Interpreting Economic and Social Data
Othmar W. Winkler
Georgetown University, Washington, DC, United States

Statistics courses for economists and social scientists ignore the interpretation of socio-economic data. This discussion of common pitfalls and problems highlights the need for such teaching. First off, data users ignore the fact that socio-economic data are quite different from those in the natural sciences. Our data are three-dimensional aggregates, of subjective, self-reported facts, not, as is mistakenly assumed, point-like objective measurements like those in the natural sciences, made by expert scientists. The row-entries in published tabulations and their totals are such aggregates. Because of their different sizes+ they are of different degrees of vagueness which ought to be considered when interpreting them. Then 'adjusting' such data with a seasonal index for seasonality or 'deflating' their monetary value with as a Price Index does not change them but creates something new: ratios that are to be interpreted differently. Then consider official Price Index Numbers with a fixed 'market- basket of products/services'. Multiplying current prices/unit with these fixed quantities creates the mistaken impression to obtain the 'true' change in the variable p by isolating and holding constant the variable q. If one really could freeze the quantities q at a base-period level, actual prices would differ from those used in the Price Index. Analysts seem unaware of this. An unexpected paradox arose when attempting to interpret the regressions between 'years of service' and 'salary' in a government agency, computed separately for the male and female professionals of equal qualification. This stunning paradox led to the important insight that data in regression analysis are static cross-sections, not to be interpreted as dynamic- longitudinal. Finally, the ogive of frequency distributions of socio-economic data reveals features of society that are not noticed otherwise. These examples should alert the statistical community to the need for greater emphasis on the proper interpretation of socio-economic data.

Keywords: Descriptive statistics; Socio-economic data

Biography: Othmar W. Winkler is a Professor emeritus, Diplom Volkswirt, (MA), and Dr.rer.pol. (Doctor rerum politicarum), Universitaet Wien, 1948. Since then having taught Statistics for 60 years. Beginning at the Facultad Economia, Universidad Central de Venezuela, Caracas 1948-1954; Teaching Price, Labor and Production Statistics at the CIEF (Centro Interamericano de Enseñanza Estadística Econόmica y Financiera), Santiago, Chile, 1954-1959. Teaching Economic and Business Statistics at Marquette University, Milwaukee. 1959-1961. Teaching Business and other Statistics courses at Georgetown University, Washington, D.C. 1961-1993 and committee work til 2007. He also did extensive statistical consulting for research in academics, industry and government.