Hi. This is the tutorial on observational studies. So what we're going to start with is two research questions. Question one, are women more likely than men to support democratic candidates? And question two, does a new interactive computer program improve the reading ability of kindergarten students?
So if we're going to go about answering these two research questions, we would need some data. So the question is now, how do we collect data to help answer each question? So let's just go back and think of a way of collecting data for each question.
So if we looked at question one, to answer this question, we would need to start with two samples-- a sample of men and a sample of women. And we would start with the sample of women and just simply ask each participant, do you support democratic candidates? You'd just record yes or no. And then you would go to the men, and you'd ask them the same question, do you support democratic candidates? And then, again, select yes or no.
We would need to then figure out what proportion of women support democratic candidates, what proportion of men support democratic candidates. And whichever one has the higher proportion, we could then use them to answer this question. So if men are higher than women, then yes, they are more likely to support democratic candidates than men. So would give us a good way of answering question one.
Question two, we're going to have to collect data using, really, a different means of statistical study. So what we're going to need to do is go in to a kindergarten classroom. And the first thing I would do is give all of the kindergarten students a pre-test. So we would do a pre-test on their reading skills.
After the pre-test, we would somehow randomly put them into two different groups. One group would represent the-- would be taught reading kind of just in the traditional manner that the school's always used to teach reading. And the second group would get their reading instruction via this interactive computer program.
So then, after maybe a couple weeks of instruction, we would then test all of the students again. We would then need to see, well-- so then we would test the traditional instruction group and then also the computer program group. Then we would take the pre-test scores, compare them to the post-test scores for each group. Then whichever group had bigger gains in reading would be the group that did better. So if the computer program group did better than the traditional group, then we could say, yes, it does seem like this computer program improves reading ability.
So both of these two statistical studies are pretty different. But only one of them is what we call an observational study. And what an observational study is, is a study that observes individuals and measures variables of interest, but does not attempt to influence the response, OK? So in this case, the response would be the reading ability.
So now, based on that definition, which of the two research questions could be answered using an observational study, question one or question two? And the answer to that is question one. So it would be question one, because we're just simply trying to determine-- just observe people's political inclinations, political beliefs. We're not trying to influence whether they want to vote democrat or not. We're just simply observing what's true.
Whereas, in that second research question, we are trying to influence the response. We're trying to see if students read better using this computer program. So question two would definitely not be an observational study, because we are trying to influence the response there.
But now, because of that, an observational study is a poor way to gauge the effect of an intervention. So it's a poor way to gauge an effect. So when our goal is to understand a cause-and-effect relationship, we cannot conduct an observational study. We can observe what's happening, but we can't definitively conclude that there is a cause-and-effect relationship.
So when we are conducting our observational study to answer question one, the people that were surveyed about their political inclinations were called the participants in the study. Whereas, animals or things that are part of an observational study are generally called subjects. So if they're people, they're participants. Really, anything else, we call them subjects.
Now, there are really two common types of observational studies. One's called a retrospective study, the prefix there being "retro-". And we also have a prospective study, where the prefix here is "pro-". Sometimes you'll hear a prospective study called a longitudinal study.
So a retrospective study is a study that looks backwards and examines exposures to suspected risk or protection factors in relation to an outcome that is established at the start of a study. A very famous retrospective study was done in the early 1900s involving the link between cigarette smoking and lung cancer. So, basically, what they did is they looked at two different groups. They took a group of smokers and a group of nonsmokers.
And basically, they just, from each sample, determined what was the incidence of lung cancer in each of those groups. And they were able to conclude that there is a higher incidence of lung cancer in the smoker group. So in many retrospective medical studies, a case control study is used to compare an exposed group to an unexposed group. So in that historical example, the exposed group was the smokers. The unexposed group was the nonsmokers.
Now, a prospective study is a study that watches for outcomes, such as the development of a disease, during the study period and relates this to other factors, such as suspected risk or protection factors. The study also, called a longitudinal study, usually involves taking a cohort of subjects and watching them over a long period of time.
So perhaps if a researcher is interested in looking at exposure to something like asbestos, maybe they would go into a factory and recruit a group, a cohort, of subjects that work in that factory and then do tests on them to see what their exposure to asbestos might be. So that's just an example of where a prospective study might be used. All right, well, that is your tutorial on observational studies. Thanks for watching.
Hi. This is the tutorial that covers experiments. So, to start, we're going to take a look at two research questions that we could use a statistical study to help us answer. Question number one, are women more likely than men to support democratic candidates? And question two, does a new interactive computer program improve the reading ability of kindergarten students?
So really to answer both of those questions, we're going to need some data. So the question is, how would we collect data to help answer each question? So let's just talk a little bit about how we would go about conducting our study to get data in both cases.
So in question number one, what I would need is I'd need a sample of men, and I'd need a sample of women. Then, really, I just need to survey them. I'd ask the group of women, are you likely to support democratic candidates in the next election, yes or no? I record their responses. And then I'd do the same thing for men. I would figure out, then, what proportion of women support democratic candidates, what proportion of men support democratic candidates, and then this question will be pretty easy to answer. So that's how I'd go about collecting data for question one.
Question two is going to be-- we're going to have to collect data in a little different way. What I would need to do first is I would need to go into a kindergarten classroom. And the first thing I would do is start by giving all of the kindergartners a pre-test. So I'd start with that pre-test. And then what I would need to do is I'd need to randomly split the classroom into two groups, OK? And this split should be done randomly.
I would need to then teach one group reading using just the traditional means, so however that school generally teaches reading. The teachers would go about teaching this group in the traditional manner. And the second group would be the ones using the computer instruction, so the interactive computer program.
So once I have them split up, then they would be exposed to each type of instruction. This group would be only taught using traditional instruction. This group would only be taught using computer instruction. And after a certain amount, then we would need to give them each a post-test.
Then what we could do is we could compare each group's post-test scores to the pre-test scores. If the traditional group had a bigger change from post-test to pre-test than the computer group, then the traditional group would end up being more effective. If the computer group's post-test minus the pre-test scores were bigger, then we could say yes to this question. Yes, it does seem that it improves the reading ability of the students.
So both of these are considered statistical studies, but only one of them is what we call an experiment. So an experiment is a study that deliberately imposes some treatment, or what's called a placebo treatment, on individuals in order to observe their responses. So, again, we need to deliberately impose a treatment.
So which of the two research questions could be answered using an experiment? And that's going to be question two. So question two could be answered using an experiment, because we actually imposed the treatment. We imposed either the traditional instruction treatment or the computer treatment in order to figure out which one has a bigger effect on the reading ability.
So an experiment is a good way to gauge the effect of an intervention. So our intervention here was the type of reading instruction. We thought that would be important to the response, which was the reading ability. So if the experiment is well designed, it is a good way to gauge the effect of that intervention.
So when our goal is to understand a cause-and-effect relationship, we can conduct an experiment. An observational study, something different than an experiment, cannot be used to gauge the effect of cause and effect, but an experiment can.
A couple quick things about an experiment. If we're going back to that research question number two, the kindergarten students are called participants in the experiment. So, generally, if they're humans, they're going to be called participants. Whereas, animals or things that are part of an experiment are called either subjects or experimental units. Sometimes people are referred to as experimental units, but generally, people are participants. And then anything else is usually a subject or an experiment.
And the last thing about experiments is that because you are able to gauge a cause-and-effect relationship with an experiment, it's very important that your experiment is well designed. So designing an experiment is a different subject on its own, but this is just your basic overview of what an experiment is. So thanks for watching.