Given the relative difference in the population sizes, we opted for a 25% sample in The Fountain (116) and a 20% (262) sample in Gobnascale. We increased our sample size slightly to allow for non-responses or non-valid responses to individual questions in the questionnaire. The final sample sizes were 264 in Gobnascale and 121 in The Fountain. This was based on a selection of 160 households in The Fountain and 430 in Gobnascale to 430 in order to allow for non-responses.

The survey was designed to yield a representative sample of adults over the age of sixteen, of the total population, living in private households. The sample was drawn randomly from a list of addresses which was drawn up manually by listing the addresses within an agreed community boundary, taken from the Northern Ireland Ordinance Survey Maps for each area. (Surveyed 1962, revised 1992; Fountain Grids: C432166 - C436166, C432162 - C436162; Gobnascale Grids C435159 - C445159, C435150 - C445150).

1.The list was agreed and updated in consultation with a community worker in each area. The addresses eliminated at this stage, were non-residential units. The map for each area did not show premises under construction, but did show the various types of units, such as houses, flats, maisonettes and business premises. Premises which were later discovered to be no longer in existence, vacant or derelict were also shown. The total number of addresses on the community list for Area 1/Gobnascale was 676, and for Area 2/Fountain was 240.

2. The total population of Gobnascale is 1312, according to our own small area statistics drawn from the census of population for 1991. Based on the same small area statistics, we calculated that the population of the Fountain is 467. It was calculated that a 20% sample for Area 1/Gobnascale and a 25% sample for Area 2/Fountain, would yield sufficiently precise findings. Since the total population of the Fountain area is quite small, it was decided a 25% sample would be required in order to produce the required precision in the findings.

3. A simple random sample of households was obtained by selecting one in three households from the list. A sample list was drawn up in this manner. The total sample drawn for Gobnascale was 451 addresses and 231 addresses for the Fountain.

4. The use of the Kish Grid in each house, represented a further stage of sampling. This is required to select an individual, from each household address, as the questionnaire respondent. As the list of addresses does not include any information on the number of individuals resident, the fieldworkers randomly select one individual, over the age of 16, using a Kish Grid. This procedure was carried out on delivery of the questionnaire.

The sampling error for any percentage, p, sample can be calculated using the formula:

where n is the number of respondents. the sampling error of any percentage, p, can be calculated by the above formula. We have calculated the sampling error using the return for both areas on age and employment.

Prior to the commencement of the fieldwork in both areas, letters were delivered in every household in the area and were distributed using existing networks of communication. The letter informed each household that Templegrove Action Research Limited would be in the area in the following weeks, carrying out a survey. It described the purpose of the survey and explained the procedures for the random selection of an individual. The letter also contained a telephone number and contact names if any individual had any queries they wished to make.

The fieldwork in Gobnascale began on the 9 October, and was completed on the 29 October, 1995, over a period of three weeks. The fieldwork in the Fountain began on the 23 October, 1995 and was completed by the 20 November, 1995, - a period of four weeks. Three staff from Templegrove Action Research carried out this exercise. The questionnaires clearly identified the research organisation, all fieldworkers were fully briefed, and dealt with residents' queries on both delivery and collection days.

451 addresses in Gobnascale and 231 addresses in the Fountain were selected. The two areas were divided up into three patches. Fieldworkers were required to make at least three calls at each address, at different times of the day (including evenings and weekends) before declaring an address as a nil return. Similarly, on collection of the questionnaire, the fieldworkers were instructed to call at least three times before declaring it a nil return.

An overall response rate of 78.25% was achieved, based on the total number of questionnaires delivered. This represented a response rate of 77.94% in the Gobnascale area and 78.95% in the Fountain area.

The overall refusal rate was 14.84%, based on the total number of questionnaires delivered. The refusal rate for the Gobnascale area is 10.59% and for the Fountain area is 24.34%. Reasons for refusal were ascertained where possible, and reasons given for refusal included: no interest in the questions; no time; had already filled out a similar questionnaire in the previous weeks; had filled out a different questionnaire in the previous weeks; and the belief that the questionnaire was a waste of time. There were a greater number of outright refusals in the Fountain area. This could well be due to the Fountain having had three different research organisations conducting surveys in the two months prior to this survey. Questionnaires which were taken, and handed back had not been completed in some cases. In a number of cases, the questionnaires were not handed back at all. A variety of reasons were given by respondents. These included: the opinion that the questions were too personal; the questions "cut too close to the bone"; that residents did not have enough knowledge of the community since they had only lived in the area for a short time; and some residents said that they had lost the questionnaire. In this last case, some respondents decided that they did not want to fill out another questionnaire.

Wherever possible, the fieldworkers introduced the research with the randomly selected respondent and agreed a time and date suitable for collection of the questionnaire. If not possible, this was discussed and agreed with the individual who answered the door. Assistance in completing the questionnaire was offered at this point, which would be made available on collection, and the questionnaire was left for the selected respondent to fill out in his or her own time. On average, 3 days were allowed before collection. On a number of occasions, assistance was availed of at the day of delivery. Usually, this occurred where the selected respondent was elderly or disabled, and unable to avail of help from other sources. On a number of occasions, assistance was availed of on the day of collection. Respondents seeking assistance varied.

Data entry was subcontracted to The Analysis Bureau, on an agreed set of coding practices. Data was entered and analysed using custom software to produce a data list and basic cross-tabulations in the first instance. Data was subsequently entered and cross tabulated and tested statistically using SPSS 6.1 for Windows. Tests conducted include Pearson's and Spearman's chi square; Continuity correction; Likelihood ratio; Mantel-Haenszel test for linear association, Phi and Cramer's V: Fisher's Exact Test: One tailed and two tailed; Pearson's Rank and Spearman's Correlation.

Due to the size of the data set, this report has limited itself to reporting on a comparisons of data from each of the two enclave areas. Further analysis, examining in more detail sub-sets of data on issues such as gender, or deprivation may be undertaken at a later stage. In addition, the full data set will be archived at the National Data Archive at the end of the project, and thus will be available for secondary analysis.

It may be useful to note that this survey was part of a larger project, which employed a range of methods, approaches and techniques. These are discussed in more detail in other publications of the project (see inside back cover). Here, it is sufficient to summarise the overall research design thus:

The overall design of the questionnaire was derived from the analysis of qualitative work in both areas. (This work is published separately: Hemmed In and Hacking It: Words and Images from The Fountain and Gobnascale. Derry Londonderry: Guildhall Press.) The preliminary in-depth interviews with residents in the Fountain and Gobnascale were tape recorded, transcribed and the contents analysed.

The questions were designed to gather data on the themes which emerged from in-depth interviews. These were: movement into and out of the area, including reasons for moving into the area; problems and difficulties of living in the area; the factors which influence decisions to stay or go; attitudes and beliefs about minority experience, population balance and segregation; and a socio-economic profile of respondents. The questions which elicit demographic details were based on the Social Attitudes Survey '95, and amended to suit the specific requirements of this survey. Other questions were amended versions of various surveys including Brendan Murtagh's surveys of Belfast interface areas.

The questionnaire was drafted and advice was sought from both the Board of Directors and Advisory Group to the project, notably Brendan Murtagh and Dr Denis McCoy. Advice on coding was given by Tom Wright of The Analysis Centre and Trevor McMullan of the University of Ulster. The questionnaire was amended and then piloted in Gobnascale and the Fountain. The completed pilots were examined, the questionnaire was further reduced in length and a number of other minor amendments were made, before finalising the design. The scheme for these interviews on which the questionnaire was based was as follows: