According to the empiricist tradition, science starts with observation. The observer should in an unprejudiced fashion faithfully record what he sees, hears, etc., about him. From his (admittedly) limited number of observation statements, the laws and theories of science can be established by generalization. Chalmers summaries the view when he writes, If a large number of A's have been observed under a wide variety of conditions, and all those observed A's without exception possessed the property B, then all A's have the property B. (1976,5) There are more sophisticated versions of the empirical account, but the core of them is still covered by the above descriptions. But a significant problem derives from what is to count as "a wide variety of conditions". What is to count as a significant variation in conditions which will allow us to decide that all A's have the property B? Unless unimportant variations are eliminated, the number of possible variations in which we could test for B is infinite. The answer is that the experimenter uses his current theoretical understanding of the situation. But this of course is to admit the Trojan Horse of prior theory; it shows the key role that theory must play prior to observation. As Hanson says, all data are already theory·laden. I think that this general point about prior theory can be generalized right across the activities of life. Clearly it is the case in education. Understanding of educational - situations and activities does not begin with observation; it begins with present values and current theory of a commonsense or more sophisticated sort, dictated by the history of educational ideas.
Chambers, J. H. (1983). Scientific Theory and Teacher Education Theory. Australian Journal of Teacher Education, 8(1). https://doi.org/10.14221/ajte.1983v8n1.4