[StATS]: What’s New in SPSS version 14.0 (August 17, 2005)
I attended a web seminar, What’s New in SPSS version 14.0, taught by Meta Brown, a Senior Systems Engineer with SPSS, Inc. Version 14 will ship in October 2005. The major improvements in version 14.0 include
- more powerful data management (multiple data sets open in a single SPSS session) and
- enhanced reporting (a new chart builder interface, and a graphics production language, GPL).
in the SPSS base product, and
- easy data validation (validate across multiple variables),
- easier more powerful forecasting (an expert modeler offers sophisticated fitting of time series models that can now incorporate independent variables and forecast many series at once),
- extended programmability (more complex functions like flow control and extensibility), and
- expanded structural equation modeling options (bayesian estimation and data imputation).
in the SPSS product family.
The licensing allows first access to advanced modules to “power users.” There is also a new commuter license that allows you to borrow a license from the network version to place on a laptop that is taken away from the network for up to 30 days. The network license is decreased by one user during this time, and the laptop can run SPSS remotely. When the user returns to the network, the license is taken off the laptop and restored to the network.
The SPSS output is now almost completely converted to the newer pivot table format. Output Management System also allows you to export the output in a variety of different ways.
Here are my thoughts and opinions on these improvements. I was impressed by the data validation modules which includes a lot of common sense checks you can apply such as
- dichotomy (0-1),
- dichotomy (1-2),
- likert scale (1-5),
- non-negative number,
- non-negative integer,
- range (0 to 100),
and you can add additional rules. Simply tell SPSS what rules apply to which variables and then SPSS will highlight how many violations occur and where they occur.
The chart builder seemed a bit gimmicky to me and the process seemed to encourage the use of pseudo 3D effects that most statisticians despise. I am also cautious about “expert modelers” because computers can’t match the efficiency and insights of a human. This is slowly changing though, and when the computer “expert modelers” catch up with their human counterparts, I will probably have to retire.
There wasn’t a lot on the graphics programming language but this looks like something I would enjoy working with.
You can also browse for pages similar to this one at Category: SPSS software.
You can find an earlier version of this page on my website.