World Cultures eJournal

A measure of modernization m ≥ 0 is created applicable to both preindustrial and contemporary societies. A sample of 174 preindustrial societies are coded for m ≥ 0, time t, 1800 ≤ t ≤ 1965 AD, 25 binary constructs X = 1, 0 (e.g. X = high, low divorce rate) and one continuous construct X > 0 (population density). A sample of 189 contemporary countries at 2000 AD is coded for the same constructs. For the 25 binary constructs parametric logistic regression functions P(X = 1) = f(m, t) are fitted to the combined sample of 363 societies. The candidate predictor set is powers from 1 to 3 of m, t and mt. Backward selection (α = 0.05) is used to reduce the candidate predictor set where appropriate. Since the 174 preindustrial societies are equally distributed over the 19th and 20th centuries it is assumed that the fitted models hold over 1800 AD ≤ t ≤ 2000 AD, although not necessarily prior to 1800 AD. Functions P(X = 1) = f(m) are fitted to the sample of 189 countries at 2000 AD. For the continuous construct X > 0 regression models X = h(m, t), h(m) are fitted. Tests for monotonic trend over m are fitted to both samples for all 26 constructs - either a Cochran-Armitage test (for binary constructs) or a Spearman rho test (for the continuous construct). Forecasts are made by extrapolating the fitted models to out of sample values of the predictor variable(s) up to 2050 AD, or by linear approximation. Internal checks are created to enhance forecast validity. Extension of the forecast method to new subject matters is considered. Business and policy applications are suggested. Strengths and limitations of the