Source: Group; Public Domain http://commons.wikimedia.org/wiki/File:1223_Brindis.JPG
Hello, class. So when we're creating or examining any form of psychological research, it's important to identify exactly what population is being studied within that research. So population is a category of people that are being studied. OK. So oftentimes the population we're talking about within psychological research is the generalized idea of all human beings. So we're talking about everybody and their general mental processes.
But sometimes psychological research is more specific. And so those populations would be a little bit smaller. For example, we might talk about infant girls. That would be one population. Middle-aged men or businessmen between 20 and 24 years of age. OK. So ideally, if we were talking about a certain population, we would want to test everyone within those populations to provide the best results. But realistically, we can't test everybody, especially if we're talking about everyone in the entire world.
And so that's when we decide to take a sample. And a sample is a small selection of people from a population. So the population is the bigger group and we take just a selection of them. And the researcher uses the results from the research to make statements about the population as a whole from the sample itself. It's important to get accurate appropriate samples so that way we can get the correct results to talk about the population as a whole.
Now a sampling error is when we have some kind of result in the study that's changed or affected as a result of the sample that was chosen. So we didn't get the correct results because the people that we studied aren't necessarily representative of the population. OK.
And there are two types of sampling errors we talk about. The first one is a random error. And this means normal errors that occur regardless of the research and the methods and everything that goes on. In other words, we always have some kind of random sampling error within our studies. We just can't help it because not everybody is talking about everybody else.
And generally random sampling errors affect the results when the groups of people that are being used are too small. So the differences within those people become more drastic as a result of the small size. So that's why it's important when we talk about research that bigger is better statistically speaking. The more people we have, the better our results within our study.
The second type of sampling error is biased sample, which is to say when the differences in a sample make it inaccurately representative of the population itself. And it's something specific that happens. And it invalidates the results themselves.
For example, in a study about memory in people and the people as a whole, you want to study or you decide to study college students on a Friday night who are more likely to go out and drink on that time. So if let's say your experiment is the next day on a Saturday morning, then your population is more likely to have impaired memory during that time.
So you see your sample that you chose, college students, who are more likely to drink on one of those days as well as the day you chose are going to affect the results. So your sample is inaccurate for everybody as a whole. So it's important to make sure that we consider these things when we create or when we review psychological research.
So there are many different techniques that researchers use to select samples that prevent biased samples from occurring. And we're going to take a look at two of the main ones for today. So the first way is to choose a random sample, which is the way to use some random technique to select participants from the entire population as a whole. So for example, if we could put everybody into a machine and have them spit out a few random people or names from that, then that would be probably the most ideal way to select a sample of a population.
This is because it removes the chance for bias within the person who's choosing when the people from a certain group or with certain characteristics are chosen by that person. So all the differences tend to cancel out when we choose randomly when we have a big enough sample of the population statistically. And that's just something that generally occurs. So the more people we have, the less chance we have of having a biased sample.
Still it's difficult because not everyone from a population is available. If we put everybody from the US into a machine and have it spit out names, not everybody would decide to do it. And only certain people would actually respond. So researchers are often constrained by these realistic situations and problems. So they need to work with what's available to them.
This is why it's also important that we replicate research over and over so that way we can get lots of different sources of information to see if it actually is true. So besides random sampling, which again we said is one of the most ideal but not necessarily realistic, we have a representative sample.
And a representative sample is a sample of the population that's chosen specifically to more accurately represent the population as a whole. So research can identify particular aspects of what they want to test or what would affect the research. And that way they can weed out those kinds of people that would give an inaccurate response.
You can also choose from particular subgroups. You can make it a proportional representation, which is to say that instead of relying on chance, you can make sure you get the right numbers of certain types of people.
For example, you want a certain percentage of males or females within those groups. To make sure that you have the right proportion to the actual population. So while there can still be unforeseen factors within a representative sample, this is again another way that we can help to control for any kinds of errors.