TABLE OF CONTENTS There are many occasions on which it is of interest to investigate the possible relationship between two variables, where one variable, the predictor variable (often denoted X), is thought of as driving the second variable, the response, Y. For example, in Chapter 5 several examples referred to the influence of quantitative variables such as injection temperature, split ratio, or injection volume, on the peak areas produced by a GC system. There the concern was exploratory - to determine if moderate changes to particular variables influenced the response of the system, and whether any influences were dependent on the levels of other variables. If the intention were to investigate in detail the influence of, say, split ratio, then instead of just two levels, the study might involve five or six different split ratio values. The analysis would involve fitting a graphical or numerical relation between the response, peak area, and the predictor, split ratio. Either of these would provide a description of how peak area changes as split ratio is varied, and would allow prediction of the peak area that would be expected at any given split ratio level.
This chapter will be concerned with the investigation of, mainly, straight-line relationships. The first step in any such investigation should be to plot the data. In many cases this will answer the questions of interest: does there appear to be a relationship between the variables? If yes, is it linear? Do the variables increase together or does one decrease...
