DOCUMENT RESUME

Distributions and linear regressions are discussed. The section dealing with the former topic emphasizes the usefulness of computer programs in statistics, demonstrating their ability to handle tedious and time-consuming tasks. The normal curve is stressed since the assumption of a normal distribution is common. An example of 200 data points is presented which illustrates a computer program's ability to do descriptive statistics for data not grouped, data which is distributed into any number of classes of equal width and testing if the 200 points are normally distributed. The section of linear regression emphasizes how computer programs can lead to erroneous results. In the area linear regression there are in current use "canned programs " which display the t-values corresponding to the coefficients of the least square estimate, but do not take autocorrelation into account. In the presence of autocorrelation it is incorrect to use these t-values. An example is presented in which autocorrelation is present but can be removed by a suitable transformation of variables. (Author