Psych 610 Handout # 1, p. 1Prof. MooreDocumenting Your Statistical PrintoutsIn the era of “clicky box” user-friendly statistical programs, it is often difficult to remember the origin ofa particular printout -- what data set, what subsample of observations within the data set, whattransformations of variables, etc. Here are some recommendations for keeping track of what you did. Most of this advice is just fundamental scientific practice. Along with the steps described here, you shouldkeep a permanently bound lab notebook with notes in ink of what you did and when you did it. Recordingresearch protocols is not covered here -- my job is to teach you to keep your data analyses straight.1. Type the date into the output file (including the year) so it is printed right on the results. If you forgetto type it in before printing it, write the date by hand when it is still hot out of the laser printer.-- Why include the year? Believe it or not, you might be working on publishing these results forlonger than a year.-- Why date it at all? (1) Dating results is fundamental scientific practice. (2) You may find an errorin your data file, correct it, and rerun the same analysis. Which printout is the correct one?2. Keep a record of all data transformations. It is not enough to say to oneself, “I know this variable isthe sum of two others because I named it ‘XPLUSY’.” If you transform data with a program youwrote (C, Pascal, etc.), save the code and the date you ran it, and document the output file. If youtransform in a statistical package like SPSS, “Paste” your syntax, run the transformation from thesyntax file, and save the syntax file. The “Paste syntax” box is the most important clicky-box in SPSS! I like to print copies of those syntax files and save them in my notebook for the project. You shouldalso copy the appropriate syntax file into the output file so that it is printed with the output. All ofthese steps will allow you to re-trace your steps and find errors (or verify that you have done whatyou wanted to do correctly). Remember, a typo (or erroneous click) in the name of a variable whencomputing a new variable implies that the new variable is not what you think it is.-- In SPSS, I almost never use the “recode” option because the program will alter your original datawithout warning you. My recommendation is that you always create a new variable in SPSS whentransforming your data. This advice is sound for other programs as well.-- Some programs will create temporary variables from your transformation instructions and putthose instructions in the output file. I like this best because what you did is written right there inthe output file. You can look at it whenever the printout is in your hand.3. What data file was used should be stated as part of the output file. (SPSS does not automatically dothis.) Type this in.-- Did you filter the data to omit certain cases? Perhaps you eliminated outliers, perhaps in separateruns you used different criteria for eliminating outliers. Document this on the printout while youare working, not later. One way is to create a separate data file for each different filtering anddocument what is in each data file. This makes it possible to re-trace one’s steps fully to verifyan analysis.Handout #1, p. 24. Page numbers, who needs page numbers? How about stapling or a three-ring binder? Some programs(SPSS) do not automatically put numbers on the pages when the output file is printed. The problemis that when you run several analyses that are similar, you can mix up the pages (“Is this the post-hoctest for analysis A or analysis B?”). In my lab we have two methods of keeping analyses together: (a)staple the pages hot out of the laser printer, (b) three-hole punch and put the analysis in a ring binder,c) electronic filing with comments embedded.. In both cases the analysis should be labeled as describedabove.5. Electronic filing tips: Some people don’t print statistical output files, but just file them electronically. If you do that, the documentation above is even more important. Plus, be careful if you open twosuch documents in your computer simultaneously. Flipping through pages on a screen can beconfusing -- are you sure that the Fs and other statistics you are copying into your manuscript are thecorrect ones? Embedded comments will help you