On non-western locale fast tables display less variables than selected on any procedure when Unicode mode is off
You work with IBM SPSS Statistics 20 on non-Western locale. Additionally you use fast tables option (SET TABLERENDER = LIGHT) which is the default setting in Release 20. In Edit - Options menu you select to use locale system setting for character encoding (SET UNICODE = OFF). In your file there are variables with non-western characters in either their names, string content, or value labels which you use on any procedure dialogs.
However, when you look at the tables created you may encounter incorrect results. For example you may notice that not all variables are included in the tables, some may left. The incorrect result depends on the table, but the most common scenario is that if you have a table with rows "X" and "Y", the "X" row's values will be overwritten with the "Y" row's values, and the "Y" row will be empty.
For example, instead of:
Male 1.23 3.45
Female 8.12 9.45
You will see:
Male 8.12 3.45
This issue was reported to IBM SPSS Development as APAR PM47177. As a workaround you could use UNICODE = ON before you create your fast rendered tables.
A Flash Alert was published on the issue. Please refer to 1515210 for more information.
An Interim Fix IBM SPSS Statistics 20.0 Interim Fix 5 has been released to resolve the issue.
This Interim Fix can be accessed 2 ways:
1. IBM Support page (recommended):
- Click on 'Manage my product list' on the upper left under "Choose your product'
· Search for Product (SPSS...)
· Click on 'Downloads' on the left under "Choose your page' (within <Products>)
· Select <Product> Downloads; filter and search as desired
2. Fix Central (same as link from IBM Support Page) http://www.ibm.com/support/fixcentral/
· After selecting the product and fix pack number, click 'More Information' to show the Fix Pack IBM Support page