Analyzing the Aircrew Stress Study data
For the record, I used to teach research methodology and statistics classes at a university, and I tend to be a fuss-budget about all that. But don't be alarmed: I'm not going to clutter things up here with tables and arcane statistics.
For the present purposes, I'm going to assume that the readers of this blog neither know nor care much about the fine details of the analysis process. They just want to know what I found out about people who fly for a living.
Nevertheless, I would like to mention a few things about how I sifted through the Aircrew Stress Study data in order to arrive at the results I'm going to report here, just so you'll know where I'm coming from.
The Aircrew Stress Study survey questionnaire was very complex. In addition to demographic data and information about their specific jobs, participants answered questions about their family situations, their habits, their physical and mental health, and about stressful things on the job and in their personal lives. If you saw all of that data from over a thousand participants you might wonder where to start to analyze it.
I begin simply by counting things up: How many people said this as opposed to that? I do that for every single item on the questionnaire.
Some sets of related items are combined to create indexes. I count up those scores as well, to see how many people scored high or low on each index.
Next, I compute the average scores for the various items and indexes. I also look at how much variability there was, to answer the question, "Did the scores vary 'all over the map,' or were they more or less clustered around the average?"
By this point, I have a bit of a feel for what's going on in the data. But simply knowing what percentage of the people said X as opposed to Y is of limited value. So, I move onto the next phase to see 'what goes with what.' You compare groups.
Did the pilots' responses differ from the flight attendants'? Were there any gender or age differences in response patterns? Did the airline crews differ from the corporate crews? Did the Europeans differ from the Americans? You get the idea.
Finally, I analyze for complex patterns in the data -- a sort of multiple what-goes-with-what. I build special equations that combine multiple elements, in order to come up with some set of things that will explain an outcome.
As I write in this blog about results from the survey, I'll mention outcomes from all of the above steps, but in plain language. I will talk about percentages, and maybe some averages, but I'll stay away from statistics that are bewildering to anyone except other researchers. I'll save those for academic-type papers and presentations elsewhere. In other words, I'll try my best to keep the whole thing 'reader friendly.'
And now, on with the show...
0 comments:
Post a Comment