Research Questions and Designs

Bohannon, Richard W. (1999). Research Questions and Designs. Topics in Geriatric Rehabilitation. Activities Related to Clinical Research. 14(3):53-59.

This article provided an overview of research theory and design.  When someone begins their own research, what questions do they ask and how do they use these questions to develop a viable research plan?  Any research activity will be an expensive undertaking.  Most will require institutional support or specific measurable results that are viable for product development or sales.  Measurable results are the key element in any case.   Thus, developing a suitable research plan that will yield specific measurable results is fundamental for any research activity design.

Some very basic research questions are given in the article and discussed including:

1. How good are the measurements of the variable or problem of interest?
2. What is the extent and nature of the problem?
3. To what is the variable or problem of interest related?
4. What are the effects of interventions on the problem?

These very basic questions seemingly will apply to everyone everywhere doing research.  The key elements are the variables of measurement for the problem of interest.  What can be measured regarding the problem?  How does the measurement vary with how big or severe the problem is?  And more importantly for researching: how are the measurements related to the problem, and how can these measurements change with an outside influence or implementing an intervention that affects the problem.  These questions further bring out the importance of measurements and how variable measurements are central to an analysis.

The specific qualities of measurements introduced included reliability, validity, sensitivity, and responsiveness. These terms each have significant meanings related to measurements. I found it interesting how it took me a while to find links that defined these points similar to the article definitions.   Everywhere we can find definitions online, but finding something appropriate to what this author was trying to say took a bit of an effort. 

This discussion leads us further into the basic design elements that were also addressed in the article.  Each of these elements can be used to derive different measurements.  They provide different ways of looking at the same problem and finding the measurements differently.  The options specifically defined and reviewed with examples in the health industry include:

1. Quantitative versus qualitative - measurements
2. Prospective versus retrospective - analysis
3. Cross-sectional versus longitudinal - periods of data reviews
4. Non-experimental versus experimental - research design elements

These options are discussed and defined similar to the measurement qualities linked above.   Instead of going to the web and trying to find links that fit each word as Bohannon discussed in the article, I will simply paraphrase the main points he made below. 

1) Quantitative research is looking at the specific numbers measured for the defined variable only, while qualitative is looking more at the big picture and looking for the overall meaning and conceptual understanding the research shows instead of just a measurement quantity.  Qualitative is often open-ended observations and perceptions like using a survey to getting peoples' feelings, believes and values.  2) Prospective is designing a planned experiment to gather data specifically during an event or time period, while retrospective it reviewing data someone else has already collected elsewhere.  3) A cross-sectional analysis would be done at one moment in time, while the longitudinal investigations are carried out over a period of time.  4)  Non-experimental options defined in the article included "case-studies, surveys and secondary analysis" while experiments are specific actions done to measure the results.

"Case-studies, surveys and secondary analysis" were explained in more detail again with examples.  Here is another quick clip directly from the article:

1. Case studies are in-depth descriptions of one or more subjects.
2. Surveys are used to obtain information from respondents via written questionnaires or interviews.
3. Secondary analysis of archived material - or other sources collected by others.

Finally the article discusses the "randomized controlled trial" as the "gold standard" for experimental research.  This is when a random representative sample is taken from the population where measurements are taken before and after the intervention.  The sampled populations are blind to whether they are in the control group or not.  This is the research type that is most respected.  It should be designed to reduce the variables and outside effects so that the only thing the sample is affected by is the variables affected by the intervention.  All the other pieces reviewed in the article lead up to this and this is what researchers should be heading for.

The article did describe a few other techniques that can be used.  The random control trial is the best though, so I will only list the others here:

• quasi-experimental designs
• sequential medical trial
• explicatory experiment
• single-case experiment