I completed a factorial survey process for my dissertation research and now I am working on another health behavior research project with a group although we are using the same general design. There are a couple of potential challenges that have driven me to look toward a quota or unrandomized vignette design. While I was doing my initial literature review for my dissertation, I started to have a bad feeling about the fractional factorial design (participants get a randomized sample of the vignette 'universe') because it seemed as if the more sophisticated researchers were using not just quota, but d optimal or d efficient designs. I stuck with what I had planned - the fully random vignette design and found ample support to do so, but filed away in the back of my brain the idea that I was going to need to work my way through quota designs in the future. Well, the future is here and I am compiling readings from the university library and the internet. This has taken me into what is for me a new world - the world of "design of experiments (DoE)." After completing some preliminary reading, I would conclude that most of my prior DoE exposure has focused on one element: sample size calculations. I actually think we are getting to the point in most academic social science/health science departments that most people have heard of and use G Power (http://www.gpower.hhu.de/en.html). I personally question whether everyone who is doing power analysis knows what he/she is looking for - you need to know what you mean when you decide what effect size you are interested in, which often means finding a precedent in prior research, but that may be subject matter for another post!
When I think about research design, I tend to think about methods, methodologies, data types and gathering, and the general organization. My previous factorial survey was a quantitative cross sectional survey research design - delivered to purposely selected respondents at one time point - and data were analyzed by me using HLM regression. As I said before, I used a fractional factorial - fully random design. When I started to look at the optimal design literature, I saw that range of possibilities presented by some elements of that design although usually expressed in terms of engineering, agricultural or other non human, non social science examples. The essence is the same, however - a primary condition is that it is impossible to assess all possible combinations of variables. So a lot of the focus of DoE - or at least the component of DoE that I am focusing on - is how to select one or a few optimal combinations in order to end up with the best possible assessment.
2 Comments
12/27/2020 10:24:43 pm
Incredibly useful and detailed information. You are growing rapidly and it shows in your blogs.
Reply
1/18/2022 04:46:09 am
Very informational blog!! I am very interested and I really liked this. I wanted to give a huge thumbs up for the great information posted on the blog.
Reply
Leave a Reply. |
AuthorI am Sheryl L. Chatfield, Ph.D, C.T.R.S. I am a member of the faculty in the College of Public Health at Kent State University. I also Co-coordinate the Graduate Certificate in Qualitative Research and I am a member of the Design Innovation Team at Kent State. Archives
February 2024
Categories
|