using the sample information to determine the factor–level combination that will give the maximum yield. From the table, we see that crop yield increases when the nitrogen application is increased from 40 to 60 (holding phosphorus at 10). Yield also increases when the phosphorus setting is changed from 10 to 20 (at a fixed nitrogen setting of 40). Thus, it might seem logical to predict that increasing both the nitrogen and phosphorus applications to the soil will result in a larger crop yield. The fallacy in this argument is that our prediction is based on the assumption that the effect of one factor is the
same for both levels of the other factor. We know from our investigation what happens to yield when the nitrogen
application is increased from 40 to 60 for a phosphorus setting of 10. But will the yieldalso increase by approximately 20 units when the nitrogen application is changed from 40 to 60 at a setting of 20 for phosphorus? To answer this question, we could apply the factor–level combination of 60 nitrogen–20 phosphorus to another experimental plot and observe the crop yield. If the yield is 180, then the information obtained from the three factor–level combinations would be correct and would have been useful in predicting the factor–level combination that produces the greatest yield. However, suppose the yield obtained from the high settings of nitrogen and phosphorus turns out to be 110. If this happens, the two factors nitrogen and phosphorus are said to interact. That is, the effect of one factor on the response does not remain the same for different levels of the second factor, and the information obtained from the one-at-a-time approach would lead to a faulty prediction. The two outcomes just discussed for the crop yield at the 60–20 setting are displayed in Figure 14.2, along with the yields at the three initial design points. Figure 14.2(a) illustrates a situation with no interaction between the two factors. The effect of nitrogen on yield is the same for both levels of phosphorus. In contrast, Figure 14.2(b) illustrates a case in which the two factors nitrogen and phosphorus do interact.