The present study finds that shareholder returns, in terms of ROE, are superior when there is CEO duality. This might be interpreted that shareholder returns cause CEO duality. But even if there is a positive effect of performance on duality, the agency theory explanation would hold only if the correlation between duality and performance was the net result of that positive correlation and the negative effect of duality on performance hypothesised by agency theory. Normally, one would expect that the net result of a negative and positive effect would cancel each other out leaving a correlation about nil. Here the observed correlation is positive between duality and performance, meaning that if there was a positive effect of performance on duality this would have to be more than the supposed negative effect of duality on performance. Only then is there a masking effect which can be used to explain why the agency theory negative effect of duality on performance fails to show the expected negative correlation herein between duality and performance. In order to examine this question there is a need for future research to go beyond the present cross-sectional design to a longitudinal research design into causality.