Maturation
Maturation may threaten the internal validity of a study when there is a passage of time where biological or psychological changes take place. Say we took a test of a child’s articulation skill at 5 years of age and again at 8 years of age after receiving 3 years of an articulation improvement program. The problem is that we may expect a child’s articulation skills to improve with the passage of time as the child matures.
Maturation may also be a threat depending on the time at which measures were taken. For example, if we were to take measures early in the morning and again in the afternoon, we may see changes in the individuals' responses simply due to the passage of time. Likewise, if we took measures prior to lunch and then after lunch, we may see which changes had taken place. Maturation is always a concern when we see that measures were taken at different times for two different groups. Say we wanted to test the effectiveness of a new reading curriculum. The difference between two groups of students in second grade was measured. One group received the new reading curriculum and the other received the standard reading curriculum. We noticed that the group who received the new reading curriculum outperformed the group who received the standard reading curriculum. However, the time the measurement was applied must be taken into account. The measurement was given to the students who received the standard reading curriculum just before they left for home and was given to the students who received the new reading curriculum the next morning. Therefore, we could make the conclusion that at least some of those observed differences may have been due to the fact that students are usually tired at the end of the day as compared to when they arrive at school in the morning.
Selection
Selection may threaten the internal validity of a study if individuals are assigned to groups in a differential manner. What makes groups of individuals different from one another is how they were selected in the first place. This threat is common when we select pre-existing groups to be involved in an experiment and do not attempt to demonstrate how they are similar on critical variables before the independent variable is provided.
Say we wanted to compare two types of reading curricula. To do this, one group of individuals who received the standard reading curriculum was compared to individuals who received the new reading curriculum. Now suppose that the first group of students was in a remedial classroom for students who had difficulty in reading and the second group of students was in a classroom for children who had above average reading skills. After the instruction was provided, it was found that the second group who received the new reading curriculum outscored the first group. Obviously, the second group would be expected to outscore the first group because the students were initially better at reading. In this case, we would be unable to assess which reading curriculum was more effective because the groups were different from the outset.
Selection by Maturation Interaction
This threat to the internal validity of a study is similar to the maturation and selection threats described above; however, the main concern here is the maturation of the individuals selected for involvement in the study.
For example, say we worked with several children with head injuries. Participants were assigned to two groups (i.e., experimental and control) on a non-random basis. The average post-injury time was 2 years for the experimental group; the average post-injury time for the control group was 8 years. The experimental group was provided with a specialized reading curriculum that was designed to re-teach comprehension. The control group received a standard reading curriculum. The comprehension skills of participants in both groups were measured 1 year later; it was found that the experimental group made greater gains in their comprehension skills when compared to the control group. Our conclusion that the specialized reading curriculum improved the experimental participants' comprehension may have been erroneous since we failed to see if the groups were initially equivalent on critical variables. They were similar to each other except on the basis of time post injury. We would expect children 2 years post injury to potentially make greater gains as compared to children who were injured 8 years ago because the advancement of skills typically slows down over time. Thus, much of the gains made may have been due to the maturation time at which the children received the specialized reading program, not the program itself.