Although a great deal can be learnt from univariate analysis of the kind discussed in previous chapters, it is desirable, particularly when dealing with economic data, to try to consider sets of variables in combination. Firstly, as Singh pointed out,1 two firms with the same profitability could show different rates of return if they had different growth rates. Secondly, equally healthy firms faced with a trade-off between growth and profitability might well choose different points on the trade-off; and the choice might not be (though it could be) systematically related to competition policy. Thirdly, a poor performance by a firm on a set of related variables may discriminate better between firms than a poor performance on a single variable considered in isolation.