Internal validity is raising and answering the question how well the exhibited research has been carried out.
In science researches always have to argue about the quality of the performed research. Mistakes are easily made and should be notified in the research report. And even though researchers do not like to exhibit their mistakes (as a matter of fact, I don’t think anyone likes to show their mistakes), it is also hard to imagine that there were no mistakes at all. Sometimes they are not even noticed till the end. Nevertheless, the authors must devote some words on internal validity in the discussion at the end of the research paper. So let's take a closer look to find out what is meant with internal validity.
Validity in general has to do with giving valid arguments. In content and construct validity it has to do with the meaning of the terms. In predictive validity it is all about giving arguments that a test can predict (or post predict) a situation in the future or in the past. External validity is about the generalizability of the research results to everyday practice.
Internal validity has to do with the quality of the executed research. The quality of the research depends on four aspects: conducting the correct research design, drawing a sample with enough subjects or objects to get a representative response, using valid and reliable instruments and applying the correct techniques to analyse the data. These four aspects are the main aspects of methodology. Therefore I prefer the term methodological validity over internal validity. Unfortunately, no one else does. So I will try not be too recalcitrant and try to be cooperative and confirmative with the majority and try to keep talking about interval validity while I mean methodological validity.
There is quite a lot to tell about these four aspects of methodology. I suggest you to read more about these subjects on the pages of this site. To give you an impression about the mistakes that can be made that lower the internal validity, I present a small list here:
- The research design has no control group;
- The sample size was too small;
- The sample size was too big;
- The sample is not representative;
- The used test is not valid;
- The used test is not reliable;
- The data contain outliers;
- Respondents have been removed (with or without doubtful reasons);
- The wrong tests for analysing the data have been used;
- The wrong conclusions were drawn.
I do not even wat to try to make a complete list. I only want to make clear that arguments can be valid or invalid. Not only the authors of the research should raise questions about the quality of the research, also peers can question this. The more questions remain unanswered or disputable, the more invalid the research is.