Experiences of Food Poverty: Methods of Study by Samuel Lindskog

This inductive (Bryman 2012:8,12,111) research project was based on nine in-depth and unstructured interviews carried out between February and May 2014, resulting in rich (Charmaz 2006:10) qualitative data. Seven of the interviews were transcribed verbatim, one was lost due to the audio file being corrupted and another was excluded as the participant was deemed to not have experienced food poverty.

Food Poverty

The sample was kept small, partly due to time restraints (Mcneil and Chapman 2005:25-26) and partly because the data collected was felt to be rich enough to compensate for  the lack of generalizability. The research did not reach saturation (Bryman 2012:426). ‘Thick’ description revealed feelings, attitudes, views and structures through personal accounts (Charmaz 2006:14), achieving high validity of data.
This research has drawn on elements of grounded theory in being guided by ‘emergence’ (Charmaz 2006), but has not prioritised theory creation, but rather has remained descriptive in nature (McNeil and Chapman 2005:7). This emergence has caused the research to shift away from a focus on solutions, towards exclusive attention on experiences of food poverty.

Sampling and Access

The purposively selected (Bryman 2012:416) sample is not representative of the sample population (people in food poverty in Brighton and Hove), but displayed a great deal of diversity. Dowler and O’Connor identified seven key groups (2011:47-48) affected by food poverty, all of which were represented by my sample, with some participants belonging to several groupings. Two participants belonged to the group Dowler calls the ‘new working poor’ (ibid), those that are insecurely employed, low-waged or migrants. Participants were not sampled on an ‘a priori’ basis (Bryman 2012:418), but excluded from the data analysis if the person was not deemed to have experienced food poverty, as was one the case with one interview.

The project was subject to severe access and cooperation problems. One implication of this was that time did not allow for conducting focus group interviews. Snowball sampling (Bryman 2012:424) was attempted twice at different sampling sites, but failed. What follows is a brief description of the sampling process.

The sampling process began long before attending a service (such as food banks), by building a contact network of relevant services and community workers in Brighton and Hove. These contacts often granted access to sampling sites after approval was sought (Wanat 2008:199). For one site, the ‘known sponsor model’ (Wanat 2008:194) was used, drawing on an already established relationship with my former live-in landlady. Parallel to building my contact network, I also put up research posters in community centres across the city, which resulted in three interviews.


It became clear that being granted formal access to a sampling site was not sufficient  to gain participants, this cooperation was only gained by building trust and familiarity (Mcneil and Chapman 2005:106-107; Narayan et al 2000:11) through ‘empathetic relationships’ (Wanat 2008:200). These relationships were built through a high degree of self-disclosure about my own struggles to eat well on a low budget, and my personal strategies around food acquisition, such as finding food in supermarket bins.

I deem this role-construction process to have been highly successful, since I was asked more than once if I didn’t “want a food parcel too?”. I did however experience some difficulty in having to switch between several front-stage roles (Susie Scott 2009:25), and felt very self-conscious about ‘breaking character’ in front of participants in order to talk to service coordinators in the role of ‘professional’ researcher. Despite making it clear that I was on site to recruit participants, no-one approached me for interviewing and I had to actively initiate in order to gain this interviews.

Apart from the participants recruited through posters, I used two main sampling sites. The first one was the Brighton Unemployed Centre Families Project (BUCFP), from which I was asked to withdraw due to unknowingly having gained access through informal channels and their concern about vulnerable individuals that used the services. A higher-level ethics approval would have facilitated continued access to this site. The second sampling site was the Bevendean food bank and community café, whose service I attended on five occasions and have continuously returned to in order to review my findings with participants. Both sampling sites fulfilled the criteria of being both homogenous enough to give insight into food poverty in Brighton and Hove specifically; and heterogeneous enough to represent a variety of experiences (Bryman 2012:417).


The concern that people in food poverty were not being consulted about solutions to their situation is what prompted this project and therefore ethics have remained very central to the research process. In addition to an information sheet and consent form with optional opt-in for being asked personal questions, the participants were given a resource and advice sheet with information about services available to them in the city.

The project has been as transparent as possible throughout. There has been freedom of access to information and freedom to not disclose personal information, made possible by the design of the interview prompts.. All participants chose to disclose this type of personal information about their experience. All but one participants wanted a copy of the final report when offered.

The research was originally aspiring to be participatory, an ideal that had to be compromised due to resource and access constraints. The project has however continued to be guided by participatory ideals, even when the principles could not be followed (Pain and Francis 2003; VSO 2009). “(The) essence of [the participatory approach] is changes and reversals… Outsiders do not dominate and lecture; they facilitate, sit down, listen and learn… Outsiders do not impose their reality; they encourage and enable local people to express their own” (Chambers 1997:103). I avoided imposing my perceived reality about poverty on participants, but I did assert my political stance that the individual isn’t to blame, in order to make the participant feel morally safe with me. This did sometimes result in debate, which in all  but one case positively contributed to the breaking down the ‘researcher’/’research subject’ role dichotomy. In the one exception, a disagreement over the impact of immigration made the participant less open, resulting in less rich data.

Some principles of participatory approaches (Chambers 1997:105) that have been retained are: the reversals of ‘hierarchies of credibility’ (Becker 1967), emergence as an ethics of handing over the power and self-reflexivity (Frank 2000).


Hierarchies of credibility exist in any system of ranked groups where “participants take it as given that members of the highest group have the right to define the way things really are” (Becker 1967:241). In the case of poverty this means that those writing policy, the so-called ‘professionals’ (Chambers 1997:33), have the power to define what poverty is, its causes and its solutions. So far, this has resulted in the complete absence of input from people in food poverty about what needs to be done. Frank (2000) aims to reverse these hierarchies of credibility, clarifying that this ‘deconstructive move’ does not claim that the stories or narratives of people at the bottom of the hierarchy are ‘more valid’ than those of professionals, only that they are valid (Frank 2000:360). The hierarchy of credibility around food poverty is mostly apolitical (Becker 1967:240): Popular discourse fixes people’s ideas of what poverty looks like and I found that they are not likely to question this unless their experience directly contradicts it.
Allowing emergence guide the research has led me to question the literature and my own political stance on poverty. My political background led me to the belief that individuals were not responsible for their poverty, a stance that I still hold on to, but with more nuance. There are individual and structural barriers to food security, the latter often being more important but the former not easily dismissed. It is however important to remember that what caused the radical increase in food poverty (Trussell Trust 2014) it is not a lack of budgeting skills that has gained epidemic proportions.
At this point, it becomes very important for me to disclose my standpoint (Frank 2000). As Becker put it “the question is not whether we should take sides, since we inevitably will, but rather whose side we’re on” (Becker 1967:239). A standpoint is a non-optional, political and ethical act of self-reflection” (Frank 2000:356). I am highly critical of the liberalised austerity approach to welfare reform and am a believer in local people-initiated solutions with national funding and support. A very pragmatic outcome of my self-reflexive stance is that the term ‘(food) poor person’ in relation to people in food poverty will not be used, as a person formerly in food poverty pointed out that the term implied that a person in poverty is a person of lower quality.


Data Collection

Interviews were carried out in a location that was as neutral or familiar as possible, usually chosen by the participant. Interviews varied in their degree of structure, depending to a large extent on my familiarity with the participant. Since structuring can give the impression of interrogation, making participants defensive (Mcneil and Chapman 2005:57), a listening approach was used to create interest and reduce self-consciousness felt by participants (Narayan et a 2000:14-15). This has been very effective in terms of generating extremely rich data and people revealing a great deal of information on personal circumstances without displaying signs of anxiety.

It became necessary to emphasise the story-telling element at the outset of the interview to break participants out of their expectation of an oral survey. These role expectations of participants hindered story-telling and on three different occasions resulted in the interviews becoming semi-structured.

The flexibility allowed me to adjust my questions to the interaction context specific to each interview. Overall, I approached the interview with an ‘active’ attitude (Rapley 2001:316) of starting conversation and being self-disclosing, as opposed to ‘doing’ neutral and facilitative (Rapley 2001:310). A necessary trade-off in creating stronger relationships with participants was asking leading questions. The answers to these questions have for the most part been excluded from the data analysis, as answers very clearly reflect the nature of the question.
Five out of seven interviews were transcribed verbatim, and care was taken to preserve the voice of participants in all quotations by not excluding colloquial expressions and slang. One participant asked not to be recorded, for which detailed notes were taken during the interview and typed up after. Transcription increases the reliability of the study, since transcripts can be shared and re-coded.
In addition to interviews, a ranking exercise using pre-defined solutions from the literature was used. The exercise sparked interesting discussions, but was limited by its pre-defined nature, leading to few original points being raised b participants. The data from this ranking exercise has been excluded from this report but will be used for further research with the Food Partnership.


Data Analysis

The richness of the data resulted in an unruly and messy data analysis process. The transcripts were coded line-by-line (Charmaz 2006:52) and codes were allowed to emerge from the data. I found that preparatory reading did influence which codes I picked up on or not and I tried to compensate for this by re-coding the transcripts several times so that interviews informed each other, rather than being informed by the literature.
Codes were printed off onto slips of paper which were then grouped under sub-themes and photographed. Often codes fitted in under multiple headings, such as ‘food banks dent your pride’  fitting both under ‘getting help’ and under ‘swallowing your pride’.

Sub-themes were consequently organised under meta-themes, more for accessibility of data than to separate the experiences in discrete analytical categories. Actions often triggered affective responses and feelings often informed actions.

I returned to transcripts for choosing quotations, recoding them for the identified sub-themes. I found that the meaning of the code sometimes emerged from the context, rather than the words which made it hard to confirm a certain claim. Coding also resulted in a certain reductiveness of the data (Frank 2000:355), which I hope to be able to better convey in a second non-academic report, for the Brighton and Hove Food Partnership.
The coding process is illustrated in Appendix B.


Further recommendations

For further research on food poverty, in Brighton and Hove in particular, methods could be improved upon in a number of ways. Firstly, the research would benefit from an established contact network of services and community workers, or a longer time-frame. This would enable the project to use a greater diversity of sampling sites. The researcher should reach out to contacts and ask about using community facilities for interview purposes, since being able to access neutral and local interview sites is incredibly important. I found that data collected in interviews done in participants’ homes was on average more rich than those done in public spaces.

To build on the findings of this study, a genuinely participatory study would be beneficial. Participatory focus groups in  particular would enable participants to build on each others’ ideas for solutions and prompt deeper reflection on their own experience, rather than suggestions being made by outsiders, as has been the case with this project.

Further research on food poverty in Brighton and Hove should still be of high priority.

Read Next Part

Read Previous Part