Academic journal article Social Behavior and Personality: an international journal

Lay and Professionjal Perspectives of the Causes of Health: A Comparative Network Approach

Academic journal article Social Behavior and Personality: an international journal

Lay and Professionjal Perspectives of the Causes of Health: A Comparative Network Approach

Article excerpt

Lay and professional interpretations of the structure of causal factors previously demonstrated to be influencing health were examined through network analysis. Fifty-three first level registered nurses and sixty-three social science students were asked to complete network adjacency grids for factors that affect health. There is considerable agreement between the lay and professional view - with stress, lifestyle, health knowledge, work and home environment forming a nexus of causes, along with physical constitution and exposure to illness which have direct links to health. The impacts of both a belief in alternative medicine and religiosity, which have been suggested to influence beliefs about health, were also investigated. Religiosity had no major effect on the model. Believers in alternative medicine rated causal links involving stress higher, but showed no difference in their ratings of the importance of medical treatment. The results are compared and contrasted with previous work which used factor analysis.

It has been argued on many occasions that lay representations of the causes of health and illness have important consequences (Blaxter, 1983; Lau, 1982; Skelton & Croyle, 1991; Farnham, 1994). Previous work looking at representations of social phenomena has relied on analysis of spontaneous discourse (Blaxter, 1983; Campbell & Muncer, 1987), semi-structured interviews (Litton & Potter, 1985), Q sorts (Stainton-Rogers, 1991) or factor analysis of questionnaire responses about the importance of perceived causes (Farnham, 1994). While these approaches are useful for revealing the causal factors involved, they do not reveal the relationships between them.

Network analysis has an advantage for this type of research in that it can reveal patterns of causal sequence. The method specifically asks participants to state if one cause leads to another and vice versa, by asking them to directly rate the strength or likelihood of causal links between nominated causes of the phenomenon. In the most commonly used procedure, participants are given a directional adjacency matrix with causes listed on the top and sides of the matrix, and asked to indicate the strength of the causal link between each pair of causes (row to column). The lay interpretations of examination failure (Lunt, 1988), personal debt (Lunt & Livingstone, 1991, crime (Campbell & Muncer, 1990) drug use (Muncer, Sidorowicz, Epro & Campbell, 1992) date rape (Gillen & Muncer, 1995) loneliness (Lunt, 1991) and poverty (Heaven, 1994) have been investigated using this method of data collection.

The original form of analysis, however, has been criticised for being atheoretical and has been shown to produce non-consensual and unreliable networks (Muncer & Gillen, 1997). To correct this, Muncer and Gillen (1997) devised inductive eliminative analysis in which a network is constructed adding causal links in order of their ratings and, as each new cause is added, the entire network is checked for endorsement. This produces a consensual network that is both more reliable and interpretable.

Inductive eliminative analysis was first developed to deal with adjacency matrices that asked subjects to state whether there was a link (in which case they put 1) or no link existed (in which case they put 0). In this way, each network could be checked for endorsement and as soon as a participant put 0 for any link he/she would be considered to be not endorsing the network. Recent network studies have used a five-point scale for rating the strength of perceived link. When a five-point scale is used the analysis becomes more complicated (Muncer & Gillen, 1997; Taylor, White & Muncer, 1999).

In the present paper, the authors adapted the inductive eliminative analysis method so that a participant's average level of endorsement of a network was the deciding factor. Hence, if a participant had an average endorsement of the entire network of over four then he/she would be considered to endorse the entire network. …

Search by... Author
Show... All Results Primary Sources Peer-reviewed

Oops!

An unknown error has occurred. Please click the button below to reload the page. If the problem persists, please try again in a little while.