A History of Distribution Sampling Prior to the Era of the Computer and its Relevance to Simulation

On the Effects of Non-Normality on the Distribution of the Sample Product-Moment Correlation Coefficient

“…The amount of disagreement exhibited above may be reduced by the observation that the distribution of r is in fact quite robust to certain types of non-normality and a realization of the difficulties (c.f., Teicheroew, 1965) inherent in simulating nonnormal distributions before the advent of electronic computers. Another important factor was the difficulty of estimating the distribution of r. This was done either by simple histograms or by fitting a Pearson-type curve to the observed distribution of r. Only in rare instances could the distribution of r be derived explicitly and hence it was necessary to rely on Monte Carlo experiments for the bulk of the theory at a time when this approach required massive quantities of data to get a reliable estimate of the density of r.…”

“…Good discussions of bivariate simulation are given by Mardia (1967) and Tocher (1963). Further references may be found in Shubick (1960) and Teicheroew (1965).…”

Quasi-Monte Carlo methods and pseudo-random numbers

“…Statisticians have intuitively used its principle long before that (cf. [334]), but it was only the computer age that could turn it into a systematic and viable technique. A number of excellent monographs explore the whole range of the method (cf.…”

Early Computational Statistics

“…A good deal of work was being done in the area of testing the generators for randomness (Teichroew 1960). Contemporaneously, and in one of the first caveats about the use of random number generators in statistical practice, Teichroew (1960) noted that "[.…”

“…Up until that point, random deviates were obtained from physical objects, such as Geiger counters; the article by Teichroew (1960) lists some interesting ways in which random deviates were obtained before the Los Alamos methodology became well known.…”