# Difference between revisions of "Guide to statistics software"

(→price overview) |
(→several) |
||

Line 32: | Line 32: | ||

=== several === | === several === | ||

"five <u>user friendly, menu-driven programs</u> that are intended mainly for users with <u>beginner to intermediate</u> levels of statistical expertise: <u>JMP, NCSS, Sigmastat, Statistica and True EPISTAT</u>. Of these, the first <u>four are general purpose</u> data analysis packages while True <u>EPISTAT is intended for medical scientists</u>. All have large user manual full of simple examples. <u>None offer anything like the same scope for power analysis as they do for statistical hypothesis testing</u>. This means that a researcher relying solely on one of these packages could analyze a study without being able to properly plan it or check its power. Sigmastat offers the widest scope for planning power or sample size (Table 3), although the ANOVA option doesn’t go beyond one-way. JMP is the most comprehensive in this respect, offering power calculations, tables, and power vs. sample size graphs for almost any fixed effect ANOVA design (but for nothing else). True EPISTAT can calculate approximate power and sample size for a limited assortment of tests and can produce graphs and some very limited tables. In Statistica, power and sample size can be calculated for some one-sample tests using the Process Analysis module, and OC (power vs. sample size) curves can be produced." [http://www.zoology.ubc.ca/~krebs/power.html] | "five <u>user friendly, menu-driven programs</u> that are intended mainly for users with <u>beginner to intermediate</u> levels of statistical expertise: <u>JMP, NCSS, Sigmastat, Statistica and True EPISTAT</u>. Of these, the first <u>four are general purpose</u> data analysis packages while True <u>EPISTAT is intended for medical scientists</u>. All have large user manual full of simple examples. <u>None offer anything like the same scope for power analysis as they do for statistical hypothesis testing</u>. This means that a researcher relying solely on one of these packages could analyze a study without being able to properly plan it or check its power. Sigmastat offers the widest scope for planning power or sample size (Table 3), although the ANOVA option doesn’t go beyond one-way. JMP is the most comprehensive in this respect, offering power calculations, tables, and power vs. sample size graphs for almost any fixed effect ANOVA design (but for nothing else). True EPISTAT can calculate approximate power and sample size for a limited assortment of tests and can produce graphs and some very limited tables. In Statistica, power and sample size can be calculated for some one-sample tests using the Process Analysis module, and OC (power vs. sample size) curves can be produced." [http://www.zoology.ubc.ca/~krebs/power.html] | ||

+ | |||

+ | = Reviews of free statistical packages = | ||

+ | |||

+ | * Graphical Interfaces to R [http://www.decisionstats.com/interfaces-to-r/] Posting on DecisionStat website, October 5, 2010 By Ajay Ohri. | ||

+ | * Journal of Statistical Software, special issue devoted to R, June 2012 [http://www.jstatsoft.org/v49] | ||

+ | * Comparisons of statistical output of multiple free statistical programs [http://gsociology.icaap.org/methods/soft.html] | ||

=== SigmaStat === | === SigmaStat === |

## Revision as of 20:55, 31 August 2012

back to stats portal |

## Contents

## price overview

### free of charge

- R software package [1]

### commercial

All price are rounded estimates as of Spring 2007 for a single, educational license of the latest version (not upgrades). Prices vary with reseller. Follow the links for exact information. Items are roughly ordered starting with the cheapest price in Euros.

- SigmaStat v3.5 €445 [2] / v3.5 $480 [3]
- SigmaPlot (usually bundled) [4] / v10 $480 [5]
- MatLab GBP370/€500 [6] (requires login)
- SysStat v? €700 [7] / v12 $490 [8]
- Origin 7.5 €720 [9] / $490 [10]
- OriginPro 7.5 €950 [11] / $700 [12]
- JMP v6 €920 [13]
- SPSS v15 Base €1720 [14]

commercial but pricing not yet entered:

- SAS

## software by license type

### open source

- R

### commercial

- SPSS
- SysStat, SigmaStat, SigmaPlot
- SAS
- Origin

## review extracts

### several

"five __user friendly, menu-driven programs__ that are intended mainly for users with __beginner to intermediate__ levels of statistical expertise: __JMP, NCSS, Sigmastat, Statistica and True EPISTAT__. Of these, the first __four are general purpose__ data analysis packages while True __EPISTAT is intended for medical scientists__. All have large user manual full of simple examples. __None offer anything like the same scope for power analysis as they do for statistical hypothesis testing__. This means that a researcher relying solely on one of these packages could analyze a study without being able to properly plan it or check its power. Sigmastat offers the widest scope for planning power or sample size (Table 3), although the ANOVA option doesn’t go beyond one-way. JMP is the most comprehensive in this respect, offering power calculations, tables, and power vs. sample size graphs for almost any fixed effect ANOVA design (but for nothing else). True EPISTAT can calculate approximate power and sample size for a limited assortment of tests and can produce graphs and some very limited tables. In Statistica, power and sample size can be calculated for some one-sample tests using the Process Analysis module, and OC (power vs. sample size) curves can be produced." [15]

# Reviews of free statistical packages

- Graphical Interfaces to R [16] Posting on DecisionStat website, October 5, 2010 By Ajay Ohri.
- Journal of Statistical Software, special issue devoted to R, June 2012 [17]
- Comparisons of statistical output of multiple free statistical programs [18]

### SigmaStat

"This is a __good starter package__. __Not very powerful__, and not all that flexible, but __simple and instructive__" [19]

### MatLab

"Particular advantages were listed as: ease of __customisation, informative error messages, the ability to handle large matrices__. The main __disadvantage was the time taken to learn__ the application. The UNIX version is command-line driven."

## links

- R Project page, R Statistics on OWW
- Review of visualisation tools in statistics by the Queen's Uni Belfast (Social Sciences)
- Statistical (and other) Packages Available as Free Software
- John C. Pezzullo's list of free statistical software