<p>Chapter 6 of Barron’s: Again, not too much news for this crowd. Definitions: census, sample survey, experiment, double-blind study, observational studies.</p>
<p>How could a sample ever be better than a census? What type of study can indicate cause-effect relations? Maybe ethical issues forcing choice of observational studies vs. experiments.</p>
<p>
</p>
<p>Um I’m pretty sure we never talked about this. Is it worth looking up?</p>
<p>More vocab: population parameter vs. sample statistic, sampling error
Requirements for a simple random sample
Types of bias. Nonresponse bias is generally said to occur when people do not respond. With nonresponse bias, there is obviously a problem if the nonresponders would have a different distribution of answers than the responders. Also, I’m not sure whether AP Stats touches on the issue of people responding, but not responding truthfully? This happens in real life.<br>
It’s worthwhile to think over other sample issues, aside from whether the subjects will choose to respond. One of these is availability–for example, think about people who work the night shift at GM. (There is one now, at some of the plants.) Another issue is the wording of the questions. Surveys often find quite different %'s of the population holding certain views, because of relatively subtle shifts in the phrasing of the question.</p>
<p>A key to identifying possible biases: think outside your own realm of experience, to detect the biases. Think of people who are very poor, or homeless, or living in shelters. Think of people who are not native English speakers. There are also some biases that could affect the representation of the very wealthy in the sample. (“Oh, we’re never at home in the winter. After skiing in Gstaad, we generally go to Bora Bora for three months.”) Note: you will probably get more favorable attention from the graders by thinking about the marginalized members of society than by thinking about the rich!</p>
<p>Simpson’s paradox is quite interesting: It occurs sometimes when several different groups are combined to form one large group, and a comparison is reversed as a consequence. A classic example comes from a study of graduate admissions at UC Berkeley, in the 1970’s. It was found that women were noticeably less likely to be admitted to grad study than men. An investigation was launched to find the discriminatory departments. As it turned out, the differences in admissions likelihood between men and women were quite small, on a department-by-department basis. UC Berkeley had admissions probabilities for any applicant that varied a lot from department to department. The departments that required the most mathematics also tended to have the highest admissions rates for their applicants. The humanities and social sciences had much lower admissions rates. The women applicants were concentrated (at that time) in the departments that had lower admissions rates. So each unit treated women (more or less) fairly, but the outcomes were different, because of the “lurking variable,” of department choice, with differential admissions rates.</p>
<p>A question that might illuminate the difference between the standard deviation of an individual measurement and the standard deviation of the mean:</p>
<p>Suppose that the 8th graders all across a large state take a standardized test, and the mean and standard deviation of the scores are obtained. Then the averages for each middle school are computed. What do you expect–qualitatively!–for the standard deviation of the averages?</p>
<p>Or, it’s sometimes argued on other threads that since grades in many schools are assigned with no more precision than 0.3 on a 4.0 scale (think A, A-, B+ B . . .), then GPA’s should only be reported with the same level of precision. Is this right?</p>
<p>Bayesian probability analysis is in Barron’s. Is it in the AP Stats curriculum? I could look it up, but I figure you know the curriculum, collectively–so it’s probably faster just to ask.</p>
<p>I just took a full length practice exam for Stats today, released from the College Board. I think I feel a 5 coming on :)</p>
<p>Good to hear, byubound! And good luck to the other posters on this thread!</p>
<p>Thanks for the helpful information.</p>
<p>I took the course last semester at a CC, and they did a few things differently. I have 6 more practice tests to go, but the only thing I have to get used to is difference in variables (There are several discrepancies between Barrons and my old textbook).</p>
<p>Preparing for this seemed harder than AP Physics C, because it is relearning something differently, but I’m still going to get a 5 (I won’t let myself not get a 5 that is).</p>
<p>I just now realized that in the past FRQ gave students tables and formulas, whereas I had to memorize them and calculate them with my calculator before. Oddly enough memorizing formulas (mostly memorizing concepts and understanding derivations of other formulas) and them forgetting them makes it harder for me to keep track of them when I’m relearning, so it is a relief if they actually put the formulas on the exams. </p>
<p>Is this the case or do they just put that on the previous exams, as Barrons doesn’t seem to have that for the exams (Or I’m missing them).</p>
<p>Thanks and good luck to everyone.</p>
<p>
</p>
<p>Same here. 36/40 on a MC exam. This exam is more of a “you lose if you mess up” type of exam rather than a “you lose because you can’t figure it out” exam.</p>
<p>excellent guide. thank you so much; i can usually get the right answers but it is very important to know this kind of stuff.</p>