Structural view independence: A criterion for judging the objectivity of economic parameters measured by opinion survey

Economic measurements have a great influence over all our lives, but as with other soft measurements, significant effort is needed to ensure their objectivity. This is particularly true of the process of consolidating into a single representative figure different people's valuations of a non-market good (“views”), as measured by opinion survey. Sometimes the views are transformed and averaged before being returned to the original domain as the back-transformed mean. Examples are the geometric mean resulting from a logarithmic transformation and the root-mean-squared (r.m.s.) value from a square transformation. Such transformations are tested for objectivity using the new criterion of structural view independence. It is shown that an analyst using any general, nonlinear, increasing and differentiable transformation other than the linear transformation can know at the outset that he is giving greater weight to views of his choosing, meaning that he has no claim to objectivity. Of all such transformations, only the linear transformation possesses the desirable property of structural view independence. The resultant sample mean is objective and the only consolidated figure recommended for human views.