**9. Data analysis**

The completed questionnaires were collated and coded using a statistical package for the social sciences [69]. In phase 1 of the data analysis, frequencies of responses and cross tabulation of individual questionnaire items were calculated for the purpose of specific demographic information and the results were displayed using tables. In phase 2 of the data analysis, observations from the micro-neighbourhood and the random selection were compared. The chisquare inferential statistical method was used to examine the potential association between categorical variables, that is, between the micro-neighbourhood sample and the random sample, in each of the following categories: demographic details; social networks and activities; contact; problems or difficulties for a neighbourhood; benefits for neighbourhood; knowledge of local centres for people with learning disabilities; type of people who attended these centres; knowledge of community-homes for people with learning disabilities; community care policy; weekend schemes and general comments. These tests were conducted to identify associations that existed between samples and each of the individual factors in the outlined categories. Where are cell sizes were too small in '2 × 2', tables, that is, when the expected cell count was found to be less than five, Fishers exact probability test was applied. For all two by two tables, Yates [70] continuity correction was used in order to accommodate for the use of a continuous probability distribution as an approximation to the discrete probability distribution [71]. The qualitative data from the verbatim comments made by respondents were not content analysed due to time constraints, but verbatim comments were included in the results and appendices. Where percentages did not add to 100%, this was due to rounding, multiple answers and exclusions of the 'don't knows'. Only the main findings will be discussed in the current chapter.
