Q. What did Exponent do in its, what you call 24 statistical variability analysis? And if you have 25 an understanding as to why they did this analysis? DIRECT/SNYDER/GREENSPANPage 181 1 A. Well, this is a relatively brief commentary. 2 My belief is, and this is a bit of a speculation, is 3 that they wanted to look at the variants, the 4 dispersion in the measurements between the Colts' 5 balls at halftime and the Patriots' balls at 6 halftime to see if there was a contrast. And if 7 there was a contrast so that the Patriots' balls had 8 more dispersion, more variance in their 9 measurements, that would lend support to the idea 10 that they didn't have a common starting value. Why 11 wouldn't they have a common starting value? Hasty 12 deflation. 13 Q. Let's go to the next slide. And what did 14 Exponent conclude as a statistical matter about 15 variability? 16 A. No statistical -- no statistically 17 significant difference. 18 Q. Did they stop there? 19 A. No. They continued, which is striking, 20 because, whereas in the difference in difference 21 analysis, they adopted the standard five percent as 22 the benchmark, here, they said, no, we will just 23 continue on and reach conclusions. And it's right 24 here at the bottom. 25 So without having found anything that's 06/25/2015 03:43:11 PM Page 182 to 185 of 457 50 of 172 sheets DIRECT/SNYDER/GREENSPANPage 182 1 statistically significant, nevertheless they have a 2 statement that begins in their report, "therefore." 3 Q. And in your experience, as a statistical 4 matter, is it a sound practice to draw conclusions 5 from an analysis which doesn't reach statistical 6 significance? 7 A. No.