Statistical Computing

A True Basic Program for Hypothesis Testing with Linear Regressions using Permutation Resampling by Xuejun Dong
This program is written with True Basic (www.truebasic.com) in order to test hypotheses of differences in slopes, elevations, arbitrary y values, or all of them, of two regression lines using permutation resampling of residuals. The program (PERMREG.exe) is a stand-alone application for a PC with Windows operating system installed.

A True Basic Program for Estimating Confidence Intervals of Linear Regressions Using Bootstrap Resampling by Xuejun Dong
This program is written with True Basic (www.truebasic.com) in order to construct the 95% confidence intervals for the slope, elevation, or the y value corresponding to an arbitrary x of a linear regression. The program (CI95_B.exe) is a stand-alone application for a PC with Windows operating system installed.

Probability Simulations using MINITAB and True Basic: with Selected Examples
by Xuejun Dong

In stochastic modeling, it is useful to be able to simulate a statistical distribution on the computer. In this short note, I will illustrate how to simulate simple probability distributions using MINITAB, an easy-to-use statistical program widely available to college students.

Calculating Wavelet Variance Associated with Discrete Wavelet Transform (DWT)
by Xuejun Dong and Haitao Li

The purpose of this program is to generate discrete wavelet transform and wavelet variance for one-dimensional data series with a length of one of the following choices: 64, 128, 256, 512, 1024, 2048, 4096, 8192, 16384, 32768, or 65536. The input data file must have the same format as test series 'pdsi.txt' and needs to be in the same folder as the executable program file. The outputs include (a) results of the multi-resolution analysis (MRA) based on different scales (the first three output files); (b) wavelet coefficients and scaling coefficients (the forth output file); and (c) the wavelet and Allan variances (the fifth output file). The executable True Basic Program and associated files are found in the following zipped folder: WVA.zip

MINITAB Macro for Statistical Comparison of Linear Regressions by Xuejun Dong

This document is a macro written for MINITAB that statistically compares up to 17 regressions. Statistically comparing slopes and elevations of more than 2 regression lines is not readily available from the standard output of statistical packages, but is useful in research. Using MINITAB, the author compiled a macro to conduct the analysis based on Jerrold H. Zar (1984): Bio-statistical Analysis (2nd Edition). The program accomplishes the following:

(1) compares overall differences for k1 linear regression lines in terms of slopes and y-intercepts.
(2) If some differences exist, conducts multiple comparisons (Tukey test) to detect which pairs are different.
(3) Calculates 95% confidence intervals for slopes, Y-intercepts, and Y-predictions corresponding to arbitrarily chosen X-values, or all of them at once.