The statistical procedures available in StatTools come in the following natural groups.

Statistical Inference: This group performs the most common statistical inference procedures of confidence intervals and hypothesis tests.

Forecasting: StatTools gives you several methods for forecasting a time series variable. You can also deseasonalize the data first, using the ratio-to-moving-averages method and a multiplicative seasonality model. Then use a forecasting method to forecast your deseasonalized data, and finally “reseasonalize” the forecasts to return to original units.

The outputs include a set of new columns to show the various calculations (for example, the smoothed levels and trends for Holt’s method, the seasonal factors from the ratio-to-moving-averages method, and so on), the forecasts, and the forecast errors. Summary measures such as MAE, RMSE and MAPE are also included for tracking the fit of the model to the observed data. Finally, several time series plots are available, including a plot of the original series, a plot of the series with forecasts superimposed, and a plot of the forecast errors. In cases using deseasonalized data, these plots are available for the original and deseasonalized series.

Classification Analysis: StatTools provides both discriminant analysis and logistic regression. Discriminant analysis predicts which of several groups a variable will fall in, and logistic regression is a nonlinear type of regression analysis where the response variable is 0 or 1 for “failure” or “success.” You can then estimate the probability of success.

Data Management: This group allows you to manipulate your data set in various ways, either by rearranging the data or by creating new variables. These operations are typically performed before running a statistical analysis.

Summary Analyses: This group allows you to calculate several numerical summary measures for single variables or pairs of variables.

Tests for Normality: Because so many statistical procedures assume that a set of data are normally distributed, it is useful to have methods for checking this assumption. StatTools provides three commonly used checks: Chi-square, Lilliefors, and Q-Q plot.

Regression Analysis: For each of these analyses, the following outputs are given: summary measures of each regression equation run, an ANOVA table for each regression, and a table of estimated regression coefficients and other statistics. In addition, StatTools gives you the option of creating two new variables: the fitted values and residuals. Plus, you can create a number of diagnostic scatterplots.

Quality Control Charts: This set of procedures produces control charts that allow you to see whether a process is in statistical control. Each of the procedures takes time series data and plots them in a control chart. This allows you to see whether the data stay within the control limits on the chart. You can also tell if other nonrandom behavior is present, such as long runs above or below the centerline. Each of these procedures provides the option of using all the data or only part of the data for constructing the chart. Furthermore, each lets you base the control limits on the given data or on limits from previous data.

Nonparametric Tests: Nonparametric tests are statistical procedures which can be used to make successful inferences when there is little available data. They are more robust than many of the widely known parametric hypothesis tests. Nonparametric tests do not always need the parametric assumptions—such as normality—or generalized assumptions regarding the underlying distribution. In most cases, the nonparametric tests are much easier to apply and provide clearer interpretation than traditional parametric tests.