8 Factor Analysis Two
It was clear in the analysis of questionnaire one that questions listed as important in a
principal component can include things that do not fit easily into any general
interpretation. Nor does ‘rotation’ necessarily produce results that are readily intelligible.
Since in comparison with the first questionnaire the second was more influenced by
comments made during interviews and less influenced by prior interests of the research
team, the data was much less ‘tidy’ and previously used methods of analysis were less
successful. A more effective approach was devised here for identifying main themes or
structures in the data. The following paragraph outlines the technical procedure used to
arrive at clusters of variables.
In technical terms, PCA was used to locate each question within and abstract multi-
dimensional space, using component loadings as Euclidean co-ordinates. A simple cluster
analysis was then used to find a structure for the questions. In less technical terms, the
analysis aimed, as far as possible, to show which questions were most closely associated
with which, but also to show how each of these small groups of questions was related to
others.
Nine clusters of questions emerged. Three items English and European nationality and
Q3 Nothing really changes where I live were not closely associated with anything else
and do not appear in this analysis. Clusters are presented in order of appearance in the
SPSS output. This has no particular significance.
Clusters 1 to 4 and Clusters 5 to 8 form two groups with relatively close internal
relationships. Cluster 9 and Cluster 10 are isolated.
Cluster 1
The first cluster (see Appendix 8e) contains two main subgroups. The first deals with
religion and ethnic background. The three questions on religion were especially closely
related.
Q19d I am taught how to have good character by my religion.
Q22 Religion is very important in my life.
Q23 Religion helps me be a better person.
Q21 I am proud of my ethnic background.
The second might be seen as dealing indirectly with the same issues.
Q46 Getting married is important.
Q48 I sometimes give to charity.
Q50 I would like to see more black and Asian MPs in Parliament.
A final item is not more closely associated with either one of these two subgroups, but is
clearly part of the first cluster. Note that this item is reversed. A high score indicates
disagreement.
82