Confidence Intervals—right, we can repeat Example One by finding the Bonferroni-corrected intervals for all the comparisons. First question is: what should the C-level be? Scroll down too soon—right. So generally, the C-level is one minus alpha. But if we're doing a Bonferroni-corrected interval, that's going to affect our alpha. C-level should be 1 minus 0.0167, or 0.9833. We will once more take these three cities. So the CLT already holds, right? And we'll just do a Two-SampTInterval for each of these. Let me go ahead and show you that on my graphing calculator. We already have our data in our list, so we don't have to re-enter it. We just go into my graphing calculator—let me just clear all of this—Stat → Tests → 2-SampTInt. I don't see it here... It is actually just the next one at the bottom. So you could arrow down to get there, or I kind of like to arrow up. So there's 2-SampTInt. Yes, we're working with data. We want to do Denver versus Houston, which is List 1 versus List 2. So let's make that correction real quick... Nope, not the right C-level. Supposed to be 0.9833. Nope, not pooled. Calculate. Just write down that interval, (-0.0104,0.09038) Do this again: Stat → Tests → 2-SampTInt. This time we want to do Houston versus Cleveland, which is List 2 versus List 3. Leave everything else alone. Interval is (-0.3422, -0.0618). Last one: Stat → Tests → 2-SampTInt. List 1 versus List 3. And then (-0.3079, -0.0161). All right. Each of these intervals is supposed to represent mu1 minus mu2. Okay. When we are looking at two-sample t-intervals, we're really just asking ourselves first the question: does this interval capture zero? And for one of them, yes, it does. Zero is in between the two numbers. We know that zero is always between negatives and positive numbers. So this one—the interval captures zero—so there is no significant difference, which matches what we got up here. There's no significant difference in gas price between Denver and Houston. Our other two intervals, on the other hand, though—they are completely negative. So mu1 minus mu 2 is less than zero. And if that is true, then that means that mu2 must be bigger. And in both of these cases, the second group was Cleveland. Again, Cleveland has significantly higher gas prices than Houston and Denver. It's between 6 and 34 cents, or like 2 and 31 cents more. We are about 98% confident average difference in price—you could say gas prices—is between blank and blank dollars. Just fill this part in down here as needed. So the difference is again either between 6 and 34 cents, or 2 and 31 cents. Notice I did not talk about the negative. Right—the negative really just helped us see that mu2 was bigger. If this was a completely positive interval, then it would have meant that the first group was bigger. Oh, realize I spelled "Denver" wrong—don’t know how many of you might have caught that earlier. Sorry—you've been spelling it hopefully correctly elsewhere. All right. We've already been doing this. Part C: Compare your conclusions with the results of Example One. Yeah, it's all checking out. It all matches with what we expect—[mumbling]—everything matches. Also, confidence intervals are much harder to interpret when you're working with two samples than hypothesis tests. Another way to say it: hypothesis tests are easier comparisons when working work with two samples than confidence intervals. Done!