AP STATISTICS - how was it?

Test Preparation AP Test Preparation
121

<p>What could you possibly have put for 1a? It like it was essentially saying "define the Standard Devation"... I put something along the lines of "The average distance from the mean is 2.141. The variance is 2.141^2." I'm happy that I did 6c right (or atleast, I think I did). I didn't pay attention (at all) on the day we learned linear regression testing... I was programming on my calculator :P.</p>

<p>I'm worried about the ones where you needed to state conditions, though. If you got all of a question all correct but didn't state the conditions at all (worst case scenario, I think I actually had some right), what would you get?</p>

122

<p>whats the difference between a 2 sample z test and a 2 proportion z test?</p>

123

<p>sample tests uses means. proportion tests use proportions.</p>

124

<p>so how do you do a 2 sample-Z test...?</p>

<p>our textbook doesn't even mention this</p>

125

<p>Jarn i put the same thing for 1a</p>

<p>2 sample Z is a difference of proportions test</p>

126

<p>there's no 2 sample Z test for means. The test statistic formula for proportions is:</p>

<p>z = (p1 - p2) / sqrt(Pc(1-Pc)((1 / n1) + (1 / n2)))</p>

<p>where p1 is the observed proportion 1, p2 is the observed proportion 2, Pc is the pooled proportion, n1 is sample size 1, and n2 is sample size 2.</p>

<p>But that was definitly not on the exam</p>

127

<p>5c. According to the calculator there is a two sample z test that tests for difference in means.</p>

128

<p>this threw me off a bit on the test as well.</p>

<p>the 2 samp ztest for mean is often abbrev. 2 sample test. BUT THINK, means would be ridiculous if they gave you proportions to begin with. It simply states that 2 samples were taken, but proportions are measured, not means</p>

129

<p>It was matched pairs, right? For number 4?</p>

130

<p>yup! matched pairs!</p>

131

<p>worstbadboy, isnt it 1.02 for noncontact wearers and 1.17 for contact wearers?</p>

132

<p>This test didn't go well for me. I bombed totally part e. Secondly, in part c, I also put B=0 instead of B=1. this sucks.</p>

133

<p>It's okay, tushar. I haven't found anyone else at my school who knew to put the null hypothesis as b=1. If people were to mess up anything in the test, it would be that.</p>

<p>Can someone remind me what question 4 was? I don't remember writing about matched pairs, heh, uh-oh.</p>

134

<p>It was the problem with pieces of meat being tested for E.Coli with two different tests, but the same piece of meat. So you had to see if there was a significant difference in the amount of E.coli found by each of the two tests.</p>

135

<p>Tell me if I'm right...</p>

<p>3a) more likely to have 15 fish than 50 greater than 10 inches if mean is 8
3b) normal distribution therefore P(z<(7.5-8)/0.3)
3c) yes, normal distribution is still ok because of CLT n>30</p>

<p>4) paired TWO TAILED t-test (because they asked if there was any difference, not greater than or less than)
so Ho: ud=0
Ha: ud=/=0</p>

<p>I think I got a p value of more than .1, so no significant difference</p>

<p>5a) experiment not observation
5b) Ho: PI1-PI2=0
Ha: PI1-PI2 > 0
5c) 2 samp z test not appropriate because each group has to be large (n>30) whereas each group here is only 24. Besides it's a PROPORTION TEST. Number of drivers out of 24 who miss an exit - how can you measure the means of that?
5d) P value means that cell phone is significantly worse than passenger</p>

<p>6c) Testing the model:</p>

<p>Ho: Beta=1
Ha: Beta>1</p>

<p>P(t>(1.102-1)/.393)
with df = n-2 = 40-2 = 38</p>

<p>P value came out to be obscenely high, I think around .3 or .4, so fail to reject null hypothesis. Can't prove that they overestimate distance.</p>

136

<p>Ok so this is driving me insane. Was #4 a matched pairs test or a two sample t-test? </p>

<p>My teacher said the way they phrased the question would make you think it was a two sample t-test though she believes it was actually a matched pairs test. A lot of kids in my class did two sample t's and she even recommended that we consider writing letters to collegeboard since the question was iffy.</p>

<p>And say it was a matched pairs, but I did a two sample t, how many points would I get? 0?</p>

137

<p>No, they're not that dumb. That problem had no part a b c stuff, so I assume if you mess up the "matched pairs" thing (like I did, saying 2 sample T) you would get a 3 because it is still a "substantial response" if you do everything else right.</p>

138

<p>i dont see how that could possibly have been 2 sample t-test. i think they'll give you 3 out of 4 if you did the 2 sample test correctly.</p>

139

<p>It was matched pairs... my teacher went over it yesterday. He actually went over all of the FR questions.</p>

140

<p>part a) How did people out there interpret the slope for #6? I'm not sure if I interpreted the slopes correctly - "slope = the overestimation of the actual distance by this much percentage."<br>
part b) I would prefer Model 2 because of a larger standard deviation so that the test would be harder to reject because of a larger spread in the data. The data would then cover more people.<br>
part c) I know I did the test wrong by saying beta = 0 in the test, but I carried out the test correctly otherwise. I just divided the coefficient 1.102 by the standard error 0.393. I don't think anyone in my class noticed that we were supposed to subtract 1 from the coefficient; some even skipped #6 because it was too "abstract" for them to answer.<br>
part d) Then I put y = 1.05x and y = 17x and graphed the two lines.<br>
part e) Same thing as part a) except I changed the numbers.</p>

<p>So what do you guys think I got on #6? Thanks...</p>