Basic Statistics

The COUNTS command

The P-STAT COUNTS command produces frequency distributions and percentages for all unique values of all character and numeric variables in the input file. There may be an unlimited number of different values of each variable. Character variables may be any length. Cumulative counts and percentages are calculated, as are various means and measures of variation. If weighting is requested, both unweighted and weighted frequencies and statistics are computed. Multiple response or related variables may be combined and their values tallied as if they were single variables. Percentages of combined variables can be based on good or total responses or respondents. Value labels may be supplied. A single BY variable can be used to get subgroup results.

More than 40 different univariate statistics ranging from the mean and median to harmonic and geometric means, percentiles, moments and tests of normality. The output can be displayed as a nicely formatted table with selected contents or printed out in complete detail with maximum precision.

Other Statistics

TURF is a statistical model which can be used to answer questions such as:

• Where should we place our ads to reach the widest possible audience?
• If we add a new model to our line will we gain enough new customers to justify the expense?

For a complete information about the P-STAT implementation of TURF analysis see TURF.

The TTEST Command compares the means of two sample groups and tests the significance of the difference between the groups.

There are 21 non-parametric tests -- tests which make no assumptions about the distribution of the data. Included are one sample tests such as chi-square and Kolmogorov-Smirnov; two-independent-sample tests such as the Mann-Whitney U test; two-paired-sample tests such as the McNemar test. K-Independent and K-paired sample tests are also included. Both Spearman and Kendall rank correlation coefficients can be requested.

Exploratory data analysis includes nine procedures including stem and leaf and box plots. A separate BOX.PLOT command produces box plots using PostScript controls for presentation quality graphics.

There are commands to do correlation, step-wise regression, non-linear regression, analysis of variance, discriminant analysis, factor analysis, survival analysis and forecasting. Matrix operations combined with the programming language permit many other statistical techniques to be performed. These can be stored as macros so that they are easy to use.