Hi, everybody. I know you're all dying to hear what the right answers were to the final exam. So here's the Screencast I promised you. I'll go through the first ten questions really quickly. Question number one, the described study is, correlational, experimental or quasi-experimental? Well, the right answer here was experimental. Because in the description of the research study, it says that all students are randomly assigned to receive either comparative or general feedback. So we have random assignment. That's a requirement for an experimental study. And we have two groups that we're comparing on a variable that we manipulated ourselves. So we also have manipulation and comparison. All the requirements for an experimental study. It's not correlational, because we have a clear causal hypothesis, manipulation, randomization in comparison. And it's not quasi-experimental, because then we wouldn't have had the random assignment. Okay, moving on to question number two. The lecturer could have performed a manipulation check by checking if. Well, the right answer was, the students open and read their emails. Because the manipulation check is about seeing whether the manipulation that you performed, actually had the desired effect. Not on the dependent variable, but on the independent variable. Did your manipulation work, were you successful? Did students actually receive feedback that was experienced as comparative, for example? Well, one requirement is that students open and read their email. Because if they didn't, then no feedback would have reached them. The other two options are just plain wrong. Because they're about engagement. And engagement is the dependent variable. So it has nothing to do with the manipulation. Moving on to question number three. The independent variable. Well, we've already talked about this. The independent variable of course, is type of feedback. Not type of student. That could maybe be a background variable, or maybe even a control variable. But it's not the variable that we manipulated. And that we expect will have an effect on another one, namely, engagement. The dependent variable here. Question number four. The research design aims to prevent a threat to internal validity caused by maturation, selection, or both? Well, in this case, it's both. Because random assignment mitigates the threat of selection. But the comparison of two groups mitigates, mitigates the effect of maturation. So a randomized control group design mitigates against both of these. Question number five, suppose the engagement questionnaire is very unreliable. As a result, an estimate of the following type of validity will be low. Well, the right answer here was construct validity. Because if our questionnaire is very unreliable. Then even if, in, in, in reality, it has good construct validity, it's actually measuring the quality, quality that we want to measure. If our questionnaire is very unreliable, then we won't be able to see this in an estimate of validity. Consider this, if you bring to mind the, the dart board. And even if your, your validity is high. If you're aimed at the bullseye in that, in that board. If your questionnaires are unreliable, your scores will be all over that board. And you won't be able to determine whether you're actually accurately aimed at the center. At the bullseye or somewhere to the left or the right of the board. So to be able to spot construct validity, you first need a reliable questionnaire. Of course, ecological validity has nothing to do with the reliability of the questionnaire whatsoever. And confirmative validity is a term that well, I made up. Maybe it exists, but we certainly didn't discuss it in this course. Question number six. The dependent variable. Well, we've already, gave, have given the answer to this question as well. It's engagement. Not type of feedback, thus the independent variable. And not type of student. Question number seven. This research has the following design. Well, is the randomized design anyway? And is two group design? Not the pre-test, post-test. Because the description of the research doesn't include any information on a pre-test. And of course this option refers to the quasi-experimental version of a pre-test post-test design. Question number eight. The external validity could have been improved by using, random multi-stage cluster sampling, stratified quota sampling, or neither? Well, in this case, you might be, might have been put off by the fact that I included random here. Which was actually meant to help you. Let's look at the description. Here it says, the lecturer investigates the effect in a group of premaster and regular students. In a course that she teaches. So you can assume here that, she's using these students as participants because she has easy access to them. So the sample here is a convenience sample. Now how could this be improved upon? Well by using random sampling. And even though multi-stage cluster sampling is a very complicated and often expensive affair. It's, it's cheaper than a normal random sample. And it's way better than stratified quota sampling. Which does not use the, the principle of random sampling. So in this case, the right answer was random multi-stage cluster sampling. Question Number 9. If the sample would have been twice as large, then the margin of error would have been? Well, in this case it would have been smaller. Because as the sample size grows, it better resembles the population. Or to look at it differently,. Chance outliers will be less, have less of an influence on any statistic that you're calculating. That is used to estimate a certain parameter in the population. So, a bigger sample size will give you better, higher precision. Which corresponds to smaller margin of error. Finally, question number 10. If the lecturer does not document the study in enough detail, then this compromises well, of course the answer here is transparency. Doesn't have anything to do with logical consistency or empirical testability. Those are substantive issues to do with the internal logic of the hypothesis and the methods chosen. Or, whether the hypothesis is testable at all. So, the right answer here was transparency.