learning curve and utility of various free statistics programs

[Esther]

I hope this is the correct forum for this question. If not, perhaps someone could suggest another place to ask it.

How do the following programs compare on (a) learning curve, and (b) utility for basic-ish stats procedures such as regression, ANOVA, MANOVA, and (c) ease of data handling, e.g., importing from Excel.

I currently use SPSS and work with social science statistics.

Here's the list (other suggestions also welcome):

R
OpenStat
PSPP
Osiris

Thank you.

Esther

[David Duffy]

Post by Esther
How do the following programs compare on (a) learning curve, and
(b) utility for basic-ish stats procedures such as regression, ANOVA,
MANOVA, and (c) ease of data handling, e.g., importing from Excel.
R
OpenStat
PSPP
Osiris

R via the Rcmdr package might be worth a look
http://socserv.mcmaster.ca/jfox/Misc/Rcmdr/

Cheers, David Duffy.

[David Jones]

Post by Esther
I hope this is the correct forum for this question. If not, perhaps
someone could suggest another place to ask it.
How do the following programs compare on (a) learning curve, and (b)
utility for basic-ish stats procedures such as regression, ANOVA,
MANOVA, and (c) ease of data handling, e.g., importing from Excel.
I currently use SPSS and work with social science statistics.
R
OpenStat
PSPP
Osiris
Thank you.
Esther

Of these, I only have experience with R and then only to a limited extent.
The only relevant things I can say are:
(i) R has some extensive material available online, in particular
learning material.
(ii) There are many published books having "using R" in the title (e.g
http://www.amazon.co.uk/Statistics-An-Introduction-Using-R/dp/1118941098/ref=dp_ob_title_bk),
whereas I have seen none based on these other packages.

[Rich Ulrich]

Post by Esther
I hope this is the correct forum for this question. If not, perhaps someone could suggest another place to ask it.
How do the following programs compare on (a) learning curve, and (b) utility for basic-ish stats procedures such as regression, ANOVA, MANOVA, and (c) ease of data handling, e.g., importing from Excel.
I currently use SPSS and work with social science statistics.
R
OpenStat
PSPP
Osiris

It looks like a nice over-view at
https://en.wikipedia.org/wiki/Free_statistical_software

I don't remember anything about OpenStat or Osiris.
In addition to what I see in the Wikip article --

PSPP was mentioned a umpteen years ago in the SPSS
group -- It was (originally) provided as a free version
of SPSS, so it was similar to SPSS but always behind.
My wild guess would be that IO would be behind, as
would the fanciest new procedures. Since Wikip says
that there is a List, you should browse there.

IIRC, R was initially (30+ years ago) a similar sort of substitute
for an expensive and sophisticated commercial package
called S. I haven't heard S mentioned in a long time, but
R has thrived. It now has a collection of programs and
procedures contributed by a large number of specialists,
which are, not-infrequently, state-of-the-art. There are
programs that SPSS makes available /only/ by using the
plugin for accessing R. There is a mailing list, and R has
a good reputation for its maintainers being good about
tackling reported errors.

Compared to using SPSS, coding data manipulations in
R is more like writing for the Matrix parser than writing
ordinary syntax. I think.

I think R does not have the nice features of SPSS
(or SAS) for easily editing and labeling variables; nor, the
features for editing output files, if you want those.
[Does it create plain text?]

[Esther]

Thanks to all who wrote.
Now I have a followup - does anyone know which one of the free programs would be good for path analysis?
Thanks again.

-Esther

Post by Rich Ulrich

Post by Esther
I hope this is the correct forum for this question. If not, perhaps someone could suggest another place to ask it.
How do the following programs compare on (a) learning curve, and (b) utility for basic-ish stats procedures such as regression, ANOVA, MANOVA, and (c) ease of data handling, e.g., importing from Excel.
I currently use SPSS and work with social science statistics.
R
OpenStat
PSPP
Osiris

It looks like a nice over-view at
https://en.wikipedia.org/wiki/Free_statistical_software
I don't remember anything about OpenStat or Osiris.
In addition to what I see in the Wikip article --
PSPP was mentioned a umpteen years ago in the SPSS
group -- It was (originally) provided as a free version
of SPSS, so it was similar to SPSS but always behind.
My wild guess would be that IO would be behind, as
would the fanciest new procedures. Since Wikip says
that there is a List, you should browse there.
IIRC, R was initially (30+ years ago) a similar sort of substitute
for an expensive and sophisticated commercial package
called S. I haven't heard S mentioned in a long time, but
R has thrived. It now has a collection of programs and
procedures contributed by a large number of specialists,
which are, not-infrequently, state-of-the-art. There are
programs that SPSS makes available /only/ by using the
plugin for accessing R. There is a mailing list, and R has
a good reputation for its maintainers being good about
tackling reported errors.
Compared to using SPSS, coding data manipulations in
R is more like writing for the Matrix parser than writing
ordinary syntax. I think.
I think R does not have the nice features of SPSS
(or SAS) for easily editing and labeling variables; nor, the
features for editing output files, if you want those.
[Does it create plain text?]
--
Rich Ulrich

[Bruce Weaver]

Post by Esther
Thanks to all who wrote.
Now I have a followup - does anyone know which one of the free programs would be good for path analysis?
Thanks again.
-Esther

The lavaan package for R is becoming quite popular for estimating structural equation models (SEM), including path analysis & confirmatory factor analysis. See the links below for more information.

http://lavaan.ugent.be/
http://lavaan.ugent.be/tutorial/sem.html
http://blogs.baylor.edu/rlatentvariable/sample-page/r-syntax/
https://groups.google.com/forum/#!forum/lavaan

HTH.

[Ezra Boyd]

Here are a few types from my experience:

- Most programs these days can read .xls, xlsx. In base R, it's not straightforward, but there are a few packages (both command line & gui) to simplify it. On the other hand, it's usually just as easy to export the spreadsheet to .cvs, which all programs can read without struggle.
- SPSS's coach feature was a life saver when I was working toward my MS. When I tried PSPP many years ago, it did not the coach but did have decent documentation.
- If you're just doing correlation, ANOVA, linear regression, etc. PSPP should work fine.
- If you want to do anything beyond the basics, you'd probably want to use R
- When I first started using R (when I started my PhD program) I printed out the manual, spent one day reading it, and by the next was able to figure out enough to run a linear regression. I had some, but not a lot of prior programming experience.
- When it was time to move beyond a linear regression, I found R's approach very intuitive and easy to use.
- Just about everything in R is well documented. Sitting down to read the documentation before jumping in is not the most exciting part of the job, but it always pays off