Module 3: Identification of Experimental Conditions 

Essential questions 

What experimental conditions are applied to the subject of studies?

How to select a representative group for experimental work?

In most cases experimental object is a sample of group or population. A population is a set of individuals that have a common observable characteristic. 

A sample of population for experimental studies can be created by one of several sampling procedures:

• Convenience sampling (volunteers, members of the same group, individuals of the same age/occupation/residence, etc.). While this sampling method if often used, it is not random and introduces more bias than any of the following sampling procedures.
• Random sampling: any individual from the same population has equal opportunity to be selected in a representative sample (either by the random table method or by a random computer selection). 
• Systematic sampling: the first subject is selected randomly, then every 10^th individual is selected from the list of subject population. For example, each 10^th student is selected to participate in the study.
• Stratified sampling: the sample represents various subpopulations either proportionately or equally. For example, it would represent males and females equally, or the male/female representation would reflect the ratio of males and females in a given population.
• Cluster sampling: the selection of sample from the population strata is random. In each randomly selected sample, all individuals are tested. For example, 10 hospitals from the same state are selected; each patient in those 10 hospitals participates in the study.

To ensure that the sample accurately represents the subject population, random sampling is preferred to all other methods. 

Another important point in experimental design is a random assignment of subjects to treatment group (experimental group) and comparison group (control group). This ensures the equivalency of the two groups, so that the unknown variables equally affect both groups, and do now skew the comparison between the outcomes in control and treatment groups.

Experimental conditions in all subject groups should be the same, with the exception of different treatments that are received by the experimental group(s) to test the effect of independent variable. This ensures that the observed effect of treatment is only the result of this independent variable, and that there is a strong causation between treatment and outcome that can be used to validate the hypothesis that is being tested.

In biomedical research the outcome of treatment should be evaluated in such a manner as to prevent a possibility of placebo effect. Placebo is a treatment that has no active components, yet it completely resembles the actual treatment so that subjects believe they are receiving the treatment. Use of a placebo mitigates the effect of “the power of suggestion” on the individual benefits of treatment. 

Additionally, the objectivity of outcomes is improved by setting up a single- or double-blind experiments.

In single-blind studies the subjects do not know what kind of treatment was received (true intervention, placebo, or control). A single-blind experiment helps to avoid the unconscious expectations of the human subjects of the specific treatment outcomes.

In double-blind experiments neither the subject nor the person recording the response know what treatment was received. This removes a bias caused by the unconscious expectations of the subjects and of the researcher.

Suggested readings

Harris G., Farley S., Russell G.J., Butler M.J., Selinger J. (2013) Sampling designs matching species biology produce accurate and affordable abundance indices. Peer J.;1: e227. Published 2013 Dec 17. doi:10.7717/peerj.227