A true experimental design uses random assignment when dividing research participants into the different groups. This ensures that every participant has an equal chance of assignment to either the experimental groups or the control group. This also ensures that the outcome of the experiment is due to the variable the researcher studies and not because one group comprises participants who share similar traits. Random assignment also reduces the amount of confounding variables as it helps ensure that variables do not cause the outcome that are other than the variable the researcher studies.
A researcher must include a control group in the research for it to be a purely experimental design. Participants in the control group will be subject to the variable the researcher studies. Therefore, any changes in the control group will not be due to the manipulation of the variable the researcher studies. The researcher can then know that any changes in the experimental groups not seen in the control group was due to the manipulation of that variable. This is another way that the experimental design isolates variables and controls for confounding variables, making sure that other factors are not the reason for the outcome.
A study may have statistically significant results, but this is still only the result of one study. Experimental design calls for researchers to repeat studies on different research participants to see if they obtain the same statistically significant results each time. In order for other researchers to replicate a study, the original researchers must operationally define each step so that the next researchers may be able to replicate the study exactly as the first researchers conducted it.
Researchers create each experimental design to identify a cause and effect relationship. A truly experimental design isolates and controls variables to discover exactly which variables cause the outcome. This is more than just descriptive research, noting the relationship between variables; rather, it is inferential research, explaining how the results occurred and identifying what caused them.
Statistical significance indicates the level of probability that the results of the study are not due to chance. An alpha level of p=.05 is what researchers most commonly use for the significance level. This means that 5 percent of the time the results are due to chance alone, and researchers are 95 percent sure that the results are due to the manipulation of the experimental variable and not to chance.
A researcher states a hypothesis at the beginning of an experiment. This states that the researcher will find a statistically significant difference between the experimental groups and the control group after manipulation of the researched variable. The null hypothesis states that the groups will not differ. The goal of experimental research is for the researcher to reject the null hypothesis by evaluating any differences between the experimental groups and control group after manipulation of the experimental variable. A successful study is one that finds a statistically significant difference between the experimental groups and control group, thus rejecting the null hypothesis.
The principle of variance in an experimental design maintains that the variance between the experimental groups and control group should be more than the variance within any one of the groups. Homogeneity should exist within the groups. After the manipulation of the researched variable, the variance between the participants in a single group should be low. The results should be more similar between the participants within a single group than between the members of that group and the participants in the other groups.