Variable Reduction
Often the large volume of data available is not just because of the number of records but because there are many variables held within each record. Often these are just subtly different measures of the same item or attitude. These techniques are used to reduce the number of variables by combining the information into new variables that can then be used in further analysis. By themselves they can be used to identify, say, the main attributes of respondent attitudes.
Factor Analysis
There are a number of factor analysis techniques that are designed to reduce the number of variables to a number of underlying variables. They analyze interrelationships among a large number of variables and ‘explain’ these in terms of common underlying dimensions, known as factors. The aim is to produce the smallest number of factors, or new variables, with least loss of data.
Principal Component Analysis
Though in use this is very similar to factor analysis, the underlying principles are different. Large numbers of variables are reduced to a smaller number of principal components (new variables). However, the technique retains all of the original variability in the data. Often only the first principal components are used, so that in use this is very similar to factor analysis. |