New in Version 8

• Firth's Penalized Maximum Likelihood Estimates (PMLE): Osteogenic Sarcoma
• Data with Missing Categorical Covariates: Logistic Regression Model(1)
• Data with Missing Categorical Covariates:
  Binary Logistic Regression Model (2).
• Quality of Life Data: Missing Categorical Covariates Method
  (Model: Multiple Linear Regression)
• Polytomous Regression : Unordered Response Model
• Polytomous Regression : Ordered Response Model
• Stratified Poisson Regression.

Examples Library

• Insurance Fraud
• Osteogenic Sarcoma
• Adenocarcinoma Tumors in the Small Intestine
• Repeated Measures Data: Cross-over Clinical Trial
  StatXact & LogXact:
• StatXact® & LogXact®: Is the New Analgesic More Effective than Aspirin?

Cytel's LogXact vs. SAS's PROC LOGISTIC

We invite you to compare LogXact, and LogXact PROCs, to SAS's PROC LOGISTIC. We have posted six problems given to us by statistical analysts in a variety of fields that we believe SAS 9's PROC LOGISTIC cannot handle. Download the data and run the SAS code that comes with it. When you are convinced that SAS can't solve these problems, email us at sales@cytel.com and ask for a free 30-day trial of LogXact or LogXact PROCs.

Execution Times for Cytel's LogXact and SAS 9's PROC-LOGISTIC

Data Set	SAS's PROC LOGISTIC	Cytel's LogXact® 4	Cytel's LogXact® 5
Crash	Can't solve	32 min	2 min
UTI	Can't solve	Can't solve	11 min
Death Penalty	Can't solve	Can't solve	15 min
Gene	Can't solve	Can't solve	6 min
Titanic	Can't solve	Can't solve	4 min
Diarrhea	Can't solve	Can't solve	12 min

All problems were run on a Pentium 4, XP operating system, with 1GB RAM.

The links below include an explanation of the problem, results, plus SAS code and data files that you can download.

1. CRASH — Calibration of crash dummies in automobile safety tests
2. UTI — Urinary tract infection and contraceptive use
3. DEATH PENALTY — Racial bias in death sentencing in New Jersey
4. GENE — Relationship between genetic mutation and colon cancer
5. TITANIC — Relationship between survival and status on the ship
6. DIARRHEA — Antibiotics and diarrhea in hospitalized patients

The universe of datasets for which exact logistic regression is needed is potentially limitless. While no logistic regression software can claim to solve all such problems, LogXact solves a substantially larger number of these problems compared to SAS's PROC LOGISTIC. Moreover, Cytel has unmatched experience and expertise in developing algorithms for exact inference and a tradition of advanced research and development.