Comments on HOMEWORK 1

There are some topics mentioned in the first homework that a few of you may not have heard of before (or in the last few years anyway), and that we obviously won't cover before it's due since that's actually our next class meeting.

So I feel compelled to provide a little explanation for each question, in case that's helpful, but skip this if you've already looked at the homework and don't have these questions. This is really just for people who may have no experience with these concepts yet.

#1: You probably know what means and standard deviations are. You're just reporting them from the program.

#2: In case you don't know what kurtosis and skewness are: Kurtosis is the degree to which a distribution's shape is either thinner and taller than a Normal distribution, or shorter and squatter than a Normal distribution. Wait, what, you don't know what a Normal distribution is? That's the standard symmetrical bell-shaped curve we'll be dealing with a lot -- don't worry, you'll understand it plenty when you need to. And Skewness is also a departure from the Normal distribution, in the sense that the peak of the curve is pushed more to one side or the other: positive skew means the right tail is stretched out, pointing in the positive direction, and most scores are piled up to the left; whereas negative skew would be the opposite, with scores piled up to the right on the high end and the the left tail dragged way out in the negative direction. There are pictures of these terms in the posted PowerPoint slides we started on Wednesday -- not to mention you'd find plenty of images by googling. Best part though, is that instead of images, you're getting SPSS to report numbers that quantify these characteristics JUST BECAUSE YOU CAN, because soon enough I'll tell you those numbers are next to useless and the real way to judge these things is visually anyway. I mean, the assignment also asks for the "standard errors" of the kurtosis and skewness, and I'm pretty sure you don't know what that means, but you'll report them anyway because I asked you to, and so now you at least know one thing about them, which is that they exist.

#3: The Pearson correlation is the usual correlation coefficient (little 'r') you've probably seen before, and if you haven't seen it, we'll get to it, don't worry. It expresses the degree to which two variables change together, as in, one tends to increase when the other increases (a positive correlation) or one tends to DEcrease when the other increases (a negative correlation); the strongest that tendency can be is 1.0 in either direction (+1.0 or -1.0) and the weakest is 0 (the variables change independently of one another so neither provides any information about the other). The question asks you to report it along with its corresponding p-value which the program labels "Sig." for "significance", because 'r' is said to be "significant" if that p-value is less than .05. Which means what, again? Literally? Precisely? Can you tell me that? Well save it for class, that's not part of the homework. Just say it's significant if p is less than .05, we'll get to what it means. We'll get to what the correlation itself means too, if you aren't familiar with it. Note that significance isn't about whether 'r' is less than .05, only whether 'p' is.

#4: A scatterplot shows where each participant falls on two variables simultaneously, in this case with Age on the horizontal axis and Memory on the vertical. Each participant is one point in the scatterplot. When you follow the instructions and produce the plot, it will be self-explanatory. The "best-fit line" is the line that best represents the relationship between the two variables, which will also probably be self-evident upon seeing it. The "trend" just means, does Memory tend to INCREASE with Age, or does it tend to DECREASE with Age, or does it stay the same (showing no trend)? The line, incidentally, has a slope and intercept that are found through regression analysis, which will occupy you in the Spring class for like 15 minutes, cause it's easy. People usually say you shouldn't claim there's a trend present at all unless it's statistically significant (p<.05 for the correlation or the regression equation), but I don't care about that for this question.

#5: A histogram divides the full range of scores into categories like 20-29 and 30-39 etc and shows how many scores fall into each, so it produces a frequency distribution that you can check for skewness and kurtosis -- only this time visually, instead of using the numbers reported in question #2. As I showed in class, SPSS can superimpose a Normal distribution over the data, for you to evaluate how Normal the data look. Don't be the ridiculous person I was making fun of and just say it fits fine no matter what!

#6: Same for the Memory variable, only you're supposed to use the SPSS syntax just to get a taste of what syntax looks like and how to run it.