February 05, 2004

Science confirms and expands our research findings

Most analysis of qualitative research is impressionistic – skilled interviewers interpret what they see and hear and report on it. As a result the interviewer’s knowledge and background can often influence the report. Until recently it had to be this way. Because qualitative research gives subjects the widest possible range of responses, they are impossible to quantify conventionally. Leximancer, a software tool developed by the Key Centre for Human Factors and Applied Cognitive Psychology at the University of Queensland changes that. This analysis of the first two weeks of our focus groups has been compiled using Leximancer.

Leximancer is a computer analysis of text which reconfigures a conversation as a map or scatter graph, assigning significance to a range of concepts within that conversation over an X-Y axis. The graph not only shows which concepts in the conversation cluster together, and which stand apart, but also indicates the strength of their relationships.

The conceptual map is dynamic: that is, the analyst can move variables using sliders and gain a richer insight into the text.

It charts those patterns of significance over a range of 100, and allows the reader to play with a number of sliding scales. The scales may open the map at the 6th percentile, 10th and so on: anything up to the 100th. The higher the percentile, the more central and significant the concepts. That is a set of concepts in the 10th percentile are more central and significant, than the concepts within the 50th percentile. However, the 50th would indicate a broader scope of issues.

There were some significant variations between the first and second weeks. In the first week at the 6th percentile the most central concepts which arose out of the discussion on the first night were Peter Beattie’s name and the Labor Party being strongly associated with voting preferences and key issues. This suggests that voting, issues and Beattie are strongly linked as central concepts in the discussion. However, much of the discussion seemed to be treating Beattie with a measure of ‘presidential’ style: that is, as the leader of Labor he, rather than his ministers, appeared to be gaining all the attention and responsibility for what the government had achieved so far, and what it was likely to achieve in the future.

Springborg also featured strongly as a central concept. His inclusion occurred on the 11th percentile. Although there was no sense in which government would be attributed to him or his party, he was generally admired as a ‘young hopeful’. He was significant for what he might achieve the next time round – in three years time. Respondents in the focus groups generally said they would ‘hesitate strongly’ before voting for him.

Whereas Beattie was considered a consummate politician, Springborg was considered to be a ‘thinking’ one.

Generally the issues were widely scattered and did not appear to cluster round particular parties or politicians. That is, the environment (tree clearing), and health (medicare) in particular. They were important issues in themselves, not because of particular party policies. That accords with our other observations that while these issues were important to particular groups of voters they are not actually vote movers.

In the second week attention was moving away from Beattie. Within the first ten percentiles of the Conceptual Map this time there was less mention of specific political parties. The central propositions were VOTE, LABOR and ELECTION.with Beattie coming well down the ladder on central significance factors – entering the conceptual map at the 13th percentile. Interestingly, and in contrast to the first groups, there was no strong association between these central concepts and Peter Beattie. The group seemed to be suggesting a Labor victory was inevitable, with Beattie towards the end of the first week featuring less strongly.

The Greens were the next political group to attain significance after Labor. However, they did not enter the map until the 24th percentile.

The National Party achieved no significance until the 50th percentile. National’s leader, Springborg, did not enter the map’s significance until the 70th percentile – scarcely rating in relative terms. This underlines our impressionistic interpretation that the Coalition has not reached minimum expectations, even in their role as Opposition. Opinion polls at this stage were showing increasing favourability for Springborg, and this was echoed in our count of responses to this question on our questionnaire. The Leximancer analysis suggests that while Springborg was climbing in favourability he was actually declining in significance.

The Leximancer analysis also supports our reading of the protest vote potential. It lies more with the Greens and Independents than it does with the Coalition, but while voters are not happy, there is little propensity for them to protest.

Posted by at February 5, 2004 09:37 PM
Graham Young
John Black
Mark Bahnisch
Michael Lee