At the time I was deciding on a thesis topic, peak oil and climate change were becoming more widely accepted as threats to our food supply. The issues of food crisis, food miles, and the carbon footprint of food were prominent in the media. Allotments are thoroughly researched but I found very little investigation of home-based food-growing. The majority of the population has no access to an allotment for one reason or another. I believe that more and more people will turn to growing in their own home gardens as the economy worsens.
On my own I was unlikely to get enough data to be statistically significant, therefore qualitative social research was required rather than a more quantitative approach. Because of the small sample group, I expected no more than for the data to point out relevant trends and issues. I knew little about the topic at the outset and needed the flexibility to alter my approach as I learned. I thought it more useful to discover the reality on the ground rather than formulating a hypothesis in advance and attempting to prove it. As Grounded Theory is designed for just such a situation I settled on that approach, utilizing questionnaires as well as interviews for data gathering.
“A grounded theory study seeks to generate a theory which relates to the particular situation forming the focus of the study. This theory is ‘grounded’ in data obtained during the study, particularly in the actions, interactions and processes of the people involved.” (Robson 2002 p.190)
Based on study of Robson’s Real World Research I decided to form an informal peer review panel and keep a research diary. (Robson 2002 p.1) The first entry is March 7th 2008 and the final entry is October 28th 2008. Many of the entries mirror the updates I sent to Alison Pooley, my thesis supervisor, every two weeks. See appendix #2.
I also consulted Gilham’s Developing a Questionnaire and The Research Interview during the formative stages of my research design and decided to triangulate my data through questionnaires, interviews, and direct observation. (Gilham 2007, 2004)
“Data triangulation … to enhance the rigour of the research … involves the use of more than one method of data collection.” (Robson 2002 p.174)
The returned questionnaires would inform the design of the interviews. I rejected a sample questionnaire due to time constraints and instead planned follow-up telephone interviews if required.
While the SW side of Sheffield is home to affluent ethnically homogenous neighbourhoods it also contains both ethnically and financially diverse neighbourhoods. To ensure a level of diversity across and within the groups and thus increase the relevance of the research, I picked four very different sample neighbourhoods.
Initially, I hoped to not only learn about home growing but also to be instrumental in promoting and supporting growers in some way. Options for financing support was only available for one neighbourhood so I discarded this idea other than to provide a list of resources to all who returned the questionnaire and wished further contact. See appendix #12. I also rejected assessing yields from active growers, as it was not likely to give me data that would be comparable to other growers due to the heterogenous nature of the sample group. It would also be misleading in terms of food security or health of diet, as I would not be following the produce to table or doing a study on household diet.
My peer review panel consisted of a professional psychologist, several professional activists, the most prominent food grower in Sheffield, several activist amateur growers, and a doctoral student at Oxford. We discussed, separately and in groups, home-based food-growing and factors that might be involved in its promotion and support as well as social research technique. Advice regarding social research texts, questionnaire design, neighbourhoods of interest, and the activities of organisations and activists centered on food in the Sheffield area was gratefully received. I practised my interview technique, using a laptop and USB microphone, by recording and transcribing several of the individual discussions.
After careful consideration of the issues I believed to be most pertinent, I formulated a set of preliminary closed questions and began with the layout, keeping it brief and simple, avoiding jargon and technical terms or too great a reliance on opinion based responses. (Gilham 2007 p.26) I created a questionnaire divided into 6 sections, each with 3 to 6 questions; Demographics, Environmental/Resource Concerns, Household Practises, Gardening, Personal Health and Other Factors. The 5 Environmental/Resource Concerns questions are the only ones that are belief or opinion based. I used pre-selected scaled responses to facilitate analysis.
Each of the 136 questionnaires included a letter of introduction explaining my research as well as a SASE to ensure its return. I personally delivered 95% of them to households by knocking on doors. Except for 7 questionnaires delivered by friends I never left a questionnaire without talking to the resident. Expecting no better than 30% return, I was delighted to receive 50% back. See appendix #3.
Questionnaire Analysis Procedure
I intended to set up a spreadsheet for one question at a time beginning with income. All responses were sorted by income and neighbourhood. See appendix #4. Upon completion I realised that due to 9 recipients choosing not to answer the income question on the survey, the data on this spreadsheet was useless except for analyzing factors correlating to income alone. I therefore created spreadsheets that produced totals for each response in each neighbourhood but these proved too cumbersome for determining correlations between factors. See appendix #5.
At this point I considered using statistical analysis software. My supervisor, Alison Pooley procured some statistical software for me to try out. After spending two days sourcing a PC, reformatting it, and loading windows and the software, I discovered the activation key for the software that had been provided had expired and I was faced with trying to get another one from UEL. I had already lost time on this and I decided not to pursue it any further as I had little faith in getting it up and running expediently. Had I been familiar with this resource prior to starting on my questionnaire design I believe I could have made good use of it. Instead, I was left to hand count and correlate responses.
For example, I separated the surveys based on what food types each household grew. I would then go through each pile looking for high numbers of responses to other questions. I discovered, for instance, that all but 1 of the respondents growing all 4 food types have a first degree or higher educational qualification. I further subdivided the groups looking for well-educated growers who grow the most who also expressed a high degree of concern for climate change. I did this for every question. All these numbers were recorded on a data sheet that grew to 44 pages. See appendix #6. In compiling the thesis, I used the spreadsheets and data sheets to assemble charts.
Anyone returning a questionnaire and expressing a willingness to participate in further research was contacted again. See appendix #3. The majority of them agreed to be interviewed. Only 3 either later cancelled or did not show up. I compiled a set of questions that complemented the questionnaire and then customised each interview based on personal responses. For example, I did not ask interviewee 13 what prevented him from getting an allotment because he indicated he had an allotment. 29 interviews were completed. Most interviews occurred in the their homes where I also measured the gardens, assessed solar resource, and noted gardening infrastructure. Interviews were recorded and all interview responses were transcribed onto the question sheet produced for the interview.
Interview Analysis Procedure
I grouped the questions into categories. I then grouped relevant interview questions under the category. For example,
Category A. Involvement in community
How long have you lived in this neighbourhood?
How well do you know your neighbours? any that grow food?
Are you familiar with any neighbourhood community activism since you have lived here?
Would you participate in neighbourhood growing?
Would you be interested in gardening in someone else’s yard?
Would you allow others to grow food in your garden?
See appendix #7. Next, I copied the answers to each question from the text of the transcriptions into the analysis document being careful to label each response with the interview number and color code by neighbourhood. See appendix #8.
By picking key concepts mentioned in the responses and counting the number of times a respondent used language relating to that concept I found and charted trends.
For example, the level of community activism known about and/or participated in varied drastically from neighbourhood to neighbourhood. I established an index of community activism based on key words given in the interviews. I did the same for social activities. I listed key words and indicated the percentage of the interviewees in each neighbourhood who mentioned either a word or a concept that implied neighborhood social activities. For instance, when one individual mentioned a “new mums” group and another mentioned ”Sure Start” I put a 1 next to parents/kids in the list. Similarly, I put a 1 next to community whenever an individual used that word or mentioned an organisation that is about community building; I only indicated one tick per category per person. If no-one in N#1 ever mentioned a category that someone else in another neighbourhood did, then N#1 received a 0 for that category. I converted these numbers to percentages; if a neighbourhood had 2 mentions of a category out of 5 total interviews they received 40% for the category. Averaging the percentages, including all the percentages for unmentioned categories, delivered a number to be used for comparison. For instance, N#1 received a 19% score for community activism, which is the highest of all four neighbourhoods, while it scored 21% for social activities. N#4 scored 9% for community activism while it scored 22% for social activities. While these scores seem low, N#4 is actually the most socially active. A neighbourhood could only receive a 100% score if every person interviewed mentioned every category. So the numbers are only relevant as comparisons between neighbourhoods.