Internal validity asserts that variations in the dependent variable originate from variations in the independent variable(s) – not from other confounding factors. In experiments, internal validity is also based on how much control has been attained in the research while collecting data.
Examples of Threats to Internal Validity in Quantitative Research Work
History: the happening of events or conditions which are irrelevant to the treatment but which take place during the research to a group of people and produce changes in the outcome measure.
Maturation: Subjects change over the course of the test or even between measurements. The chance that a variation between the pre- and post-tests could be the outcome of the physical or mental maturation of the individuals rather than of variations in the independent variable. Example: the performance of first graders in a learning test starts decreasing after Forty-five minutes due to fatigue.
Statistical regression: It is the result of a tendency for topics chosen on the basis of extreme scores to regress towards the mean on subsequent tests. When measurement of the dependent variable is not perfectly reliable, there exists a tendency for extreme scores to regress or move toward the mean. The level of statistical regression is inversely associated with the reliability of the test.
Testing: could cause modifications to the participant’s scores obtained in the second administration as a result of having taken a pre-intervention test. Example: In an experiment where performance on a logical reasoning test is the dependent variable, a pre-test hints the subjects regarding the post-test.
Mortality: It means the situation where individuals chosen either fail to participate in the research study at all or don’t take part in each and every phase of the study. This, therefore, may or may not create a bias. For instance, the proportion of group participants having stopped smoking at post-test was discovered a lot higher in a group having received a quit-smoking training course in comparison to the control group. Nevertheless, in the experimental group only 60% have finished the program. If this attrition is systematically associated with any feature of the study, the administration of the independent variable, the instrumentation, or if dropping out contributes to relevant bias among groups, an entire class of alternative explanations is possible that account for the observed differences.
Design contamination: Did the comparison group know regarding the experimental group? Did either group have a reason to want to make the research succeed or fail? Usually, researchers must interview subjects after the experiment finishes to find out if design contamination happened.
Evaluation anxiety: anxiety experienced when one’s behavior or accomplishments are being evaluated.
Restricted range: missing out on the knowledge which almost all parametric analyses represent the general linear model, investigators may unnaturally classify variables in non-experimental design using ANOVA, even though it leads to relevant variance being dumped.
Confirmation bias: the tendency for interpretations and conclusions according to new data to be overly consistent with preliminary hypotheses.
Matching bias: variables not used to match the groups could be more linked to the observed findings than is the independent variable.
Demoralization: Realizing they’re in a control group, and not receiving an intervention which they desire, a few people in the control group look even worse with time than they otherwise would. This could lead to differences between the experimental and control group being mistaken for the results of the intervention.
Multiple-treatment interference: carryover influences from a previous intervention will make it challenging to measure the usefulness of a later treatment.
Illusory correlation: identification and interpretation of associations which aren’t true but statistical artifacts.
Instrumentation: is challenging when scores produced from a measure don’t have the suitable degree of consistency or don’t produce valid scores (because of inadequate content, criterion and/or construct validity).
Threats to internal validity in quantitative research compromise our confidence in stating that a relationship exists between the independent and dependent variables.