The within-subjects design, which is also known as repeated-measures design, is a research procedure characterized by submitting the same group of individuals to different treatment conditions and then make scores comparisons between them.  This kind of design must ensure all the same basic principles of the experimental research as the between-subjects design does.

As an example, we can say that a researcher is interested in the effect of how the number of students in a class can influence on student concentration capacity to understand the subject. Exposing the same group of students to different situations like a full classroom with 40 students in University A and the same group in a smaller class of 5 students at University B, the researcher can access different scores of following tests based on the lectures that they have received at both places. Through the scores, the research can compare how the number of students in the class can interfere in students’ concentration during class.

It is pretty obvious to see that the major strength of this method is to limit any possible individual variance between participants that can compromise results based on differences and personal characteristics as between-subjects design use to have. In the within-subjects design, we can estimate that any factor that affects the performance on the dependent variable, must be the same as we use the same group on the determined treatment conditions. 

However, the high level of similarity hold by the group can produce some problems when the participants are submitted to different treatments where confounding variables as time -related variables and environmental variables can be introduced to the scene and influence their scores. This factor must be seen as a threat to internal validity if for example the group of the students above is tested on the same day. The results of the test would be not so accurate due to the fatigue of participants or by the distress caused by two tests on the same day, for example. The score results can be ambiguous if the low scores of the tests could be attributed to the exaustion of the participants and not to the excess of people in one classroom according to the researcher's theory, for instance.  It is necessary that the researcher has control over time-related threats that can interfere negatively in an expected result. In this specific situation, it would be productive if the experiment could be split into two separate days to dissipate the effects caused but the first treatment condition and then increase the chances to further clear results derived by the second treatment condition. 

As we see, Internal Validity is a big issue to be considered in this kind of design. It is essential that other aspects like as History, Maturation, and Instrumentation among others being addressed to produce reliable results.