Abstract
Marketing research targeting poor consumers has been developing recently. The specific characteristics of these consumers make it difficult to conduct empirical research. This difficulty has raised the question of how to adapt the methods and techniques usually used in marketing. This article provides a multidisciplinary review of the difficulties encountered in addressing people living in poverty and enables researchers to assess this population and anticipate their needs. In addition, based on a literature review of the methods used in 83 empirical marketing studies published between 1990 and 2021, our study provides researchers with a vade-mecum to guide them in their choice of research design. Concrete and detailed recommendations are provided to facilitate their involvement in the field of poor consumers.
Introduction
With the exception of a few researchers, such as Caplovitz (1963), Andreasen (1975) and Hill (from 1990 1 ), marketing researchers have only become interested in the issue of poverty relatively recently. It was mainly from the 2000s that the amount of research increased and driven by three research streams: (1) the multidisciplinary Bottom of the Pyramid (BOP) stream, (2) the Subsistence Markets (SMts) stream and (3) the Transformative Consumer Research (TCR) stream (Blocker et al., 2013; Gorge et al., 2015; Mick, 2006).
The consumers who were traditionally studied in marketing, namely, the ‘have’ – and the more recently studied population, namely, poor consumers – the ‘have-nots’ – (Hill and Gaines, 2007: 81) differ in their purchasing decision processes and their cultural codes (Crockett et al., 2013; Ingenbleek et al., 2013). The literature also describes the low level of literacy among poor consumers, and therefore, their limited abstraction, reading and calculation skills (Crockett et al., 2013; Gau et al., 2012; Hamilton and Catterall, 2006; Hill and Adrangi, 1999; Ingenbleek et al., 2013) in addition to their social or geographical isolation (Crockett et al., 2013; Ingenbleek et al., 2013). As with other vulnerable consumer populations, literature studies highlight this group of consumers’ particular reluctance to be interviewed (Bordiec and Marnet, 2017; Crockett et al., 2013; Gau et al., 2012; Hill and Stamey, 1990; Hill, 1995; Ozanne and Anderson, 2010), probably because they feel stigmatised (Piacentini and Hamilton, 2013) and have a fear of being confronted with their low literacy levels.
These specific features give rise to a number of difficulties that hinder empirical research concerning poverty and could even discourage it. Our research aims to help marketing researchers discover more knowledge concerning poverty in light of these difficulties. This work is important for two reasons. While poverty remains widespread throughout the world, marketing research on poverty is likely to help reduce it. This idea is indeed the goal of studies of offers or services designed specifically for poor markets for the BOP stream (Prahalad and Hart, 2002), the study of the characteristics of markets that help or hinder the process of satisfying basic consumption needs for the SMts stream (Viswanathan et al., 2012), and/or the way in which consumption experiences shape the individual and collective well-being of poor consumers for the TCR stream (Blocker et al., 2011, 2013). Moreover, the knowledge about poverty that our research produced is new. Indeed, other Human and Social Sciences (HSS), such as sociology (Simmel, 1998) and economics (Banerjee and Duflo, 2009), have been studying the phenomenon of poverty for a long time. However, with the exception of a few works, such as those by Lazarus and Colombi in France (Colombi, 2020; Lazarus, 2006, 2018), other areas of social sciences and humanities do not focus much on the role of consumption and the market in the lives of poor people, even though they have a central role in consumption (Blocker et al., 2011; Prahalad, 2006).
The article is organised into several sections. First, we detail the difficulties associated with the empirical study of poverty as described by the researchers who have experienced such difficulties. In line with the recommendations of Borraz et al. (2021) and Ingenbleek et al. (2013), this initial survey, which was carried out from a multidisciplinary perspective, should enable marketing researchers who have not yet or have only rarely worked in the field of poverty to assess and anticipate the difficulties they will encounter while studying this topic. On the basis of a literature review (Littell et al., 2008) of the research methods, techniques and practices used in marketing over more than 30 years (1990–2021), we then present the way in which marketing researchers produce knowledge on poverty. Apart from the fact that no state-of-the-art study integrating the three aforementioned streams of research has been conducted, our results may constitute an initial source of inspiration for those wishing to engage in the empirical study of poverty. We also used some of the research identified in this empirical literature review to illustrate the vade-mecum for marketing researchers that we propose in the third part of the article.
This vade-mecum includes questions to guide the choice of a research design and concrete recommendations for approaching this field. It was designed based on a multidisciplinary perspective so as to benefit not only from the empirical experience of marketing in the field of poverty as we have identified it but also from research in other human and social sciences, such as sociology and economics.
Difficulties in implementing the poverty initiative: A multidisciplinary perspective
The specific characteristics of poor consumers that are highlighted in the introduction are not the only reason for the obstacles encountered by researchers in the empirical study of poverty. These characteristics are combined with other factors, in particular the complexity of the phenomenon of poverty, which is widely emphasised in all disciplines (Banerjee and Duflo, 2009; Davis et al., 2016), the harshness of poverty situations (Besozzi, 2021; Mercier, 1995), and the marked differences in the living conditions and cultural codes of the researchers and poor people (Blocker et al., 2013; Chakravarti, 2006; Downey et al., 2007; Hamilton et al., 2006). Researchers face numerous difficulties of various kinds when attempting to study poor people. Some difficulties are related to research methods. For example, economists have come up against the multiple causal chains at the root of poverty when assessing the impact of development programmes (Banerjee and Duflo, 2009) while researchers using laboratory experiments have found it virtually impossible to reproduce situations of poverty. Other difficulties relate to the poverty field itself. These difficulties are scattered throughout the multidisciplinary literature, and we sought to identify and classify them so as to offer an integrative reading. We identified four types of difficulties (Figure 1), which we described in detail: (1) conceptually and operationally defining who to interview; (2) obtaining access to the targeted field; (3) brushing with poor people (Blanchet, 2022), particularly for qualitative researchers; and (4) finally, dealing with the low level of literacy and reluctance of these poor consumers to express themselves.
Difficulties when studying poverty.
Defining poverty conceptually and operationally
Conceptually defining the object of the research in question and operationalising this definition are essential steps in any empirical research project. These steps are recommended by Farrell and Hill (2018) for identifying the target of empirical research on poverty because frames of reference ‘dictate the way [research] is conducted and the conclusions [researchers] draw from it’ (Farrell and Hill, 2018: 772) 2 . However, when discussing poverty, this approach encounters two difficulties that are partly linked.
The first difficulty is defining poverty conceptually. It has been revealed by the many debates that have stimulated economists and sociologists (Bisiaux, 2011; Bresson, 2007; Mercier, 1995), the co-existence of multiple definitions within their discipline, and above all, the multidimensional nature of these definitions if we exclude the neo-classical monetary approach of poverty (Lewis and Ulph, 1988; Ravallion, 1992). For example, among economists, Sen (1981, 1983, 1985) defines poverty as a lack of capabilities in the sense of capacities to be but also capacities to do, and Rawls (1971, 1982) as a lack of ‘primary goods’, such as freedom, equality, health and the social foundations of self-respect. Among sociologists, Simmel (1998) sees poverty as relational and relative and therefore, dependent on social groups, while Bourdieu (1993) defines it as both a misery of condition that is linked to insufficient financial resources, and a misery of position associated with a relative lowering of status or social role.
The marketing discipline is no exception to this difficulty. Researchers within this discipline use a plurality of definitions for poverty (Gorge and Özçağlar-Toulouse, 2017). Recently, some authors have formalised conceptual reflections or definitions of poverty from a disciplinary perspective (Farrell and Hill, 2018; Venugopal and Viswanathan, 2017; Yurdakul et al., 2017). These definitions differ but are all multidimensional. For example, proponents of the SMts stream see poverty ‘as a subjective experience, structured by a complex and variable concatenation of factors that generate a threat to the satisfaction of basic consumption needs’ (Venugopal and Viswanathan, 2017: 351). Other researchers mobilise the intersectional approach to poverty, which considers the ‘multiple and overlapping marginalisations at individual and institutional levels within the same social group experiencing poverty (Saatcioglu and Corus, 2014: 123). Thus, multiple experiences of poverty may be experienced within the same group depending on the gender or ethnicity of its members. Some poor consumers are thus marginalised in a ‘multiplicative way’ (p. 123) and also in a lasting way.
This first difficulty in conceptualising poverty gives rise to the second, namely the difficulty in operationalising a conceptual definition of poverty if multidimensional. This idea is demonstrated by studies in economics to identify poor people as reported by Bisiaux (2011) in Delhi (India) using Rawl’s (1971, 1982) and Sen’s (1981, 1983, 1985) definitions of poverty and by the more recent marketing proposals by Farrell and Hill (2018). After constructing a three-dimensional conceptual definition of poverty, the latter provides a precise methodological guide for operationalising it (Farrell and Hill, 2018: 786). The operationalisation work to be conducted appears very substantial as it includes one or two qualitative research studies, a questionnaire survey and an analysis of secondary data.
Access to the poverty field
Access to the poverty field is understood in a broad sense from the individual identification of poor people to be interviewed or observed to obtaining agreements to conduct interviews or surveys. Difficulties exist in both emerging (EC) and industrialised countries (IC), although they do not necessarily arise in the same way or for the same reason. Researchers are initially hampered by the lack of data on the target populations or databases containing their details. In defining sampling strategies and targets, development economists have come up against the absence or inadequate updating of statistics on EC populations, for example on education, work and access to services. Some states do not have sufficient resources to collect or update such data, while others do not wish to let evaluate their public policies (Banerjee and Duflo, 2009). Similarly, in some ECs, few or no up-to-date databases providing contact details for all or part of the population are available. This obstacle is not specific to the ECs. It echoes the difficulty encountered by Demazière and Zune (2019) in identifying their target for their research on into the experiences of exclusion for unemploymed in an IC. The Belgian citizens who made up their empirical target were in fact administratively removed from any institutional affiliation and the social assistance, and employment databases no longer contained any data concerning them. ‘Invisibility’ constitutes an obstacle to access to the population’ (Demazière and Zune, 2019: 9).
Being able to compile a list of names of people to be interviewed or to identify a community precisely does not, however, reduce all obstacles to accessing the field. Researchers may encounter material problems. Some EC populations are not very settled (Thomas et al., 2001) or are even nomadic (Ingenbleek et al., 2013). In the ICs, the local environments of certain neighbourhoods constitute physical barriers to encountering and meeting poor people. For example, in her sociological work on pathways out of poverty, Auzuret (2017) was unable to contact the interviewees she had identified via the French Family Allowance Fund; she had their names and addresses but was unable to locate them on intercoms and letterboxes.
Finally, whatever the discipline, the literature is unanimous in pointing out that when researchers do manage to have contact with the people to be interviewed, they have difficulty in obtaining interviews (Crockett et al., 2013; Ingenbleek et al., 2013; Ozanne and Anderson, 2010). Of course, this reluctance is not specific to research on poverty and is also particularly marked in other sensitive areas (Hill, 1995).
Brushing with poor people, particularly in the case of qualitative research 3
Ethnologists of poverty often stress the difficulty in collecting data when they are out in the field. For example, Besozzi (2021, 2022) explains how he was constantly regarded as an intruder by some homeless people during his ethnographic immersion. Lepoutre (2001) recounted the frequent refusals and terse responses he received from young people on housing estates during his ethnographic work, ‘so many ways of resisting him when he tried to impose a relationship that was too unequal’ (Lepoutre, 2001: 91). In his study, data collection was hampered by the asymmetry of position between researchers and poor people as widely pointed out by sociologists and ethnologists. It is sometimes in favour of the former, sometimes in favour of the latter. ‘The survey shapes processes that weaken and strengthen both the interviewer and the interviewee’ (Blanchet, 2022: 45).
Data collection is also made difficult by the researcher’s vulnerability to poverty. Poverty leaves no one indifferent (Cortambert, 2022; Mercier, 1995). Confronting poverty arouses strong emotions, which are highlighted in the literature in all disciplines. These emotions are particularly strong for street ethnologists. Besozzi (2021) describes ‘a form of disgust mixed with pity’ when confronted with the uncomfortable and unhygienic conditions of his informants. Cortambert (2022) recounts shedding tears when meeting families with young children or pregnant women who were sleeping outdoors. These emotions are not the only prerogative of immersive investigations. Bashir (2018) evokes them through the lens of the narrative of six emblematic interviews conducted in the homes of people living in poverty. In marketing, Jafari et al. (2013) describe them on the basis of their survey experiences in which feelings of anxiety and sadness coexist with feelings of frustration, guilt and grief during the research process. These emotions are sources of vulnerability for researchers (Downey et al., 2007; Jafari et al., 2013) as is their insertion into social spaces that sometimes contradict their own symbolic or even ideological universe (Hamilton et al., 2006). These emotions entail psycho-social risks for the researcher and are sometimes compounded by physical risks (Hamilton et al., 2006), which causes disconnection from the field.
Finally, researchers encounter another difficulty when confronted with poverty, namely having to deal with new ethical dilemmas. Cortambert (2022) points out that she disguises her status in relation to the homeless people on whom she is conducting ethnographic research and does not make her research approach explicit to them so as to gain their acceptance. This ‘non-transparency’ (p. 63) may raise ethical questions. Bashir (2018) explains how the vulnerability of certain interviewees increases with the survey. Should we continue? Furthermore, some researchers develop empathic relationships with interviewees (Grignon and Passeron, 1989), triggering, for example, the desire to help after an interview. It can be asked whether offering such help is the role of the researcher. Furthermore, doesn’t such help run the risk of create a situation of dependency between the interviewee and the researcher. What should be done?
Dealing with low literacy levels and the reluctance of poor consumers to express themselves
The literature highlights the difficulties that poor consumers have in expressing themselves during interviews. Two main reasons for these difficulties exist. Some poor people experience a feeling of non-recognition (Beaud et al., 2006) and do not feel up to the task (Beaud and Weber, 1997). In such cases, they prefer not to accept interviews or to answer tersely because they do not feel capable of providing relevant answers or are afraid of being scorned (Beaud et al., 2006). Moreover, poverty is often associated with illiteracy. In fact, more than half of the 750 million men and women in the world who are unable to read and write live in South Asia (49%) and more than a quarter in sub-Saharan Africa 4 (UNESCO, 2016). The rate of illiteracy is over 50% in 20 developing countries. This rate is most often due to a lack of schooling; however, some people who have been in school for a long time cannot read or write enough to be fully independent in everyday life, for example, when filling in forms. These illiterate people are, of course, very well represented among the population in the ECs, but are significant numbers in the ICs among the poor are also present. For example, the French National Agency for the Fight Against Illiteracy has estimated their number at 2.5 million inhabitants 5 and stresses that ‘the poverty map and the illiteracy map overlap perfectly’ (Le Paon and Morange, 2018: 132).
Not surprisingly, the difficulties encountered by these poor people also constitute obstacles for researchers working in the field of poverty. The most obvious difficulty is the impossibility of administering self-administered questionnaires (Gau et al., 2012; Ingenbleek et al., 2013). However, other barriers also exist; people with low literacy skills have a limited vocabulary with which to express themselves in interviews, for example (Gau et al., 2012). Some research suggests the impact of illiteracy on people’s ways of thinking and cognitive architecture (David, 2015). Researchers then come up against the interviewees’ difficulties in abstracting (Viswanathan et al., 2008) and their concrete thinking based on a single rather than plural pieces of information (Gau et al., 2012). Finally, this population’s difficulties with reading and writing are often associated with identical difficulties with numbers (Feigenson et al., 2004). The limited level literacy as defined by UNESCO is also an obstacle to the use of standard measurement scales in research (Ingenbleek et al., 2013).
This multidisciplinary state-of-the-art report discusses the main difficulties encountered by social science and marketing researchers in empirically addressing poverty. Our report enables marketing researchers with little or no experience in this field to anticipate such difficulties. In the next section, we identify and analyse the research methods, techniques and practices used by marketing researchers in the difficult field of poverty based on a review of empirical research conducted in marketing between 1990 and 2021 with a view to establishing guidelines that we call a vade-mecum.
A review of the methods and techniques used in empirical marketing research on poverty (1990–2021): Methodology and data
A detailed presentation of the processes and criteria used to collect and analyse the bibliographical data we have selected is given below.
Collection and description of bibliographic data
The database was compiled in two main stages. The first consisted of identifying articles potentially relevant to our target, which we did by successively using three selection criteria, each of which corresponded to the presence of keywords (Appendix 1). With the first criterion, we sought to specifically select research on poverty (criterion #1). We found that poor consumers were described using a variety of terms in the literature, sometimes even within the same article (Gupta and Srivasta, 2016; Hill, 2008). These descriptions led us to compile as comprehensive a list as possible of the terms used to describe consumers in the marketing literature. We compiled it starting from the founding articles in marketing research concerning poverty (Blocker et al., 2013; Hill, 2008; Hill and Stamey, 1990), state-of-the-art and/or retrospective articles on the BOP and SMts currents (Dembeck et al., 2020; Venugopal and Viswanthan, 2017), reflections on the intersectionality paradigm (Corus et al., 2016) and the concept of vulnerability (Baker et al., 2005), which are sometimes used to understand poverty. The terms used to describe poor people are listed: (1) low-income, (2) low income, (3) poverty, (4) poor, (5) low socioeconomic status, (6) bottom of the pyramid, (7) base of the pyramid, (8) subsistence, (9) low socioeconomic class, (10) emerging countries, (11) emerging market economies, (12) underserved, (13) disadvantaged, (14) marginalised working class, (15) economic vulnerability, (16) economically vulnerable, (17) BOP, (18) impoverished and (19) homeless. Using the other two selection criteria, we sought to select those of the articles identified above that reflected empirical research (criterion #2) conducted in the marketing discipline (criterion #3).
Searches were identified using the Business Source Premier database. We used other databases, in particular Science Direct, for articles that the Business Source Premier database returned in keyword searches but did not make available in their full format. We chose the period 1990–2021 as the study period. The seminal article on poverty by Ronald Paul Hill, who is one of the pioneering marketing researchers on the subject, was published in 1990. This first stage in the creation of the database resulted in the identification of 1,065 articles.
The next step was to sort these articles. The first exclusion criterion was the publication journal. Many articles published in journals not referenced in the Harzing list of academic marketing journals (Journal Quality List 2021) were excluded. This list, already used in state-of-the-art studies in marketing (Chabowski and Mena, 2017), guarantees the quality of the journals without being too restrictive, which was essential given the relatively recent nature of research on poor consumers in the discipline. We then excluded a much smaller number of articles that had not given rise to an empirical study of poor consumers but were present in the database despite selection criteria #1 and #3. Eighty-six articles written in English were then found to be eligible for analysis. Four of them were excluded because the full text was not available. One article was added following a specific search of the two French-language academic marketing journals (Appendix 1). The resulting database contains 83 articles from 23 journals. The number of publications is increasing over time as more than half of the articles (44) were published in the last 6 years (2016–2021).
Content analysis of articles
We performed a content analysis of the articles as had been done in previous reviews (Andriopoulos and Slater, 2013; Çelik et al., 2023; Eryigit, 2017). We began by compiling an analysis grid of the criteria usually used in the literature to carry out state-of-the-art studies from a methodological perspective for qualitative, quantitative and mixed research. We then supplemented our analysis with criteria specific to our research objective that was identified in particular on the basis of our literature review on the difficulties associated with the empirical study of poverty. The list of these second criteria can be found in Appendix 2.
Each of the two authors analysed eight different articles on qualitative, quantitative and mixed research. Successive pooling of the analyses made it possible to resolve differences in interpretation of the article content, create a common frame of reference for interpretation (Aguinis et al., 2009) and develop the analytical grids. All articles were then analysed using double coding. This analysis phase involved reading the articles in their entirety, not just the methods section. The search for a conceptual definition of poverty, for example, meant reading the introduction and the literature review of the articles and searching for the theoretical frameworks used to understand poverty.
Poverty marketing research methods, techniques and practices (1990–2021): results
More than two-thirds of empirical marketing research on poor consumers (60 out of 83) is conducted in emerging countries (EC) with India representing by far the largest research area (27 out of 83). The vast majority of these studies (51 out of 83) are qualitative, and almost one-third (27 out of 83) are quantitative. Research using a mixed research design is a large minority (5 out of 83). We do not comment further on this division between these three types of methodological approach or on sampling strategies for quantitative studies. Our focus was on the methods, techniques and practices used by researchers to examine poverty.
In this respect, our analysis led us to highlight that marketing researchers produce knowledge: (1) with or about poor consumers, (2) from targets or samples rooted in specific situations of poverty or a-situational and (3) by giving a differentiated role to third parties in the conduct of the research. However, we are in no way presuming the way in which these choices were made, probably upstream of the research in most cases, but possibly along the way as the practice of research does not always follow a totally linear temporal trajectory.
Research methods and techniques: Producing knowledge about or with poor consumers
In our database, we distinguish between two types of methods that depended on whether or not the researchers give consumers an active role in the research. In one case, the researchers used the methods and techniques usually employed in marketing to understand the behaviour of ‘have’ consumers. They sometimes adapt this first set of methods to ‘have-nots’, in particular to deal with the latter’s low level of literacy; in other words, a researcher who alone develops knowledge that sheds light on a distanced phenomenon, and therefore produces knowledge about poor consumers. In the second case, which is much less common in our database, researchers opt for a participatory method, namely, cooperation or co-construction of research with consumers. This method is certainly not specific to the field of poverty; however, what distinguishes it from the methods mentioned above and also seems interesting to us from the point of view of the poverty experience is that the researcher produces knowledge that actually involves the poor consumers.
Producing knowledge about poor consumers: Replicating and adapting standard methods and techniques
This standard research logic concerns all quantitative research and 46 of the 51 qualitative research studies. For each of these two types of research, we present the protocols and data collection techniques used, their possible adaptations to the field of poverty, and/or their specificities.
For quantitative research, the researchers used techniques commonly used in marketing. In the majority of cases (25 out of 27 studies), data were collected via questionnaire surveys based on commonly used marketing protocols. Researchers sometimes adopt these survey materials or methods to poor consumers. Some adjust their questionnaires or the format of the measurement scales to match the consumers’ low level of literacy (Table 1).
Adjustment of data collection materials to the limited literacy of poor consumers.
Other researchers conduct their surveys face-to-face rather than self-administered, arguing that this compensates for limited consumer literacy (Hasan et al., 2019, 2020; Venugopal et al., 2015). This method of administering questionnaires was relatively common in our database (22 out of 32 quantitative studies and quantitative parts of mixed studies). Although the reasons for this choice are rarely explained by the authors, we can hypothesise that in a number of these cases, face-to-face questionnaires are a response to the difficulties posed by the low literacy levels of poor consumers. Indeed, surveys are now most often administered in the form of self-administered surveys in quantitative marketing research.
Only two of the quantitative studies in which researchers produced knowledge about poor consumers stand out, one because of the data collection technique used (Soman and Cheema, 2011) and the other because of the research protocol (Viswanathan et al., 2021e). Soman and Cheema (2011) observe consumers’ savings behaviours by recording the amount they put in envelopes distributed for this purpose rather than by conducting a questionnaire survey. It could be hypothesised that this choice is a way for researchers to circumvent the difficulties associated with consumers’ low level of literacy.
Viswanathan et al (2021e) performed a series of three experiments. Marketing researchers use the experimental method, but mainly under laboratory conditions (Gau et al., 2012). In these situations, the researchers were conducting field experiments. They involve 1,000 people in 34 villages. Their research echoes the experimental approaches used in development economics to assess the impact of development programmes (Banerjee and Duflo, 2009). Not only are they field experiments, but the dependent variable–market education–is manipulated on the basis of the implementation of an educational programme in which one group of consumers participates while another group is subjected to a placebo programme. The educational programme designed and evaluated by the research team as part of its previous research (Viswanathan et al., 2008, 2009) was also consistent in that the educational sessions last from 4 to 8 hours a day for 1 to 6 days.
In qualitative research in which researchers produce knowledge about poor consumers, they opt for conventional approaches, individual and/or group interviews, as they would in other fields, except in seven cases in which they chose to use the ethnographic method. This method uses a variety of data-gathering devices (interviews, participant observation, visual note-taking) to get as close as possible to the material environment and the symbolic registers and actions of consumers. The seven studies differ in their aims, which can sometimes influence the type of data-gathering techniques used and the way in which both the conduct of the research and the implications of its findings are presented. Some have a reflexive aim, which is aimed at researchers. Studies by Hill and Stamey (1990) and Hill (1991) are characterised by their desire to describe the difficulties encountered by researchers in the field in great detail and in narrative form, the ethical questions they have to face, and the ‘tricks of the trade’ of the ethnologist who gradually gains the trust of the informants. The presentation of the methodology used in this research, which lasted between 1 and 4 years, is a veritable narrative.
Other research is more comprehensive. Immersion in the field lasts less time than in previous research, often only a few months (Chikweche and Fletcher, 2010). This research gives access to a detailed understanding of the uses of certain products or services and the experiences the researchers undergo (Dey et al., 2016; Çekirdekci, 2020). Finally, two other groups of researchers, Saatcioglu and Corus (2014) and Venugopal et al. (2019), use the ethnographic method with a transformative aim, although at two different levels. While Saatcioglu and Corus (2014) simply call on the public authorities and researchers to move away from traditional categorisations of poverty and rethink the boundaries of oppression, Venugopal et al. (2019) not only employ techniques explicitly aimed at giving a voice to the most vulnerable (role-playing to make the research issues concrete for those concerned and to obtain ‘the voice of the least powerful members of the communities’) but also produce very concrete and varied recommendations for the public authorities.
Producing knowledge with poor consumers: The use of participatory methods
This logic of producing knowledge with, which exists in a limited number of research projects (5), results in the involvement of those concerned in the research protocol itself and in the service of their well-being or emancipation. Their involvement is not systematic throughout the research. First, we identified two qualitative research projects that were under the heading of participatory action research. The team of researchers led by Viswanathan (Viswanathan et al., 2021c) involves consumers or their representatives in producing proposals for the implementation of action plans resulting from data analysis. The authors proposed the Symbiotic Academic–Social Enterprise (SASE) method, which transforms the role of marketing into a discipline serving the development of emerging countries. The method could be broken down into several stages involving poor consumers or their representatives, leading to concrete solutions for improving the living conditions of poor consumers.
Research by Schmidtke et al. (2021), rooted in social marketing and focusing on the design of a programme to reduce school drop-outs in a community in South Africa, also falls into this category. The researchers describe a research design consisting of three studies, one of which involved students and parents in designing the programme during the ideation phase only (the final study involved educational experts in assessing the feasibility and relevance of the programme ideas devised by the participants), using materials made available to them.
Less systematically, in two of the ethnographic studies mentioned above (Hill and Stamey, 1990; Venugopal et al., 2019), the authors involved poor consumers only during the data collection stage. Venugopal et al. (2019), claiming to be community action research (Ozanne and Anderson, 2010; Ozanne and Saatcioglu, 2008), for example, gave cameras to key informants in the community that they were observing and asked them to film key moments in their fishing activities and community life themselves. Similarly, Hill and Stamey (1990) describe a ‘member checking’ protocol in which homeless people who are recognised as having observational skills were given a deliberately inaccurate description of their peers so that they could amend or add to it. Viswanathan et al. (2021b) worked with a group of target informants (refugee entrepreneurs) to recruit other informants and facilitate interviews by acting as translators.
The populations studied: Consumers rooted in situations of poverty or ‘a-situational’
We report the characteristics of the targets or samples interviewed from the point of view of poverty and not in relation to the object and/or questions of the research. Our analysis shows that these targets or samples were represented in two main ways. The first brought together individuals in the same poverty situation within a target or sample. The consumer targets or samples were then anchored in this contingent and inevitably in a multidimensional situation. The second consisted of identifying a population on the basis of criteria that assigns substance to objective poverty. Contexts were sometimes described, but they remained in the background when defining the sample. This process is the reason we call these targets or samples ‘a-situational’.
Samples of consumers in situations of poverty
In research that uses such targets or samples, they are made up of consumers who share the same life situation or who benefit from a service or organisation that helps the most disadvantaged. The life situations were varied (Table 2). This situation is the case, for example, in a study by Banister et al. (2016) who sought to understand how young women who are primiparous and have low incomes seek to secure their social value and legitimacy through the market. But this is not systematically the case. Thus, for example, to explore the status construction of poor individuals through the use of moral repertoires of action, Saatcioglu and Ozanne (2013) interviewed individuals who were all living in the same mobile home campsite.
Targets or samples rooted in situations of poverty.
For research mobilising targets or samples that bring together populations cared for by or receiving support from a service or organisation helping the poor (Table 2), the choice of partner may be dictated by the object of the research as is the case in a study by Du et al. (2008) or may be unrelated as is the case in a study by Jose et al. (2015).
In the research carried out on the basis of these two types of target/sample, the populations studied are embedded in a situation, either a living situation or an assistance situation, which in itself helps to contextualise the production of knowledge. This setup was not the case in the other research projects in our database.
Samples of a-situational consumers
These targets or samples are based on objective characteristics of poverty, whether in terms of income or geographical location (Table 3). When using an income criterion, researchers refer to standards published by international or national bodies, which most often refer to absolute poverty. When they use a location criterion to define their samples/target, without expressing it, researchers consider location as a proxy for poverty. This is the case, for example, of Kamran and Uusitalo (2019) on a fieldwork in Pakistan or Talkudar (2008) for research on poor consumers in the United States. The choice of location is justified by the recognised poverty of certain districts or geographical areas, including those within the ECs.
A-situational samples.
In research conducted on the basis of these two types of target, the studied populations consisted of consumers who were homogeneous in terms of their poverty as understood either through their income levels or through their locations but who may experience very different situations of poverty. The contexts in which these consumer populations live can sometimes be described (description of the socioeconomic characteristics of the region inhabited), but these contexts were considered ‘part of the background’. Consumers were studied outside their specific living situation. Our analysis did not reveal any marked regularities in the adoption of one type of sample as anchored in a specific situation or a-situational versus another type, depending on the selected methodological approaches. A difference may emerge depending on whether the research is conducted in ICs or ECs (Table 4), but we feel that this result needs to be confirmed.
Consumer populations studied.
Taking support on third parties to implement empirical studies
In nearly one-third of the research projects in our database, the teams used third parties to conduct fieldwork, whether in ECs (20 out of 63 projects) or in ICs (12 out of 20 projects). The nature of these third parties varied. They included private companies, such as market research companies, (Hamilton and Catterall, 2006), public or para-public players, such as social services in the ICs or development policy deployment services in the ECs, and associations or non-government organisations (NGOs) in which the researchers may or may not be involved. The researchers are sometimes individuals with strong social ties to the targeted consumer communities (Kamran and Usitalo, 2019). Some research teams involved several third parties in the same research project. This situation is the case, for example, in Du et al. (2008) who relied on both an NGO and a person who was very familiar with American Hispanic communities they were studying.
The use of third parties is quite common in marketing research. For example, research teams ask companies to identify lists of possible contacts among their customers or prospects or to be put in touch with them. This process is also the case for some of the research in our database. For example, Fang et al. (2014) cooperated with a telephone operator to study the impact of mobile money services on consumer welfare in Cambodia. However, in our database used in our research, we identified a wider use of third parties and coverage of the different stages of conducting research from identifying the target to data collection (Figure 2).
Involvement of third parties in the field.
In some research, third parties identify the target on behalf of the research teams. For example, Ranjan et al. (2021) used a private research company to identify villages in which to collect data, and Weaver et al. (2019) asked a local resident with decades of experience in the fight against poverty to identify both target communities and potential respondents within them. Third parties thus play a major role in the research, both because they enable it and because they can lead to possible biases induced by the possible filters of the consumers interviewed. In other research, third parties promoted the research to a target audience identified in advance by the research teams. For example, it was a local leader working in an NGO who guaranteed the seriousness of the research team gathered around Venugopal (Venugopal et al., 2019) to a community that the team identified. This step is important for removing the community’s mistrust of stakeholders and enabling the recruitment of the target group, which the researchers then assume.
However, the third-party recruitment phase is also sometimes entrusted to third parties. For example, in their research in Ireland, Hamilton and Catterall (2006) used a market research company to recruit the target group and arrange appointments for poor consumers. The authors pointed out that this is a costly but facilitating approach because the research company’s employees belonged to the same local communities as the target families. In this situation, the aim was to reduce the social or cultural distance between researchers and interviewees. Finally, some teams outsource data collection to third parties. Dadzie et al (2013) and Ingenbleek et al (2013), for example, had their questionnaires administered by public service employees who were in regular contact with the target population while Du et al (2008) asked an employee of a research company who was already working with the community under study to act as moderator for group interviews.
While the use of third parties is relatively frequent and varied, they are used for instrumental purposes. Research teams do not use them in the data analysis phase, for example to reduce the difficulty of interpreting data due to differences in living conditions and asymmetries in position between researchers and poor consumers (Blocker et al., 2013; Lepoutre, 2001).
We present the results of a literature review of empirical research that addresses poverty in marketing during the period from 1990 to 2021. These studies provide an insight into the types of methods, samples or targets, and implementation methods used by researchers to produce knowledge on poverty. They also provide examples to illustrate the proposed vade-mecum in the following section for addressing the sensitive terrain of poverty.
Vade-mecum for conducting empirical research on poverty
This section aims to support marketing researchers who wish to undertake empirical research on poor consumers with a questioning tool and five concrete recommendations for approaching the field (Figure 3). This tool makes it possible to carry out ‘a reflexive activity’ on the possible ways of producing knowledge’ (Allard-Poesi and Perret, 2014: 15) on poverty phenomena, beyond the established epistemological paradigms (positivism, interpretativism, constructivism). No researcher working in the field of poverty is insensitive to misery (Mercier, 1995). Moreover, in our corpus of empirical articles, we observed a common ambition among researchers to help reduce poverty as stated in the introduction or in the implications of the research. However, while this ambition is shared, researchers do not envisage contributing to reducing poverty in the same way. An analysis of our database shows the variety of contributions that can be made. Therefore, the first question we recommend that they ask themselves in the vade-mecum is: ‘How does the researcher envisage contributing to a reduction in poverty?’. This question is not usual in the context of a marketing research project as described by Delacroix et al. (2021). We see it as a question about the ‘destination’ of the results of empirical research or its ‘social impact’, which comes down to identifying at whom the research is aimed and its purpose.
A vade-mecum for conducting empirical research on poverty.
Second, we encourage researchers to question the ways in which knowledge is produced (question 2), first through the prism of research methods and techniques (question 2.1) and second through the prism of the delimitation of the field of poverty, which raises questions about the type of validity of the knowledge produced (question 2.2). This questioning tool and the associated concrete recommendations (from R1 to R5) in addition to a few chapters in books on specific subjects (Baker and Gentry 2007; Gorge and Özçağlar-Toulouse, 2017; Mick et al., 2012; Ozanne and Fischer, 2012; Viswanathan, 2012; Wansink, 2012) should enable researchers to make choices in terms of methods, research techniques and targets or samples suited to the empirical poverty field. These choices are, of course, also determined by the objects and questions of the research as shown in Figure 3. We would also like to indicate that, while these recommendations seem to us to be essential in the context of research on poverty in the light of our state of the art, some of them are not necessarily exclusive to the objectives of this study.
1. How does the researcher plan to contribute to poverty reduction?
Although our database does not always explicitly and systematically reveal the contributions to poverty reduction targeted by marketing researchers (which is why we have decided not to code such contribution), a body of research point us in the direction of four complementary types of social impact that offer varied possibilities for contributing to the fight against poverty. The first, Çekirdekci (2020), is an example that can be described as ‘comprehensive’. Research helps produce a reduction in poverty by revealing the hidden, invisible and/or invisibilised realities experienced by poor people and bringing these to the attention of those with the power to regulate, supervise and improve such conditions. The second approach can be described as ‘inclusion through the market’. Researchers, for example, Hasan et al. (2019), produce recommendations for businesses on how to design innovative products or services that can give poor people easier access to consumption.
Finally, marketing researchers can play a ‘transformative’ role in two different ways. On one hand, they can help to reduce poverty by making the results they produce and their operational implications for improving welfare or protecting consumers available to businesses, public authorities and/or market regulators (see Figure 3, boxed social transformation target). This is, for example, the case of Venugopal et al. (2019) and also in research published in the Journal of Public Policy and Marketing. Alternatively, as Schmidtke et al. (2021) did, researchers may consider the research process itself as a vehicle for the emancipation of poor populations with the conduct of research becoming a space for expression reflecting consumers who gradually regain control over their consumption (see Figure 3, emancipatory social transformation objective).
The answer to this question about the contribution and purpose of the research only has a partial impact on the choice of methods as shown in Figure 3. This occurs only when researchers aims at contributing to poverty reduction via recommendations for emancipatory social transformation. In this type of case, the researcher should use participatory research methods. In other cases, the range of possible methods is wide and the choice is closely linked to the research question. It should be noted that this preliminary questioning about the type of contribution to the reduction of harmful phenomena for the consumer can be replicated for any other subject who design and establish public policies on ageing, disability and/or health, for example.
2. Which knowledge production methods should be chosen?
We consider the conduct of research in the field, which covers all research methods and techniques (2.1), and the delimitation of the field of poverty, which influences the type of validity of the knowledge produced (2.2).
2.1. How to study poor consumers?
Our results show that marketing researchers produce two types of knowledge about poverty: (1) about and (2) with poor consumers. While the production of knowledge about constitutes most of the research, which is reconcilable with the aims of comprehensiveness, inclusion by the market and controlled social transformation, and authorises the use of a diversity of methods customary in marketing research, the production of knowledge with those concerned should be underpinned by the emancipatory transformative aim of the research referred to above and calls for the mobilisation of participatory methods. These methods are based on the premise that poor people are the first to know about their condition and what is causing problems as Schmidtke et al. (2021) states. The aim is to move people experiencing poverty from the status of objects to that of subjects of knowledge, a status that makes it easier to reappropriate their problems and even to organise forms of protest or emancipation.
These methods resonate with the objectives of the cross-fertilisation of knowledge approaches initiated by the ATD Fourth World Network (Carrel et al., 2017), which works with people living in poverty and recognises that they have ‘knowledge of [their] experience of poverty, certainly; but, more profoundly, knowledge of what poverty does to people, what people become in the lasting conditions of poverty or fragility that they endure’ (Brun, 2007: 12). This approach to producing ‘situated’ knowledge is widely used in other disciplines that have been researching poverty for longer (such as the sociologists Grignon and Passeron, 1989; Schwartz, 1990). The latter attempted to reconstruct the multifaceted experiences of poverty. The aim is to collect ‘situational discourse’ (Lepoutre, 2001) and produce knowledge from the point of view of those concerned as also advocated by some feminist researchers (see Harding, 1991).
These participatory methods, the value of which has recently been highlighted in marketing for tackling the sensitive terrain of poverty (e.g., Blocker et al., 2013; Crockett et al., 2013; Gorge and Özçağlar-Toulouse, 2017; Saatcioglu and Corus, 2018) and also raises the question of the status of the knowledge produced as shown by a body of work in sociology. How does the experiential knowledge of poor people acquire the status of scientific knowledge (Carrel et al., 2017)? Is explaining how this knowledge is produced enough to classify it as scientific knowledge? Should we engage in the co-production of knowledge by recognising knowledge rooted in the lived experience of poverty in the service of action to reduce poverty and thus move from the ‘role of evaluator to that of co-experimenter’ (Banerjee and Duflo, 2009: 696)? On the other hand, should the aim be to produce scientific knowledge that enhances current poverty-related knowledge by treating the participatory approach as a means of collecting additional data (Carrel et al., 2017)?
Whatever the researchers’ responses to this question about research methods and techniques are (2.1), we would like to make four initial concrete recommendations for fieldwork. These recommendations relate to subjects that we distinguish for presentation purposes but which are sometimes linked. Thus, for example, carrying out a full-fledged pre-field study (recommendation no. 1) may contribute to the design of appropriate data collection procedures (recommendation no. 2) or to the introduction of tools or approaches that enable a balance to be struck between being too close to poor consumers and being too far away (recommendation no. 4).
Recommendation 1 (R1) is to carry out a preliminary study in its own right. It can be very tempting for researchers to rush out into the field, given their limited grasp of the conditions in which poor consumers live and their understanding of such conditions (Blocker et al., 2013). However, carrying out preliminary work appears to be an essential step in its own right. We are not referring to studies designed to test the validity of interview guides, questionnaires and/or measurement tools previously adapted to the target group of poor consumers to which we will return in recommendation 4. In this case, we stress the importance of a preliminary phase in line with the recommendations of Ingenbleek et al. (2013).
Ingenbleek et al. (2013) recommend that quantitative researchers in EC carry out preliminary work to gain a better understanding of the cultural and social context in which their empirical research will occur and to identify the research protocols that have been developed there. This work includes, as does most research, a bibliographical search, which it is recommended from a multidisciplinary perspective. It also includes documentary research, interviews with experts in the local culture and language and/or researchers who have experience in similar areas, or with NGOs who have extensive knowledge of the local populations (Gau et al., 2012; Ingenbleek et al., 2013). This preliminary phase should enable researchers to determine the most relevant unit of observation and assess the availability of data on the populations. In particular, researchers should establish an appropriate sampling strategy by identify the social, cultural, and physical barriers to accessing respondents, the means of circumventing them or their consequences, and the existence of measurement tools developed specifically by other researchers (Ingenbleek et al., 2013). We extend this recommendation to qualitative research, on the one hand, and on the other, to research conducted in fields located in ICs. Indeed, it is the understanding of rich empirical contexts that enables qualitative researchers to contribute on the basis of their results to the construction of theoretical knowledge (Crick, 2021). However, as we have already indicated, researchers are struggling to gain knowledge about poor consumers, given the differences in the life situations between the researchers and their poor subjects (Blocker et al., 2013; Chakravarti, 2006; Downey et al., 2007). For Çekirdekci (2020), who conducted qualitative research concerning how poor migrant consumers in Turkey transform their squats into a home, this preliminary work, which consisted of observations and a group interview with informants, constituted an ‘introduction to the subject before immersing oneself in the field’ (Çekirdekci, 2020: 255). As such, such conditions can initially pose emotional difficulties for researchers. These emotional issues are why we recommend that they mobilize multiple sources of information and, at the same time, take a reflective approach to their relationship with the field of poverty as set out in recommendation 4.
Our second recommendation (R2) to researchers is that they should adapt the techniques and practices used to collect data using questionnaires and interviews or find innovative alternative methods of data collection. No researcher wishing to work in the field of poverty can ignore the low level of literacy among poor consumers, whether in the EC or the IC. We recommend that marketing researchers innovate with data collection techniques that have already been tried and tested in marketing or that are newer. First of all, projective techniques, particularly collage (Gorge, 2014), can facilitate the expression of difficulties encountered by people experiencing poverty. Our database highlights a number of new approaches, some of which are fairly experimental or use tools that are not yet commonplace, such as participatory photography or videography. These types of approaches can help empower poor consumers by enabling them to express significant aspects of their experiences themselves. The example of Venugopal et al. (2019) who equipped informants with cameras is one typical example. We recommend that marketing researchers spend time evaluating these techniques. In this respect, the work of Meyer (2017) on photography, that of Cléret et al. (2018) on the videographic method, and that of Boch (2023) are interesting guides for choosing these type of techniques.
For researchers who prefer to use data collection techniques that they have already mastered, such as questionnaires or interviews, we recommend adapting both the questions to the concrete way of thinking of individuals and their restricted vocabulary (as described by Gau et al. [2012]) and the measurement scales and evaluation tools for quantitative research. While the exercise of adapting questions requires time prior to conducting the fieldwork for both quantitative and qualitative research, it is probably easier to carry out for quantitative research. Questionnaires can be pre-tested as Ingenbleek et al. (2013) did (and described in our database) and adjusted iteratively. For qualitative research, even if the researcher has worked upstream on interview guide questions formulated in a simple way, he must learn to adjust his vocabulary, for example, for follow-ups as Besozzi (2021) did. Finally, we recommend that marketing researchers carry out face-to-face interviews to reduce the bias associated with insufficient understanding of the questions or the use of digital tools. In the case of large samples, interviewers should be called in (recommendation 3) and carefully briefed (Ingenbleek et al., 2013).
Our third recommendation (R3) is to rely on third parties and/or work with associations to identify and recruit poor consumers and even to manage data collection. Working with partners is a recommendation for gaining access to the field of poverty, whether in economics (Banerjee and Duflo, 2009), sociology (Morrissette and Baribeau, 2019) and marketing (Davis et al., 2016; Gorge and Özçağlar-Toulouse, 2017). Public institutions in charge of anti-poverty programmes and NGOs, charities and social workers are key partners in development economics and sociology as we have described in the results. These institutions are all avenues for marketing researchers to explore when approaching a field, whether in ECs or ICs. Nevertheless, on the basis of our results, we encourage researchers to broaden the spectrum of possible partners to include individuals and market research companies.
The opportunity to use third parties should not lead us to believe that access to the field of poverty is easy. We warn marketing researchers that the process of recruiting poor consumers can be lengthy. For example, Ranjan et al. (2021) successively recruited three third parties to select households to be interviewed. The recruitment process can also be costly. As Hamilton and Catterall (2006) point out, they use a market research agency whose staff come from the same local communities in Northern Ireland as the target families to recruit them. Finally, we suggest that marketing researchers consider the possible support of third parties beyond the mere access to the field with which they are most often associated. Third parties can be invaluable in collecting data. This is particularly the case for quantitative research, given the need to administer questionnaires face-to-face.
An alternative to using third parties to identify and recruit poor consumers is to use the voluntary sector to gradually come into contact with poor individuals. Although we have seen little evidence of this in our database, this practice is widespread among qualitative sociologists and ethnologists and has been used in recent marketing dissertations in France. This practice was described in Anis Jounaid’s thesis (Jounaid, 2021), whose ethnographic approach was based on voluntary work with two French associations « Les Restos du Cœur » and «Le Secours Populaire ». However, we invite marketing researchers to take note of the way in which sociologists work their way into the associative fabric in addition to the investment and involvement to which they consent, for example, in the research of Arnal (2014) and Arnal and Haegel (2019) or Bordiec and Marnet (2017).
Our fourth recommendation (R4) concerns the tools or approaches that researchers should choose so as to strike a balance between being too close or too far from poor consumers. Contact between researchers and poor consumers is not systematic. Analysis of our database demonstrated that some research teams entrust data collection to third parties. However, when this process is not the case, how can researchers guard against the vulnerability they may develop through contact with poverty? While none of the articles in our database mentions any vulnerabilities developed through contact with interviewees let alone suggests ways of dealing with them, the marketing literature does offer a few recommendations beyond preparing for strong emotions as we discussed earlier. In terms of more tactical solutions, Downey et al. (2007) and Hamilton et al. (2006) recommend that researchers have a list of support organisations for poor consumers to give to interviewees if necessary, as these organisations would demonstrate the necessary empathy without the risk of getting too close. Borraz et al. (2021) recommended formulating evasive reminders to create an emotional delay and thus manage emotions or discomforts in these situations (Borraz et al., 2021: 72).
The authors also agree on the importance of not working in isolation. Conducting fieldwork with another researcher helps share or reduce the emotional burden (Borraz et al., 2021). For a researcher, providing support in the form of regular debriefs with peers, which is a widespread practice in SHS health teams (Dickson-Swift et al., 2007), is a way of combating vulnerability as a researcher (Hamilton et al., 2006; Jafari et al., 2013).
At the same time, however, some of these authors stress the importance of the researcher accepting his or her own vulnerability so as to be capable of listening to the vulnerability of the interviewee (Downey et al., 2007; Hamilton et al., 2006). In this case, we find the importance, indeed the necessity as emphasised by sociologists, ethnologists and anthropologists of striking a balance between being too close to and too distant from respondents or seeking excessive objectivity (Robert-Demontrond et al., 2013). Arnal (2014), for his part, suggests finding ‘the right balance between personal proximity and the distance specific to the sociological approach’ (p. 110).
We recommend that researchers engage in reflexivity and self-analysis of their emotions in the light of what ethnologists, sociologists and anthropologists do (Gobatto, 2020; Laplantine, 2005; Tcholakova, 2020). Sociologists describe this approach as the production of ‘a critical, reasoned and systematic analysis of the conditions of enquiry carried out at the various stages of research in order to develop rigorously constructed knowledge’ (Buscatto, 2012: 11). Arnal (2014) offers an example of this self analysis in the field of street outreach work in Paris.
Finally, the search for the ‘right’ distance from the field should guide the interpretation of data, which is likely to be biased by the asymmetry of positions between researchers and people experiencing poverty. In this respect, Rémy (2017) recalls the recommendations of Grignon and Passeron (1989), who suggest avoiding the pitfalls of populism (‘in recognising and above all valuing the cultural autonomy of the working class while denying the relationships of domination’ p. 78) or on the other hand, the pitfalls of miserabilism (‘in seeing in the working class only domination and heteronomy linked to difficult living conditions’, p. 78). Encouraging ‘alternation’ (increasing the number of observation sites, developing comparative research protocols in order to move away from a populist stance) or ‘ambivalence’ (accepting the idea of reading into the data the elements of domination and freedom that are at play in any experience of poverty) are therefore appropriate methodological directions.
2.2. Which poor consumers should be studied?
Our review of empirical marketing research concerning poverty highlights two types of target or sample from the point of view of poverty. Some are ‘anchored in a specific poverty situation’, which are not necessarily linked to the research objects or questions, while others are ‘a-situational’. In the first case, poverty is first and foremost a culturally embedded social experience; therefore, the perspective is therefore situational. The aim is to take account of real-life experiences, tensions and adjustment strategies. In the second case, the study population is identified on the basis of criteria that give substance to objective poverty. In this case, the researchers adopt an a-contextual, a-historical, a-temporal ‘heterodefinition’ of poverty (Martinet and Payaud, 2010). From this point of view, these samples are ‘a-situational’, and even when the contexts are described by the researchers, they remain in the background. The populations that make up the samples are considered to be representative of the category of ‘low-income consumers’, and the aim is to bring out the obstacles, motivations or other cognitive drivers, and structuring collective action logics.
These two types of sampling can be reconciled with the range of methods mentioned, although the choice of an ethnographic method corresponds to the desire to explore singular communities. They raise the question of the validity of the resulting knowledge, which goes beyond the debates surrounding qualitative versus quantitative studies. Can the results of a study of Indian consumers earning less than two dollars a day be generalised to other ECs? Can the results of a study of a home for young single mothers in an IC be transposed to other similar situations of poverty?
The critical work of Favereau (2014) concerns the experimental random assignment method developed in development economics by Banerjee and Duflo (2009) and sheds some interesting light on these questions. While the internal validity of the results of research using this method has been widely recognised, because it is, in particular, associated with the observation of populations rooted in their situation of poverty, its external validity was strongly criticised (Favereau, 2014). Favereau (2014) poses two questions and refers to two levels of external validity; can the results of the evaluation of a development programme in the context of these experiments be extended to other regions or even generalised? In other words, can the results obtained be transposed to other contexts? The answer is no. Banerjee and Duflo (2009) agree and recommend that experiments should be replicated rather than the results transposed. The reasons for this choice are not found in the debates between qualitative and quantitative approaches. Random assignment experiments are by the very nature part of quantitative approaches. The limited validity of the results is largely due to the fact that these methods are not capable of establishing a causal link to explain the factors behind the success of development programmes; it is not possible to isolate the contextual factors that could be behind them and that the poverty contexts are different. With regard to ECs, Favereau stresses that ‘there is not one poverty, but several, which are linked to different economic, social and environmental contexts’ (p. 214). In the specific case of poor consumers, we invited marketing researchers to show the diversity of empirical cases that poverty covers without any intention of generalising. Along the same line, the managerial implications and/or those aimed at public authorities should be limited to the studied groups and include recommendations for replicating the results.
From a practical point of view, since access to the field often presents multiple obstacles as mentioned in the literature review on the difficulties of this sensitive terrain, we would like to make a fifth concrete recommendation (R5) to marketing researchers, that is to constitute their samples ‘as they go along’ and in a pragmatic manner, namely, with a dose of methodical opportunism and flexibility. That is what Ranjan et al. (2021) did when relying on third parties with relational networks and knowledge of the environment to identify interviewees. Ranjan et al. (2021) provided a very detailed account of their approach for which they first used the services of a qualified private data collection agency to identify the villages of the target profiles. The research team then met with the head of the villagers’ group, health service providers and other child development services (linked to their research topic) and asked for their help in selecting people to interview. The research team then went door-to-door to each selected household.
The example of Mukherjee et al. (2018) is also interesting as researchers constitute a group of respondents progressively, as the field was explored through a system of questioning that makes it possible to identify (and in so doing retain only) the people best able to answer the research questions. Pragmatic sampling based on a monetary approach to poverty can also be envisaged depending on the areas that are explored as recommended by Bisiaux (2011). At the end of her work on operationalising conceptual definitions of poverty, she concludes that in the case of India, the monetary approach constitutes ‘an ideal measurement tool, sufficiently sophisticated to avoid recourse to complex operationalisations of theories such as those of Sen or Rawls’ (Bisiaux, 2011: 11).
Conclusion
Our research offers three contributions to help marketing researchers engage in empirical work addressing poverty. The first contribution is a vade-mecum that we propose in response to the difficulties inherent in the field of poverty to which we have alerted them. The vade-mecum consists of a set of questions that can be used for designing a research plan and five concrete recommendations for approaching the field (Figure 3). It is illustrated for ease of use by researchers in the community.
Our second contribution is based on our literature review of empirical marketing research on poverty conducted between 1990 and 2021. By gathering and explaining the different strands of marketing research on poverty, this literature review makes it possible to propose a tool for researchers to conduct preliminary work prior to any engagement with the poverty empirical literature, a step that we recommend in line with the recommendations of Ingenbleek et al. (2013). Finally, the use of a variety of terms by marketing researchers to designate poor consumers makes identifying research on poverty challenging and presents a real risk for conducting incomplete investigations. Our third contribution is the list of terms that we have drawn up for our database and used to designate poor consumers (Appendix 1). It should make it easier for marketing researchers to identify forthcoming publications concerning poverty and to update or supplement our state of the art in a few years, which we feel is important given the acceleration in empirical marketing research on poverty as demonstrated by the composition of our database.
However, our research is not without its limitations. Our analysis is based exclusively on bibliographical data. As a result, we re-constructed the methodological choices made by the research teams. This limitation could be overcome by conducting qualitative research using interviews by the teams. Moreover, our research is only the first step in addressing poverty research in that we have focused our attention on empirical research in the discipline of marketing, even though it was relatively late to take up the theme of poverty. In this article, we put forward elements from other social sciences and humanities (essentially economics and sociology), which have been involved in the study of poverty for a long time. It would be interesting to conduct more investigations and broaden the scope of our review of empirical literature to include these other disciplines. This research focused exclusively on the methods and practices of conducting research in poverty-related fields. Formalisation and dissemination of scientific production in these fields represent another fertile area for reflection. In terms of formalisation, how can researchers produce critical and un-biased knowledge without objectifying their subjects and betraying their trust? In terms of dissemination, is it primarily a question of restoring the voice of the most vulnerable and giving them back a say, in particular to influence the public policies that affect them or simply to inform their decisions? Finally, the resulting knowledge is likely to have a stigmatising effect on the groups of people who are studied. So what format should be adopted for presenting this knowledge and encouraging people to adopt it? In sociology based on the example of Nicolas-Le Strat (2023), some researchers are experimenting with short, ‘autonomous and undisciplined publication’ formats (fanzines, brochures) designed to be distributed as close to the action as possible in the working-class neighbourhoods under study.
Finally, we would like to point out that despite the difficulties inherent in the empirical study of poverty that we have identified, in the empirical research that makes up our database we saw almost no mention of the difficulties encountered by researchers, the conditions for the success of their fieldwork, or the compromises that they had to do to conduct their research. Unlike sociologists, marketing researchers provide very little behind-the-scenes information about their fieldwork. Similarly, in general, they say very little about the limitations associated with any pragmatic choices made to adapt to the conditions of poverty. We wish to emphasise that the weakness of such feedback not only hinders sharing the ways of overcoming the difficulties associated with the empirical nature of poverty but may also hinder the progress of a necessary reflection within the community on the trade-offs to be made between rigour and feasibility (Crockett et al., 2013). In response to this challenge, the TCR research community has embarked on new practices concerning research conferences (Davis et al., 2016). It seems that reflection within the discipline in conjunction to an identical effort concerning the expectations of writing practices could complement this work which is already largely engaged.
Footnotes
Appendix 1
Process for compiling the bibliographic database.
1
2 Four articles were not available, so only 82 were included in the bibliographic database.
3 The database was completed following a search for articles published in French journals. The data were collected using Google Scholar based on the list of poverty keywords used to search for articles in English that were translated into French in addition to the names of the journals. The search resulted in two articles published in Décisions Marketing. Only one article was retained as the second did not give rise to empirical research on poor consumers. The two journals Recherche et Applications en Marketing (RAM) and Décisions Marketing (DM) were included as a source of data, even though they do not appear in the Harzing list, and given their importance for the French community, Recherche et Applications en Marketing for the period 1990–2006 and Décisions Marketing for the period 1990–2021.
Appendix 2
| Analysis criteria | Definition or example |
|---|---|
|
|
|
| Conceptual definition of poverty | Definition of poor individuals with reference to an existing theoretical approach to poverty or proposal of a definition specific to the author(s) |
| Operational definition of poverty | Criterion used to determine empirically whether a consumer is poor |
| Involvement of a third party in the research | |
| Nature of the third party or parties | NGO (non-government organisation), public service, private company, individual third party |
| Contribution of the third party(ies) to the research process | Promotion of the research to the target audience, identification of respondents, administration of questionnaires, interpretation of data |
| Flexibility in building the site | Any mention of a change in the target/sample, the research design, the data collection technique and/or the data collection itself during the fieldwork. |
| Difficulties and ways of overcoming them | |
| Difficulties encountered | Any mention of a choice justified by a difficulty encountered in the field and linked to the poverty of the target group |
| Choice in response to a difficulty linked to the specific characteristics of the site | Any mention of a choice that makes it possible to overcome a difficulty listed above, whether it relates to method, technique(s), research practice, among others. |
|
|
|
| Stakeholders involved in the research process | Consumers, experts |
| Nature and methods of stakeholder involvement | Data collection, data analysis |
|
|
|
| Sampling strategies | |
| Adapting materials to consumers’ literacy and their local consumption context | Any mention of changes to existing scales or measuring instruments to make them easier to use for poor consumers. |
| Pre-testing of questionnaires | |
| Questionnaire administration | |