The last of these questions is very important. Has a slowdown actually occured? The data summarized in the previous section are highly aggregated, and, as such, are subject to error. One might ask: "Are we dealing with a sick patient or a sick stethoscope?" [145, p. 123]. Darby [73], among others, contends that the stethoscope is sick. 37 As noted earlier, a number of micro-level production studies have indicated clearly that the estimated residual A(t) diminishes when input quality adjustments are made. But as Rees [271] observed, for such data problems to have been the sole contributor to the measured slowdown implies that downwardbiasing quality mismeasurements must have been growing substantially during the 1970s. The opposite is more likely.