Your AP Statistics Test is Thursday afternoon. Be outside York Lecture Hall by 11:50 with your calculator and pencils (non-mechanical, remember!). Do not bring your backpack, bag, or your cell phone (or any other “smart” device like watches); find somewhere to put them. You may bring a snack or water, but it must be in clear packaging and you may not have it at your desk with you. Do not expect to leave until at least 4:00!
See this post from before with the archive of review materials.
Things you should be sure to remember (these are my thoughts, see pages 608-610 of your review book for some further points):
- The formulas for the mean/standard deviation of a sampling distribution of means and a sampling distribution of proportions are on your formula sheet. Whenever you’re being asked to calculate a mean or a standard deviation, check to see if it’s one of these.
- Related: Pay attention to which standard deviation you are being asked to calculate. The standard deviation of a sample is not the same thing as the standard deviation of a sampling distribution.
- Probability questions will come in three categories
- Direct calculation based on a finite sample space (like rolling dice). Write out all the possible outcomes and count which ones are the ones you want. This also includes two-way tables. Don’t try to use fancy formulas for those; just count!
- Number of successes questions. For example: “how many model E cars would you expect to see in a 2000 car sample?” or “what’s the probability that at least one out of the four drivers is speeding?” These are applications of the binomial probability model and should be calculated appropriately using the formulas on your sheet.
- Number of trials questions. For example: “how many chips would you expect to need to check before finding a defective one?” or “what’s the probability that the 5th donor is the first with Type-B blood?” These are applications of the geometric probability model. The formulas are not on your sheet, and so you should memorize them: q^(K-1)*p for a general probability and 1/p for the expected value.
- Know how to combine the standard deviations of two or more random variables. Remember: variances always add.
- Know how to read the regression output of a linear association. What do “constant,” “estimate/coefficient,” “prob” all mean? Refer to pages 111-112 in your review book or page 729 in your textbook.
- Know the difference between stratified and cluster sampling and how to identify/use them. Also remember that for a simple random sample every possible sample has the same chance of being chosen.
- Know the important conditions for each of the inference procedures:
- Everything: Random Condition
- Any z-test: Success/failure condition
- Any t-test: Nearly Normal/Large Enough condition
- Any chi-square test: Expected cell frequency condition
- Any 2-sample/2-proportion test: groups are independent from each other.
- Also: know when to use a 2-sample t-test vs. a matched pairs t-test
- Be mindful of notation. Don’t use x-bar when you should be using µ. Don’t use p when you should use p-hat.
- Related: No Naked Numbers! Every number included in any response should have a label somewhere (mean, standard deviation, p-value, etc.)
- Context, context, context. Always frame your answers in the context of the situation. Don’t say, “I am 95% confident that the true proportion of success is between 22% and 37%.” Say instead: “I am 95% confident that the explosive sensor will accurately detect hidden explosives 22% to 37% of the time.”
- Be complete with your responses. When choosing between two possibilities in a free response question, explain why one possibility is wrong and why the other one is correct. When asked to compare two distributions, make explicit comparative statements, e.g, “Group 1 is centered at around 12, which is greater than Group 2, which is centered around 7”
- At the same time, don’t give more information than is requested. If the question asks you to describe the shape of a distribution of data from a histogram, don’t discuss center and spread. If the question asks you to explain your randomization procedure for an experiment, you don’t need to explain what the response variable is, how you’re going to measure it, and what a statistically significant outcome would look like. If the question asks you to merely calculate a confidence interval, you don’t have to interpret it.
