A scientific investigation or study is undertaken to answer some specific question or hypothesis. Such questions may concern whether certain conditions, events or situations cause particular behaviors or events or if certain conditions or events occur together in time. Scientific methods are designed to minimize the biases that affect subjective opinion. In this article, I have explained the difference between experimental and non-experimental studies.
What are Experimental Studies?
An experimental study is one in which the subjects (people or social systems) and conditions (events or situations) to be studied are manipulated by the investigator. In other words, the investigator does something to affect the subjects studied and then determines the effects of those manipulations. Go through this
http://www.nationaltechcenter.org/index.php/products/at-research-matters/experimental-study-design/ for helpful information on experimental studies.
Elements of a good Experimental Study
Describes all steps in procedure clearly and completely
Keeps all variables, except the one being tested, the same
Includes a control for comparison
Can be reproduced by other investigators to give similar results
Describes all data to be collected
Issues in Experimental Study
- Researchers’ and participants’ expectations about an experiment can bias the results
- Constant attempts to minimize the error variance in an experiment may lower the study’s external validity
- Greater count of independent variables included in the study makes the analysis and interpretation complex
What are Non-experimental studies?
A Non-experimental study is one in which the investigator doesn’t have complete control over the conditions of study. In other words, the researcher doesn’t assign subjects to conditions or levels of the independent variable. Unlike an experimental study, a non-experimental study itself isn’t concerned with who created the independent variable.
Elements of a good Non-experimental study
Compares the same group of participants before and after the program.
Helps identify important variable related to the study’s success
Evaluator is able to collect data at regular interval including daily, weekly, monthly or at staggered time points
Issues in Non-experimental Study
- Over-the-time improvements of participants are mistakenly attributed to the program under evaluation
- Frequent changes in events affect the research problem, being analyzed.
Although, both experimental and non-experimental studies are based on certain assumptions which may not always be true, it is possible to statistically adjust for or control the effects of certain variables. While design involves the structure of a study, statistics deals with analysis of the data generated and the drawing of conclusions from that data.
It is often taught that only experiments can establish casual relationships among variables and that observational or correlational study can only establish that relationships exist without specifying casual direction. While in practice, this is often true, one should be cautious in assuming that experimental designs always establish and observational studies do not.
For the majority of researches, experimental studies are considered to be more powerful than non-experimental designs in uncovering casual relationships among variables. This is due to the fact that through control and randomization, potential confounding effects can be eliminated from a study. On the contrary, a non-experimental study merely established that relationships exist among variable. Through systematic observation over time, and collection of data on several variables, it is indeed possible to determine the cause and effect. Experimental studies which involve direct manipulation are more frequently conclusive because they involve principles of control, randomization and comparison. To put it more simply; unlike a non-experimental design an experimental design can be viewed as a trade-off among comparison, randomization and control. Some variables are set at different levels and compared, others are held at a constant level and controlled, and still others may be free to vary with the hope that randomization will average out confounds.