Furthermore, Lunt’s method did not allow participants to make direct links between potential causes and the target under scrutiny. It has been argued elsewhere (e.g., Heffernan, Green, McManus, & Muncer, 1998) that this does not allow any understanding of the relationship of causal factors to the target phenomenon and may also lead to overly elaborate networks as participants seek to produce repre- sentations without being able to relate to the central concept. In our study however, participants were able to make links with the target of exam failure. This had two important effects. First, the resultant representation was simpler than that produced by the binary method, in that it contained both fewer links and fewer causes, and second, the causal paths with the highest levels of agreement all connected directly with exam failure.