by Dr. Caelus » Tue Aug 21, 2018 9:09 am
So, my education is *finally* coming to an end. I'm scheduled to defend my dissertation September 14, 2018, after which point I will either be Dr. Caelus or a broken wreck of shame and disappointment. Either way, this will be my last semester associated with a university, so once my SPSS (Statistical Package for the Social Sciences) subscription expires December 31st, that will be it - I won't be able to renew it because I'll have to pay for the home user license, which I cannot afford. I also don't have Qualtrics access, anymore, but SurveyMonkey still allows me to create short surveys for free.
So, that has me wondering - before I lose my stats software in four and a half months - if anyone would be interested in participating in some sort of short survey to learn about the Transformers fandom? It wouldn't be something publishable in an academic journal, because I wouldn't be collecting data in association with a university or institutional review board, but I thought it might make for a neat featured article on Seibertron.com. Thing is (assuming there's any interest and assuming Ryan would be onboard with it) I'm not sure what to ask.
The first thing that leaps to mind is the question, 'Who are we?' I could easily throw up a comprehensive demographic survey, and write a fairly interesting break down of the results. I could also put in some open-ended questions (probably wouldn't want to do more than one, actually) and perform some sort of content analysis on the responses.
Descriptive statistics are something that can be done by anyone who can add and divide, though - it doesn't exercise any of my more advanced skills as a social scientist, or really take advantage of my access to SPSS, which seems like it would be a shame.
======== JARGON AHEAD =========
I'd like to do something that would entail some sort of inferential analyses - t-tests, ANOVA, Chi-Square, etc. My bread and butter has become linear regression, which - besides being what I know best - would probably make for the most interesting results to read.
Linear regression, put simply, is used to build a model in which one or more continuous variables predicts a dependent variable. Basically, if I want to predict z from x and y, I ask people to report their attitudes about z, x, and y, and then I can use that data to create a model that will then predict z for people who didn't take the survey, based on their x and y.
In the 'real world' it's used for things like predicting the likelihood a particular kid will drop out of high school. In the context of being a Transformers fan, I had thought about using it to create a model for forecasting toy distribution, but couldn't spare the time that would be involved in mining news articles and the sightings page for data. While I probably can't use survey results to turn out anything that functionally useful, I can still use them to create something people might find interesting, as linear regression can be used to detect complex relationships between variables (e.g., the how y affects the relationship between x and z).
Now, I can use categorical predictor variables in linear regression, but it's pretty messy, so I generally want to limit it to one yes/no proposition in a model. I much prefer to use interval or ratio variables for the meat of the model, and of course, the outcome (dependent variable) has to be something continuous.
======== END JARGON =========
Simply put, besides the obvious check-in-the-box demographic questions, I would need some ideas for things I could ask that would have numbers as answers.
For example, I could ask a person's age, length of time collecting, number of stores frequented, etc. I could ask about estimated size of collection or estimated value of collection, but I think that many of us wouldn't be able to answer that. I could ask about annual spending, but I think most of us wouldn't want to think about it (he says as he looks at the shelf with 12 Prime Wars combiners on it). For the most fun, I could ask people to rate different things on a series of Likert scales; think "strongly disagree" to "strongly agree" or "never" to "Very often" - that sort of thing.
And obviously the whole thing has to be short too. Without offering compensation, I can't expect people to spend more than 5-10 min on it, and if I use Survey Monkey's free account, I get a maximum of 10 questions (though it might still be possible to work around those limitations so that it's more like 10 measures). (I could potentially do multiple surveys, though linking the data between them would be quite challenging.)
Finally, my research background is pretty diverse, but it's primarily in the realm of romantic relationships (currently focused on relational maintenance communication, jealousy, and compersion). So, if anyone has any research questions that somehow tie this hobby to close relationships, I could definitely run somewhere with that.
Addendum: Since I don't have an IRB to review my work, bear in mind that the questions need to stay really tame. No questions about Rule 34 or anything else that might be considered a 'sensitive' topic.