The validity and reliability of the data in a Moodle site varies based on how the institution is using Moodle. Some kinds of data are very reliable, e.g. login dates. Others are harder to predict. For example, if you are studying final grades, the validity and reliability depend on the instructors who assign the grades. If you are studying quiz questions, the validity and reliability are dependent on the design of the quiz questions.
As you know, validity refers to how well the data (in Moodle) represent something you care about. I don't know your research question, but let's say you are trying to find out if login patterns represent "student engagement." To test this, you would need a definition of "student engagement" and a way to establish that the way you are counting logins is a good representation of student engagement. (I believe logins alone are not a good representation, by the way.)
Reliability refers to how consistent the data are as a measurement of something. You can check this with your data by looking at measures of variation.
In my own research, I am checking the validity of the "Community of inquiry" model of engagement by using factor analysis on the variables within Moodle that I think will align with the factors of the Community of Inquiry. This is a common way of checking construct validity.
If you have more specific questions, please feel free to post them here and I will do my best to answer, or you can send me a private message.