contribution of the Study: PAGEREF _Toc523863699 h 71.5Conceptual Model and Framework: PAGEREF _Toc523863700 h 81.6Research Design and Methods: PAGEREF _Toc523863701 h 91.7Structure of the dissertation: PAGEREF _Toc523863702 h 141.8Ethical Considerations: PAGEREF _Toc523863703 h 15Chapter 2: Literature Review PAGEREF _Toc523863704 h 152.1Social Media Marketing Theories: PAGEREF _Toc523863705 h 162.2Non-Profit Marketing – Theories of Philanthropy PAGEREF _Toc523863706 h 192.3Research variables PAGEREF _Toc523863707 h 22Chapter 3: Conceptual Model and Hypotheses Development PAGEREF _Toc523863708 h 253.1Introduction PAGEREF _Toc523863709 h 253.2Conceptual Model PAGEREF _Toc523863710 h 263.3Hypothesis Development PAGEREF _Toc523863711 h 263.1) The relationship between Social Media Marketing and Customer Engagement PAGEREF _Toc523863712 h 273.2) The relationship between Social Media Marketing and Willingness to Donate PAGEREF _Toc523863713 h 273.3) The relationship between Social Media Marketing and Brand Awareness PAGEREF _Toc523863714 h 283.4) The relationship between Customer Engagement and Repeat Behaviour PAGEREF _Toc523863715 h 283.5) The relationship between Willingness to Donate and Repeat Behaviour PAGEREF _Toc523863716 h 283.6) The relationship between Repeat Behaviour and Brand Image PAGEREF _Toc523863717 h 293.7) The relationship between Brand Awareness and Brand Image PAGEREF _Toc523863718 h 293.4Conclusion PAGEREF _Toc523863719 h 30Chapter 4: Research Methodology and Design PAGEREF _Toc523863720 h 304.1 Introduction PAGEREF _Toc523863721 h 304.2Research Strategy PAGEREF _Toc523863722 h 304.3Sampling Design PAGEREF _Toc523863723 h 324.4Data Collection Method (Questionnaire Design, Pre-testing the instrument, Ethical considerations) PAGEREF _Toc523863724 h 344.5Data Analysis Procedure (Data Coding and Cleaning, Descriptive Statistics, SEM – measurement model and structural model -) PAGEREF _Toc523863725 h 374.6Confirmatory Factor Analysis (with each index explained and observed) PAGEREF _Toc523863726 h 394.7Reliability (Cronbach Alpha, Composite Reliability, Average Variance Extracted) PAGEREF _Toc523863727 h 424.8Validity (Convergent, Discriminant) PAGEREF _Toc523863728 h 43Chapter 5: Data Analysis and Survey Results PAGEREF _Toc523863729 h 445.1Introduction PAGEREF _Toc523863730 h 445.2Demographic Statistics PAGEREF _Toc523863731 h 445.3Measurement Model Assessment: Measurement Reliability and Validity PAGEREF _Toc523863732 h 475.4Confirmatory Factor Analysis (CFA) PAGEREF _Toc523863733 h 515.5Structural Model PAGEREF _Toc523863734 h 535.6Hypothesis testing PAGEREF _Toc523863735 h 555.7Proposed Hypotheses with Academic and Organisational Implications PAGEREF _Toc523863736 h 57Chapter 6: Discussion, Conclusion, Recommendations and Future Research PAGEREF _Toc523863737 h 596.1Discussion of the Findings PAGEREF _Toc523863738 h 596.2Recommendations and Contributions of this Study PAGEREF _Toc523863739 h 626.3Limitations and Future Research PAGEREF _Toc523863740 h 636.4Conclusion PAGEREF _Toc523863741 h 64References PAGEREF _Toc523863742 h 64
Chapter 1 : Introduction to the StudyBackground/Overview of the Study:1.1Introduction:
Non-profit organisations have, with the emergence and later domination of social media on all markets, needed to adapt their strategy, especially in terms of marketing. With the wealth and quantity of knowledge available to the average consumer, it is essential for non-profit organisations to reach and target the most likely individuals to engage with their cause.
According to Salamon and Anheier, there are three main sectors enveloping all institutions and organisations: the first is the private sector, or the market, which houses all those companies striving to operate at a profit in today’s capitalist economy. The second sector is the public sector, which is generally associated with the State, and encompasses all those public services and enterprises which are available to all and made available by the State. Finally, the third sector refers to the non-profit sector, which occupies a space which is separate from both the market and the state, since its functions do not affiliate with making a profit and are not seen as ‘public’ and available to all (Eikenberry and Kluver, 2004).
This third sector then is irrevocably present and of vital significance, but holds an obscure space positioned in society’s perception, with it neither being market nor state-directed. Organisations operating in that sector are thus disadvantaged from the get-go, considering the fact that consumers don’t really have a clear image of their purpose, operations or even presence. These organisations, also referred to as ‘charities’ by the average consumer, essentially react to lack of action in specific areas by the State or the market and incite change through fundraising, generally speaking.
There is however a blurring of the boundaries between these three sectors (Eikenberry and Kluver, 2004). The non-profit sector has gradually shifted to integrate some of the philosophies and actions of the more market oriented firms. This could be explained by referring to the concept of resource dependency theory, whereby an organisation needs resources to operate and survive (Eikenberry and Kluver, 2004). In the marketing context, this is especially important since it involves a lack of flexibility and funding to reach audiences and engage with consumers (Helmig, Jegers and Lapsley, 2004). The purpose of marketing for non-profit organisations is then to collect resources, but needs to spend some resources to do so.
In South Africa, the Non-profit sector forms a large part of the number of organisations, due to the high amount of diversity and large wealth inequalities present across the country (Stuart, 2013). Many of these organisations have managed to shift to a more consumer and relationship-based approach, with partnerships with the public and private sectors that have shifted their ideologies and processes. However, due in part to the global economic crisis, most non-profit organisations have trouble accessing funding (Stuart, 2013), which associated with the significant decrease in donations may lead to these organisations losing touch with their consumer base.
This is where social media marketing is of vital import. Social networking sites and other platforms are accessible, free, and most importantly interactive, which results in engagement with the consumer (Lovejoy and Saxton, 2012). Social media gives a solution to organisations struggling to come up with ideas on how to interact with their target audience, and because of the increase in use of platforms such as Facebook or Twitter, gives organisations a reliable way to find and attract consumers to their cause. There are still issues with this alternative, however: consumers can easily decide to ignore or block any information that they find uninteresting, which results in those organisations needing to create worthwhile content all the while targeting the population most likely to engage with them optimally.
One organisation which is doing this effectively is the Praekelt Foundation, a South African- based non-profit organisation which uses mobile technology to solve issues in emerging countries, spanning across multiple continents and with a human-centred approach aiming to solve issues sustainably (Praekelt.org, n.d.).
1.2Statement of the research problem:
A significant amount of literature exists on the growing non-profit sector, and how it should fashion its strategy towards its target market. With the ever-important technological advances being made in terms of communication and marketing, organisations have had to adapt and use the advances as opportunities to avoid being left behind and unable to adjust. With the arrival of social media at the forefront of consumer behaviour patterns, organisations have had to shift their focus from websites to those platforms to initiate more contact with consumers and the general public. This has led to organisations needing to change their marketing strategy as a result, creating a more balanced platform for non-profit organisations compared to for-profit companies.

However, there is not an extensive amount of research on the topic of social media marketing for non-profit organisations, and how they can use it to increase their brand image. In a society such as South Africa, where according to a report by Qwerty Digital from 2017, 15 million individuals were qualified as active or daily users of social media (Qwerty Digital, 2017), this would be a vital platform that should be understood and integrated by any organisation to their operations.

Organisations thus need to capitalise on the growing importance of social media in the average consumer’s life, with the aim of engaging with them and getting something back, be it through donations, volunteers, repeat behaviour, or an increase in brand awareness.
There needs to be some research conducted for the South African market as well, since comparatively to other countries it has a very large amount of diversity and a high level of income inequality between the higher and lower classes. This will result in a fragmented consumer base, with varying behaviours between the extremes of society.
Non-profit organisations need to be able to understand how to use social media effectively to engage with consumers and increase their brand image. This study will attempt to analyse the different elements of social media marketing and determine how non-profit organisations can utilise these to increase their brand image. With the focus of the study being the Praekelt Foundation, a non-profit organisation which manages to effectively integrate all of their digital marketing and has presence on a number of social media platforms such as Facebook, Twitter and a blog on Medium, this should provide insight on what other organisations can do to emulate this approach and increase their brand image with the public through social media.
Purpose of the Study:The purpose of this study is to assess the effects of social media marketing on brand image, through assessing factors such as customer engagement and willingness to donate, and how these valuables result in repeat customers; as well as brand awareness in the context of non-profit organisations in South Africa. This will be done through the case study of the Praekelt Foundation as a successful South African-based global non-profit organisation and its use of social media and more specifically Facebook, Twitter and blogging platform Medium.

2.1Research objective:
The primary objective of this study is to investigate the impact of social media marketing on a non-profit organisation’s brand image.

The secondary objectives will be referred to below under Theoretical and Empirical objectives.
2.2Theoretical objectives:
To investigate existing studies on Social Media Marketing.

To investigate existing studies on Customer Engagement.

To investigate existing studies on Repeat Behaviour.

To investigate existing studies on Willingness to Donate.

To investigate existing studies on Brand Awareness.

To investigate existing studies on Brand Image.

2.3Empirical objectives:
To investigate the relationship between Social Media Marketing and Customer Engagement
To investigate the relationship between Social Media Marketing and Willingness to Donate
To investigate the relationship between Social Media Marketing and Brand Awareness
To investigate the relationship between Customer Engagement and Repeat Behaviour
To investigate the relationship between Willingness to Donate and Repeat Behaviour
To investigate the relationship between Repeat Behaviour and Brand Image
To investigate the relationship between Brand Awareness and Brand Image
Research Questions:3.1Key Question:
What is the effect of Social Media Marketing and its elements on an organisation’s Brand Image?
3.2Sub-questions:
Is there a positive relationship between Social Media Marketing and Customer Engagement?
Is there a positive relationship between Social Media Marketing and Willingness to Donate?
Is there a positive relationship between Social Media Marketing and Brand Awareness?
Is there a positive relationship between Customer Engagement and Repeat Behaviour?
Is there a positive relationship between Willingness to Donate and Repeat Behaviour?
Is there a positive relationship between Repeat Behaviour and Brand Image?
Is there a positive relationship between Brand Awareness and Brand Image?
Significance and Contribution of the Study:Theoretically, this study will increase the knowledge concerning the effect of social media marketing on a brand’s image, and will serve to assist non-profit organisations in using social media effectively for that purpose. This will be especially significant in the South African context, the country where the study will take place and the country of origin for the organisation being observed, the Praekelt Foundation. The study will also evaluate the different relationships between the variables obtained through the use of social media marketing, and the subsequent increase in brand image for the organisation in question.

Other significant theoretical results the study will assist in will be in giving more insight on the behaviour of students at the University of the Witwatersrand, which will be where the sampling for this study will take place. It will educate on the perception the students have of both social media and non-profit organisations, with the added benefit of inferring how social media could be used optimally to increase brand image for non-profit organisations.
This study should also be of interest in other markets with similar characteristics to South Africa, to those non-profit companies struggling to connect or engage with their target market. It will be of particular significance to those organisations who similarly to the Praekelt Foundation are trying to offer services or products based primarily through digital means, be it mobile technologies, social media, or the Internet.
Conceptual Model and Framework:Predictor variable: Social Media Marketing
Moderator variables: Customer engagement, Repeat Business, Willingness to Donate, Brand Awareness
Outcome variable: Brand Image
Figure 1 – Conceptual model of the effect of Social Media Marketing on Brand Image-33020306705
202882537465Customer Engagement
Customer Engagement

338645580010H4
00H4
1348105365760H1
0H1
14528802794033864551422403596005142240Repeat Behaviour
Repeat Behaviour

90805317500Social Media
Marketing
0Social Media
Marketing
486283021589933864551587502028825318770Willingness to Donate
Willingness to Donate

486283066040H6
00H6
3386455123190H5
00H5
161925076200H2
0H2
1452880346710145288034670900
486283077470Brand Image
Brand Image

3748405156210H7
00H7
1452880156210H3
0H3
33864541136652024380260985Brand Awareness
Brand Awareness

Hypothesis statement:
H1: The relationship between Social Media Marketing and Customer Engagement is positive.

H2: The relationship between Social Media Marketing and Willingness to Donate is positive.

H3: The relationship between Social Media Marketing and Brand Awareness is positive.

H4: The relationship between Customer Engagement and Repeat Behaviour is positive.

H5: The relationship between Willingness to Donate and Repeat Behaviour is positive.

H6: The relationship between Repeat Behaviour and Brand Image is positive.

H7: The relationship between Brand Awareness and Brand Image is positive.

Research Design and Methods:This section will give information on the type of research undertaken and the methods used to get the results as required. The elements discussed will be Sampling Design, Questionnaire Design, Data Collection Approach and Data Analysis Approach. The research will be quantitative in its method, which will require a larger sample than if the study used qualitative methods. The research is of explanatory nature, and not exploratory or descriptive, and will thus seek to establish causal relationships between the variables highlighted previously through the use of hypotheses which have also been defined above.
7.1Sampling Design:
This element comprises of four important characteristics that need to be established: target population, sample size and sampling method.

Target population describes the collection of individuals from which the sample is selected. The target population used for this study will be Wits students that use social media, since they are representative of the market non-profit organisations are trying to target through social media marketing. University students form part of the age groups that have the most presence on social media in South Africa, and will thus be the most engaging consumers on those platforms. (Business Tech, 2017) It will also be valuable for non-profit marketers to know whether students successful with their education are aware of marketing attempts made in their direction. The age of the students questioned for the purpose of the study will be between of a minimum of 18 years old for ethical considerations.
Sample size involves the size of the sample selected for the research questions, and when using quantitative methods it is needed to use a larger sample size. For this study, a sample size of 300 would be adequate, from a variety of subcultures and backgrounds. This amount of samples should be satisfactory to make allowance for there being potential errors or non-responses in the survey results. This sample size was determined by using a rule of thumb put forward by Bentler and Chou, whereby there must be 5 or 10 observations per parameter (Wolf et al., 2013). Since this study contains a total of 30 parameters, using 10 observations per parameter gives us a sample size of 300 respondents.

Sampling method denotes the procedure used to select the samples, and differentiates between two potential method types: Probability sampling and Non-probability sampling. For the purpose of this study and considering the research will use Quantitative methods the method used will be probability sampling. There are then four different types of probability sampling to choose from: simple probability sampling, systematic probability sampling, stratified probability sampling and cluster probability sampling.
The sampling method chosen for the purpose of this study is cluster sampling, since it is more efficient in terms of data collection. It is in essence a method where the researcher divides the target population into clusters and then a random set of clusters is selected for sample collection. It is a very efficient method of data collection but has drawbacks such as reduced accuracy and representativeness.

7.2Questionnaire Design:
This section refers to the specificities of the questionnaire, and mentions elements such as the Likert scale used, and the different sections of the questionnaire which need to be completed by the respondent. In this case the questionnaire will use a five-point Likert scale, which consists of five options: 1. Strongly Disagree, 2. Disagree, 3. Neutral, 4. Agree and 5. Strongly Agree. There will be seven sections to the questionnaire, sections A to G. Section A will be formed of questions giving general demographic and background information about the respondent, such as Age, Gender and Level of Education. Sections B to G will then ask questions specific to each variable, with section B relating to Social Media Marketing, section C relating to Customer Engagement, section D relating to Willingness to Donate, section E relating to Brand Awareness, section F relating to Repeat Behaviour, and section G relating to Brand Image.
Below are some of the measurement instruments that would be used in the research, with the questions being modified from original sources to better fit the type of the organisation being referred to. In the preliminary questions the student will have to pick one NPO they feel strongly about, and if the respondent can’t think of a South African organisation then they would be allowed to pick an international NPO.
Instruments for Section A: (Vinerean et al., 2013)
Gender (M/F/Other/Prefer not to say)
Age class (18-23 / 24-29 / 30-35 / 36-41)
Level of education (Undergraduate / Honours / Masters / PhD)
For how long have you been using social media websites? (1 – 6 months / 6 months – 1 year / 1 – 2 years / 2 – 3 years / More than 3 years)
How would you describe your log in pattern on social media sites? (Always connected / Several times a day / Every three days / Once a week / Occasionally – Less than once a week)
Instruments for Section B: (Vinerean et al., 2013) (Godey et al., 2016)
The ads that appear on my profile are relevant for my personal interests and I enjoy seeing them.

Quite often I access the ads that I see on my social media profile.
I do experience concern regarding the confidentiality and privacy of my personal information.
Content of X brand’s social media seems interesting.

Using X brand’s social media is fun.

Content of X brand’s social media is the newest information.

Using X brand’s social media is very trendy.

Instruments for section C: (Godey et al., 2016)
X brand’s social media enable information-sharing with others.

Conversation or opinion exchange with others is possible through X brand’s social media.

It is easy to provide my opinion through X brand’s social media.

I would like to pass information on brand, product, or services from X brand’s social media to my friends.

Instruments for section D: (Van der Heijden and Verhagen, 2004)
I am positive towards donating on X’s website.

The thought of donating on the website of X is appealing to me.

I think it is a good idea to donate on the website of X.

Instruments for section E: (Godey et al., 2016)
I am always aware of X brand.

Characteristics of X brand come to my mind quickly.

I can quickly recall the symbol or logo of X brand.

Instruments for section F: (Godey et al., 2016)
Although another brand has the same features as X, I would prefer to donate to X.

If another brand does not differ from X, it seems smarter to donate to X.

Although there is another brand as good as X, I prefer to donate X.

I will suggest X brand to other consumers.

I would love to recommend X brand to my friends.

I regularly visit X brand’s social media page.

I intend to visit X brand’s social media page again.

I am satisfied with X brand with every visit to their social media page.

X brand would be my first choice.

Instruments for section G: (Godey et al., 2016)
X brand is a leading NPO.

X brand has extensive experience in the industry.

X brand is a leading representative of the NPO industry.

X brand is a customer-oriented company.

7.3Data Collection Approach:
This section describes and designates how the data collection will be undertaken and through which medium the respondent will have to complete the questionnaire. This could be either through personal direct distribution, through post, through telephone communications, or online through the use of the Internet.
For this study the methods used will mainly be personal direct distribution with a little bit of online distribution, since these methods are the most effective ways to collect the data. The benefit of direct distribution is that respondents are more likely to agree to do the questionnaire, but the drawback is that it takes more time for the marketer to collect the data. On the other hand, online surveys can be more practical for respondents since they are more flexible in terms of when they decide to complete the questionnaire. The respondent doesn’t however have the pressure of the physical presence of the marketer which might take away any incentive to complete the survey. Using these two types of collection methods should increase the response rate and the reach of the questionnaire, and thus is more desirable for the study. The type of data to be collected is primary data, data that is original and that has been created for the purpose of the research.

7.4Data Analysis Approach:
The data collected will then need to be analysed. The approach used to do so will use a combination of structural equation modelling (SEM), descriptive statistics and path modelling. This will use software such as SPSS (Statistical Package for the Social Sciences) and AMoS (Analysis of Moment Structures). SPSS will be used to perform descriptive statistics on the data collected, whereas AMoS will be used to measure and test that data using SEM.
Two elements will have to be tested relative to the measurement instruments utilised: reliability and validity. Reliability refers to the quality of the measurement procedure that must be repeatable, consistent and accurate. It is most often estimated using coefficients, either using Cronbach’s Alpha value or the Composite reliability value. On the other hand, validity relates to whether the collection methods were meaningful and valid through application of the correct procedures to gather the required results. This would indicate that the instruments used do in fact gather the information they were chosen to acquire. There are different types of validity which would determine which coefficients and values would be used to quantify results. Convergent validity would result in either use of Item loading, Item-to-total correlation values and Average Variance Extended (AVE), whereas discriminate validity would require either inter-construct correlation matrix or one of the variance indicators. This would be either the AVE or the shared variance.

The next stage would be hypothesis testing, which would result in either evidence of support in favour of the hypothesis or evidence against the hypothesis being observed. The signage and the significance of the relationship scrutinised will need to be ascertained in clear vocabulary, with findings to assist in the explanation of the results. The model fit results will also have to be decoded and explained in clear terms alongside signage and significance of the relationships.
Structure of the dissertation:Chapter 1: Overview of the study
Chapter 2: Literature Review
Chapter 3: Research Methodology
Chapter 4: Data Analysis and Results
Chapter 5: Conclusion and Recommendations
Ethical Considerations:When assessing whether or not a study can be undertaken, some ethical considerations need to be assessed. This is generally monitored by the ethical committee of each university, who will be able to whitelist the different research projects if deemed suitable. This study will only collect data from respondents over the age of 18, which will mean they can take comprehensive decisions themselves without needing any guardian or parent’s agreement as well. Another element that needs to be clarified is that questionnaires will be completed anonymously, so as to not capture any potentially personal or intrusive information relative to the respondent. The respondent also has the choice of whether they will complete the survey or not, and thus has freedom of choice on the matter. To make this clear respondents will sign at the beginning of the survey to symbolise their acceptance to being part of the data collection process. And lastly the information collected through this study will be used solely for the research purposes stated previously, and will be collected responsibly and legally.
Chapter 2: Literature ReviewThis chapter will review all relevant literature on theories that are needed for the purpose of understanding the important elements of the study, which are categorised under non-profit marketing theories, social media theories and the literature used to evaluate the 6 constructs from the conceptual model. In terms of non-profit marketing, the theories evaluated will be the theory of change and some theories of philanthropy. The social media theories observed are social exchange theory, social network theory, and finally McLuhan’s media theory. Finally, the constructs that will be examined are Brand Image, Social Media Marketing, Customer Engagement, Willingness to Donate, Brand Awareness, and Repeat Behaviour.

Social Media Marketing Theories:1.1) Electronic word of mouth theory
In the current environment, social media marketing and understanding how it functions and changes is one of the most essential tasks for the non-profit organisation. The first theory that will be assessed is the theory of ‘word of mouth’, which can be related to the theory of psychological ownership and perceived control (Pan and Crotts, 2012). Word of mouth refers to “the act of exchanging marketing information among consumers” (Chu, 2009), and is generally considered to be communication that is held person-to-person. However, with the arrival of the Internet and the change in behaviour of society that accompanied it, word of mouth also migrated onto the Web, and particularly the multitude of social media platforms that allow communication between its users. Termed electronic word of mouth by Chu (2009), the free flow of information has given users unlimited access to any information they desire or need, with opinions on every service and product waiting to be discovered by any undecided consumer. With information and opinions more accessible than ever online, word of mouth must be monitored with care by marketers, since any negative feedback can rapidly spiral out of control at a moment’s notice, and become an “online firestorm” as qualified by Pfeffer, Zorbach and Carley (2014). Accordingly, it must be noted that in case of negative word of mouth an organisation’s best bet would be to try and deal with the feedback and not just ignoring it, which would be perceived as extremely negative by consumers (Pfeffer, Zorbach and Carley, 2014). For this, the organisation needs to have a long-term strategy of counteracting the negative information as it comes, which necessitates employing Social Network Analysis (Pfeffer, Zorbach and Carley, 2014) and gaining the necessary traction online to appear more trustworthy than the criticising party.

1.2) Psychological ownership and perceived control theory
On the other hand, when analysing psychological ownership theory, there needs to be an understanding of the basis of the theory, that individuals have an innate needs to possess, with the growth in possessions having a positive effect on the mood of an individual while a loss in possessions will have the opposite effect on personality (Karahanna, Xu and Zang, 2015). This theory can be used in social media as well, since an individual’s possessions on the virtual platforms could extend to one’s contributions, communities and networks that they have spent time and energy into. According to Karahanna, Xu and Zang (2015), social media fulfils all three characteristics of psychological ownership theory, them being the need for self-efficacy, the need for having a place or belonging somewhere and finally the need for self-identity. An individual might give feedback or contribute to an organisation’s outgoing communication on social media since this would give them a higher sense of perceived control, and would then be tied to those necessary and beneficial characteristics such as self-efficacy and self-identity. This would thus be a beneficial scenario for both organisation and consumer, both on a more personal level and through sheer satisfaction of helping others (Pan and Crotts, 2012).
1.3) Social Exchange Theory
When seeking to evaluate social media marketing, it is necessary to examine the quality of the relationship between consumer and organisation, since in most cases individuals will attempt to create relationships that they perceive as rewarding (Clark and Melancon, 2013). Indeed, in creating a relationship the individual will internally make a cost-benefit analysis that will allow the avoidance of those connections that are felt to be too costly. The benefits expected to be received from a relationship through social media by the consumer would range from an anticipated reciprocity from the beneficiary, an expected gain in reputation or influence, altruism or a direct reward resulting from the relationship (Pan and Crotts, 2012). In the aim of better understanding consumer groupings in terms of what they want from their social media relationships, the Global Web Index (TrendsStream Limited, 2009) has grouped social media users in categories: watchers, sharers, commenters and producers. The only category that does not reciprocate any direct benefits from the relationships created by their social media usage are the watchers, who receive benefits but do not share, comment or create any content in return (Pan and Crotts, 2012).
1.4) Social Network Theory
Social network theory involves the viewing of individuals in a community as nodes, with communications between those nodes being established at multiple levels from personal, within a family context, within a community or within a nation’s society (Pan and Crotts, 2012). Social networking has established itself as crucial in the context of social media. Social networking has transitioned from occurring through interacting in a more personal and face-to-face environment to a more impersonal type of networking, using the Internet as a mediator and as a means to communicate with each other (Michaelidou, Siamagka and Christodoulides, 2011). The main benefit however is that all geographical barriers have been neutralised which means that any desire to communicate can be satisfied instantaneously. This means that organisations should try and exploit the fact that consumers will be making use of social media and make sure that communications to those consumers are effective and opportune. Because of the social networking shift online, marketers have the opportunity to target their market using one-to-many or even one-to-one approaches, which are both incredibly effective and practical for the marketer (Dervan, 2015). The one-to-many approach increases brand awareness and allows diffusion of general communications in a controlled and specific environment, whereas the one-to-one approach is more focused on improving brand equity, which might be considered more valuable to the brand. Indeed, whilst the former is more interested in targeting those individuals that aren’t yet considered consumers, brand equity is what is more valuable to the brand since this is what generates profit for them through the targeting of those individuals that can already be considered consumers.

1.5) McLuhan’s Media Theory
McLuhan was a Canadian philosopher and professor that is most famous for saying “the media is the message”, implying that it isn’t the message itself that transforms society but the media platform itself (Minton et al., 2014). This can be directly applied to the shift of all media towards the Internet and most specifically social media platforms, since according to this theory social media will influence society and change it through individuals’ use of it. According to Pan and Crotts (2012), this will most likely be achieved – if it hasn’t been already – through the “interactivity” and “frequency” patterns that accompany this new type of media. One example would be the popularity of Twitter, which is limiting its users’ character count to 140 characters, making messages and communications achieved through its platform more concise, direct and instantaneous (Johnston, 2015). For marketers to optimise these types of media platforms, there should be a clear appreciation of how social media has affected our communications and interactions, and will continue to change them in the near future (McKenzie, 2013).
Non-Profit Marketing – Theories of PhilanthropyPhilanthropy can be defined as the desire to promote the welfare of others, especially through monetary donations. It thus implies a necessary level of altruism and “generosity of spirit” (Fioravante, 2010).
Donating due to solicitations
When observing philanthropy, and non-profit marketing in general, the first thing that needs to be assessed is what drives consumers to “give” to a cause. The first and most important reason would be if the “giver” is prompted to do so, which constitutes the main reason why individuals are driven to donate (Bekkers and Wiepking, 2011). This is essentially presenting an opportunity for targeted individuals to give, instead of remaining as a passive opportunity for them, which might not lead to anything if they are not aware of that opportunity. The actual contribution will also increase the amount of donations received, since the individual that has contributed once is then more likely to contribute in the future (Leliveld and Risselada, 2017). This will lead NPOs to target those individuals that have donated in the past more aggressively than those that have never donated.
Another element to take into account is the amount requested in the solicitation to donate. According to Goswami and Urminsky (2016), presenting a list of default or suggested donation amounts on a webpage from which the donor can choose from can be both positive and negative, since although they give flexibility to donors in their donation amounts, they may also reduce the amount of funds that are raised. Lower default donation amounts increase the participation percentage, but at a possible decrease in total donation amount, whilst setting a higher default donation amount would discourage a larger section of donor-base than if there was no default amount. (Goswami and Urminsky, 2016).
Donating due to awareness of the cause
Another crucial factor affecting the reasons why individuals ‘give’ is the necessity of the individual being aware of the need before they can support and donate (Bekkers and Wiepking, 2011). A company’s reputation may affect whether the donor will be more likely to donate to the cause put forward by the NPO (Snipes and Oswald, 2010). This is because potential contributors will be more assured that their effort is being put to good use, where a lack of transparency about incoming donations will decrease the likelihood of an individual contributing, with projected results being crucial as a reassurance for those potential donors (Buechel, n.d.). The beneficiary’s deservedness is also of vital import, and will affect the amount of contributions radically. Knowing a beneficiary or potential beneficiary will greatly increase an individual’s willingness to donate, whilst on the flip side thinking that the beneficiaries in question are not deserving of the contribution will most likely decrease the chance of a donation (Bekkers and Wiepking, 2011). Thus, an organisation that is asking for contributions for children will most likely be more successful because it will statistically provoke more empathy in the potential donors, who in most cases have or will have children of their own and make their wellbeing their priority (Dickert, Sagara and Slovic; 2010).

Donating due to perceived costs or benefits
Cost and benefit are also large factors that donors need to consider when donating to a NPO. As mentioned previously, reducing the amount needed per donor increases the amount of donations. Another element of significance is how the solicitation is worded, since it alters the donors’ perception on what the value of the donation will be. For example, a solicitation announcing that each dollar donated will be accompanied by the NPO’s own 1$ donation (Bekkers and Wiepking, 2011).
Cost can be accompanied by benefit in some scenarios, with more and more donations acting as purchases for either products or services with some or all of the money used in purchase going to charity. However, according to a Stanford study, individuals were more likely to donate if they perceived the donation as an economic transaction, rather than a purely altruistic act (Miller, 2014). This is because the donor felt generally worse getting benefits from the donation if they perceived their donation as a “worse altruistic act” than if it were a “better version of a transaction”. Thus framing is of crucial import for donors and their perception when deciding if they should donate or not. Another alternative when deciding on rewards for donors could be a lottery system, where a few of the donors chosen randomly at a later date would receive larger rewards, inciting those potential donors to participate for the rewards in question.
Donating to increase one’s reputation
Reputation is another driver when it comes to donations. A donator will gain more repute if peers know that they donated and conversely will lose recognition if others know they chose to not donate. This is described by Reinstein and Riener (2011) as the “Reputation seeking effect” or “Repseek effect”. This implies that the donations should be made public or observable to peers by the organisation to encourage donor behaviour. Having peers that have closer-knitted relationships with the donor also helps, since it may lead those more influential individuals to be peer-pressured by the donor to contribute to the NPO, or another charity (Meer, 2011). Any individual is more likely to behave similarly to one’s peers in any social circles they are present in, so a donation made by a peer will most likely increase likelihood of contribution by others within that social group.

It is also important to note relative to reputation that social status is a large factor of whether individuals donate or not. Self-perceived social status constitutes one of the 8 drivers of charitable giving according to Yao (2015), with social position being defined as “combination of number of connections and societal involvements, reciprocity of those relationships, political participation, and attitudes and perceptions about the local community”. Social status should however also include the quality of lifestyle conditions, wealth and other income-based measurements that change how an individual is perceived by their peers.

Donating to better one’s self-image
Another factor to consider when assessing why people ‘give’ is how contributing to a charitable cause is may yield considerable psychological benefits to the donator and increase their self-image (Bekkers and Wiepking, 2011). According to a study analysing the effects of charitable or kind behaviour on students over a period of 6 weeks (Konrath, 2013), results showed there was a general increase in wellbeing on those students involved in the study, which included increased levels of self-acceptance, happiness, and general reductions in stress induced in their daily lives. Other psychological effects of contributing to an NPO’s activities could be an increase in positivity, neutralisation or appeasement of guilt, and satisfaction of one’s desire to give their gratitude. There is also another behavioural reason why donors would want to consider philanthropy as a possibility. Indeed, an individual might want to donate if they feel they are doing their part in bettering society as a whole (Bekkers and Wiepking, 2011), depending on their values and how they idealise society to be.

Research variables3.1) Brand Image
Brand Image will be the outcome variable used in the conceptual model and for the general purpose of the study. Brand Image should be considered the key driver to brand equity, which is the consumers’ general perception of a brand and affects their behaviour relative to it (Zhang, 2015). For a NPO wanting to alter their targets’ attitude relative to the brand, this element is vital to consider and improve and should be considered a priority for the marketer. The concept of Brand Image has been constantly evolving but generally refers to the consumer’s perception of the brand (Zhang, 2015). The relationship between Brand Image and a consumer’s buying intention has been measured using two concepts, consumer satisfaction and customer loyalty, with Brand Image proving to be heavily influential on both of these elements as proved by Zhang (2015).
3.2) Social Media Marketing
In section 2 of this literature review some of the theories related to social media marketing were discussed, explaining how social media marketing should be approached by NPOs. Now there will be more of a focus on social media marketing as a variable, with a definition and some other concepts being explained to further the understanding of the variable. Social media marketing is the marketing type that uses social media and its platforms to establish communication channels with its consumers and target market (Saravanakumar and Suganthalakshmi, 2012). Some of the social media platforms used are Facebook, Twitter, Youtube, and Instagram; which each have different benefits to brands using them. These mediums allow users to network with each other, make blog posts, share content or give out their opinions (Saravanakumar and Suganthalakshmi, 2012). Brands should take each medium into account when deciding on a social media strategy, whether they use the medium to advertise or for its own functionalities.
There are some concepts that a marketer needs to grasp before commencing any activities on social media. The concept of Virtual Brand Community (VBC) refers to an aggregation of consumers that hold the same interests in terms of a brand or product (Paquette, 2013). This implies that they may have similar purchasing behaviour relative to those similar interests they hold, which should be recognised and utilised by the marketer if possible. Being a part of such communities could also give members a channel to share their opinions with each other, thus giving the brand a better idea on consumer attitudes and desires. Finally, another concept which should be recognised is viral advertising. Viral advertising is defined as “unpaid peer to peer communication of provocative content originating from an identified sponsor using the Internet”, in the purpose of influencing an audience to use word of mouth to share that content to others (Paquette, 2013). This type of advertising is highly effective today since something highly popular on social media will most likely spread across the Internet very fluidly and rapidly, thus increasing its reach without much effort on the marketer’s part. It banks on consumer-to-consumer communication in the hope that the advertisement will spread across social media.
3.3) Customer Engagement
Customer Engagement is the third variable which will be observed in this literature review. There have been many conceptualisations of the term ‘customer engagement’, however there is a general consensus that it relates to the participation and activities consumers have with the brand (Poorrezaei, 2016). Views then differ on what should be assessed, whether it be focused on facets such as the commitment demonstrated, the perception consumers have of their participation with the brand or the consumers’ behaviour resulting from the the activites undertaken with the brand. The working definition used in this study is that consumer engagement is the sum of all “interactive experiences between consumers and the brand” (Poorrezaei, 2016). The concept has strong ties to relationship marketing, with relationships between the consumer and the brand being crucial to understanding how and why consumers engage and participate with the brand.

3.4) Willingness to Donate
Willingness to Donate illustrates the consumers’ willingness to give to a beneficiary in the form of a donation of resources (Leskovec, 2010). Donations generally are either monetary or time-based, in cases of volunteering. Due to this study focusing more on donations in a monetary sense, volunteering will be ignored in this section, to allow the focus to be more directed on monetary-based donations. There are some other results other than the direct benefits received from the donation, with one being an increase in brand loyalty following the donation. In terms of the behaviour and reasoning triggering the donation, Leskovec has created multiple conceptual frameworks to better understand the process and behaviour behind donations from consumers’, with consumer motives to donate being self-interest, altruism, a guilty conscience, pity for the beneficiary, social justice, empathy or sympathy, tax benefits or finally gaining prestige or making a difference (Leskovec, 2010). All of these are relevant factors that influence the consumers’ willingness to donate.

3.5) Brand Awareness
Brand awareness describes the strength of a brand in the consumer’s mind, and is generally recognised through observing the percentage of the population that knows of the brand’s name (Subhani and Osman, 2016). Brand awareness is beneficial to the marketer since it makes the brand familiar to consumers, which is the starting point to any activities targeting the aimed market. The objective is that the name of the brand presents itself at key decision-making stages in the purchasing process (Subhani and Osman, 2016). Awareness could thus be determinant on consumers choosing to donate to a specific NPO over another, if that NPO has taken the necessary course of action in increasing market awareness relative to its competitor. Other concepts that should be understood include brand recognition and recall, which relate to whether the consumer can recognise or recall the brand from prompts and cues, or from memory, respectively (Subhani and Osman, 2016). A brand that is ‘Top-of-the-mind’ is a brand that comes up first when consumers make a purchase, which is a significant advantage to have over competition. Additionally, a dominant brand is a brand that has total dominance over its consumers’ awareness, to the point that it is the only brand that consumers can recall off of memory.

3.6) Repeat Behaviour
Repeat behaviour consists in the situation whereby the consumer repeats the beneficial behaviour for the brand, which has been performed in the past. In this study, repeat behaviour will refer to any positive word of mouth made by the consumers, as well as any donations generated by willingness to donate. Those donors or consumers with an existing positive relationship with the organisation are easier to target since they constitute a known target for the brand, with easier to predict behaviour and motives than an unknown individual (O’Connor, 2016). Due to their past interest in the brand, it is more likely that they will feel inclined to help the NPO again in the future, which constitutes a large advantage for the brand both in terms of cost and effort (O’Connor, 2016). NPOs should therefore attempt to establish long-term relationships with their consumers, and facilitate more repeat behaviour on their part. This will also allow the brand to have more information on their consumer base, which should assist in developing more consumer satisfaction, another crucial factor on repeat behaviour (Pansari, 2016). This is due to the fact that if a consumer feels satisfied donating to a NPO, through either personal satisfaction or through direct perceived results as shown by the organisation, then they will be more likely to engage in a repeat donation to experience that same expected satisfaction as before.

Chapter 3: Conceptual Model and Hypotheses DevelopmentIntroductionThis chapter will firstly present the conceptual model which will be utilised for the purpose of this study, with secondly an evaluation of the hypotheses and how they were drawn from the literature review.
Provided below in Figure 3 is the conceptual model used in this study, which consists of one predictor variable, Social Media Marketing (SMM); four moderator variables, being Customer Engagement (CE), Repeat Business (RB), Willingness to Donate (WD) and Brand Awareness (BA); and finally one outcome variable, Brand Image (BI). The relationships between these different variables is observed in the model below.

Conceptual ModelFigure 4 – Conceptual model -33020306705used in this study
202882537465Customer Engagement
Customer Engagement

338645580010H4
00H4
1348105365760H1
0H1
14528802794033864551422403596005142240Repeat Behaviour
Repeat Behaviour

90805317500Social Media
Marketing
0Social Media
Marketing
486283021589933864551587502028825318770Willingness to Donate
Willingness to Donate

161480561595H2
00H2
486283066040H6
00H6
3386455123190H5
00H5
1452880346710145288034670900
486283077470Brand Image
Brand Image

3748405156210H7
00H7
1452880156210H3
0H3
33864541136652024380260985Brand Awareness
Brand Awareness

Hypothesis DevelopmentBased on the conceptual model above, below are the research hypotheses that can be formulated.

H1: The relationship between Social Media Marketing and Customer Engagement is positive.

H2: The relationship between Social Media Marketing and Willingness to Donate is positive.

H3: The relationship between Social Media Marketing and Brand Awareness is positive.

H4: The relationship between Customer Engagement and Repeat Behaviour is positive.

H5: The relationship between Willingness to Donate and Repeat Behaviour is positive.

H6: The relationship between Repeat Behaviour and Brand Image is positive.

H7: The relationship between Brand Awareness and Brand Image is positive.

3.1) The relationship between Social Media Marketing and Customer EngagementOne of the objectives of social media marketing is to increase customer engagement on any social media platforms used by the marketer. This relationship was observed by Farook and Abeysekera (2016) in detail, using Facebook as their main source of data in their study. The study in question analyses whether the extent invested in social media marketing on Facebook was impactful or not for the brand, which was measured by the number of consumers revisiting the brand’s page. The findings were that social media marketing has a significant impact on customer engagement, with an emphasis on a positive relationship between the two, given that more social media marketing was used, the more customer engagement resulted from the promotion (Farook and Abeysekera, 2016).

Thus, the hypothesis that will be tested is the following:
H1: The relationship between Social Media Marketing and Customer Engagement is positive.

3.2) The relationship between Social Media Marketing and Willingness to DonateAnother of the main goals of social media marketing is to increase the consumers’ willingness to buy, or donate in the context of non-profit organisations. The relationship between the two was investigated by Hajli (2015), who was using Trust as a moderator variable between Social Media and Intention to Buy. According to the findings, social media significantly affects intention to buy in terms of trust and should thus be used by non-profit marketers in their objective to change willingness to donate. Another study undertaken by Toor, Husnain and Hussain (2017) aimed to establish a positive relationship between Social Network Marketing and Consumer Purchase Intention. Findings in this study were also underlining the fact that social media had a significant effect on intention to buy, thus reinforcing Hajli’s findings considerably.
Thus the hypothesis that will be tested is the following:
H2: The relationship between Social Media Marketing and Willingness to Donate is positive.

3.3) The relationship between Social Media Marketing and Brand AwarenessFinally, social media also aims to increase consumers’ brand awareness through its social media content. The aim here is to attempt to prove a positive relationship between social media marketing and brand awareness variables. This was proven by Hutter et al. (2013) in their study on social media interactions relative to brand awareness and purchase intention, although social media marketing was recorded as Brand Page Commitment. Indeed, according to findings social media marketing positively influences brand awareness, and is fully supported as a hypothetical relationship.

Thus the hypothesis that will be tested is the following:
H3: The relationship between Social Media Marketing and Brand Awareness is positive.

3.4) The relationship between Customer Engagement and Repeat BehaviourIn theory, once customers have engaged on the social media of a brand, they are more likely to engage in the same type of behaviour in the future, which can also lead to brand loyalty in the future. A study observing customer engagement and loyalty (Fernandes and Esteves, 2016) established a relationship between customer engagement and loyalty behaviour, which includes amongst other factors the propensity to engage in repeat behaviour, whether it be through purchase or other meaningful contributions to the brand. Findings on that particular relationship were that customer engagement has a positive influence on repeat behaviour, and is one of the most impacted factors alongside word-of-mouth, and compared to factors such as complaints and price sensitivity.

Thus the hypothesis that will be tested is the following:
H4: The relationship between Customer Engagement and Repeat Behaviour is positive.

3.5) The relationship between Willingness to Donate and Repeat BehaviourWillingness to Donate is a variable that should influence Repeat Behaviour positively. Indeed, if a customer is more willing to donate then the customer will be more likely to contribute with the brand repeatedly. A study undertaken by Wallace, Buil and de Chernatony (2017) observing the moment when ‘liking’ a charity would shift to donation behaviour. According to their findings, conspicuous donation behaviour, which is based on one’s self interests has a positive and significant relationship with donation behaviour in terms of both time and money. One important thing to note is that in assessing conspicuous donation behaviour, both intention to volunteer time and to donate money were taken into account. There were more mixed findings relative to the impressing of others and acquisition of social status, which was distinguished from self-centred donation behaviour.

Thus the hypothesis that will be tested is the following:
H5: The relationship between Willingness to Donate and Repeat Behaviour is positive.

3.6) The relationship between Repeat Behaviour and Brand ImageRepeat behaviour should positively influence brand image. Indeed, for donors to come back and donate or contribute repeatedly, this implies that they are satisfied with their contribution enough to do it again. This should theoretically increase brand image for other consumers, since repeat behaviour will give them a more positive image of the brand. A study undertaken by Pappu and Quester (2006) attempted to prove a relationship between customer satisfaction and perceived quality, and managed to successfully establish a positive relationship between the two. It can thus be logically conceived that the same relationship would apply for the Repeat Behaviour, considering the fact that it is directly linked to consumer satisfaction.

Thus the hypothesis that will be tested is the following:
H6: The relationship between Repeat Behaviour and Brand Image is positive.

3.7) The relationship between Brand Awareness and Brand ImageBrand awareness and brand image are the two elements that form together the concept of brand equity. The two however differ since brand awareness matters earlier in the relationship with consumers, whereas brand image is more important in the later stages once the brand is more known and settled. This relationship was tested by Mudzakkir and Nurfarida (2015), and established successfully that brand awareness has a significant and positive influence on brand image.
Thus the hypothesis that will be tested is the following:
H7: The relationship between Brand Awareness and Brand Image is positive.

ConclusionChapter 3 has established the conceptual model for the variables and has detailed all of the hypotheses that will be tested, as well as the reasoning behind each hypothesis linking the different variables. Additionally, this was done using the support of previous literature relevant to each proposed hypothesis. The next chapter will further elaborate on how each hypothesis will be tested and what methodology will be used for that purpose.
Chapter 4: Research Methodology and Design4.1 IntroductionThis chapter explores in detail what alternatives have been chosen in terms of research methodology and design, to optimally suit and initiate this study. The different elements observed will specify decisions relative to research strategy, sampling design, data collection method, data analysis procedure, confirmatory factor indices, reliability of the study, and validity of the study.

Research Strategy
This section will cover decisions relative to research philosophy, research approach and research methodology in the context of this study.
2.1) Research Philosophy
Research philosophy refers to the way of method through which the data should be collected for the purpose of the research. There are three main schools of thought which can be chosen to undertake the data collection, with positivism, realism and interpretivism.
Positivism is a philosophy which strives to understand the social behaviour of individuals through the application of natural sciences, through in particular laws that can be generalised and applied to all individuals at any time (Vosloo, 2014). It is thus a philosophy that prioritises the collection of objective, measurable facts that are more suited for quantitative research.

Realism is a philosophy that has some similarities with positivism in that it also focuses on a scientific approach to gather and collect its information. The difference with positivism is that it believes that reality is independent from what we perceive is reality, and thus the only truth is what is shown by our senses (Saunders, Lewis and Thornhill, 2009).

Finally, interpretivism is a philosophy that bases its knowledge around the meanings individuals attribute to the physical world (Saunders, Lewis and Thornhill, 2009). Thus according to interpretivism reality is socially constructed, and cannot be studied the same way as with natural sciences, but rather independently (Saunders, Lewis and Thornhill, 2009). Thus with this philosophy knowledge will differ from individual to individual, and will need to be understood and observed as a singularity rather than a generality.
The research philosophy which will be applied in this study is the positivism approach to collecting data.
2.2) Research Approach
Research approach refers to how the researcher interacts proceeds with the study in terms of steps and procedure, and is divided between two types of approaches: deductive theory and inductive theory. Inductive theory begins with observation, and concludes with a theory inferred by the implications of the findings, whereas deductive theory begins with a theory and concludes with conclusion after hypothesis testing has been undertaken (Zhalaghi and Khazaei, 2016).
The type of research approach that will be used for the purpose of this study is deductive theory, since this approach is more applicable to the procedure of the study and is striving to confirm a theory that has been established early on in the study. This approach is more in line with the positivist philosophy that was preferred and will be supported by a quantitative method of analysis.
2.3) Research Methodology
There are two main types of research methodology that can be used for the purpose of this study, the qualitative and quantitative methodologies of research.
Qualitative research focuses on subjective data and the understanding of the meaning individuals associate with their experiences, thus giving a better idea on why individuals behave a certain way and what drives them to do so (Sutton and Austin, 2015). Data will be collected through interviews in a more personal and direct manner, with a dialogue between the researcher and the subject being recorded either with audio or video (Sutton and Austin, 2015).
On the other hand, quantitative research will focus on establishing objective generalisations that will give a better understanding of social and general behaviour (McLeod, 2008). The underlying purpose is to test a theory which will result in either a confirmation or a rejection of the theory. The types of tools used for data collection are varied, and range from experiments to questionnaires and observations (McLeod, 2008).
The type of methodology used is quantitative research for the purpose of this study, since its purpose is more similar to our line of enquiry, with the testing of a theory that will yield either positive or negative results.

Sampling DesignThis section comprises of three important characteristics that need to be established: target population, sample size and sampling method.

Target population describes the collection of individuals from which the sample is selected. The target population used for this study will be Wits students that use social media, since they are representative of the market non-profit organisations are trying to target through social media marketing. University students form part of the age groups that have the most presence on social media in South Africa, and will thus be the most engaging consumers on those platforms. (Business Tech, 2017) It will also be valuable for non-profit marketers to know whether students successful with their education are aware of marketing attempts made in their direction. The age of the students questioned for the purpose of the study will be between of a minimum of 18 years old for ethical considerations.
Sample size involves the size of the sample selected for the research questions, and when using quantitative methods it is needed to use a larger sample size. For this study, a sample size of 300 would be adequate, from a variety of subcultures and backgrounds. This amount of samples should be satisfactory to make allowance for there being potential errors or non-responses in the survey results. This sample size was determined by using a rule of thumb put forward by Bentler and Chou, whereby there must be 5 or 10 observations per parameter (Wolf et al., 2013). Since this study contains a total of 30 parameters, using 10 observations per parameter gives us a sample size of 300 respondents.

Sampling method denotes the procedure used to select the samples, and differentiates between two potential method types: Probability sampling and Non-probability sampling. For the purpose of this study and considering the research will use Quantitative methods the method used will be probability sampling. There are then four different types of probability sampling to choose from: simple probability sampling, systematic probability sampling, stratified probability sampling and cluster probability sampling.
The sampling method chosen for the purpose of this study is cluster sampling, since it is more efficient in terms of data collection. It is in essence a method where the researcher divides the target population into clusters and then a random set of clusters is selected for sample collection. It is a very efficient method of data collection but has drawbacks such as reduced accuracy and representativeness.

Data Collection Method (Questionnaire Design, Pre-testing the instrument, Ethical considerations)This segment relates to the different element of data collection, and will observe questionnaire design decisions, the pre-testing of the instrument and ethical considerations that need to be undertaken when approaching research participants.

4.1) Questionnaire Design
This section refers to the specificities of the questionnaire, and mentions elements such as the Likert scale used, and the different sections of the questionnaire which need to be completed by the respondent. In this case the questionnaire will use a five-point Likert scale, which consists of five options: 1. Strongly Disagree, 2. Disagree, 3. Neutral, 4. Agree and 5. Strongly Agree. There will be seven sections to the questionnaire, sections A to G. Section A will be formed of questions giving general demographic and background information about the respondent, such as Age, Gender and Level of Education. Sections B to G will then ask questions specific to each variable, with section B relating to Social Media Marketing, section C relating to Customer Engagement, section D relating to Willingness to Donate, section E relating to Brand Awareness, section F relating to Repeat Behaviour, and section G relating to Brand Image.
Below are some of the measurement instruments that would be used in the research, with the questions being modified from original sources to better fit the type of the organisation being referred to. In the preliminary questions the student will have to pick one NPO they feel strongly about, and if the respondent can’t think of a South African organisation then they would be allowed to pick an international NPO.
Instruments for Section A: (Vinerean et al., 2013)
Gender (M/F/Other/Prefer not to say)
Age class (18-23 / 24-29 / 30-35 / 36-41)
Level of education (Undergraduate / Honours / Masters / PhD)
For how long have you been using social media websites? (1 – 6 months / 6 months – 1 year / 1 – 2 years / 2 – 3 years / More than 3 years)
How would you describe your log in pattern on social media sites? (Always connected / Several times a day / Every three days / Once a week / Occasionally – Less than once a week)
Instruments for Section B: (Vinerean et al., 2013) (Godey et al., 2016)
The ads that appear on my profile are relevant for my personal interests and I enjoy seeing them.

Quite often I access the ads that I see on my social media profile.
I do experience concern regarding the confidentiality and privacy of my personal information.
Content of X brand’s social media seems interesting.

Using X brand’s social media is fun.

Content of X brand’s social media is the newest information.

Using X brand’s social media is very trendy.

Instruments for section C: (Godey et al., 2016)
X brand’s social media enable information-sharing with others.

Conversation or opinion exchange with others is possible through X brand’s social media.

It is easy to provide my opinion through X brand’s social media.

I would like to pass information on brand, product, or services from X brand’s social media to my friends.

Instruments for section D: (Van der Heijden and Verhagen, 2004)
I am positive towards donating on X’s website.

The thought of donating on the website of X is appealing to me.

I think it is a good idea to donate on the website of X.

Instruments for section E: (Godey et al., 2016)
I am always aware of X brand.

Characteristics of X brand come to my mind quickly.

I can quickly recall the symbol or logo of X brand.

Instruments for section F: (Godey et al., 2016)
Although another brand has the same features as X, I would prefer to donate to X.

If another brand does not differ from X, it seems smarter to donate to X.

Although there is another brand as good as X, I prefer to donate X.

I will suggest X brand to other consumers.

I would love to recommend X brand to my friends.

I regularly visit X brand’s social media page.

I intend to visit X brand’s social media page again.

I am satisfied with X brand with every visit to their social media page.

X brand would be my first choice.

Instruments for section G: (Godey et al., 2016)
X brand is a leading NPO.

X brand has extensive experience in the industry.

X brand is a leading representative of the NPO industry.

X brand is a customer-oriented company.

4.2) Pre-testing of the instrument
It is generally preferred to have the instrument tested prior to the distribution of the questionnaires, through a trial run with a small sample of participants which form part of the target population. This pilot survey will inform the researcher on any errors or problems with the questions, and assess whether the questions are easily understood by research participants. This will also assist in establishing a general time estimate it would take to complete the questionnaire, which will help in informing respondents on whether they have the time availability to complete it. For the purpose of pre-testing, we will use a sample size of 15 respondents, which corresponds to 5 % of our questionnaire’s sample size of 300 respondents (Chaudhary and Israel, 2014).
4.3) Ethical considerations
When assessing whether or not a study can be undertaken, some ethical considerations need to be assessed. This is generally monitored by the ethical committee of each university, who will be able to whitelist the different research projects if deemed suitable. This study will only collect data from respondents over the age of 18, which will mean they can take comprehensive decisions themselves without needing any guardian or parent’s agreement as well. Another element that needs to be clarified is that questionnaires will be completed anonymously, so as to not capture any potentially personal or intrusive information relative to the respondent. The respondent also has the choice of whether they will complete the survey or not, and thus has freedom of choice on the matter. To make this clear respondents will sign at the beginning of the survey to symbolise their acceptance to being part of the data collection process. And lastly the information collected through this study will be used solely for the research purposes stated previously, and will be collected responsibly and legally.
Data Analysis Procedure (Data Coding and Cleaning, Descriptive Statistics, SEM – measurement model and structural model -)5.1) Data Coding and Cleaning
Before being able to analyse and observe the data collected through the survey, it is necessary to check certain factors that might have occurred in the data collection. It is for this reason that data capture is so important, since it allows the researcher to engage in an initial screening of the data to ensure that there are no discrepancies such as errors, which could be either because of the respondents or the researcher (Jones et al., 2013). The types of errors that could be encountered can be categorised as “missing values, systematic errors, random errors, in?uential errors, and outliers” (Jones et al., 2013)and represent respectively errors that are relative to either uncompleted questions in the questionnaire, systematic vagueness or misunderstanding with a specific question, accidental mistakes occurred in data capture by either respondent or researcher, unusually large recorded values, or finally observations that have a large distance from the rest of the observations.
Data coding refers to the act of transferring all data into code which can both be “summarised and analysed” (Jones et al., 2013). This is done through evaluating how many possible answers exist for each question, and then labelling an answer as a number. For example for a 5-point Likert scale, answers would range from 1 to 5, with 1 representing “Strongly Disagree” and 5 representing “Strongly Agree”.
According to de Jonge and van der Loo (2013), data cleaning is the act of transforming raw data into data that can be consistently and accurately analysed. This thus aims to correct the errors observed in the data coding phase of data analysis, which is also called error localisation (de Jonge and van der Loo, 2013). Remedies will most likely require correction, transformation or standardisation depending on the type of error. In some cases, a respondent’s entire questionnaire will have to be eliminated if there is too much missing data on their questionnaire, and is referred to as “case deletion” (Kang, 2013), and is the most common method of remedying to missing data. That is why it is important to have a sufficiently large sample size that won’t be affected easily by case deletions.
5.2) Descriptive Statistics
Descriptive statistics can be referred to as the use of tables and graphs to provide detailed and descriptive information about specific variables, and include percentage distributions, means, medians and standard deviations (Glewwe and Levin, 2005). Descriptive statistics will be used for the purpose of illustrating and analysing the demographic characteristics of the respondents involved in the study.
5.3) Structural Equation Modelling
Structural Equation Modelling is a technique that allows for a set of variables and the relationships between them to be observed, using two different models, the measurement model and the structural model (Ullman and Bentler, 2013). For the purpose of this study, it will be used to observe and analyse the relationships between our variables, using the hypotheses that were established and theorised in Chapter 3. The reason why this technique is so effective is that it uses both manifest variables, which are observed and visible; and latent variables, which are unobserved and help in optimising the accuracy of the statistical estimation of the tested hypotheses (Ullman and Bentler, 2013). Thus the latent variables will be the variables from our conceptual model in Chapter 3, whereas our manifest variables will be the different questions from the questionnaire which have been answered by respondents. For the purpose of this study, both the SPSS and the AMOS software will be used, the former to transfer the data from Excel to AMOS and the latter to create the measurement and structural models.

The first step of this approach is the measurement model, and uses confirmatory factor analysis (CFA) to identify latent variables and demonstrate which manifest variables are determinant in measuring them (Hatcher and O’Rourke, 2013). Instead of creating a model with directional relationships between latent constructs, the CFA model allows correlations between each latent variable, and will attempt to establish high levels of goodness of fit and identify where changes are needed to ensure an optimal model fit (O’Rourke and Hatcher, 2013).

The second model that will be used is the structural model, which will use path analysis to verify and validate relationships between the latent variables. This is the model that uses directional relationships between the latent constructs, as opposed to the previous measurement model (O’Rourke and Hatcher, 2013). Instead of identifying relationships between latent variables, the purpose here is to test the strength and type of each relationship (O’Rourke and Hatcher, 2013).

Confirmatory Factor Analysis (with each index explained and observed)As was explained in the previous section, CFA is a technique used to identify the different relationships between variables, and test the strength of the model fit for each relationship. This is done through the analysis of the differing model fit indices, which will be used for both the measurement model and the structural model. Another piece of information that is provided by CFA and is essential to the study are the standard regression weights between the variables.

Model fit indices can be categorised in four groups: absolute fit indices, relative fit indices, parsimonious fit indices, and noncentrality-based indices (Newsom, 2018). Absolute fit indices refer to those indices that don’t use another model for comparison, and act as a more direct measurement of the fit between the model and the collected data (Newsom, 2018). Relative fit indices on the other hand are a group of indices that use a null model that is used to compare with the tested model (Newsom, 2018). Parsimonious fit indices are relative fit indices that are adjusted versions of the indices mentioned prior, and ensure simplicity in the model rather than complexity (Newsom, 2018). Finally, noncentrality-based indices are indices that attempts to reject the possibility that chi square is false, rather than testing to prove its truth (Newsom, 2018).

The indices examined for the purpose of this study are CMIN/df or the Chi-square index, the Comparative Fit Index (CFI), the Goodness of Fit Index (GFI), the Normed Fit Index (NFI), the Incremental Fit Index (IFI), the Tucker – Lewis Index (TLI), and finally the Root Mean Square Error of Approximation (RMSEA). We thus have some absolute fit indices, which are the Chi-square index (?2) and the GFI; some relative fit indices, which are the NFI, the TLI, and the IFI; and finally some noncentrality-based indices, which are the CFI and the RMSEA.

The Chi-square test, or ?2, is used for when one nominal variable is tested to measure whether its observations fit theoretical expectations relative to its sample size (McDonald, 2015). For the purpose of this test then, it is crucial to have a large sample size which will maximise the accuracy of the test. There needs to be two or more values associated with the nominal variable (McDonald, 2015), which works for our variables that have five different possible values that can result from it, since the 5-point Likert scale is used in our questionnaire. According to Moss, Lawson and White (2015), a relative chi-square (CMIN/df) of ; 3 indicates an acceptable fit between the hypothetical model and the observations from the sample data, while a CMIN/df of 3 ; 5 indicates a reasonable fit between the model and the sample data (Moss, Lawson and White, 2015).

The Comparative Fit Index (CFI) compares the tested model to an alternative, which is determined through the “manifest covariance matrix” evaluation (Cangur and Ercan, 2015). The CFI provides values which are between 0 and 1, with high values demonstrating a better fit between the two models. A reasonable fit would be ; 0.9 and a good fit would be ; 0.95 (Moss, Lawson and White, 2015).
The Goodness of Fit Index (GFI) is an index that was created as an alternative to the ?2 test, and demonstrates the fit between the proportion of variance and the estimated population covariance matrix (Hooper, Coughlan and Mullen, 2008). The GFI is a value that ranges between 0 and 1, with high values demonstrating a better fit. A reasonable fit would be ; 0.9 and a good fit would be ; 0.95 (Moss, Lawson and White, 2015).

The Normed Fit Index (NFI) is an index that corresponds to the difference between the chi-square of the null model and that of the tested model (Moss 2016). Since values range between 0 and 1 – with 1 being an ideal fit – then the higher the value will be, the better fit the tested model has relative to the null model (Moss 2016). A reasonable fit would be ; 0.9, and a good fit would be ; 0.95 (Moss, Lawson and White, 2015).

The Incremental Fit Index (IFI) is an index that is calculated by first computing the difference between the chi square value of the independent model, which uses uncorrelated variables; and the chi square value of the tested model (Moss 2016). The next stage necessitates the computing of the difference between the chi square value of the tested model and the df for the tested model. These two values are then used to provide a ratio which corresponds to the IFI. The IFI is seen as a reasonable fit if ; 0.9 and a good fit if ; 0.95 (Moss, Lawson and White, 2015), however it isn’t necessarily a value that ranges from 0 to 1 since it sometimes exceeds that ceiling (Moss, 2016).

The Tucker – Lewis Index (TLI) is an index that is calculated by dividing the chi square values of the target model and of null model by their respective df values, which results in relative chi square values for each model. The difference in these relative chi square values is then computed, and then divided by the value of the relative chi square for the null model, minus 1 (Moss, 2016). It is used to counterbalance the NFI since it is non-normalised and less susceptible to sample size (Cangur and Ercan, 2015). A reasonable fit for the TLI would be > 0.9, and a good fit would be > 0.95 (Moss, Lawson and White, 2015), but it isn’t required to be between 0 and 1 (Cangur and Ercan, 2015).

The Root Mean Square Error of Approximation (RMSEA) is an index that computes “the difference between the observed covariance matrix per degree of freedom and the hypothesized covariance matrix which denotes the model” (Cangur and Ercan, 2015). Acceptable norms for the RMSEA are that the value needs to be ; 0.8 to be acceptable, ; 0.5 to be good and ; 0.1 to be excellent (Moss, Lawson and White, 2015).
Reliability (Cronbach Alpha, Composite Reliability, Average Variance Extracted)There are multiple tools used to test the reliability of the research methods used for the purpose of this study, such as the Cronbach Alpha test, Composite Reliability and Average Variance Extracted, which all will be expanded upon in this section. All of these tools are used through the SPSS and AMOS software, similarly to the model fit indices mentioned previously.
Cronbach Alpha is a measure that allows an evaluation of the reliability and consistency of a test, and is a value that ranges between 0 and 1 (Tavakol and Dennick, 2011). It essentially measures the interrelatedness of the items within a test, and whether these items all measure the same elements (Tavakol and Dennick, 2011). Accordingly, the Cronbach Alpha value should be ; .70 to be acceptable, and it is also recommended that the value be ; .90 (Tavakol and Dennick, 2011). This is because although having a high value in terms of reliability is good, a too high value might correspond to redundant values that measure the same or similar elements using different wording (Tavakol and Dennick, 2011).
Composite reliability (CR) acts as an assessment that focuses on estimating an item’s internal consistency (Hair, Ringle and Sarstedt, 2011). Accordingly, and similarly to Cronbach Alpha, values between 0.70 and 0.90 are deemed acceptable, with high values demonstrating a higher reliability (Hair, Ringle and Sarsteldt, 2011). Values > 0.90 are not recommended, since once again it might just suggest redundancy in the items due to their similarity. The formula used for the purpose of testing composite reliability of the different items is as follows:
CR? = (??yi) 2/ (??yi) 2+ (??i)
Composite Reliability = (square of the summation of the factor loadings)/ {(square of the summation of the factor loadings) + (summation of error variances)}.

The Average Variance Extracted (AVE) is a measure used to calculate the “average variance shared between a given construct and its indicators” (Dean et al., 2008). Typically, the AVE of a given construct should be greater than the value of the variance between the construct in question and any other constructs (Dean et al., 2008). The value of the AVE should also be > 0.5 since values above that threshold indicate that more than 50 % of the item’s variance is encapsulated by that construct (Dean et al., 2008). The formula used for the purpose of testing the AVE of the different constructs is as follows:
V?=??yi2/ (??yi2+??i)
AVE = {(summation of the squared of factor loadings)/ {(summation of the squared of factor loadings) + (summation of error variances)}.

Validity (Convergent, Discriminant)Validity is another element alongside reliability that needs to be maintained when assessing variables and their relationships. Validity is essential in any study, since it is basically demonstrating that its results and findings are sound and logical, or at least seem to be so (Becker, Rai and Rigdon, 2013). There are two different types of validity that will be used to test the constructs, which are convergent and discriminant validity.
Convergent validity is a concept that refers to the comparison of two instruments, the tested instrument and another that measures a similar but different construct (Godwin et al., 2013). This allows a comparison between the two similar and related constructs, which gives an idea on whether one depends on the other or whether they exist concurrently (Godwin et al., 2013). Results from convergent validity should be ideally ; 0.70 to indicate a strong relationship, although a value ; 0.50 could be sufficient according to some researchers (Godwin et al., 2013).

On the other hand, discriminant validity ensures the uniqueness of a construct measure to guarantee the fact that the utilised measures do not overlap on what they are capturing (Henseler, Ringle and Sarsteldt, 2015). This assures that distinct measures that need to make different contributions of interest and aren’t supposed to be assessing the same objects don’t have too high a correlation (Henseler, Ringle and Sarseldt, 2015). In this study, we will be utilising two main methods for discriminant validity, AVE which was mentioned previously and the construct correlation matrix. This matrix establishes through its calculations whether the constructs that are tested have a high correlation or not, with values ; 0.90 demonstrating an extremely high correlation (Ashraf et al., 2017). A correlation that is too high implies a common method bias, which means that values should ideally be kept ; 0.90 to ensure discriminant validity between the different variables (Ashraf et al., 2017).
Chapter 5: Data Analysis and Survey ResultsIntroductionThis chapter provides a discussion and analysis of the results formed using the data collected, which was done through the SPSS and AMOS software. In the first stage demographic statistics of the respondents will be focused on, before assessing the measurement model using reliability and validity instruments. Information will also be given on the variables, their correlations, the strength of their relationships and on what that means for each hypothesis tested in the study.
Demographic StatisticsThis section will give an overview and outline of the demographic information relative to the 300 respondents that participated in this study. The demographic data that was collected ascertain elements such as gender, age, level of education, duration of social media usage and finally frequency of social media usage.
Gender
Table 5.1 Frequency Percent Cumulative Percent
Valid Male 121 40.3 % 40.3 %
Female 174 58.0 % 98.3 %
Prefer not to say 5 1.7 % 100.0 %
Total 300 100.0 % Thus the majority of respondents were female (58%), with 40.3% of males and 1.6% of respondents that preferred not to say. This might signify that there are a majority of female students studying at Wits University.

Age
Table 5.2 Frequency Percent Cumulative Percent
Valid 18-23 262 87.3 % 87.3 %
24-29 29 9.7 % 97.0 %
30-35 7 2.3 % 99.3 %
36-41 2 .7 % 100.0 %
Total 300 100.0 % Table 6.2 and its corresponding pie chart give a representation of the age demographic that responded to the study, with a large majority of respondents being between 18 and 23 years of age (87.3 %). This was intended and expected, since the study was aiming to target students of that age range. In contrast there was 9.7 % of respondents that were between 24 and 29 years of age, 2.3 % of students between 30 and 35 years of age, and finally 0.7% of respondents that were between 36 and 41 years of age, with no respondents above 41.
Current Level of Education
Table 5.3 Frequency Percent Cumulative Percent
Valid Undergraduate 218 72.7 % 72.7 %
Honours 67 22.3 % 95.0 %
Masters 1 .3 % 95.3 %
Ph.D 1 .3 % 95.7 %
Other 13 4.3 % 100.0 %
Total 300 100.0 % The majority of respondents indicated they had a current level of education of Undergraduate (72.7 %), whilst there was 22.3 % of respondents with a current education level of Honours, 4.3 % of respondents who indicated their current education level as “Other”, 0.3% of respondents indicated being at a Masters level and finally 0.3 % of respondents indicated being at a Ph.D level. The respondents that preferred the option of “Other” might have chosen it since their degree or qualification didn’t match to any of the options in the question, since there are many degrees that do not conform to the typical degree of 3 years and another year of Honours, such as the LLB degree for Law students.

Duration of Social Media Usage
Table 5.4 Frequency Percent Cumulative Percent
Valid Started this year 5 1.7 % 1.7 %
1-3 years 22 7.3 % 9.0 %
3-5 years 56 18.7 % 27.7 %
5+ years 212 70.7 % 98.3 %
Never 5 1.7 % 100.0 %
Total 300 100.0 % The above question was aimed to understand how much experience the respondents had with social media, which was most obvious in terms of years of usage. Most students have had extensive experience with social media, since 70.7 % of respondents indicated that they have had more than 5 years of social media experience, whilst 18.7 % indicated they had between 3 and 5 years of social media experience. On the other hand, 7.3 % of respondents indicated having between 1 and 3 years of social media experience, 1.7 % admitted having started using social media this year and another 1.7 % indicated not being on social media at all.
Frequency of Social Media Usage
Table 5.5 Frequency Percent Cumulative Percent
Valid Always connected 78 26.0 % 26.0 %
Several times per day 156 52.0 % 78.0 %
Once per day 18 6.0 % 84.0 %
Several times per week 27 9.0 % 93.0 %
Once per week 8 2.7 % 95.7 %
Less than once per week 7 2.3 % 98.0 %
I’m not on social media 6 2.0 % 100.0 %
Total 300 100.0 % The above table and its pie chart are relative to the frequency of social media usage question, and aims to shed light on how frequently respondents use social media on average. 52 % of respondents indicated that they’d check their social media platforms several times per day, whilst another 26 % of respondents indicated being always connected to social media, most likely through their smartphones. On the other hand, 9 % of respondents indicated using social media several times per week, 6 % of respondents indicating they used social media once per day, 2.7 % indicated their usage was once per week, 2.3 % indicated usage of less than once per week, and finally 2 % indicated not using social media on a weekly basis. What is interesting is that the 2 % of non-users of social media does not correlate to the previous percentage of non-users (which was 1.7 %). Perhaps a reason could be that some experienced users had decided to give up social media, which would thus explain the discrepancy between the two questions.

Measurement Model Assessment: Measurement Reliability and ValidityThis section will comprise of information relative to the measurement model’s assessment, and most notably its reliability and validity, using a variety of statistics such as Cronbach Alpha, Composite Reliability, and Average Variance Extracted.

Below is a table of accuracy analysis statistics for each measurement instrument as well as for the latent variables.

Table 5.6: Accuracy Analysis Statistics for Reliability and Validity
Research constructs Scale item Cronbach’s test CR AVE Factor loadings
Mean
SD
Item-total value SMM SMM4 3.18 .971 .644 .795 0.82 0.53 0.775
SMM5 3.04 1.045 .662 0.801
SMM6 3.21 .974 .542 0.668
SMM7 3.01 1.093 .580 0.655
CE CE1 3.62 1.030 .595 .790 0.76 0.44 0.688
CE2 3.47 1.006 .663 0.563
CE3 3.39 1.034 .623 0.617
CE4 3.19 1.154 .525 0.773
WD WD1 2.87 1.221 .697 .853 0.85 0.66 0.773
WD2 2.91 1.243 .790 0.908
WD3 3.25 1.243 .689 0.754
BA BA1 3.07 1.177 .636 .768 0.77 0.53 0.757
BA2 3.16 1.138 .608 0.756
BA3 3.64 1.204 .561 0.659
RB
RB3 3.31 1.019 .547 .858 0.84 0.44 0.657
RB4 3.56 1.008 .681 0.763
RB5 3.51 1.000 .690 0.767
RB6 2.69 1.131 .582 0.505
RB7 3.21 1.102 .635 0.623
RB8 3.15 1.017 .662 0.645
RB9 3.38 1.030 .581 0.628
BI BI1 3.44 1.041 .696 .840 0.84 0.64 0.784
BI2 3.78 .961 .701 0.798
BI3 3.52 1.026 .718 0.815
Notes: SMM = Social Media Marketing, CE = Consumer Engagement, WD = Willingness to Donate, BA = Brand Awareness, RB = Repeat Behaviour, BI = Brand Image
Mean, Standard Deviation & Item-to-Total Correlation
The presented results in Table 5.6 provide information on the descriptive statistics and the measurement model assessment statistics, which are used to identify whether the model is accurate in terms of both validity and reliability. The values which are shown under the mean column indicate the average answer for each question, using the 5-point Likert scale as was discussed previously. Most mean values are between 3 and 4, with only instruments WD1, WD2 and RB6 that are under 3 with mean values of respectively 2.87, 2.91 and 2.69. Considering that the standard deviations are all less than 2, this clarifies a correct reflection of the majority average perceptions of the mean values. The item total correlation is a correlation between the measurement’s score and the study’s score (Pope, 2009). For this study, items which depicted results of below 0.5 were removed from the study as they are considered to be incorrect results. The measurement instruments that were removed are, in order, SMM1, SMM2, SMM3, RB1, RB2, and BI4, due to either Item-Total Correlation values that were below 0.5, factor loadings that were below 0.5, or both. The Item-to-total correlation values from the individual results of the remaining measurement instruments range from 0.525 to 0.79. This implies that these remaining values are acceptable for the purpose of this study.

Cronbach Alpha
Cronbach Alpha was developed to provide a measure of the internal consistency of a test or scale, and is expressed as a number between 0 and 1. Internal consistency describes the extent to which all the items in a test measure the same concept or construct and hence is an indicator of the inter-relatedness of the items within the test (Tavalok & Dennick, 2011). There are different reports about the acceptable values of an alpha value, with values ranging from 0.70 to 0.95 which can be deemed acceptable (DeVellis, 2003). The lowest value that appears in this studies result is 0.76, and the highest is 0.858. This therefore indicates that the provided results are acceptable.

Composite Reliability (CR) and Average Variance Extracted (AVE)
Composite Reliability (CR) is a measure of scale reliability, and assesses the internal consistency of a measure (Fornell & Larcker 1981). The composite reliability values must exceed 0.7, and is tested using the formula bellow:
CR? = (??yi) 2/ (??yi) 2+ (??i)
Composite Reliability = (square of the summation of the factor loadings)/ {(square of the summation of the factor loadings) + (summation of error variances)}.

Table 6.6 indicates that all values are above 0.7, the lowest being 0.76 and the highest is 0.85. Composite Reliability for the variables is thus acceptable.

Average Variance Extracted (AVE) shows the percentage of variance interpreted by the latent factors from measurement error. The larger average variance extracted is, the larger indicator variance could be interpreted by the latent variables and the smaller relative measured error is. Generally speaking, the criteria for average variance extracted should be greater than 0.5. However, values between 0.4 and 0.5 can still be deemed acceptable if composite reliability for that variable is above 0.6 (Huang et al., 2013). The following formula is used to calculate (AVE):
V?=??yi2/ (??yi2+??i)
AVE = {(summation of the squared of factor loadings)/ {(summation of the squared of factor loadings) + (summation of error variances)}.

The results of Table 6.6 reveal that variables SMM, WD, BA and BI are above 0.5. The two variables that are beneath the threshold of 0.5 are CE and RB, which both have values of 0.44. Since both variables have Composite Reliability values of more than 0.6 (respectively 0.76 and 0.84), these AVE values are deemed acceptable. The highest AVE value is WS which has a value of 0.66.
Inter-Construct Correlation Matrix
A correlation matrix basically demonstrates the interconnections between a series of variables. It computes the correlation coefficients between variables which are represented in the same sequence of rows and columns (White, Korotayev & Khaltourina, 2004). These values range from 0-1, with a lower value being an indicator of a more independant relationship, while the higher values will indicate the absence of discriminant validity and show more relation between variables. According to Table 6.7, the highest value shown in this study is 0.575 while the lowest value is 0.145. Since these values are not very high this demonstrates that there is in fact discriminant validity present in the relationships between the 6 variables used in the study.

Table 5.7: Inter-Construct Correlation Matrix
CE WD BA RB BI SMM
CE Pearson Correlation 1 WD Pearson Correlation .212 1 BA Pearson Correlation .477 .153 1 RB Pearson Correlation .575 .400 .513 1 BI Pearson Correlation .263 .309 .369 .330 1 SMM Pearson Correlation .490 .209 .359 .497 .145 1
Confirmatory Factor Analysis (CFA)Figure 5.1 illustrates the Confirmatory Factor Analysis model used for the study utilizing AMOS software. It was used firstly to obtain the standardized regression weights for the 5 constructs in the model, which were then used to obtain the composite reliability (CR) values as well as the average variance extracted (AVE) values for each variable, which were discussed in the previous section. The CFA model is used to identify the strength of the relationships between the different variables, which are represented by the double sided arrows between the variables, on the left of Figure 5.1.

13716071755Figure 5.1 : CFA Model
00Figure 5.1 : CFA Model

Note: SMA refers to Social Media Marketing, CEN refers to Customer Engagement, WDO refers to Willingness to Donate, BAW refers to Brand Awareness, RBE refers to Repeat Behaviour, and BIM refers to Brand Image.

With the results obtained through the CFA model, there were a variety of model fit indices that the study needed to conform to. The indices examined for the purpose of this study are CMIN/df or the Chi-square index, the Comparative Fit Index (CFI), the Goodness of Fit Index (GFI), the Normed Fit Index (NFI), the Incremental Fit Index (IFI), the Tucker – Lewis Index (TLI), and finally the Root Mean Square Error of Approximation (RSMEA).
The model fit indices were as follows. The Chi-square (CMIN/DF) for the model was 1.094, falling within the recommended threshold of below 3 (Kline, 1998). The Comparative Fit Index (CFI) was of 0.994 and surpassed the acceptable level of 0.900 (Bentler, 1990), the Goodness of Fit Index (GFI) was 0.946, and surpassed the acceptable 0.900 level according to (Bentler 1990). In addition, the normed fit index (NFI) was of 0.941 and exceeded the 0.900 acceptable threshold (Bentler, 1990).The Incremental Fit Index (IFI) was 0.995 and also exceeded the 0.900 recommended threshold (Clarke, 2010). The Tucker-Lewis Index (TLI) was of 0.992, which was above the recommended threshold of 0.900 (Moss, 2016). Finally, the Root Mean Square Error of Approximation (RMSEA) was below the acceptable threshold of 0.08 at 0.018 (MacCallum, Browne & Sugawara, 1996).

Table 5.8: Model Fit Summary – CFA Model
Model Fit Indices Current Study Threshold Acceptable Threshold Decision
CMIN/df 1.094 < 3 Acceptable
CFI 0.994 > 0.900 Acceptable
GFI 0.946 > 0.900 Acceptable
NFI 0.941 > 0.900 Acceptable
IFI 0.995 > 0.900 Acceptable
TLI 0.992 > 0.900 Acceptable
RMSEA 0.018 < 0.08 Acceptable
As seen in Table 5.8 above, all model fit results are within the recommended and accepted thresholds as required. This means that it is fine to proceed to the hypothesis testing and path modelling stages of the analysis, with the data that was collected for the purpose of the study.

Structural ModelFigure 5.2 below is the structural model used for the purpose of this study, which is a representation of the conceptual model using AMOS software. Each retained measurement instrument is represented by a rectangle, whereas the latent variables are repesented by circles. Error terms, which are represented by circles and labelled e1 to e31 are associated to each variable, be it observed or latent. The one-way arrows from latent variable to latent variable are used to represent causal relationships.

51435300355Figure 5.2 : Structural Measurement Model
00Figure 5.2 : Structural Measurement Model

Note: SMA refers to Social Media Marketing, CEN refers to Customer Engagement, WDO refers to Willingness to Donate, BAW refers to Brand Awareness, RBE refers to Repeat Behaviour, and BIM refers to Brand Image.

Table 5.9: Model Fit Summary – Structural Model
Model Fit Indices Current Study Threshold Acceptable Threshold Decision
CMIN/df 0.649 < 3 Acceptable
CFI 1 > 0.900 Acceptable
GFI 0.971 > 0.900 Acceptable
NFI 0.969 > 0.900 Acceptable
IFI 1.017 > 0.900 Acceptable
TLI 1.029 > 0.900 Acceptable
RMSEA 0 < 0.08 Acceptable
The model fit for the path analysis is presented in Table 5.9 and gives results for the indices of CMIN/df = 0.649 ; CFI = 1 ; GFI = 0.971 ; NFI = 0.969 ; IFI = 1.017 ; TLI = 1.029 and RMSEA = 0, which are all within acceptable norms and within the acceptable thresholds as confirmed in the CFA table of the summary of its model fit indices.

Hypothesis testingTable 5.10: Hypothesis Testing Results
Proposed Hypothesis Relationships Hypothesis Path Coefficients (?) P-Values Rejected/
Supported
SMM CE H1 .638 *** Supported and significant
SMM WD H2 .405 *** Supported and significant
SMM BA H3 .523 *** Supported and significant
CE RB H4 .784 *** Supported and significant
WD RB H5 .253 *** Supported and significant
RB BI H6 .080 .516 Supported but insignificant
BA BI H7 .431 *** Supported and significant
Note: SMM refers to Social Media Marketing, CE refers to Customer Engagement, WD refers to Willingness to Donate, BA refers to Brand Awareness, RB refers to Repeat Behaviour, and BI refers to Brand Image.

***= 0.001 level of significance
From Table 5.10 found above, we can observe that 7 hypotheses were tested for the study. The hypotheses were named H1, H2, H3, H4, H5, H6 and H7, and their coefficients were respectively 0.638, 0.405, 0.523, 0.784, 0.253, 0.080 and 0.431. These hypotheses are all supported but one of them is insignificant at a level of 0.01 when assessing their p-value, the hypothesis in question being H6. H1, H2, H3, H4, H5 and H7 are all on the other hand significant at a level of 0.001 when assessing their p-values.

Below is the conceptual model updated with the path coefficients established in Table 5.10
-342903086100Figure 5.3 – Conceptual model used in this study with path coefficients
3394711310515H4=0.784
00H4=0.784
202882537465Customer Engagement
Customer Engagement

333756013970001089660311150H1=0.638
00H1=0.638
1452880279403596005142240Repeat Behaviour
Repeat Behaviour

48615591784350090805317500Social Media
Marketing
0Social Media
Marketing
33864551587502028825318770Willingness to Donate
Willingness to Donate

486156064770H6= 0.080
00H6= 0.080
3385185121920H5=0.253
00H5=0.253
145161064770H2=0.405
00H2=0.405
1452880346710145288034670900
486283077470Brand Image
Brand Image

3642360113665H7=0.431
00H7=0.431
1184910151765H3=0.523
00H3=0.523
33864541136652024380260985Brand Awareness
Brand Awareness

Proposed Hypotheses with Academic and Organisational ImplicationsH1 = Social Media Marketing and Consumer Engagement
The relationship between Social Media Marketing and Consumer Engagement was assessed. The result shows that Social Media Marketing has a strong and significant effect on Consumer Engagement (? = 0.638, p – value = 0.001). This result indicates that Social Media Marketing could possibly explain approximatively 63.8 % of Green Purchase Intention. Thus it can be considered that the use of social media marketing from a NPO’s perspective will increase consumer engagement behaviour on social media.
H2 = Social Media Marketing and Willingness to Donate
The relationship between Social Media Marketing and Willingness to Donate was assessed. The result shows that Social Media Marketing has a relatively strong and significant effect on Willingness to Donate (? = 0.405, p – value = 0.001). This result indicates that Social Media Marketing could possibly explain approximatively 40.5 % of Willingness to Donate behaviour, which is less of an effect than Social Media Marketing’s effect on Consumer Engagement but is still considerable. Thus it can be considered that the use of social media marketing by an NPO will increase potential donors’ willingness to donate.

H3 = Social Media Marketing and Brand Awareness
The relationship between Social Media Marketing and Brand Awareness was assessed. The result shows that Social Media Marketing has a relatively strong and significant effect on Brand Awareness (? = 0.523, p – value = 0.001). This result indicates that Social Influence could possibly explain approximatively 52.3 % of Attitudes towards Green Appliances, which is more than Social Media Marketing’s effect on Willingness to Donate but less than for Consumer Engagement. Thus it can be considered that a NPO’s use of social media marketing will positively affect their brand awareness among its consumer-base.
H4 = Consumer Engagement and Repeat Behaviour
The relationship between Consumer Engagement and Repeat Behaviour was assessed. The result shows that Consumer Engagement has a strong and significant effect on Repeat Behaviour (? = 0.784, p – value = 0.001). This result indicates that Perceived Benefit could possibly explain approximatively 78.4 % of Repeat Behaviour intentions. Thus it can be considered that consumer engagement behaviour shown by potential donors on a NPO’s social media will positively increase the likelihood of repeat behaviour with these same consumers.

H5 = Willingness to Donate and Repeat Behaviour
The relationship between Willingness to Donate and Repeat Behaviour was assessed. The result shows that Willingness to Donate has a relatively strong and significant effect on Repeat Behaviour (? = 0.253, p – value = 0.001). This result indicates that Willingness to Donate could possibly explain approximatively 25.3 % of Repeat Behaviour, which is smaller than the effect Consumer Engagement has on Repeat Behaviour whilst still significant. Thus it can be considered that willingness to donate will result in an increase in repeat behaviour by those same potential donors, who will be more likely to donate in the future.

H6 = Repeat Behaviour and Brand Image
The relationship between Repeat Behaviour and Brand Image was assessed. The result shows that Repeat Behaviour has a supported, but weaker and insignificant effect on Brand Image (? = 0.080, p – value = 0.516). This result indicates that Repeat Behaviour could possibly explain approximatively 8 % of Brand Image, which demonstrates a relatively weak and insignificant relationship, whilst still positive.
H7 = Brand Awareness and Brand Image
The relationship between Brand Awareness and Brand Image was assessed. The result shows that Brand Awareness has a relatively strong and significant effect on Brand Image (? =0.431, p-value = 0.001). This result indicates that Brand Awareness could possible explain approximatively 43.1 % of Brand Image, which is much larger and more significant than the effect Repeat Behaviour has on Brand Image. Thus it can be considered that brand awareness is likely to increase brand image of a NPO in consumers’ minds.

Chapter 6: Discussion, Conclusion, Recommendations and Future ResearchDiscussion of the FindingsBelow we will be discussing each hypothesis accordingly to their findings, including whether or not the findings were expected or were in line with what other researchers have established.

H1: The relationship between Social Media Marketing and Customer Engagement is positive.

As stated above in Chapter 5, the relationship between the Social Media Marketing and Customer Engagement variables was found to be supported, positive and strong (? = 0.638, p – value = 0.001). This is in line with what was tested by Farook and Abeysekera (2016), which was based on Facebook usage and the factors that affected social media marketing and consumer engagement. This shows that the use of social media marketing should increase consumer engagement and thus should be utilised by any organisation, but is especially important for non-profit organisations whose source of revenue is limited compared to for-profit organisations.
H2: The relationship between Social Media Marketing and Willingness to Donate is positive.

The relationship between Social Media Marketing and Willingness to Donate was found to be supported, positive and relatively strong (? = 0.405, p – value = 0.001). This relationship was tested by both Hajli (2014) and Husnain and Hussain (2017), although not in the context of willingness to donate but rather intention to buy a product. Interestingly, results show that social media marketing should increase willingness to donate much like it should increase intention to buy. This information is vital for non-profit organisations since this proves that the utilisation of social media marketing should theoretically increase willingness to donate in consumers, which is directly linked to donation behaviour.

H3: The relationship between Social Media Marketing and Brand Awareness is positive.

The relationship between Social Media Marketing and Brand Awareness was found to be supported, positive and relatively strong (? = 0.523, p – value = 0.001). This relationship was tested by Hutter et al. (2013), through the use of Facebook as the media platform of importance in their case study; however using “Brand Page Commitment” instead of Social Media Marketing. Findings were expected, and were in line with the results published by Hutter et al., showing that non-profit organisations that utilise social media marketing will increase brand awareness for their consumers and potential donors.

H4: The relationship between Customer Engagement and Repeat Behaviour is positive.

The relationship between Customer Engagement and Repeat Behaviour was found to be supported, positive and strong (? = 0.784, p – value = 0.001), and was the strongest relationship tested in the study. This is in line with parts of Fernandes’ and Estevez’s (2016) study on the impact of consumer engagement on loyalty, although it is more relevant in a for-profit context compared to this study. Results thus indicate that through customer engagement it is more likely to have repeated behaviour in the form of either more customer engagement on social media, or even possibly donations.
H5: The relationship between Willingness to Donate and Repeat Behaviour is positive.

The relationship between Willingness to Donate and Repeat Behaviour was found to be supported, positive and relatively strong (? = 0.253, p – value = 0.001), and was the weakest relationship of some significance of the study. The relationship between willingness to donate and donation behaviour was tested by Wallace, Buil and de Chernatony (2017), and found that willingness to donate based on one’s self interests has a positive relationship with donation behaviour. H5 is thus a relationship that is in accordance with prior findings, and should give non-profit organisations valuable information on how to trigger repeated donation behaviour and other time-based donations.

H6: The relationship between Repeat Behaviour and Brand Image is positive.

The relationship between Repeat Behaviour and Brand Image was found to be supported but both weak and insignificant (? = 0.080, p – value = 0.516). These results were unexpected, since prior research led by Pappu and Quester (2006) found a positive relationship between customer satisfaction and perceived quality, which could be compared to the variables of this relationship. The difference in results might be based in the fact that either prior research was based on for-profit organisations, or that customer satisfaction and perceived quality are not comparable to repeat behaviour and brand image. Either way, these findings should allow non-profit organisations to learn that repeat behaviour is not necessarily correlated to an increase in brand image, and should thus be pursued independently of each other.

H7: The relationship between Brand Awareness and Brand Image is positive.

The last tested relationship was between Brand Awareness and Brand Image, which was found to be supported, significant and relatively strong (? =0.431, p-value = 0.001). Considering that both of these concepts are linked and are relative to consumer-perception, these results were expected, as was also established by Mudzakkir and Nurfarida (2015) in their research. These results thus indicate to non-profit organisations that brand awareness is likely to increase brand image in the minds of its consumer-base and potential donors.

Recommendations and Contributions of this Study
Results of this study indicated positive relationships for all hypotheses, although the relationship between repeat behaviour and brand image was found to be insignificant. This implies that consumers that repeatedly interact with the brand through either repeat donations or repeatedly engaging with the brand on social media, this will not necessarily result in an increase in brand image, due to the low significance of the relationship. For non-profit organisations, this means that repeat behaviour and brand image should be pursued separately and independently to achieve consistent results. This could be achieved by dividing focus between two lines of action, with consumer engagement and willingness to donate on one hand and brand awareness on the other hand, which would respectively lead to repeat behaviour and brand image. One positive for NPOs is that this has successfully established a weak and insignificant relationship between repeat behaviour and brand image, which assists non-profit organisations in better formulating their strategy, depending on what outcome would rather be accomplished.

This research paper should give non-profit organisations much needed information as to how they can reach and alter their brand image through the use of social media marketing, a subject that has not been researched extensively in the past. The study also makes a valid contribution theoretically, since it establishes relationships between several variables linked to the social media marketing predictor variable. The variables that were tested consisted of social media marketing, consumer engagement, willingness to donate, brand awareness, repeat behaviour and brand image. The aim was to find how social media marketing affected each of these variables using a conceptual model, and using brand image as the outcome variable.
This research paper should thus aid non-profit organisations in their strategy formulation for social media marketing, depending on what outcomes the NPO in question is striving to achieve. Information such as the relationships between each variables can be compared and contrasted to give a general idea on the effectiveness of social media marketing for any NPOs that want to target the South African market more extensively.
This study will also serve to give extensive information about the South African market, with a focus given to students between 18 and 30 years old, which consisted of the majority of respondents tested. This target segment is vital for any NPO trying to use social media marketing effectively, since it is this generation of individuals that uses social media most prominently, and thus should be considered in priority. These students are also going to move into the working population in the next five to ten years, and thus will start earning money whilst keeping their ideals, attitudes and values.
Limitations and Future ResearchThis study focused on targeting students of the University of the Witwatersrand, in Johannesburg. This means that this study might be of significantly less relevance in other cities in South Africa, and even less so in other cities outside of South Africa. This implies that this study is of particular relevance to NPOs wanting to target Johannesburg, and to a lesser extent other South African cities. Another limitation of this study is that its sample had a particularly small age range, due to the target individual being a university student. These individuals are not typically able to donate and interact with non-profit organisations, and in some cases were not interested in their social media activity and projects. These individuals will thus not be able to donate immediately but will have to wait perhaps several years before they start earning and potentially donating, depending on how long they pursue their studies. Individuals under the age of 18 and above the age of 40 were not targeted in this study, which implies that the study does not give an accurate representation of the South African market in its entirety.

Future research could thus extend to other segments of the South African population to gather a more accurate representation of attitudes and values in the country, and could also be extended to other cities and locations in South Africa, not just Johannesburg. Something else that could be undertaken in the future is having the study trialled in other countries around the world where non-profit organisations are having issues assessing whether social media marketing would prove a good addition to their strategy, and what outcomes would come of such a development. Finally, it would be interesting to realise some future studies which test different relationships between the variables, or with some new ones. This would give more information in establishing causal relationships which could greatly assist NPOs that are thinking of using social media marketing as an influencer for the market.
ConclusionThis study has established that the following relationships are positive and significant: social media marketing and consumer engagement, social media marketing and willingness to donate, social media marketing and brand awareness, consumer engagement and repeat behaviour, willingness to donate and repeat behaviour, and finally brand awareness and brand image. The only relationship without much significance was between repeat behaviour and brand image, but still remained positive. These findings imply that NPOs should be using social media marketing if they want to improve factors such as brand image, repeat behaviour regarding donations and consumer engagement, brand awareness and willingness to donate. In today’s society where social media is a platform that involves increasing amounts of individuals and where each individual is spending more and more time; social media marketing is a crucial element of an organisation’s strategy.

ReferencesAnheier, H. (2005). Nonprofit organizations. Oxon: Routledge.Eikenberry, A. and Kluver, J. (2004). The Marketization of the Nonprofit Sector: Civil Society at Risk? Public Administration Review, 64(2), pp.132-140.

Ashraf, M., Humayon, A., Ahmad, J. and Abbas, M. (2017). Three condensed importance of interfunctional communication for the acceptance of CRM system. Cogent Business & Management, 4(1).

Becker, Jan-Michael, Arun Rai and Edward E. Rigdon (2013), “Predictive Validity and Formative Measurement in Structural Equation Modeling: Embracing Practical Relevance,” Proceedings of the International Conference on Information Systems (ICIS).

Bekkers, R. and Wiepking, P. (2010). A Literature Review of Empirical Studies of Philanthropy. Nonprofit and Voluntary Sector Quarterly, 40(5), pp.924-973.

Bentler,P. (1990). Comparative fit indexes in structural models. Psychological Bulletin. 1990;107(2):238–246. doi: 10.1037/0033-2909.107.2.238.

Bekkers, R. and Wiepking, P. (2007). Understanding Philanthropy. Default Journal. Online Available at: https://www.rug.nl/research/portal/files/2814285/BekkersR-Philantropy-2007.pdf Accessed 4 Apr. 2018.

Bekkers, R. and Wiepking, P. (2010). A Literature Review of Empirical Studies of Philanthropy. Nonprofit and Voluntary Sector Quarterly, 40(5), pp.924-973.

Buechel, K. (n.d.). The Future of Philanthropy in Uncertain Times: Five Years of Lectures. Online Available at: http://philanthropy.gspia.pitt.edu/Portals/8/SpeakerSeries/FifthPublication_Web.pdf (Accessed 7 May 2018).

Business Tech (2017). How many people use Facebook, Twitter and Instagram in South Africa Online Available at: https://businesstech.co.za/news/internet/199318/how-many-people-use-facebook-twitter-and-instagram-in-south-africa Accessed 16 Mar. 2018
Cangur, S. and Ercan, I. (2015). Comparison of Model Fit Indices Used in Structural Equation Modeling Under Multivariate Normality. Journal of Modern Applied Statistical Methods, 14(1), pp.152-167.

Chaudhary, A. and Israel, G. (2014). The Savvy Survey #8: Pilot Testing and Pretesting Questionnaires. The Savvy Survey. Online Florida: IFAS, pp.1-6. Available at: https://edis.ifas.ufl.edu/pdffiles/PD/PD07200.pdf Accessed 2 Jul. 2018.

Chu, S. (2009). Determinants of Consumer Engagement in Electronic Word-of-Mouth in Social Networking Sites. Ph.D. The University of Texas.

Clark, M. and Melancon, J. (2013). The Influence of Social Media Investment on Relational Outcomes: A Relationship Marketing Perspective. International Journal of Marketing Studies, 5(4).

Clarke, S. (2010). Computational advancements in end-user technologies. Hershey, Pa.: IGI Global.

Dawson, R. (2009). Future of news and journalism – Framework by media futurist Ross Dawson. Online Ross Dawson. Available at: https://rossdawson.com/frameworks/social-media-strategy/ Accessed 16 Mar. 2018.

Dean, G., Fahsing, I., Gottschalk, P. and Saether, H. (2008). Investigative thinking and creativity: an empirical study of police detectives in Norway. International Journal of Innovation and Learning, 5(2), p.170.

Dervan, P. (2015). Social Networking Theory and the Rise of Digital Marketing in the Light of Big Data. Ph.D. University of Southampton.

DeVellis, R. (2003). Scale development: theory and applications: theory and application. Thousand Okas, CA.
Dickert, S., Sagara, N. and Slovic, P. (2010). Affective motivations to help others: A two-stage model of donation decisions. Journal of Behavioral Decision Making, 24(4), pp.361-376.

Farook, F. and Abeysekera, N. (2016). Influence of Social Media Marketing on Customer Engagement. International Journal of Business and Management Invention, 5(12), pp.115-125.

Fernandes, T. and Esteves, F. (2016). Customer Engagement and Loyalty: A Comparative Study Between Service Contexts. Services Marketing Quarterly, 37(2), pp.125-139.

Fioravante, P. (2010). Corporate Philanthropy: A Strategic Marketing Consideration. Journal of Applied Business and Economics, 11(3).

Fornell and Larker. (1981). Evaluating Structural Equation Models with Unobservable Variables and Measurement Error. Journal of Marketing.
Glewwe, P. and Levin, M. (2005). Presenting simple descriptive statistics from household survey data. Household Sample Surveys in Developing and Transition Countries. Online pp.335-354. Available at: https://unstats.un.org/unsd/hhsurveys/pdf/chapter_16.pdf Accessed 2 Jul. 2018.

Godey, B., Manthiou, A., Pederzoli, D., Rokka, J., Aiello, G., Donvito, R. and Singh, R. (2016). Social media marketing efforts of luxury brands: Influence on brand equity and consumer behavior. Journal of Business Research, 69(12), pp.5833-5841.

Godwin, M., Pike, A., Bethune, C., Kirby, A. and Pike, A. (2013). Concurrent and Convergent Validity of the Simple Lifestyle Indicator Questionnaire. ISRN Family Medicine, 2013, pp.1-6.

Goswami, I. and Urminsky, O. (2016). When Should the Ask Be a Nudge? The Effect of Default Amounts on Charitable Donations. Journal of Marketing Research: October 2016, Vol. 53, No. 5, pp. 829-846.

Hajli, M. (2014). A study of the impact of social media on consumers. International Journal of Market Research, 56(3), pp.387-404.

Hatcher, L. and O’Rourke, N. (2014). A step-by-step approach to using SAS for factor analysis and structural equation modeling. Cary, NC: SAS Institute.

Helmig, B., Jegers, M. and Lapsley, I. (2004). Challenges in Managing Nonprofit Organizations: A Research Overview. VOLUNTAS: International Journal of Voluntary and Nonprofit Organizations, 15(2), pp.101-116.

Hooper, D., Coughlan, J. and Mullen, M. (2008). Structural Equation Modelling: Guidelines for Determining Model Fit. Electronic Journal of Business Research Methods, 6(1), pp.53-60.

Huang, C., Wang, Y., Wu, T. and Wang, P. (2013). An Empirical Analysis of the Antecedents and Performance Consequences of Using the Moodle Platform. International Journal of Information and Education Technology, 3(2), pp.217-221.

Hutter, K., Hautz, J., Dennhardt, S. and Füller, J. (2013). The impact of user interactions in social media on brand awareness and purchase intention: the case of MINI on Facebook. Journal of Product ; Brand Management, 22(5/6), pp.342-351.

Johnston, G. (2015). Marshall McLuhan’s Ideas Applied to Social Media. Blog McLuhan Galaxy. Available at: https://mcluhangalaxy.wordpress.com/2015/11/14/marshall-mcluhans-ideas-applied-to-social-media/ Accessed 7 May 2018.

Jones, J., Willimack, D., Snijkers, G. and Haraldsen, G. (2013). Designing and Conducting Business Surveys. Wiley, pp.459-504.

De Jonge, E. and van der Loo, M. (2013). An introduction to data cleaning with R. online The Hague: Statistics Netherlands. Available at: https://cran.r-project.org/doc/contrib/de_Jonge+van_der_Loo-Introduction_to_data_cleaning_with_R.pdf Accessed 2 Jul. 2018.

Kang, H. (2013). The prevention and handling of the missing data. Korean Journal of Anesthesiology, 64(5), p.402.

Karahanna, E. Xu, S. & Zhang, A. (2015) Psychological Ownership Motivation and Use of Social Media, Journal of Marketing Theory and Practice, 23:2, 185-207, DOI: 10.1080/10696679.2015.1002336
Kline, R. (1998). Principles and practice of structural equation modeling. NY: Guilford Press; 1998.

Konrath, S. (2013). The Power of Philanthropy and Volunteering. Interdisciplinary Program on Empathy and Altruism Research, 6(11), pp.387-428.

Leliveld, M. and Risselada, H. (2017). Dynamics in charity donation decisions: Insights from a large longitudinal data set. Science Advances, 3(9), p.e1700077.

Leskovec, D. (2010). How to Collect Donations: Conceptual Review and Implications for Online Non-Profit Information Goods Providers. Akademija MM, 10(15), pp.55-68.
Lovejoy, K. and Saxton, G. (2012). Information, Community, and Action: How Nonprofit Organizations Use Social Media*. Journal of Computer-Mediated Communication, 17(3), pp.337-353.

MacCallum, R., Browne, M. & Sugawara, H. (1996). Power analysis and determination of sample size for covariance structure modeling. Psychological Methods. 1996;1(2):130–149. doi: 10.1037/1082-989X.1.2.130.

McKenzie, S. (2013). McLuhan’s Relevance in Today’s Society: A Look at Social Media on Mobile Devices. Public Relations – Theses. Paper 2
McLeod, S. (2008). Qualitative vs Quantitative Data | Simply Psychology. Online Simplypsychology.org. Available at: https://www.simplypsychology.org/qualitative-quantitative.html Accessed 2 Jul. 2018.

Meer, J. (2011). Brother, can you spare a dime? Peer pressure in charitable solicitation. Journal of Public Economics, 95(7-8), pp.926-941.

Michaelidou, N., Siamagka, N. and Christodoulides, G. (2011). Usage, barriers and measurement of social media marketing: An exploratory investigation of small and medium B2B brands. Industrial Marketing Management, 40(7), pp.1153-1159.

Miller, D. (2014). How Nonprofits Make the Ask: Framing Donation Requests. Blog Insights by Stanford Business. Available at: https://www.gsb.stanford.edu/insights/how-nonprofits-make-ask-framing-donation-requests Accessed 7 May 2018.

Miller, K. (2012). The Theory of Change Behind Nonprofit Marketing – Kivi’s Nonprofit Communications Blog. Online Kivi’s Nonprofit Communications Blog. Available at: https://www.nonprofitmarketingguide.com/blog/2012/12/04/the-theory-of-change-behind-nonprofit-marketing/ Accessed 16 Mar. 2018.

Minton, E., Lee, C., Orth, U., Kim, C. and Kahle, L. (2012). Sustainable Marketing and Social Media. Journal of Advertising, 41(4), pp.69-84.

Moss, S. (2016). Fit indices for structural equation modeling. Online Sicotests.com. Available at: https://www.sicotests.com/psyarticle.asp?id=277 Accessed 14 Apr. 2018.

Moss, T., Lawson, V. and White, P. (2015). Identification of the underlying factor structure of the Derriford Appearance Scale 24. PeerJ, 3, p.e1070.

Mudzakkir, M. and Nurfarida, I. (2015). The Influence of Brand Awareness on Brand Trust Through Brand Image. Online pp.609-615. Available at: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2670597 Accessed 10 Jun. 2018.

Newsom (2018). Some Clarifications and Recommendations on Fit Indices. Online Available at: http://web.pdx.edu/~newsomj/semclass/ho_fit.pdf Accessed 2 Jul. 2018.

Novatorov, E. (2014). Nonprofit Marketing: Theory Triangulation. SSRN Electronic Journal.

O’Connor, T. (2016). Improving Nonprofit Retention through Dialogue. Masters. Gonzaga University.

Pan, B. and Crotts, J. (2012). Theoretical models of social media with marketing implications. Online Available at: https://www.researchgate.net/publication/254862367_Theoretical_models_of_social_media_with_marketing_implications Accessed 7 May 2018.

Pansari, A. (2016). A Road to Engagement. Ph.D. Georgia State University.

Pappu, R. and Quester, P. (2006). Does customer satisfaction lead to improved brand equity? An empirical examination of two categories of retail brands. Journal of Product ; Brand Management, 15(1), pp.4-14.

Paquette, H. (2013). Social Media as a Marketing Tool: A Literature Review. Textiles, Fashion Merchandising and Design. Online Available at: http://digitalcommons.uri.edu/cgi/viewcontent.cgi?article=1001;context=tmd_major_papers Accessed 7 May 2018.

Pfeffer, J., Zorbach, T. and Carley, K. (2013). Understanding online firestorms: Negative word-of-mouth dynamics in social media networks. Journal of Marketing Communications, 20(1-2), pp.117-128.

Poorrezaei, M. (2016). Customer Engagement: Conceptualisation, Measurement and Validation. Ph.D. University of Salford.

Pope, G. (2009). Psychometrics 101: Item Total Correlation. Online. Accessed April 15th 2017. From: https://blog.questionmark.com/168 .

Praekelt.org. (n.d.). Home. Online Available at: https://www.praekelt.org/ Accessed 16 Mar. 2018.

Qwerty Digital (2017). THE DIGITAL LANDSCAPE IN SOUTH AFRICA 2017 – A data driven look at South Africa’s relationship with digital.. Online p.11. Available at: https://qwertydigital.co.za/wp-content/uploads/2017/08/Digital-Statistics-in-South-Africa-2017-Report.pdf Accessed 16 Mar. 2018.

Reinstein, D. and Riener, G. (2011). Reputation and influence in charitable giving: an experiment. Theory and Decision, 72(2), pp.221-243.

Saravanakumar, M. and SuganthaLakshmi, T. (2012). Social Media Marketing. Life Science Journal, 9(4), pp.4444-4451.

Salamon, L. and Anheier, H. (1992). In search of the non-profit sector. I: The question of definitions. Voluntas, 3(Salamon and Anheier, 1999), pp.125-151.

Salamon, L. and Anheier, H. (1999). Defining the nonprofit sector. Manchester u.a.: Manchester Univ. Press, p.1.
Saunders, M., Lewis, P. and Thornhill, A. (2009). Understanding research philosophy and approaches to theory development. 1st ed. Harlow: Pearson, pp. 122-161.

Snipes, R. and Oswald, S. (2010). Charitable giving to not-for-profit organizations: factors affecting donations to non-profit organizations. Innovative Marketing, Vol. 6, No. 1, pp. 73-80
Stuart, L. (2013). The South African Nonprofit Sector: Struggling to Survive, Needing to Thrive. Online NGO Pulse. Available at: http://www.ngopulse.org/article/south-african-nonprofit-sector-struggling-survive-needing-thrive Accessed 16 Mar. 2018.

Subhani, M. and Osman, A. (2016). A Study on the Association Between Brand Awareness And Consumer/Brand Loyalty For The Packaged Milk Industry In Pakistan. Online Available at: https://mpra.ub.uni-muenchen.de/21367/1/ Accessed 16 Mar. 2018.

Sutton, J. and Austin, Z. (2015). Qualitative Research: Data Collection, Analysis, and Management. The Canadian Journal of Hospital Pharmacy, 68(3).

Tavakol, M., Dennick, R. (2011). Making sense of Cronbach’s alpha. International Journal of Medical Education. 2:53-55.
Toor, A., Husnain, M. and Hussain, T. (2017). The impact of social network marketing on consumer purchase intention in Pakistan: Consumer engagement as a mediator. Asian Journal of Business and Accounting, 10(1), pp.167-199.

TrendsStream Limited. 2010. “Global Web Index.”
Ullman, J. B., ; Bentler, P. M. (2013). Structural equation modeling. In J. A. Schinka, W. F. Velicer, ; I. B. Weiner (Eds.), Handbook of psychology: Research methods in psychology, pp. 661-690. Hoboken, NJ, US: John Wiley ; Sons Inc.

Van der Heijden, H. and Verhagen, T. (2004). Online store image: conceptual foundations and empirical measurement. Information ; Management, 41(5), pp.609-617.
Vinerean, S., Cetina, I., Dumitrescu, L. and Tichindelean, M. (2013). The Effects of Social Media Marketing on Online Consumer Behavior. International Journal of Business and Management, 8(14), pp.66-79.

Vosloo, J. (2014). A sport management programme for educator training in accordance with the diverse needs of South african schools. Ph.D. North-West University.

Uses of Social Media in Public Transportation. (2012). Online pp.6-12. Available at: https://www.nap.edu/catalog/14666/uses-of-social-media-in-public-transportation Accessed 16 Mar. 2018.

Wallace, E., Buil, I. and de Chernatony, L. (2017). When does “liking” a charity lead to donation behaviour? European Journal of Marketing, 51(11/12), pp.2002-2029.

White, D., Korotayev A., Khaltourina D. (2004). How to Analyze a Correlation Matrix. Chapter x.

Wolf, E., Harrington, K., Clarke, S. and Miller, M. (2013). Sample Size Requirements for Structural Equation Models. Educational and Psychological Measurement, 73(6), pp.913-934.

Yao, K. (2015). Who Gives? The Determinants of Charitable Giving, Volunteering and their Relationship. Wharton Research Scholars, 126.

Zalaghi, H. and Khazaei, M. (2016). The Role of Deductive and Inductive Reasoning in Accounting Research and Standard Setting. Asian Journal of Finance ; Accounting, 8(1), pp.23-37.

Zhang, Y. (2015). The Impact of Brand Image on Consumer Behavior: A Literature Review. Open Journal of Business and Management, 03(01), pp.58-62.