All great research begins with a great question. The manner in which you respond to that question is all about your research design. The design is the blueprint, which will direct your data collection, analysis, and interpretation. It will determine the strength of your findings. All too often, researchers resort to mere surveys. They do not take advantage of the fact that they can use more important methods.
The selection of the appropriate design is essential in the determination of causality and generalizability. Being one of the leading consultancies dealing with Scientific and Technical Editing, we at Fast Publications, find that most of the promising studies fail due to an inappropriate design. It is critical that you go beyond the survey and the survey and pick the design that best fits your purpose: experimental, quasi-experimental, or the correlational.
When Correlation Is Enough: Learning the Correlational Designs
The most widely and frequently available design is a correlational design. You adopt this design when your main aim is to establish the nature and the strength of the relationship between two or more variables without the manipulation of these variables.
Correlational design is selected when
- You would like to test initial relationships between variables.
- The variables you are studying cannot be ethically or practically controlled (e.g., personality traits, existing medical conditions, past events).
- You intend to make inferences concerning one variable using another.
Hypothesis: The relationship between the hours of sleep and the test scores. You just measure both the variables in their natural occurrence.
Although correlation studies are good to determine trends and come up with hypotheses, they present a major constraint: it is impossible to determine causation. Keep in mind the adage: correlation is not causation. The observed relationship could actually be brought about by a third unknown variable (a confounder).
Setting the True Cause: The Experimental Design
When your research question is a direct statement, i.e., you want to know whether X causes Y, you have to select a true experimental design. This is the gold standard of establishing causal relationships. The key characteristic of the design is the manipulation of one or several independent variables and the chance assignment of the participants to various conditions.
An experimental design is selected because:
- You must be able to come up with a definite cause and effect relationship.
- The environment and the provision of treatment/intervention can be managed.
- Participants may be ethically and practically randomised, to control and experimental groups.
The essential steps include:
- Manipulation: The researcher is the one who manipulates the independent variable (the cause).
- Random Assignment: The assignment of the participants to the conditions is made by chance so that the groups are similar in the beginning.
- Control: You exercise strict controls to reduce the bearing of extraneous variables.
The critical distinction is random assignment; this is a mathematical process that evenly allocates any possible confounding variables to groups so that you can be certain that any difference that occurs after intervention is due to the manipulated variable that you are measuring.
Bridging the Gap: The Quasi-Experimental Design
What about when you need to study cause and effect, but you cannot randomly give out participants? At this point, the quasi-experimental design comes in. You use this design when you have to examine naturally occurring groups or exploit existing groups in which randomization cannot or is not feasible.
You use quasi-experimental design because:
- It is assessing the actual effect of a program or policy under real world conditions (e.g.: a new curriculum in a school district).
- You are observing variables with respect to inherent traits which are immeasurable (e.g. gender, age, socio economic status).
- Random assignment is prohibited by ethical or logistical factors.
The quasi-experimental design most commonly used is the non-equivalent groups design, in which you are comparing two or more existing groups (e.g., two different classrooms) which you suspect are similar but of course are not identical, due to the absence of random assignment.
Limitation: Since you are not randomizing, you lack the ability to conclusively assert that you caused anything as with a true experiment. This you should take note of and consider any possible pre-existing differences between the groups in your analysis and discussion.
Furthering the Assurance of Integrity: The Task of Scientific and Technical Editing
The choice of the appropriate design is not the only step. Then you need to present your selected design excellently and share your approaches and results with impeccable clarity. This is where the experience of any consultancy, such as Fast Publications will be highly required.
Scientific and Technical Editing Services will assure you that your:
- The Methodology Section is such that it accurately and precisely depicts the requirements and limitations of the selected design.
- The appropriate data structure supports the design; researchers should use a t-test to analyze an experimental design; analysts should employ regression to analyze a correlational design.
- Discussion explains the results in the limitations of the design without exaggerating the causal relationship of correlational or quasi-experimental designs.
Through strict guidelines of design and careful Scientific and Technical Editing, you will no longer be content with trite surveys but you will come up with research that is strong, publication worthy, and actually relevant.
