1.Confounding: A tertiary variable may also ends up affecting the dependent variable inadvertently, and its effect cannot be separated easily from the independent variable. There is a chance for confounding when the investigator has not studied the situation adequately and fails to identify the existence of such a variable. It could also happen that situational factors affect the data collection process and cause unforeseen confounding in spite of the efforts of the investigator.
An example would be a situation where two groups of students are given a speeded test simultaneously in different rooms. The first group is given minimal instructions, while the second group is given detailed instructions. During the test, one room experiences an electricity failure for a few moments that distracts the students. Now the final scores of the students are also affected by the fact that the two groups had dissimilar testing experiences.
2.Selection bias: This refers to the possibility that when the researcher divides the participants into groups, each group may be more homogeneous within itself than expected, and more distinct from the other groups on a significant tertiary variable. When this happens, the groups cannot be considered equal, and the resultant differences may be skewed in favour of one group or the other.
For example, a study attempted to compare the job satisfaction of workers in two different factories. Inadvertently data was collected from one factory just after the management had
declared the results of their annual appraisal, while it was collected just before appraisal results were declared at the second factory. In this situation, the effect of declared versus anticipated results had contaminated the results.
3. Maturation: The process of being part of a study can have an effect on the individual. In long term studies, time and experiences can influence the study results. Even in smaller studies, subjects often learn to respond to a test situation, or start to experience fatigue from the testing process – both situations affecting the quality of responses. Such maturation on part of the subjects – if not an objective of the study – can affect the final analysis. For example, if a subject is exposed to similar test material regularly, s/he may start getting better scores simply due to practice. 4. Regression towards the Mean: Test scores of individuals fluctuate a little even without external influence. When study participants are chosen for their extreme scores on the basis of a single test and their scores at the end of the study are closer to the population mean than they were before; it could be due to a natural fluctuation and not the experience during the study. For example, students who are at the bottom of a class are chosen for remedial work, often score a little better a couple of weeks into the remedial classes. This change in scores may simply be due to a natural fluctuation and not due to the classes. 5. Diffusion: Sometimes it is difficult to keep the different groups being studied isolated from each other, and the resultant interaction can dilute the experiments' results. This can create a misrepresentation of the relationship between the independent and dependant variables. An example would be when an experimenter selects participants for a control and test group from the same class of students. Students who know each other may chance to discuss the experiment and exchange information that was to be controlled. At such times, the distinction in the results of the two groups would less. 6. Experimenter bias: Subjects respond even to subtle differences in the behaviour of the experimenter. If the experimenter has