Guidelines for Papers
A good statistics paper is more than a compilation of statistical
facts. It should be well-organized, demonstrate clear and sound
reasoning, contain easily interpreted data displays, and be
well-written, using complete sentences and proper grammar. Lab reports
and papers are as much an exercise in communication as they are in
statistical correctness. You should type your paper using a word
processor like Word, with SPSS graphs and tables pasted into the
document. The emphasis in your paper should be on clarity and
communicating your message effectively. Here are some specifics.
- Label your graphs clearly. Graphs should have titles
and axis labels. Captions, titles, labels, and legends must be
legible, complete, accurate, and clear. Graph the data in
context. Show the scale of your axes. Choose a scale that illuminates
the variation in the data. If two plots are to be compared, make their
scales the same.
-
Variable names should be chosen well. Do not use the default SPSS
variable names. If you're working with a height variable, call it
height
, not Var00001
. Data have
context. Always give the units of the data.
- The precision of the data should dictate the
precision that you report. If your variable is weight to the nearest 5
pounds, then the mean weight should not be given as 174.0351259
pounds. A good rule of thumb is that the reported precision should
have one or two more significant digits than the precision of the
data.
-
A picture is worth a thousand words. Use appropriate graphs to
summarize your data. Examine data carefully enough to know what they
have to say, and then say it with a minimum amount of adornment and
verbiage. Do not just reprint the output of an SPSS session with all
its tables and graphs. Never include a table or graph in your report
that you don't discuss or refer to. Similarly, don't just list all the
summary statistics that SPSS can produce. Choose the appropriate
summary statistic(s) to use for your data.
-
Do not include too much technical or mathematical detail. Do not
include the details of algebraic calculations. The focus of your paper
should be on interpretation and understanding. If you feel you must
include mathematical calculations, put them in the back of the paper
in an appendix so that they don't detract from overall readability and
clarity. Use the standard statistical symbols and notation that we use
in class. Don't invent your own. If you need to use mathematical
symbols or expressions that you can't get in Word then leave space and
write them out neatly by hand. Again, it is most important that your
report be clear and easy to read.
-
The tone of your discussion should be measured and
objective. Sometimes questions posed in a lab may be ambiguous. This
is to be expected, and reflects the nature of actual scientific
inquiry. The aim of your analysis is not to give an exact answer to a
mathematical problem, but rather to provide some understanding of the
questions posed using data and statistical evidence. There is not
always a mathematically "correct" answer to each question. What I'll
be most concerned with is the logic of your argument and how you
bolster your case using the data and the available evidence.
-
In attempting to answer the questions posed in the lab you might want
to speculate for the reasons behind the phenomenon you are
observing. This is fine and even desirable as long as it is clear to
the reader which part of your argument is an objective statement of
fact, and which part is subjective and speculative.
-
Proofread your report several times. Use the following questions to
appraise your paper. I will be using them in grading.
- Are the statistical statements correct?
- Are the data displays informative?
- Are the conclusions based on sound evidence?
- Are the grammar and sentence structure correct?
- Are the style and tone appropriate?