Art Kendall: When the asker doesn’t want to cooperate in the process of moving from a presenting problem to a researchable problem—that makes the consulting difficult.

King Douglas: Many problems can be solved in a variety of ways. Do you value parsimonious code and work to attain it? Or is parsimony irrelevant given that a computer is going to do the work?

Art Kendall: I believe it is more important to enhance the quality assurance process than the machine process. The most concise code is not necessarily the most readable to a person. We never write things as well as we would like to the first time. When I write a letter, I rewrite it. When I review for a journal, I go through the article several times. When I write code, I will go back and make the code as readable and clear as can be…not in the interest of elegance, but of clarity. The more clearly it is written, the easier it is to maintain and adapt.

Machine efficiency is no longer the bugaboo that it used to be. I write code so that it can be modified. The people time in preparing the data is 90% of the effort. A few weeks of your time or buy a new PC…which is cheaper?

King Douglas: Describe the way in which you attack a difficult problem. Do you work with pencil and paper first? stare at the wall or ceiling? work directly in the syntax editor?

Art Kendall: First of all I will decide whether to solve the problem sequentially is the best way to do it. Very often, I will build parts of the syntax I’m sure of, then work through each part. Some things can be done in the GUI, pasted into a syntax window, then polished and refined. I build the pieces I can build clearly and then fill in as required. A lot of things are pretty automatic by now, such that I’m not really conscious of how I’m doing it.

King Douglas: Among experts, the consensus is that it is important to learn and use syntax. Why is that the case when SPSS makes it possible for the user to do most things via point and click?

Art Kendall: Marketing…the GUI is marketing. SPSS would do much better service to the analytic world if they sold the GUI as a way to get the first draft of syntax. I believe that any piece of prose needs to be polished and refined through a process of continuous improvement. You program by a system of successive approximation. This is true in the analytic world when you are going to do the analysis once. It is even more important in the data processing world where you are going to set it up in production mode.

If you take Deming approach, “we don’t have time to do it right, but we have time to do it over” is anathema. Syntax allows you to walk away from your work for a brief time or a long time and know, when you return, where you left off.