Testability is a measure of whether or not data gained through empirical research can be measured and "tested" sufficiently to determine whether or not the premise that is being tested can be reliably labeled as true or false.
For instance, a chemistry experiment can be recreated in a lab and show the same results every time and this is considered a highly testable result. However, a sociological experiment can be constructed and carried out and conclusions can be drawn from the results. If someone tries to replicate this experiment, no matter how closely they try to come to the original population sample used, they will come up with somewhat different results. This would be considered a much less testable premise.