Question

1. For the case-control study in Example 9.3 (the toxoplasmosis study):
(a) Find a 99% confidence interval for a parameter which compares the probability of being infected for drivers who had accidents and drivers who did not.
(b) A test of homogeneity is designed for cohort studies, where you sample from each population and test to see if there is evidence that the distribution of the categorical variable is different on some of the populations. In a case-control study, you can apply the same test and same hypotheses, even tough the underlying experiment is very different (i.e. you are not sampling from the populations of interest). Carry out the test at level 0.01 (and also be sure to check any conditions you can check).

Answer 1

P 1 2013297 185 16 20.0864 ↑ 61 +16 O. 20 81 1857185

Paramelen hay compare the per nosasility is (Pi-P2) lennon 997, confidence interval son this rareyetor is (-) + Ex standen ennen of Cours) > (@) 2-3 * PO 4 BURU -> (0.3297-00 0.0864) ± 2.58 (0.329720.6703) 6.0364 - 185 0.9136) 185 0.2433 1 2.58 X 0.0016 y error simit lower limit 0.2474 0.23917

infated 19 Driven with accidents Drivors with out accidents Unafect of total 124 185 1697185 16 coluna 77 293 totde Contingency taske summerize nelationship between categorical variable

6 test of homogenity chir squene test O osscoured value of calli E > Exrated value of celli 121 E; K no of cell in tash G D 185x71 +33.5 E = row total & column total 970 E2 185 X 291 146.5 portado 370 22 - Ž (0; -E;)2 E 38.5 ² (0, - E,)? Ey → 146.5 . CO₂-E₂) O2 - E2 - + (O2-E2) 2 - + CO₂-ET [oy - Eye E2 E] Eu عملن -> (61-385) + (124-146-5), (16-132-3)? 38.5 146. 38.5 + (109 - 146572 1465 27 13.14 + 3.45 + 13.14 + 3.45 = 33.72 son of the best statistics x² with (R-1) (6-1) head stations the distribution of the neday degree of (2-1) + (2-1) 71 Reralbs-Reve ) p(262)

to of at levd o.o1 Persimy of extending Critical raine ) of chisquine distrisation with t degnne of force day is 6.635 aff/2001 x2 7 6.635 => lie in the critical region > There is add sustrist evidence at the level (old) to conclude that there is a relationship exist seth to two categorical variash ie being in feet of or not affect chance of accidat ,
the critical value have taken from chi square table.