Perhaps the single most important concept for both scientific reasoning and scientific literacy is what I would call the distinction and comparison of remote and proximate objects of measurement. It is very common in serious scientific reasoning, although, of course, not usually put in those terms. The basic idea is this. Every act of measurement has an object, namely, what we want to measure. But measurement in a practical sense is very often not direct: what I want to measure and what I can directly measure are often two different things. For instance, I may need to use a device for the kind of measurement I am making -- a Geiger counter, for instance. This introduces a sort of hierarchy of objects of measurement. For instance, if it's a very simply Geiger counter, I may just count clicks; these clicks in turn measure pulses in the circuits of the device; these pulses in turn measure brief events of electrical conductivity in a tube full of inert gas; and the whole device is being used to measure ionizing radiation in the device's environment. None of these objects of measurement are simply equivalent, though, and being able to conceive of a Geiger counter requires at least some basic recognition of this fact. Ionizing radiation doesn't go around inducing audible clicks; I need to know that these clicks measure something that measures something (and so forth) that measures ionizing radiation. Getting to this point required recognizing that one thing can measure another thing under certain circumstances, and assessing the quality of Geiger counter measurement (whether it is being used properly, whether it is working properly, whether it works at all as a measuring device) depends entirely on work that established a channel from the most remote object of measurement (ionizing radiation) to the more proximate objects of measurement (like electrical pulses or audible clicks) so that the more proximate objects of measurement are adequate to the more remote objects of measurement in the channel.
The reason this sort of thing is important is that massive portions of scientific inquiry depend on establishing proxy measurements. This is fundamentally why we have experiments, in fact. If one could always assume that there was no disparity between remote and proximate objects of measurement, so that all you ever needed to collect precise data was to go and look, there would be no need for experiment. Experiments become more important than mere observation to the extent that we need to establish rather than assume the adequacy of proximate objects of measurement to remote objects of measurement. Because of this, it is always important to determinate the proximateness and remoteness of the objects of measurement in an experiment in order to determine what kinds of conclusions you can draw from it.
Consider, for instance, happiness studies. Getting precise measurements of happiness by any direct means is obviously out of the question. So we need some proximate object of measurement which can be adequately linked (probably through other objects of measurement) to our remote object of measurement, which is happiness itself. One proximate object of measurement you might choose is subjective a response to a set of survey questions. Another you might choose is some measurement of activation of this or that part of the brain. There are, in fact, many you might choose; the big question is how adequate they actually are for measuring what you are trying to measure. What you're aiming for is practical equivalence. It need not be strict equivalence; you don't need to say that happiness just is having certain answers on a particular kind of survey. But you do need it to be the case that the one can be substituted for the other for the purpose of reasoning about the remote object of measurement. Achieving this substitutability requires determining (1) whether the reliability of this kind of measurement depends on specific conditions; (2) whether the measure is a partial or total measurement (i.e., whether other things need to be measured); (3) how it relates to other objects of measurement in the same channel; and (4) how it relates to other useful proximate objects of measurement that are not in the same channel. And, of course, with happiness studies people constantly worry about the adequacy of the most proximate objects of measurement to the most remote object of measurement. People who support happiness studies obviously think they can overcome such challenges; but the important thing here is that they are rational in thinking this precisely to the extent that they have good reason to think that the more proximate objects of measurement are adequate to the remote object of measurement about which they are trying to draw conclusions. Reasons to think the one object adequate to the other are a precondition even for considering the experiment relevant to the subject you are trying to talk about.
Whenever we are assessing experiments or drawing conclusions on the basis of them, then, we really are asking how the proximate objects of measurement in the experiment are related to most remote object of measurement.