Quality research -- research that efficiently produces clear, useful results -- starts with quality research questions. Every step in the research process builds upon those questions. No matter how meticulously each subsequent step is performed, fuzzy questions will lead to fuzzy results.
So what is a quality research question? Ideally, in the context of behavior, it is one that leads to a meaningful, unambiguous answer based on data that can be accurately collected from real-world populations. In behavioral research, it is rarely possible to fully achieve this ideal. There are nearly always gaps between the concept and the measurement, the population and the sample and very, very rarely can a finding exclude all but a single explanation.
The design and planning of behavioral research is, at its best, a conscious, iterative process of balancing what "needs" to be assessed against what "can" be assessed within the context of the research goals, resource constraints and maximizing cost-effectiveness.
With clear questions in mind, research design becomes a relatively straightforward technical exercise in determining the best fiscally and logistically feasible plans for sampling and measurement. From the possible research designs, one must look honestly back and forth between the plans and the question to see whether the data collected will, in the end, provide adequate answers to the questions, make adjustments as needed, then, finally, decide "is this worth doing?" After all, collecting data that fails to address the question at hand does little other than waste time and money.
We provide support services for each step in this process:
Question development:
- Helping frame research questions to meet stakeholder needs -- it is rare that questions designed to meet the needs of the basic science community also meet all of the needs of the service planning community. Questions need to be framed with the end-users of the research in mind.
- Helping to develop the logical structure of the questions so that the data collected can provide the least ambiguous answers.
- Helping to develop operational definitions of the concepts in the research questions. In other words, translating the abstract ideas in the research questions into characteristics that can be accurately and repeatably assessed in people (or animals) in the real world.
Measurement design:
- Questionnaire and/or observation schedule selection and development to maximize the efficiency with which data are collected to address the research questions
- Measurement pilot testing and validation to assure that the study accurately measures what it intends to measure with minimal bias
- Measurement sequence design
- Subject follow-up planning
Sampling design:
- Population definition to assure that the data collected refer to the same groups as the questions
- Sample selection to assure that the data collected are representative of the populations of interest
- Sample size and power calculation to optimize resource utilization
- Sample weighting design, when appropriate, so that estimates from the study accurately reflect values in the target population
Statistical design:
- Analysis planning to formally map the the data collected to the research questions
Data management design:
- Data element definition
- Database design
- Programming of double-key entry and optical mark reading (scan form) systems.
|