When I'm reading through the tables and charts from some study, I want to know exactly whom or what the outcomes are supposed to apply to. That is, I want to know the target population of the study. Although this is an obvious and essential question, many studies don't clearly answer it.
A research question usually formulates something we’d like to know. Now, a target population can be loosely defined as “the collection of objects about which we'd like to know something”. Note that the target population is usually an essential element of a research question. Sadly, the social sciences all too often fail to be explicit on what the target population for a given question is. We’ll briefly sketch how that affects research quality.
Say somebody asks you to investigate whether frustration causes aggression. An immediate counter question should be: “among whom?” Let’s first assume the question relates to human beings only. Now perhaps frustration causes aggression among men but not among women. If so, the answer depends on the target population it involves. And the latter not being specified really renders this question incomplete and thus unanswerable.
Now note that “men” is still rather vague. For instance, does “men” include male children? Does it include all men who are currently living or men who lived in the past as well? And what about the men who’ll live in the future?
If we keep things simple, we could say that we’d like to know whether frustration causes aggression among all men -regardless of age- who are currently living. Since there’s roughly 7 billion people living on our planet and roughly half are men, our population size is around 3.5 billion men, give or take some millions.
However, our question could also relate to all men - all who are living, who lived in the past and even those who’ll live in the future. Since the latter don’t ‘exist’ yet, our target population doesn’t actually exist either; our question thus relates to a hypothetical population (whose size is obviously unknown). This may sound overly philosophical at first but there’s nothing weird about hypothetical populations and they are pretty common in real world research.
Hypothetical populations, since they don’t actually exist, sometimes have infinite population sizes.
Necessity of Sampling
From this discussion it should become clear that in many cases you can't investigate the entire population about which you’d like to know something. Sometimes the population you’re interested in is hypothetical. Alternatively, it's sometimes theoretically possible but simply too much work to study an entire target population. In such cases, we usually study a sample from a population. A major chunk of inferential statistics deals with the question of how to generalize sample outcomes to target populations. We'll discuss this into much greater length later on.
Finally, note that it's by no means uncommon that a research question relates to more than one population. For example, “are men more aggressive than women?” basically asks whether the average level of aggression among the population of men is equal to that among the population of women.Note that this question is typically answered by using and independent samples t test. Again, it must be specified exactly which men and women we'd like to investigate here.