Some sampling methods offer better protection against the selection threat to external validity than others do. To understand why, you first need to be familiar with some basic sampling concepts. The two most important concepts are population and sample. The selection threat concerns the generalization of findings to other persons. Exactly what other persons are we referring to? All people in the entire world, or just people in our country or culture? Researchers should anticipate this question, by defining their target population explicitly. The term population refers to the entire collection of people or groups to whom the hypothesis is supposed to apply. Let's look at two examples. Consider the hypothesis, loneliness causes an increase in depression. This is a typical universalistic hypothesis. If the population is not explicitly mentioned, we infer the relation is assumed to hold for all people, in all cultures, in the past, now, and in the future. Another example, patriotism is steadily declining in the Netherlands over the last five years. This is a typical particularistic hypothesis. It's clear that this hypothesis applies to a specific country, and to specific time. Let's assume for a minute that the target population for a hypothesis is clearly defined. How can we determine if the results generalize to this entire population? Well, if we measure the entire population, then we're automatically sure that the results hold for the entire population. Everybody was measured. For universalistic hypotheses, it's simply impossible to measure the entire population, because it consists of all people, everywhere, including all people who were long dead and all people who have yet to be born. Even if the target population is smaller and well defined, it's almost always too complicated and too expensive to include the entire population in the study. This is why we take a sample, a sub-set of the population. The sample is used to represent, or estimate a property of the population. Of course, it's possible that this sample does not represent the population accurately. Suppose we sample mostly elderly people in our study of the effects of loneliness, on depression, and we find a strong effect here. The over-representation of a specific part of the population can weaken the studies external validity. Perhaps the strong effect of loneliness on depression is less apparent for young people. If our sample had been more representative of the entire population, we would have found a smaller effect. The same goes for our study of decreased patriotism. Suppose our sample consisted mainly of highly educated people working at a university. This might lead us to underestimate patriotic attitudes in the Netherlands. Our results will be biased. We will consider different sampling methods, and see how they deal with a selection threat to external validity. But, before we can do so, there are some terms you need to become familiar with. An element, or unit, is a single entity in the population. Together, all the elements form the population. An element most often consists of one person, but of course it depends on your hypothesis. An element can also be a group, a school, a city, a union, a country. You name it. A stratum is a subset of elements from the population that share a characteristic. In the population of currently enrolled students from the University of Amsterdam, we can distinguish a female and a male stratum for example. Of course, we could identify many different strata that may overlap, for example, male and female undergraduate and graduate students. The term census refers to an numeration or count of all elements in the population. The term could also refer to a situation where all elements in the population are actually measured. In that case, the sample consists of the entire population. The term census can also indicate a national census, a nation wide survey where demographic information on each inhabitant is collected. Of course, in many western countries this census is conducted virtually by collecting information from government databases. A final term that you need to be familiar with is the term sampling frame. A sampling frame is essentially a list of all the elements in a population that can be individually identified. A sampling frame can overlap with a census, defined as a numeration of the population. A sampling frame is more than a simple list of elements, however. A sampling frame provides a way of actually contacting elements. It could be a phone book, or a list of email addresses, for all students currently enrolled at the University of Amsterdam, for example. Also a sampling frame doesn't always include all elements of a population. This could be due to clerical errors, or using an outdated list. Okay, you now know the basic concept necessary to learn about different sampling methods.
