Reporting Statistics in APA Style
Reporting statistics means writing a story with supporting evidence.
Posted April 21, 2017
One of the many frustrating experiences my Research Methods in Psychology students face is learning the ins and outs of APA Style.
APA Style is the set of style guidelines developed by the American Psychological Association to standardized scientific writing in the field. It is widely adopted by journals in the social, behavioral, and medical sciences. The entire canon of writing standards is published in the Publication Manual of the American Psychological Association. It includes information on publication ethics, timely guidelines on how to write each section of a manuscript, and guidance on how to handle issues of authorship, and client confidentiality. It was first developed in 1929 and has evolved a long way from when manuscripts were typed, graphs were inked, and the t-test was a new and rarely used technique.
It is a fantastic guide to learning to write. But it's also pretty darn frustrating. Why? Because it's all about details.
- How many levels of headers do you have and what should each look like?
- How do you do in-text citations? With two authors? With 10?
- How do you cite a journal article? A book chapter? A Wikipedia entry?
- How should your tables be formatted?
Reporting statistics in the text can be very complicated, both because the information is complex, but also because there are just SO MANY different statistics that are used for different contexts.
Statistical Solutions published a nice, clear blog on the topic the other day. It goes through the basics of reporting the results of regression, correlation, t-tests, and ANOVAs.
Purdue OWL's excellent online writing center provides good information on punctuating statistics (another detail that frustrates many of us).
A very short story
Reporting the results of a statistical analysis has a very short story arc that can often be told in two or three sentences:
- What question were you asking?
- What statistical test did you run to answer it?
- What is the answer to the question?
- What is your supporting evidence?
This might look like:
A t-test was used to determine whether left handed people were more religious than right handed people. Means are reported in Table 1. Left-handed people reported higher scores on religiosity than right handed people (t=8.92, df=12,042, p<.001).
If I report a chi-square analysis or a multiple-regression or a latent-growth curve analysis, the details of the report will be different, but the STRUCTURE is the same:
I did ______ to learn _______. Here is where you can find the table. This is the specific, substantive answer to the question (test statistic=?, df=?, p=?).
A couple of notes:
As a journal editor I see a lot of variation in writing style. This is good—we all express ourselves in different ways. However, I argue to my students that the goal of writing a scientific paper is to write clearly enough that people can pull the key information out without too much work. It's about the IDEAS not about the STYLE.
Some common questions I am asked:
Is is okay to use the words 'I' or 'We' at the beginning of the sentence? Sure. Nothing in APA form says you can't. There are times you definitely should—constantly writing using the passive voice can get tedious for your reader. To me, the questions is should you? The focus of the readers attention should be on your ideas. In this case, what did you do? Why? What did you find?
If I begin a sentence with 'I', I am what the sentence is about. If I begin with my question and technique, that is the subject of the sentence. That's where I want my readers' attention to be. Often you can arrange your sentences to keep the focus where you want it. For similar reasons, it is often better to write about other authors' ideas and cite them in text rather than begin a sentence with an author's name.
Shouldn't I be neutral and just talk about the null hypothesis? This is somewhat controversial. I am a psychologist who teaches statistics, not a statistician. I run studies to find answers to substantive questions. If you tell me that left and right handed people differ in religiosity, my next question is 'who is more religious?'. That's why you did the study, isn't it? Saying that the test revealed them to be different and not saying in what way isn't neutral or scientific and does not add to our understanding of the phenomena.
It's just a convention
Like other conventions, APA Style can be limiting and frustrating. But it's clarity can really help your readers. Reporting stats in APA style is just writing a very very very short story.