AP Statistics Exam Review – 1st Semester ANSWER KEY
1. graph/ shape is right skewed, 200 is possible outlier, center is 137 to 140, range is from 101 to 178 ignoring outliers.
2. mode, median, mean since mean is non-resistant to outliers
3. distribution is right skewed so median would be better (resistant); IQR = 23; test for outliers shows 85 and 86 are outliers; mean and standard deviation are affected most by outliers.
4. y = 71.95 + .3833 X on average, height increases by approximately .3833 cm per month
5. smallest circumference is about 9 – only one has it; highest strength is 75 – only one has it; 3 people have
strength of 40. Linear, moderate, positive association.
6. y-hat = 1.0934 + .1890X where y is gas consumption and x is degree days
7. graph scatterplot and resid. plot. Residual plot shows definite pattern, therefore linear model is not best.
8. An experiment must have treatment imposed.
9. (i) 0.7021 (ii) 0.4524 (iii) 0.1968
10. Voluntary response sample: people volunteer to respond or self-select themselves to be in the sample.
It is not random and typically has volunteer response bias with overrepresentation of strong opinions.
11. SRS – every group of a specific size has an equal chance of being selected.
12. invnorm(.98, 150, 23) = 197.24
13. P(Y|X) = 0.32
14. Stratified – sampling procedure were everyone is grouped by common characteristic first (homogeneous
groups) then SRS is chosen from each group; typically gives very accurate results
15. Experimental group receives the experimental treatment; control group receives the placebo (dummy treatment) or no treatment.
16. yes; experiments can give good evidence for causation only if they are well-designed and well-controlled
17. control, randomization, replication
18. Blocking is an experimental procedure. Subjects are grouped by common aspect first according to the lurking variable, then randomly divided into treatment groups. Matched pairs is a form of blocking.
19. Undercoverage – method of selection that leaves out part of population
Non response – people choose not to respond or no one can get in touch with them after several attempts.
Response bias – actions of interviewer or respondent causes bias.
Wording of questions – unclear or slanted.
20. 500 heads on 1000 flips (Law of Large numbers)
21. 7/8 or 0.875
22. log y-hat = -.1094.5071 + .5558x so y-hat = 10^(-1094.5071*10^(.5558x)
23. make segmented bar graph: one bar is “both parents smoke” with 22.5% for student smokes and 77.5% for student doesn’t smoke; second bar is “one parent smokes” with 18.6% for student smokes and 81.4% for student doesn’t smoke; third bar is “neither parent smokes” with 13.9% for student smokes and 86.1% for student doesn’t smoke. In general, parents who smoke are more likely to have kids that smoke.
24. normalcdf(35000, 1E99, 32550, 2000) = 0.1103
25. mean = 715, std dev = 66
26. 0.275
27. 1 – (.70)^5 = 0.8319
28. 0.6903
29. sampling error is the difference between our sample statistic and the true population parameter. It is natural and we cannot eliminate it.
30. 0.88