Software by Alexis Dinno

Find here some of the fruits of my quantitative and computational geekery

cochranq

Cochran's Q test tests for stochastic dominance in b blocks of binary data across k groups, and is a nonparametric analog to the oneway repeated measures ANOVA. Cochran's Q can be considered a generalization of McNemar's test to an arbitrary number of groups. cochranq performs m = k(k-1)/2 multiple pairwise comparisons, and implements an array of family-wise error rate and false discovery rate methods to account for multiple comparisons.

cochranq is currently available for Stata on the web here, and from within Stata by typing:
net describe cochranq, from(http://www.alexisdinno.com/stata).

conovertest

The Conover-Iman test tests for stochastic dominance among multiple pairwise comparisons strictly following the rejection of a Kruskal-Wallis omnibus test of stochastic dominance among k groups, and is a nonparametric analog to the post hoc t tests following a oneway ANOVA. conovertest performs m = k(k-1)/2 multiple pairwise comparisons, and implements an array of family-wise error rate and false discovery rate methods to account for multiple comparisons. The Conover-Iman test is similar to Dunn's test, but is strictly more powerful given the prior rejection of the Kruskal-Wallis null hypothesis.

conovertest is currently available for Stata on the web here, and from within Stata by typing:
net describe conovertest, from(http://www.alexisdinno.com/stata)

conover.test is currently available for R on the web from CRAN.

dthaz

Discrete-time event history analysis is a powerful parametric regression technique for modeling whether and when events occur in abstracted (i.e. discrete) units of time. I have implemented methods introduced to me through the event history analysis curriculum of Judith Singer and John Willett within the context of educational and sociological research.

dthaz is currently available for Stata on the web here, and from within Stata by typing:
net describe dthaz, from(http://www.alexisdinno.com/stata).

dunntest

Dunn's test tests for stochastic dominance among multiple pairwise comparisons following a Kruskal-Wallis omnibus test of stochastic dominance among k groups, and is a nonparametric analog to the post hoc t tests following a oneway ANOVA. dunntest performs m = k(k-1)/2 multiple pairwise comparisons, and implements an array of family-wise error rate and false discovery rate methods to account for multiple comparisons.

dunntest is currently available for Stata on the web here, and from within Stata by typing:
net describe dunntest, from(http://www.alexisdinno.com/stata).

dunn.test is currently available for R on the web from CRAN.

Loop Analyst

Loop analysis is a body of methods for understanding the nature of system behavior in systems of causal feedback. Originally devised by Richard Levins as a means of creating testable hypotheses about the behavior of dynamic systems in population biology, loop analysis has a very general application in sciences concerned with modeling causal feedback.

Loop Analyst is currently available for R on the web here, and from CRAN.

A stand-alone version with GUI is in beta release, see the Loop Analyst page for details and download.

paran

Parallel analysis is a general method to reliably assess the underlying dimensionality of multicolinear data. Originally devised by John L. Horn, the method has been favorably compared to other dimensional analysis methods and modified slightly over the years.

paran is currently available as an ado for Stata on the web here, and from within Stata by typing:
net describe paran, from(http://www.alexisdinno.com/stata)

paran is currently available for R on the web from CRAN.

tost

Two One-Sided Tests (TOST) for equivalence provide evidence of no effect larger than a specified effect size. Combining the more common tests for difference with tests for equivalence is termed "relevance testing" and places both effect size and statistical power directly into the hypothesis testing process. the tost package provides a selection of TOST applications (t tests, z tests, normality tests, McNemar's test, and signed-rank and rank-sum tests.

tost is currently available as a package for Stata on the web here, and from within Stata by typing:
net describe tost, from(http://www.alexisdinno.com/stata)

Contact

As a rule, I am happy to take feedback, respond to requests for assistance, and consider feature requests for my projects. I can be contacted at: