Date: Mon, 21 Jan 2008 15:23:24 +0100
Reply-To: David Young <email@example.com>
Sender: "SAS(r) Discussion" <SAS-L@LISTSERV.UGA.EDU>
From: David Young <dyoung@TELEFONICA.NET>
Subject: Re: Need training on SAS - can I convince the boss to pay?
Content-Type: text/plain; charset=ISO-8859-1
I'm not very familiar with SPSS, but based on some comments of
associates that used it I can say that SAS handles missing data much
more logically. The people I saw using SPSS were inserting ZERO
values for all the data that was missing in order to get SPSS to do
some calculations. This obviously has the risk of skewing many
results depending upon what the calculation was.
SAS on the other hand generally handles missing values in the way that
you would want it to. Examples:
* Regression applications exclude observations that would use a
variable with a missing value.
* Frequency counts exclude missing values from percentages by default,
but can optionally include missing as a separate category not merged
with zero values.
* Functions summing across variables like MEAN,MAX,STD, etc. will use
all the non-missing values and exclude the missing ones.
By my way of thinking that is pretty much what you would want to have
happen in each case and no artificial manipulation of zero values is
needed, so it is fast, accurate, and without risk of summing incorrect
On the other hand, to be fair to SPSS, SPSS is better at reading
delimited files and external file types. SPSS normally reads external
files correctly, while SAS reads the first few observations in order
to guess at the format and then assumes that format for the rest of
the file. This often results in errors, which are correctable
manually, but you have to always check it and often fix it yourself.
At one time there was a third party program known as DBMS Engines that
would allow SAS to read and write to external files, but SAS bought
them and then trashed the program. Now you can only pay a lot extra
for SAS/ACCESS which doesn't work as well as SPSS's base package.
Hope that helps.
Monday, January 21, 2008, 2:14:18 PM, you wrote:
l> I am looking for some advice regarding the following. I currently use
l> SPSS for my work, and spend a lot of time aggregating datasets,
l> manipulating and cleaning up data, modelling continuous and
l> categorical data in regression, and standard descriptive statistics.
l> Data is a mixture of financial and customer/HR survey data.
l> Many of the analyst jobs in the market are looking for SAS experienced
l> analysts. SPSS less so. I am only on a 6-month contract and want to
l> upskill in SAS so that I have at least some exposure to the software.
l> Here's the crunch : What does SAS offer that SPSS does not? Why
l> should they pay for a course (£1400)?
l> As I've never used SAS before, it's quite difficult to argue in favour
l> of it...
l> Any ideas would be appreciated...