Methodology

No matter how compliant your hardware and software is, true year 2000 compliance requires that you examine how you do your work. People are used to using two digits for years. Look at your checkbook; unless you got it fairly recently, the date field is represented as "_______19__". This problem goes beyond preprinted forms, however, and the result (bad data) is a lot less obvious than a computer that simply won't work.

There are some things to look for in any project you do, especially ones that involve processing data.

First, you should always use date formats when they are available. For example, Excel, Access, SAS, and SPSS all have methods for dealing with date and time values. However, the programs must be told which variables are dates; they are not smart enough to know the difference between 120,590 and December 5, 1990 (or 2090!) on their own. If date formats are not available, or if you choose not to use them, you are responsible for making sure that all your calculations are Y2K compliant.

Second, you should use four-digit years whenever possible, especially when starting a new project. This is true for input, output, or storage.

Third, you should understand how your software handles two-digit years. Some will assume that all two-digit years are in the 1900s. Many of the newer software programs, however, use a windowing procedure that assumes all two-digit years fall within a particular 100 year span, such as 1920-2019. Some software will let you change this window, while others will not.

The University of California, Berkeley has some excellent articles that discuss this problem in more detail. Take a look at "Why You Should Enter Dates With Four-Digit Years Whenever Possible", "Checking your data for Y2K problems", and "Data Files: Finding & Resolving Y2K Problems".

StatUser has some advice for users of SAS and SPSS. Users of other statistical packages may find it helpful as well. Another source of information is the paper "Testing Your SAS Software Applications for Year 2000 Compliance" (formerly titled "It's Almost the Year 2000 -- Should I Panic?"), which once again was intended for users of SAS but others may also find it informative.