Scribd Logo
Upload

Welcome to Scribd!

  • Solution John Freund

    Uploaded by

    Soyeb Lalani
    391 views43 pages
    Solution

    Copyright

    © © All Rights Reserved

    Available Formats

    PDF or read online from Scribd

    Did you find this document useful?

    Is this content inappropriate?

    Report this Document
    Solution

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    Download as pdf
    391 views43 pages

    Solution John Freund

    Uploaded by

    Soyeb Lalani
    Solution

    Copyright:

    © All Rights Reserved
    Download as PDF or read online from Scribd
    Flag for inappropriate content
    Download as pdf
    of 43
    136 Chapter Mathematical Expectation Solution FXK+N= LL Ety- fe») 1 2 1 1 0+0)-440+1)- 24042240 +0)-4 0-5 D+ 5+O+2)- 36 +0+0-5 1 1 4040-2 424+0-5 140-5 +240-5 10 . 9 EXAMPLE 4.9 If the joint probability density of X and ¥ is given by 3e42y) ford f(x) = I, and 4, would be this purpose, th the moment of inertia, Ths also explains why the moments; are called moments Here, fet us met about the origin: in the analogy to physics, the length of the lever arm is in each case the distance from the origin. The analogy applics also in the continuous case, where an jes might be the x-coordinate ofthe center of gravity and the moment of inertia ofa rod of variable density Proof. When r = 0, we have ju = E(X°) = E(1) = 1 by Corollary 2 of Theorem 4.2, and this result is as it should be in accordance with Theorems 3.1 and 35. When 1 = 1,we have | = E(X), which is just the expected value of the random variable XX, and in view of its importance in statistics we give it a special symbol and a special name ‘TuroreM Derunrrion 4.3. 11; is called the mean of the distribution of X, or simply the ‘mean of X, and it is denoted by 146 Chapter 4 Mathematical Expectation ‘TuroreM 4.10. Ifa and b are constants, then L. Myya(t) = Ele] = of. My (oy; 2 Mx (t) = Ee) = My(on); van() The proof of this theorem is left to the reader in Exercise 4.39, As we shall see later, the first part of the theorem is of special importance when a = —, and the third partis of special importance when a = —j1 and b= a, in which case 3. Mxze(t) = Ble *)'] was =eF ae(2) bveneses 4417, With reference to Definition 44, show that jag = 1 and that sy random variable for which £(X) exists 4.18. Find s.,1,,and o” for the random variable X that has the probability distribution f) = }forx = 2 and x = 2 419, Find 1, and o? for the random vatiable X that has the probability density for any = ford1 0 elsewhere check whether its mean and its variance exist 425. Show the 4.26, 427. 428, 429, 430, 431. 432, forr=1 about the The symr by means Use the f following @ Fa) as; Fa) ios, Also drat negativel The exter distributi Use the 1 following Duplicate theorem f Show that x <0,the This inegt because it Use the in tute OF = Whats th that aranc (a) atlea (b) at lew Ifwe let ke probability 39, As we shall see later, na =~—p, and the third hich case and that jy = 0 for any he probability distribution the probability density robabilty density = 2K 43. {deviation 2, show that se of X by means of the s said to be in standard able, we are said to be Section 4.5 Moment-Generating Functions 147 425. Show that emt (faa ten! ([) ais enl ee heD He Daal forr = 1,2,3, ...,and use this formula to express 13 and sg in terms of moments about the origin 426. The symmetry or skewness (lack of symmetry) ofa distribution i often measured by means of the quantity Use the formula for 3 obtained in Exercise 4.25 to determine as for each of the following distributions (which have equal means and standard deviations): (@) f(1) = 005, f(2) = 0.15, FG) = 030, f(4) = 0.30, f(5) = 0.15, and f(6) 0.08; (b) /(1) = 005, /(2) = 0.20, £3) = 0.15, 4) = 045. £05) = 0.10, and f(6 os Also draw histograms of the two distributions and note that whereas the first, is symmetrical, the second has a “tail” on the left-hand side and is said to be negatively skewed. 427. The extent to which a distribution is peaked or lat, also called the kurtosis of the distribution, is often measured by means of the quantity Use the formula for j14 obtained in Exercise 425 to find ws for each of the following symmetrical distributions, of which the frst is more peaked (narrow humped) than the second: (a) f(-3) = 006, f(~2) = 0.09, F(-1) = 0.10, f(0) = 0.50, fa) = 010, f(2) = 0.09, and f3) = 0.06; (by f(-3) = 004, (2) = 0.11, F(-1) = 0.20, FO) = 0.30, fa) = 020, ‘f2) = 0.11, and f3) = 0.04 428. Dupiicate the steps used in the proof of Theorem 48 to prove Chebyshev's theorem for a discrete random variable X 429. Show that if X is a random variable with the mean y« for which f(x x <0,then for any positive constant a 0 for P(X 2a) This inequality is called Markov's inequality, and we have given it here mainly because it leads to a relatively simple alternative proof of Chebyshev's theorem, 430. Use the inequality of Exercise 4.29 to prove Chebyshev’s theorem, [Hint: Subst tute (X — w)? for X.] 431. What is the smallest value of kin Chebyshev's theorem for which the probability that a random variable will take on a value between 1 — ko and jt + ko is (@) atleast 095: {b) atleast 0.992 432. If we let ko = c in Chebyshev’s theorem, what does this theorem assert about the probability that a random variable will take on a value between js ~c and +c? 148 Chapter 4 Mathematical Expectation 433, Find the moment-generating function of the discrete random variable X that has the probability distribution and use it to determine the values of x, and 41, 434, Find the moment-generating function of the continuous random variable X whose probability density is given by jaya |! fro<2 <1 = Vo elsewhere and use it to find yu, and 0°. 435. If we let Ry(t) = In My(1), show that Ry(0) = 1 and RYO) Also, use these results to find the mean and the vafiance of a random variable X having the moment-generating function My(r) = et! 436, Explain why there can be no random variable for which Mx (1) = 4; 437. Show that ifa random variable has the probability density L fx) "foro 0, y>0 FEY= 19 ctsewhere find their joint moment-generating function and use it to determine the values of E(XY), E(X), EC), and cov(X, ¥). 448. If X,, Xo, and Xq are independent and have the means 4, 9, and 3 and the variances 3,7, and 5, find the mean and the variance of (a) Y= 2X) — 3X2 4X: (by Z = X1+2%2~ X3 449. Repeat both parts of Exercise 4.48, dropping the assumption of independence and using instead the information that cov(X;, X2) = 1, ov(X2, Xs) = ~2, and cov(Xy, Xs) = -3. 450. If the joint probability density of X and ¥ is given by 1 xty) fordb dus case that the partial fon with respect to 1 at (b) Differentiate the result of part (a) with respect to x to find the conditional tus case that the second probability density of X given a < X [too 4.9 THE THEORY IN PRACTICE usesit to determine the In Section 3.8 we discussed how empirical distributions, those arising from data, can bbe described by their shape. Here we will extend this discussion to include deseriptive ‘measures, calculated from data, that extend the methodology of describing data. These descriptive measures are based on the ideas of moments, given in Section 4.3. The analog of the first moment, i; = i, is the sample mean, x, defined as ans 4,9, and 3 and the ption of independence cov(X2, X3) = -2, and where i = 1,2,...,n and nis the number of observations. We will have much more to say about the sample mean in Chapters 8 and 10 through 13, “The usefulness of the sample mean as a description of data can be envisioned by imagining that the histogram of a data distribution has been cut out of a piece of cardboard and balanced by inserting a fulcrum along the horizontal axis. This balance point corresponds to the mean of the data. Thus, the mean can be thought of asthe centroid of the data and, as such, it describes its location interms of the variances ‘The meanisan excellent measure of location for symmetric or nearly symmetric distributions. But it can be misleading when used to measure the location of highly 3,eov(Xs, Xs) = —2, skewed data. To give an example, suppose, in a small company, the annual salaries 1242 +3X3 and ofits ten employees (rounded to the nearest $1,000) are 25,18, 36,28, 16, 20,29, 32, 41, and 150. The mean of these observations is $39,500. One of the salaries, namely $150,000, is much higher than the others (it’s what the owner pays himself) and only can and the conditional tne other employee earns as much as $39,500. Suppose the owner, in a recruiting ad, claimed that “Our company pays an average salary of $39,500.” He would be 2ectation ofthe random technically correct, but very misleading, Other deseriptive measures for the location of data should be used in cases ¢an and the conditional like the one just described. The median describes the center of the data as the ‘middle point of the observations. If the data are ranked from, say, smallest to an even integer, and itis 1 P itmisan odd integer. example is $28,000, and tis company can expect sn income” for, say, the ifthe mean here because in the United States is moderate incomes, but a ption. Give the location bservations are grouped be based on the square |, then subtraction of the is much easier to use the {in for reasons that will calculation of s requires im in common use will deviation once the data ledin a steel mill and cut 9 122 the mean a reasonable SU84121+...122 standard deviation, we ons, (11.8)? + 2.1)? + Section49 The Theory in Practice 161 + 02.2) 435.94, Then substituting into the formula for s, we obtain s® = (10)(1435.94) — (119.8)7/(10)(9) = 0.082 foot. Taking the square root, we obtain s = 0.29. The mean, 11.98 feet, seems to be a reasonable measure of location inasmuch as the data seem to be approximately symmetrically distributed. . ‘The standard deviation is not the only measure of the dispersion, or variability of data. The sample range sometimes is used for this purpose. To calculate the range find the largest and the smallest observations, x; and x, defining the range to be This measure of dispersion is used only for small samples; for larger and larger sample sizes, the range becomes a poorer and poorer measure of dispersion, APPLIED EXERCISES SECS. 4.1-4.2 4460, The probability that Ms. Brown will sll apiece of property ta profit of $3,000, is gy the probability that she will sell it at a profi of $1,500 is y, the probability that she will break even is 3, and the probability that she will lose $1,500 is 3, ‘What is her expected profit? 4461. A game of chance is considered fair, or equitable, if each player's expectation is ‘qual to zero, If someone pays us $10 each time that we roll a3 oF a4 with a balanced die, how mach should we pay that person when we roll a1, 2 5, oF 610 make the game equitable? 4.62, The manager ofa bakery knows that the number of chocolate cakes he can sellon any given day i a random variable having the probability distribution f(x) = 4 for 1 =0,1.2.3.4, and 5. He also knows that there isa profit of $1.00 for each cake that sells and a loss (due to spoilage) of $0.40 foreach eake that he does not sell. Assuming that eath cake can be sold only onthe day itis made, find the baker's expected profit fr a day on which he bakes {@) one of the cakes; @) two of the cakes {© three of the cakes; (@) four of the cakes ©) five of the cakes How many should he bake in order to maximize his expected profit? 4463, Ifa contractor’ profit on a construction job can be looked upon as a continuous random variable having the probability density 1 A(t) for-l0 fay = 44 0 elsewhere Find the mean and the variance ofthis random variable. 4.70, With reference to Exercise 392, find the mean and the variance of the random variable in question. With reference to Exercise 387, find the mean and the variance of the random variable V. 4.72, The following are some applications of the Markov inequality of Exercise 429: (a) The scores that high school juniors get on the verbal part of the PSA) NMSOF test may be looked upon as values of a random variable with the mean = 4. Find an upper bound to the probability that one of the students will get a score of 65 or more. () The weight of certain animals may be looked upon as a random variable ‘with a mean of 212 grams. none of the animals weighs less than 165 grams, find an upper bound tothe probability that such an animal will weigh atleast 50 grams. 4.73, The number of marriage licenses issued ina certain ety during the month of June may be looked upon as a random variable with ss = 124 and ¢ = 7.5. According to Chebyshev’s theorem, with what probability can we assert that between 64 and 184 marriage licenses will be issued there during the month of June? 44.74, A study of the nutritional value ofa certain kind of bread shows that the amount Of thiamine (vitamin B;) in a slice may be looked upon as a random variable with je = 0.260 milligram and = 0.005 milligram. According to Chebyshev's theorem, between what values must be the thiamine content of (a) atleast & ofall slices ofthis bread; (@) atleast {fof all slices of this bread? 4.75, With reference to Exercise 4.69, what can we assert about the amount of time it Lakes a person tobe served a the given restaurant if we use Chebyshev's theorem with k = 1.5? What isthe corresponding probability rounded to four decimals? SECS. 4.6-4.9 4:16, A quarter is bent so that the probabilities of heads and tails are 0.40 and 0.60. Itt is tossed twice, what is the covariance of Z, the number of heads obtained ‘on the first toss, and W, the total number of heads obtained inthe two tosses of the coin? 4.77. The inside diameter of a cylindrical tube is a random variable with a mean of 3 inches and a standard deviation of 0.02 inch, the thickness of the tube is a random var and the twe deviation 0 4.78. The length a standard bricks is a1 (0,03 ineh. ¥. made of 50 variables in 4.79. Ifheadsis: die, and ge (of 52 playin of successe (a) fp a well-st (b) fipat card fr 480. If we alter probability of the num 481. With refer number of 482. With refer. con gasoline 4.83, The amout telephone With refer these telep REFERENCES Further informatie mathematical staté lips of @ coin, At the start thas b dollars, at each flip dntinues until either player itable game each player's that Mr. Adams will win SECS. 4.3~- ‘e number of television at a given restaurant is a ae, hhe variance of the random he yariance of the random equality of Exercise 4.29: verbal part of the PSAT/ ‘random variable with the ity thatone of the students fapon as a random variable weighs less than 165 grams, sn animal will weigh atleast ty during the month of June 124 and « = 7.5. According Wwe assert that between 64 the month of June? read shows that the amount ‘upon as a random variable ‘According to Chebyshev’s content of about the amount of time it te use Chebyshev's theorem rounded to four decimals? SECS. 4.6-4.9 and tails are 0.40 and 0.60. + number of heads obtained ‘btained in the two tosses of om variable with a mean of, f thickness of the tube is @ References 163 random variable with a mean of 0.3 inch and a standard deviation of 0.005 inch, ‘and the two random variables are independent. Find the mean and the standard deviation of the outside diameter of the tube. 4.78. The length of certain bricks is a random variable with a mean of 8 inches and a standard deviation of 0.1 inch, and the thickness of the mortar between two bricks is a random variable with a mean of 0.5 inch and a standard deviation of (0.03 inch, What is the mean and the standard deviation of the length of a wall ‘made of 50 of these bricks laid side by side, if we can assume that all the random variables involved are independent? 4.79. If heads is a success when we flip a coin, getting a six isa success when we roll a die, and getting an ace is a success when we draw a card from an ordinary deck ff $2 playing cards, find the mean and the standard deviation of the total number of suecesses when we {(@) flip a balanced coin, roll a balanced die, and then draw a card from a well shuffled deck (b) ip a balanced coin three times, roll a balanced die twice, and then draw a ‘card from a well-shuffled deck. 480. If we alternately flip a balanced coin and a coin that is loaded so that the probability of getting heads is 0.45, what are the mean and the standard deviation ff the number of heads that we obtain in 10 flips of these coins? 481. With reference to Exercise 3.98 and part (b) of Exercise 3.103, find the expected ‘number of mathematics texts given that none of the statistics texts is selected. 482. With reference to Exercise 3.107, by how much can a salesperson who spends $12 ‘on gasoline expect to be reimbursed? 483. The amount of time (in minutes) that an executive of a certain firm talks on the telephone is a random variable having the probability density for02 0 elsewhere With reference to part (b) of Exercise 4.59, find the expected length of one of these telephone conversations that has lasted at least 1 minute. REFERENCES Further informati mathematic on about the material in this chapter may be found in the more advanced istics textbooks listed atthe end of Chapter 3 172 ChapterS Special Probability Distributions if we want to determine, say, 4’, or 1,. Actually, there exists yet an easier way of (b) Based o determining the moments of the binomial distribution; itis based on its factorial if XX moment-generating function, which is explained in Exercise 5.12. Bernoul c caauial 5.7. Prove Theor f for = 1,2,....k show that 58 When calcul: k be simplified SL. IEX has the discrete uniform distribution f(x k41 =1 (b) its variance is o? re Verity this f (Hine: Refer to Appendix A.) re ae 1 5.2. If X has the diserete uniform distribution f(x) = 1 for x = 1,2,...,.,show that 89. Use the rect its moment-generating functions given by © Serctatical : aa () amaxin va - et) Mx = Fare (b) maxima S10, IFX isa bino ‘Also find the mean of ths distribution by evaluating lim M'y(), and compare the anal result with that obtained in Exercise $1 SAL. In the proof y detail in Section 5.3, because second facto 1. Show that for the We didnot study the Bernoull distabution in itean be looked upon asa binomial distribution with n Bemoulliistibuton, = @forr = 1,2,3,.,by : Expres (a) evaluating the sum J x" f3:0) The eto 2 given by (©) lettingn = 1 in the moment-generating function ofthe binomial distribution and examining its Maclaurin's series. 54, Use the result of Exercise 53 to show that for the Bernoulli distribution, show tm (a) c= +=. where as is the measure of skewness defined in Exer- factorial mo wo} 53, With eferer cise 434 (@) the Ber =wa-6 () y= *=¥O=O here ay is the measure of peakedness defined in () the bine £85 Sad, If we let a= Exercise 427 es . Verify that (a) bee; n, 8) = bin =a: 0, 18) Also show that if B(x; ,) = $2 b(ki n,@) for x =0,1,2,....m, then Gy Seon (by bors, 8) = Blin, 8) ~ BUx Lin.) (@) Find sw (©) B(x; 0.6) = Bin — x21 - 0) — Bn ~~ 1;n,1—8); distribu (@)_Bxsn,8) =1— Bin x—I:n, 1-6). SAS. Use the rest 6. An alternative proof of Theorem 5.2 may be based on the fact that if X}, Xo, {nd X, are independent random variables having the same Bernoulli distribution with the parameter 6, then ¥ = X1 + X2-+-:-+ Xy is a random variable having the binomial distribution with the parameters n and 9. where ay is (a) Verify directly (that is, without making use of the fact that the Bernoulli conclude ab distribution isa special case of the binomial distribution) that the mean and (a 9=) the variance of the Bernoulli distribution are 4-0), (b) nislarg yet an easier way of, )ased on its factorial 2 1,2,...,k show that 2 1,2,...,kshow that {f_(t)-and compare the in Section 53, because = 1. Show that for the te binomial distribution Ali distribution, ‘mess defined in Exer- peakedness defined in 1 then 8. fe fact that if X3, X2, te Bernoulli distribution random variable having fact that the Bernoulli tion) that the mean and. ando? = 6(1 — 6). Section 5.4 The Binomial Distribution 173 (b) Based on Theorem 4.14 and its corollary on pages 153 and 154, show that if X),Xy,..., and Xp are independent random variables having the same and Y Bernoulli distribution with the paramete 1+ Xa+ +X, then EW) n@ and var(Y) ° Prove Theorem 53. When calculating all the values of a binomial distribution, the work can usually be simplified by fist calculating b(0;n, 0) and then using the recursion formula w+pa-H Verify this formula and use it to calculate the values of the binomial distribution with n = 7 and @ = 0.25. 159. Use the recursion formula of Exercise 5.8 to show that for # distribution has gs bee + 10,8) = borin.) 4 the binomial (oy macimaats = "5! and =! whenn so 5.40. If X is a binomial random variable, for what value of 6 is the probability b(x; m,@) S11. In the proof of Theorem 5.2 we determined the quantity E[X(X ~ 1)], called the second factorial moment. In general, the rth factorial moment of X is given by [X(K X= 2)-. KDI Hin Express 1) 15, and 1 in terms of factorial moments. 5.12. The factorial moment-generating function of a discrete random variable X is given by FY = De fe) Show that the rth derivative of Fy(t) with respect to r at ¢ = 1 is wi). the rth factorial moment defined in Exercise 5.11 ‘5,13, With reference to Exercise 5.12, find the factorial moment-generating function of (a) the Bernoulli distribution and show that y/,, = @ and 4/,, = 0forr > 1 (b) the binomial distribution and use it to find ye and o* Sad, If we let a =~ in the first part of Theorem 4.10, where 1 is the mean of X, we get Fei My(t) = Mx-y(t) =e" Mx) Show that the rth derivative of My-y(t) with respect to at 1 = 0 gives the rth moment about the mean of X. (b) Find such a generating function for moments about the mean of the binomial distribution, and verify that the second derivative at r = is n6(1 — 6). @ Use the result of part (b) of Exercise 5.14 to show that forthe binomial distribution 1 28 Yao — where as is the measure of skewness defined in Exercise 4.34. What can “Conclude about the skewness of the binomial distribution when @) o=4 () nis large? terval or in a specified nlapping time intervals success occurring in a tional to the length of bility of more than one such a small region is 2 number of telephone errors per page, or the ther of successes, 4, for truck depot in a certain fen day fewer than nine siven day. Then, using, ly, successes occur at a fe number of successes ed region is a Poisson ‘Therefore, the number mm of size r units has the sis per 10 square feet. If ita 15-square-foot sheet quare-foot sheet of the 0.7586 EXERCISES 5.16. sa. 518, 519. 520. 521. 5823. 524, Section 5.7 The Poisson Distribution 185 “The negative binomial distribution is sometimes defined in a different way as the distribution of the number of failures that precede the kth success. Ifthe kth success oceurs on the xth tral, it must be preceded by x ~ k failures. Thus, find the distribution of ¥ = X — k, where X has the distribution of Definition 5.4 With reference to Exercise 5.16, find expressions for yey and of. Prove Theorem 55. Prove Theorem 5.6 by first determining £(X) and E[X(X +1)} Show that the moment-generating function of the geometric distribution is siven by M0 = oe Use the moment-generating function derived in Exercise 5.20 to show that for 1-8 >. Differentiating with respect to # the expressions on both sides of the equation Yea - chs hat the mean ofthe geomet duo gen by = 3 Then differentiating again with respect to 6, show that yi = ==" and hence that If X isa random variable having a geometric distribution, show that PCa x tniX > m) (X=x) If the probability is f(«) that a product fails the xth time its being used, that i, fon the xth tral, then its failure rate at the xth trial is the probability that it will fail on the xth trial given that it has not failed on the first x ~ 1 trials; symbolically itis given by fix) Za) = i-Fe-0 where F(x) is the value of the corresponding distribution function at x. Show that if X is a geometric random variable, its failure rate is constant and equal to 6 . A variation of the binomial distribution arises when the n trials are all indepen- ‘ent, but the probability of a success on the ith trial is 6, and these probabilities fare not all equal. If X is the number of successes obtained under these conditions in m trials, show that (0) sox =n0, where = 2. n(1 = 0) —no?, whe! 1 Sa -. How do these results tie in with the discussion on page 178? ‘ 5.29, When calculating all the values of a Poisson distribution, the work can often be simplified by first calculating p(0; 2) and then using the recursion formula por + 1:3) = AS meas) Verify this formula and use it and «-? = 0.1353 to verify the values given in Table IL for 2 = 2 ‘530, Approximate the binomial probability (3; 100, 0.10) by using (the formula forthe binomial distribution and logarithms (b) Table IL 8.1. Suppose that f(x) isthe probability of getting x successes during atime interval tt length # when (i) the probability ofa suocess during a very small ime interval from to1-+ Arisa At, i) the probability of more than one success during such 4 time interval is negligible, and (i) the probability of a success during such a time interval does not depend on what happened prior to time t (a). Show that under these conditions fot +A = fl. n[l—@- Or} fee 1ne- ae and hence that atfex.o) OI alee = 1.) Fe.0] (b) Show by direct substitution that asolution ofthis infinite system of differential equations (there isone for each value of xis given by the Poisson distribution with 4 =at .e repeated integration by parts to show that 532, DE This result is important because values of the distribution function of a Poisson random variable may thus be obtained by referring to a table of incomplete ‘gamma functions. 15.33, Derive the formulas for the mean and the variance of the Poisson distribution by first evaluating E(X) and E[X(X ~ 1)] 534. Show that ift are applied te get the mome [Hine Make v 5.35. Use Theorem the measure ¢ 5.36, Differentiatis derive the fo Poisson distri tofin 5.37, Use Theoret X isa randor tse itt veri 5.8 THE MULTINOMIAL An immediate gen more than two pos the same for each for instance, when are for a candidat products are rated To treat this are n independent probabilities are @ of the first kind, tt probability of gett and x, outcon Proceeding we first find that t of the second kit on -6 a each kind in any order by according to The! stribution, the work and then using the hein.) the hypergeometric ‘Theorem 57 variance of the hyper N=n a-8 How N=1 he work can often be ursion formula ‘during a time interval y small time interval ne success during such success during such Lar tesystemof differential he Poisson distribution 1 function of Poisson a table of incomplete Poisson distribution by Section 5.8. The Multinomial Distribution 187 ‘5.34, Show that ifthe limiting conditions n —> co, 6 —+ 0, while n9 remains constant, are applied to the moment-generating function of the binomial distribution, we get the moment-generating function of the Poisson distribution. (Hin: Make wse ofthe fact that fi, (1+ 2)" =) 5.38, Use Theorem 5.9 to show that forthe Poison distribution y= where as is the measure of skewness defined in Exercise 4.26. 5.6, Differentiating with respect to the expressions on both sides ofthe equation t= Dear -* derive the following recursion formula forthe moments about the mean of the Poisson distribution ol ey ma feet 2B pros =afrinar Be] forr = 1,2,3,... Also, use this recursion formula and th fact that yo = 1 and ju; = Otofind na, and a, and verify the form given for ap in Exercise 5.35, 537, Use Theorem 59 to find the moment-generating function of Y = X— A, where X isarandom variable having the Po use itto verify that o} on distribution with the parameter 2, and 5.8 THE MULTINOMIAL DISTRIBUTION An immediate generalization of the binomial distribution arises when each trial has more than two possible outcomes, the probabilities of the respective outcomes are the same for each trial, and the trials are all independent. This would be the case, for instance, when persons interviewed by an opinion poll are asked whether they are for a candidate, against her, of undecided or when samples of manufactured products are rated excellent, above average, average, or inferior To treat this kind of problem in general, let us consider the case where there are n independent trials permitting k mutually exclusive outcomes whose respective probabilities ar 0.6, a (wine :) Referring to the outcomes as being of the first kind, the second kind, .... and the kth kind, we shall be interested in the probability of getting x; outcomes of the first kind, x2 outcomes of the second kind, and x outcomes ofthe Ath kind (rs Euan) Proceeding as in the derivation of the formula for the binomial distribution, wwe first find that the probability of getting x; outcomes of the first kind, x) outcomes of the second kind, ..., and x outcomes of the kth kind in a specific order is 6-2... of. To get the corresponding probability for that many outcomes of each kind in any order, we shall have to multiply the probability for any specific anferby sayz according to Theorem 1.8. bile acidents is due to ‘tion and roundin; tomobile accidents are seome very aggressive 1 drug, find the ed the drug will fn in 70 percent of all divorce cases filed in ool seniors capable of, this claim is true, use ‘01 seniors capable of artain Wester city will ind the probability that of Figure 5.1, find the faywhere from three to age had been 42 instead binomial distribution age had been 51 inste binomial distribution ard member claims that School for more than a ‘would be correct to say ave been about eq information we would thas to be right, what lund that 11 of 12 newly stribution by half, what distribution having the de about the standard Section 5.10 The Theory in Practice 195 SSL. A manufacturer claims that at most 5 percent of the time a given product will sustain fewer than 1,000 hours of operation before requiring service. Twenty products were selected at random from the production line and tested. It was found that three of them required service before 1,000 hours of operation. Comment on the manufacturer's claim, $52. (a) Use a computer program to calculate the probability of rolling between 14 and 18 “sevens” in 100 rolls ofa pair of dice. (b) Would it surprise you if more than 18 “sevens” were rolled? Why? 553, (a) Use a computer program to calulate the probability that more than 12.0 80 busines telephone cals las longer than five minutes fits assumed that 10 percent of such alls last that long (0) Cam this result be used as evidence that the assumption s reasonable? Why? ‘554, Use Chebyshev’s theorem and Theorem 5.3 to verify that the probability is at teast 8 that {@) in 90 flips of balanced coin the proportion of heads will be between 040 and 060 (©) in 10000 fps of a balanced coin the proportion of heads willbe between 047 and 035 (6) in 1,000,000 fips ofa balanced coin the proportion of heads will be between 0.497 and 0.503 Note that this serves to illustrate the law of large numbers. 555. You can get a feeling for the law of large numbers given on page 170 by flipping coins, Flip @ coin 100 times and plot the accumulated proportion of heads after ‘each five fips 5.56. Record the first 200 numbers encountered in a newspaper, beginning with page 1 and proceeding in any convenient, systematic fashion, Include also numbers appearing in advertisements. For each of these numbers, note the leftmost digit and record the proportions of I's, 2's, 3s, ..., and 9's, (Note that 0 cannot be a Teftmost digit. In the decimal number 0.0074, the leftmost digit is 7.) The results ‘may seem quite surprising, but the law of large numbers given on page 170 tells you that you must be estimating correctly SECS. 5.5-! 5.57. Ifthe probability is 0.75 hata person will believe a rumor about the tran of a certain politician, find the probabilities that (a) the eighth person to hear the rumor willbe the fifth to believe it {b) the fifteenth person to hear the rumor will be the tenth to believe it ‘a male or female child are both 0.50, find the 7 . If the probabilities of havi probabilities that (a) a family's fourth child is their first son; (b) a family’s seventh child is their second daughter; (©) a family’s tenth child is their fourth or fifth son. 5.59, An expert sharpshooter misses a target 5 percent of the time, Find the probability that she will miss the target for the second time on the fifteenth shot using {@) the formula for the negative binomial distribution, (b) Theorem 5.5 and Table I 5.60. When taping a television commercial, the probability is 0.30 that a certain actor will get hs lines straight on any one take. What is the probability that he will get his lines straight for the fist time on the sixth take”? 196 ChapterS Special Probability Distributions S61. Ina “torture test” a light switch is turned on and off until it fails. Ifthe probability is 0.001 that the switch wil fail any time its turned on or off, whats the probability that the switch will not fail during the first 800 times that itis turned on or off? ‘Assume that the conditions underlying the geometric distribution are met and use logarithms. 5462. Adapt the formula of Theorem 5 so that it can be used to express geometric probabilitesin terms of binomial probabilities, and use the formula and Table to (a) verily the result of Example 55; (b)_ rework Exercise 5.60. 5.63. A quality control engineer inspects a random sample of two hand-held calculators from each incoming lot of size 18 and accepts the lot if they are both in good ‘working condition; otherwise, the entire lot is inspected with the cost charged to the vendor. What are the probabilities that such a lot will be accepted without further inspection if it contains (@) four calculators that are not in good working condition; (b) eight calculators that are not in good working condition; (€) 12 calculators that are not in good working condition? |. Among the 16 applicants for a job, 10 have college degrees. If three of the applicants are randomly chosen for interviews, what are the probabilities that (@) none has a college degree: (b) one has a college degree: (©) wohave college degrees; - (@) all three have college degrees? ‘5465. Find the mean and the variance of the hypergeometric distribution with m = 3,.N = 16,and M = 10,using {@) the results of Exercise 5.6 (b) the formulas of Theorem 57. 5.66, What isthe probability that an TRS auditor will catch only two income tax returns with illegitimate deductions if she randomly selects five returns from among 15 returns, of which nine contain illegitimate deductions? ‘5.67. Check in each case whether the condition for the binomial approximation to the hypergeometric distribution is satisfied: (a) N =200andn = 12: (b) N= 500 and n = 20; (©) N = 640 and n = 30, ‘5468. A shipment of 80 burglar alarms contains four that are defective. Ifthree from the Shipment are randomly selected and shipped to a customer, find the probability that the customer will get exactly one bad unit using (a) the formula of the hypergeometric distribution: (b) the binomial distribution as an approximation. 5.469, Among the 300 employees of a company, 240 are union members, whereas the others are not. If six of the employees are chosen by lotto serve on a committ that administers the pension fund, find the probability that four of the six will be union members using, (a) the formula for the hypergeometric distribution: (b) the binomial distribution as an approximation. 5.70, A panel of 300 persons chosen for jury duty includes 30 under 25 years of age. Since the jury of 12 persons chosen from this panel to judge a nareotics violation does not include anyone under 25 years of age, the youthful defendant's attorney complains that this jury is not really representative. Indeed, he argues, if the selection were random, the probability of having one of the 12 jurors under 25 g su. 87. 873. 575. S71. 5.78, 57. 5.80, Sal. years of a 25 years Cheek in a good ap use the Pr @ nm ) @n @n With refe the perce to-approx tisknow are wrong to determ Records spending binomial the fair at In a give binomial drivers ra (a) only (b) at me With refe the proba cight ot (a) they, () thev The numt a random for the Pc complaint ‘The mumt «Poisson to find the (a) withe () with Use Table Inacerta year from distributic (a) three () atlea (©) anyw In the in ‘imperfect is. f the probability hatis the probability js turned on or off? ‘bution are met and ‘ express geometric ‘mmulaand Table Ito tand-held calculators ley are both in good bhthe cost charged to be accepted without rees. If three of the robabilties that Jistribution with two income tax returns tums from among 15 approximation to the tive, If three from the find the probability members, whereas the t four ofthe six will be under 25 years of age. se a narcotics violation il defendant's attorney eed, he argues, if the the 12 jurors under S71. 572. 573. 5.78, 8.76. S77. 578. 579. 5.80. S8L Section 5.10 The Theory in Practice 197 years of age should be many rimes the probability of having none of them under 25 years of age. Actually, what isthe ratio of these two probabilities? Check in each case whether the values of m and é satisfy the rule of thumb for 1 good approximation, an excellent approximation, or neither when we want to use the Poisson distribution to approximate binomial probabilities. (a) n= 125 and =0.10; (by n =25 nde (©) n= 120ands (@ n=40ande With reference to Example 58, determine the value of « (from 5 to 15) for which the percentage error is greatest when we use the Poisson distribution with 2 = 7.5, to approximate the binomial distribution with n = 150 and @ = 0.0. Itis known from experience that 14 percent of the cals received by a switchboard are wrong numbers. Use the Poisson approximation to the binomial distribution to determine the probability that among 150 calls received by the switchboard ‘two are wrong numbers. Records show that the probability is 0.0012 that a person will get food poisoning spending a day at a certain state fair. Use the Poisson approximation to the binomial distribution to find the probability that among 1,000 persons attending the fair at most two will get food poisoning, . In a given city, 4 percent of all licensed drivers will be involved in at least fone car accident in any given year. Use the Poisson approximation to the binomial distribution to determine the probability that among 150 licensed drivers randomly chosen in this city {(@) only five will be involved in at least one accident in any given year (b) at most three will be involved in at least one accident in any given year. With reference to Example 5.13 and the computer printout of Figure 5.4, find the probability that a 1S-square-foot sheet of the metal will have anywhere from eight to twelve defects, using (a) the values in the POX = K) column; (b) the values in the POX LESS OR = K) column, ‘The number of complaints that a dry-cleaning establishment receives per day is a random variable having a Poisson distribution with 2. = 3.3. Use the formula for the Poisson distribution to find the probability that it will receive only two complaints on any given day. The number of monthly breakdowns of a computer is a random variable having 4 Poisson distribution with 4 = 1.8. Use the formula for the Poisson distribution to find the probabilities that this computer will function (a) without a breakdown; (b) with only one breakdown. Use Table II to verify the results of Exercise 5.78. In a certain desert region the number of persons who become seriously ill each year from eating a certain poisonous plant is a random variable having a Poisson distribution with 4 = 5.2. Use Table II to find the probabilities of (a) three such illnesses in a given year (b) at east 10 such illnesses in a given year; (©) anywhere from four to six such illnesses ina given year. In the inspection of a fabric produced in continuous rolls, the number of imperfections per yard is a random variable having the Poisson distribution with sare distribution are nerating function of amma distribution is some of the lower Isewhere is a special vay. tion and itis referred llications in Bayesian inbles, and there is a ter 6 of the binomial sval from 0 to 1. By ma great variety of 2 beta distribution in Chapter 10, te curve of the beta sut in the proof of the oted Bea, p); in other ‘unction may be found Section 6.4 The Beta Distribution 207 ‘TweoREM 6.3. The mean and the variance of the beta distribution are given by ap and 0° = Ga Hat AED Proof. By definition, Te@+h) ft) cet " y= DEFP) | fy. yet af tae *-T@-TH i . Ta@+p) Pa@+t)-T6) =P@)- Pe Te+p+) ate where we recognized the integral as B(a +1, 8) and made use of the fact that Pi@+1) =a-M(a) and Pa +B +1) = (@+f)-T(a +p). Similar steps, which will be left to the reader in Exercise 6.28, yield (@+1)a @sB+D@+p) he and it follows that = arprnern (a) = cep nas thats EXERCISES ‘61. Show that if a random variable has a uniform density with the parameters « and {, the probability that it will take on a value less than a + p(B ~ a) is equal to p. 62. Prove Theorem 6.1 63. Ifa random variable X has a uniform density with the parameters e and p, find its distribution function 64, Show that if a random variable has a uniform density with the parameters a and Pathe rth moment about the mean equals (@) Owhen r is odd 1 (8 » © al 65. Use the results of Exercise 6.4 to find as and ag for the uniform density with the parameters a and f. 64. A random variable is said to have a Cauchy distribution if its density is given by ) when riseven. ‘Show that for this distribution wi, and j, do not exist 208 Chapter 6 Special Probability Densities 6.7. Use integration by parts to show that P'(@) = (a ~ 1) F(a —1) fora > 1 6.8. Perform a suitable change of variable to show that the integral defining the | we : we rq . eee aa Winn fF @QFa2{Pevad {fre vra aaf Peterae 610, Find the probabilities thatthe value ofa random variable will exceed 4 if thas a 623. A rane gamma distribution with density G@) a= 2and p= 3: © aa3andpa4 641, Show that gamma distribution with a > 1 has a relative maximum at x = Ble ~ 1) What happens when 0 2 a < land when @ = 1? aa 6.12, Prove Theorem 6.4, making the subttution y = x(3—1) in the integral @) defining Mx ° © sb 643. Expand the moment generating function of the gamma distribution as a binomial nal series and read off the values of sje 1 and 1 ie Sea re ers tokadet tae tte gamma cote, 628 te Show that ita random variable has an exponential density withthe parameter 8 the probability that twill ake on a valueless than ~9'In(1~p) is equal to p Ft fords pl Ba 616, 1X hasan exponential distribution, show that that PURE I4 TINE T= PED ( Sb ©) Sb ‘This property ofan exponential random variable parallels that of geometsic s random variable given in Exercise S23 625, Vesity 617, 1 isa vandom variable having an exponential distribution with the parameter oe 6, use Theorems 410 and 64 to find the moment-generating function of the oe random variable Y = X'~ 8. os Oa 6.18, With reference to Exercise 6.17, using the fact that the moments of Y about the origin are the corresponding moments of X about the mean, find cs and ay for z the exponential distribution with the parameter 6 ae 6.19, Show thatif v > 2, the chi-square distribution has a relative maximum at x = v—2. ors What happens when y= 200-<1 <3? Qe 6.20, Aandom variable has Rayleigh dtebution and only i is probably density is given by a « aya PE fore > 0 aia 0 elewhere ee where @ > 0. Show that for this distribution 6.28, Verify (-1)fora > 1 + integral defining the a0 [eterracay ategral, and thus show will exceed 4 if it has a ative maximum at x = tr 1) ine ine! isbons binomial jan stn ty with the parameter 8, ‘In(l — p) is equal to p ) lels that of a geometric ation with the parameter erating function of the moments of Y about the ‘mean, find a3 and a for vyemaximumat x = v=2. 1d only if its probability 621. Section 6.4 The Beta Distribution 209 o=1(1-7) A random arable haa Pareto istebutonf and ony iis probability density ispvenby wil de IS 4 pey={¥ 0 chewbere where a > 0. Show that 1; exists only ifr < a 9. With reference to Exercise 6.21, show that for the Pareto distribution w= 2; provideda > 1 .. A random variable X has a Weibull distribution if and only if its probability density is given by xP lee? forx > 0 where a > Oand 8 > 0 (a) Express kin terms of «and i (0) Show that ye = aren (145). : a Note that Weibull distributions with 6 = 1 ate exponential distributions | Ifthe random variable T is the time to failure of a commercial product and the aise art prota deg and dsbution function st ine Fare) and FA), then its fire ate at ime (se aio Exerc 524) given by 0 Ths the fare rat at time isthe probably density of fare atime gen tht fare doesnot sur prior time (2) Show that hasan exponential dstbston, he failure rte constant (2) Sow that iF has¢ Webuldstibation (ce Exerc 623) the at ate is given by apr?!) 6.28. Verify that the integral ofthe beta density from —oo to oe equals 1 for (a) @ = 2and f= () a=3andp=3 626, Show that ita = Land p > 1, the beta density has a elatve maximum a ad *"a4p-2 627, Sketch the praphs ofthe beta densities having 6.28, and p = 2 and 8 = 1; (@) a =2andp [Hine To evaluate (3) and T(3), make use of the recursion formula Pa) = (@=1)-T(@~ 1) and the result of Exercise 6.9] Verify the expression given for 1; in the proof of Theorem 6.5. the limit (provided 66 with » = 0 and tm that when n > 20 able, approaches the fer to two theorems generating functions xis. fable approaches that sity) of the first ran- fable under the same ce, but the normal seven when nis fairly when né and n(1 ~6) fon to determine the anced coin. aity correction accord: te interval from k ~ ine the area under the 0.5 = 0.75 are 0.3944 and obability of “six heads ‘onding value in Table =0,0012 and that the asedtobe applied quite ‘iated with large sets of this work is done with Section 6.6 The Normal Approximation To The Binomial Distribution 219 ox210 Number of ead FIGURE 6.9: Diagram for Example 66. ‘computers, as illustrated in Example 6.5, and we have mentioned the relationship between the binomial and normal distributions primarily because of its theoretical applications. It forms the basis for many of the statistical applications discussed in Chapters 11 and 13. EXERCISES 631, Show thatthe normal distribution has (a). arelative maximum atx = us ©) inflection points atx = yo and x= n+9. 6.32, Show that the differential equation of Exercise 6:30 with b = ¢ = Oand a > 0 yields a normal distribution, 633. In the proof of Theorem 66 we twice differentiated the moment-generating function of the normal distribution with respect to to show that E(X) = j. and var(X) ~ 02, Differentiating twice more and using the formula of Exercise 4.25, find expressions for 13 and 3. 634, If X isa random variable having a normal distribution with the me: standard deviation o, find the moment-generating function of Y tis. constant, and use it to rework Exerese 63 638. Use the results of Exercise 6.25 to show that as = 0 and a4 = 3 for normal distributions, where crs and ag areas defined in Exercises 426 and 4.27 636. If X isa random variable having a normal distribution with the mean j and the standard deviation ¢, use the third part of Theorem 4.10 and Theorem 6.6 to show that the moment-generating function of an wand the X—e, where is the moment-generating function of the standard normal distribution. Note that, together with the two theorems on page 218, this proves Theorem 6.7 6.37. If X is a random variable having the standard normal distribution and Y = X* show that cov(X, ¥) = O even though X and Y are evidently not independent. 220 Chapters Special Probability Densities 6.38, Use the Maclaurin’s series expansion of the moment-generating function of the for the margin standard normal distribution to show that r (a) 1, = Owhen ris odd ** to simp! ©) m= when r is even, pata 639. If we let Kxit) In Mx_y(t), the coefficient of “ in the Maclaurin’s series of ‘x(0) is called the rth cumulant, and itis denoted by «,. Equating coefficients of like powers, show that @ 2 er completin oe After complet (©) «5 Ft 640. With reference to Exercise 6.39, show that for normal distributions « = 0? and all other cumulants ate zero. and collecting te GAL. Show that if X is a random variable having the Poisson distribution with the parameter 2 and ), > co, then the moment-generating function of x-a 8 vi that is, that of a standardized Poisson random variable, approaches the moment -nerating function of the standard normal distribution, 6.42, Show that when « — co and f remaifis constant, the moment-generating function of a standardized gamma random variable approaches the moment-generating function of the standard normal distribution. | Finally, identifyi from ~20 t0 0, for —00 0,02 > O,and-1

    0.79) (©) POSS < Z < 1.22 (a) P(-1.90 2 2 > 0.49, 664, Find: ifthe standard-normat-curve area (a) between 0 and zis 0.4726; () to the left of: 0.9868 (©) tothe right of 280.1315 (@) between —2 and is 0.8802. > 6465. If 2 is a random variable having the standard normal distribution, find the respective values: <2. 3, and z4 such that (@) PO 2) =02912, (@) Pay © Z < zs) = 09700, 6.66, IX isa random variable having a normal distribution, what re the probabilities of getting a value (a). within one standard deviation ofthe mean: (©) within wo standard deviations of the mean; (6) within three standard deviations of the mean; (4) within four standard deviations of the mean? 667. If zis defined by fo nco.nd find its values for (a) «= 005; (b) a = 0.025; © «=001; (a) «= 0.005, 6.48. (a) Use a computer program to find the probability that a random variable hhaving the normal distribution with the mean —1.786 and the standard deviation 1.0416 will assume a value between ~2.159 and 0.5670. (b) Interpolate in Table III to find this probability and compare your result with the more exact value found in part (a). 6.69. (a) Use a computer program to find the probability that a random variable having the normal distribution with mean 5.853 and the standard deviation 1.361 will assume a value greater than 8.625, ©) Inter then 6.70. Suppose person's with = that durin will be re (a) atlee () atm (©) anyw 6.71. In a phot a random 7 = 048 the prints (a) atlea (b) atime (©) anyw 6.72. Suppose t ‘6-ounce ounce. If what must 673. A random the rando probabilit 624, Checkine may be us @ n=1 ) 1 =6 © n=1 6.75. Suppcie t tion to det (a) Basec appre () Make © lao deci part ( This serve approxime 6.76. With refer yielded at 6-71, Use the ne decimals) balanced ¢ 6.78. 1£23 perce certain k that amon, than 32 wi 6.79. Ifthe prob normal ap bank will | g(y) is the value of » probability density arly leting ¢(0) feet EXERCISES a4, 1 of this result may be | . ns +0 1 fonot Fie we get 1s . If X has an exponent Section 7.2. Distribution Function Technique 237 FIGURE 7.1: Diagram for Example 7.3. If the probability density of X is given by forx >0 elsewhere and ¥ = X?, find (a) the distribution function of ¥; (b) the probability density of ¥ al distribution with the parameter 9, use the distribution function technique tofind the probability density of the random variable ¥ = In X If X has the uniform density with the parameters « = 0 and A = 1, use the distribution function technique to find the probability density of the random variable ¥ = V¥. If the joint probability density of X and ¥ is given by Arye EH) forx > Oy >0 f= 0 elsewhere and Z = VX? +¥?, find (a) the distribution function of Z: (b) the probability density of Z: If X, and Xp are independent random variables having exponential densities with the parameters & and @, use the distribution function technique to find the probability density of ¥ = Xj + X2 when fa) 6, 2 6 tb) 6 =e (Example 73 is a special case ofthis with 6 = | and 238 Chapter? —_ Functions of Random Variables ocy=ntncl yon FIGURE 7.2: Diagram for Exercise 77. 7.6. With reference to the two random variables of Exercise 7.5, show that if @ 6 = 1, the random variable Mit hhas the uniform density with a = O and f = 1 7.7. Let X} and X> be independent random variables having the uniform density with a = 0 and f = 1. Referring to Figure 7.2, find expressions for the distribution function of ¥ =X; + X2 for (a) y 50; () D Oy >0 fx.) 0 elsewhere and Z = nique. 7.3. TRANSFORMATI Let us show hy variable can b discrete case values of X an substitution, EXAMPLE 7.4 If X is the nur probability dist Solution Usir we find that the Then, using the find that the prc If we had wante distribution with getting 80) Note thati only difference i instead of the e« se 75, show that if @) = the uniform density with sions for the distribution >0 Section 7.3. Transformation Technique: One Variable 239 x ‘TRANSFORMATION TECHNIQUE: ONE VARIABLE Let us show how the probability distribution or density of a function of a random variable can be determined without first getting its distribution function. In the discrete case there is no real problem as long as the relationship between the values of X and ¥ = u(X) is one-to-one; all we have to do is make the appropriate substitution a EXAMPLE 7.4 IX is the number of heads obigined in four tosses f a balanced coin. nd the robability distribution of ¥ = 7 a SS ee 1+X Solution Using the formula for the binomial distribution with n = 4 and @ = } we find thatthe probability distribution of X is given by 1464 16 16 16 16 16 foo) Then, using the relationship y 1 34 4 1 46 16 16 16 i If we had wanted to make the substitution directly in the formula for the binomial derbi ohn =4and = we cond be shed = =i yor=(*)(2)) ore-o12.84 woai(t)-(h)Q) ee getting Note that in the preceding example the probabilities remained unchanged: the only difference is that in the result they are associated with the various values of Y instead of the corresponding values of X. That is all there is to the transformation 252 Chapter7 Functions of Random Variables. for yy > 0; A(y1) = 0 elsewhere, Observe that we have shown that the sum 749, EX has of three independent random variables having the gamma distribution with has the @ = Land 6 = 1 isa random variable having the gamma distribution with | 720. Conside a=3andp=1 . As the reader will find in Exercise 7.39, it would have been easier to obtain the result of part (b) of Example 7.14 by using the method based on Theorem 7.1 as modified on page 246, exercises © Us ete i te variable Z = (X ~ 1)? | 722, Ifthe joi TAL. IE X has a binomial distribution with m = 3 and @ distributions of x" Je find the probability @ rae | for x = @) U=(x-14 | (@) they 742. IEX has a geometric distribution with &= 4. find the formula for the probability | ©) they | distribution of the random variable Y = 4 ~ SX. 723, With rele 743. IX is the total we roll with a pair of dice, for which the probability distribution is (@) thej given on page 72, find the probability distribution of the remainder we get when () then the values of X are divided by 3. 724, Ifthe joie 74, Use the transformation-of-variable technique to prove Theorem 6.7. AS. Use the transformation technique to rework Exercise 7.1 716. If the probability density of X is given by Fe gly cn for =1, ey fe (a) the je () thea 0 elsewhere 728, 1 X.%s find the probability density of Y = X3, Also, plot the graphs ofthe probbilty densities of X and Y and indicate the respective areas under the curves that 126. Vinnta represent P(} < X <1) and P(} < ¥ <1), (a) the pr 147. Ik the probability density of X is given by ©) thepr (©) thei kt 7.27, We X; and jo = [rae TF? with the n has the bi ° elsewhere Theorem 728, Xr and) where kis an appropriate constant, find the probability density ofthe random wth the p BR " negative variable T$2x Identify the distribution of Y, and thus determine the value 729. it and otk. distibutio. 748, IC X has a uniform density with a = 0 and f = 1, show that the random variable (Hine: Con Y =—2-InX has a gamma distribution. What are its parameters? of this non own that the sum a distribution with a distribution with . len easier to obtain Jon Theorem 7.1 as and n = 2, find 's minus the number ution of the random find the probability tule for the probability obability distribution is femainder we get when aeorem 6.7, raphs of the probability ‘Sunder the curves that . ty density of the random {thus determine the value that the random variable arameters? Section 7.4 Transformation Technique: Several Variables 253 show that ¥ =X"! with a > 0 749. If X has a uniform density with a = Oand p has the Pareto distribution of Exercise 6.21 7.20. Consider the random variable X with the probability density 8 i eget aa\2 0 elsewhere (a) Use the result of Example 7.2 to find the probability density of Y (b) Find the probability density of Z = X°(=Y°), 721. Consider the random Variable X with the uniform density having a 3. (a) Use the result of Example 7.2 to find the probability density of ¥ (b) Find the probability density of Z = x*(= ¥*), +722. Ifthe joint probability distribution of X; and X2 is given by Slam) = 3 for xy = 1,2,3and x2 = 1,2,3,find (a) the probability distribution of X1Xo: (b) the probability distribution of X1/X2. 723, With reference to Exercise 7.22, find (a) the joint distribution of Yi = X1 + Xo and Yo = Xy ~ X (b) the marginal distribution of Y 1724, If the joint probability distribution of X and ¥ is given by fix. for x = 1,2 and y =1,2,3, find {@) the joint distribution of U (b)_ the marginal distribution of U. If X;. Xs, and Xs have the multinomial distribution (see Definition 5:8) with iy = 4, & = 4, and & = &, find the joint probability distribution of Y, = Xp + Xa, Yo = Xt — Xa, and Ys = Xs. 726, With reference to Example 3.12, find 4+¥ and V (a) the probability distribution of U = X + ¥: (b) the probability distribution of V = XY; (©)_ the probability distribution of W = X ~ ¥. 7a7. IX, and X> are independent random variables having binomial distributions with the respective parameters mand @ and n and 8, show that Y = Xi + Xo has the binomial distribution with the parameters mj + m2 and 9. (Hint: Use ‘Theorem 1.12.) 728, If X; and X) are independent random variables having the geometric distribution with the parameter 8, show that ¥ = X; + X> is a random variable having the negative binomial distribution with the parameters @ and k 79. If X and Y are independent random variables having the standard normal distribution, show that the random variable Z = X+Y isalsonormaly distributed. (Hint: Complete the square in the exponent.) What are the mean and the variance of this normal distribution? asa chapter? Functions of Random Variables 130, Consider two random variables X and ¥ withthe joint probability density elsewhere pxyi-y) ford 0.y > 0x49 <2 elsewhere x by using:Theorem 7.1 as modified on rheorem 72 to determine the joint probability eid then finding the marginal density of © e esatSmtinuous random variables having the Joint probabil paym, ford isthe number of successes on the second trial... and X, isthe number of and the probabi successes onthe nth trial. As we shall se later this isa fruitful way of looking atthe binomial distribution | EXERCISES TAL. Use the moment generating function téchnique to rework Exercise 7.27 should b 742, Find the moment- generating function of the negative binomial distribution by ment inasmuch making use of the fact that if k independent random variables have geometrie the nominal valu distributions withthe same parameter 6, thei sum isa random variable having might have been the negative binomial distribution with the parameters © and k- (Hint: Use the 10 and 11). result of Exercise 529.) ‘The maxed 743, If n independent random variables have the same gamma distribution with the Parameters a and f, find the moment-generating function of their sum and, if possible, identify its distribution, ‘74, If n independent random variabies X; have normal distributions with the means 1; and the standard deviations , find the moment-generating function of their sum and identify the corresponding distribution, its mean, and its variance, 7AS, Prove the following generalization of Theorem 7.3: If Xj, X32... and Xq are independent random variables and ¥ = «)X; +a2X2 +----+a,Xq,then in Section 638, EXAMPLE 7.18 What underlying formation is used data are nonnegai , Solution sis E to write down th meio = [ [Maa transformed obse remember that th ‘where Mx, () is the value of the moment-generating function of X; a. obtain 76. Use the result of Exercise 7.45 to show that, ifn independent random variables X; have normal distributions with the means 1, and the standard deviations then ¥ = aX; +4:X2+---+ 0p, has a normal distribution, What are the mean and the variance of this distribution? The required dens 7.6 THE THEORY IN APPLICATION Examples of the need for transformations in solving practical problems abound. To ‘This distribution i illustrate these applications, we give three examples. The first example illustrates using appropriate time problems. The nentially distributed, e for the nucleus to it such a process has Iso can be described ts of a particles are hhata given substance and i +1, for i = tke place is the sum af this sum is given in he gamma distribution probability is given by Stax = 1.6671 Ally certain to occur. SECS. 7.1-7.2 in dollars) and its total ‘the joint density given ¢ to find the probability ‘00 units) that is spent on ensity of the average that a salesperson spends orshe is reimbursed. Use the distribution function variable Z = X ~ ¥, the 1 ‘ons) that a service station ‘amount that the service 751. 782. 153. 754. 158. 731. 758, Section 7.6 The Theory in Application 261 station sells during that day. Ifthe joint density of X and Y is given by L A ford 0.22 > 0, and + Fan at rns tem ae eta Oe o elsewhere tse the distribution function technique to find the probability density of Y = Xi + Xo. Also find E(Y), the expected total percentage of copper and iron in the ore SECS. 7.3-7.4 ‘According to the Maxwell-Boltzmann law of theoretical physics, the probability density of V, the velocity ofa gas molecule, i kite ® forv>0 fo) = 0 elsewhe where § depends onitsmass and the absolute temperature and ks an appropriate Constant, Show that the kinetic energy £ = JmV2, where m the mass of the molecule is a random variable having a gamma distribution, With reference to Exercise 3.100, find the probability density of the distan between the point of impact and the center of the target. With reference to Exercise 3.101, find the probability density of the random answers that a student will get on the two aptitude tests. With reference to Exercise 3.102, use Theorem 7.2 to find the joint probability density of the two random variables V = SP and W = P, and then find the ‘marginal density of V. Use a computer program to generate 10 “pseudorandom” numbers having the standard normal distribution, Describe how the probability integral transformation might have been used by the writers ofthe software that you used to produce the result of Exercise 756, SEC. 7.5 AA lawyer has an unlisted number on which she receives on the average 2.1 calls every half-hour and a listed number on which she receives on the average 10.9 alls every half-hour. Ifit can be assumed that the numbers of calls that she receives ‘on these phones are independent random variables having Poisson distributions, ‘what are the probabilities that in half an hour she will receive altogether (a) Iealls, (b) at most six calls? 262 7159. 7.60. 7982. 7163. 164, 768. 1.66. 767. 768. 709. Chapter 7 Functions of Random Variables Ina newspaper ad, a car dealer lists 2001 Chrysler Sebring, a 1998 Ford Mustang, and a 1999 Buick Century. Ifthe numbers of inquiries he will get about these cars ‘may be regarded as independent random variables having Poisson distributions with the parameters 2; = 3.6, 22 = 5.8, and iy = 46, what are the probabilities that altogether he will receive (a) fewer than 10 inquiries about these cars; (b) anywhere from 15 to 20 inquiries about these cars (6) atleast 18 inquiries about these cars? With reference to Exercise 7.59, what is the probability that the car dealer will receive six inquiries about the 1998 Ford Mustang and eight inquiries about the other two cars? . If the number of complaints a dry-cleaning establishment receives per day is a random variable having the Poisson distribution with 4 = 3.3, what are the probabilities that it will receive (a) two complaints on any given day (b) five complaints altogether on any two given days; (©) atleast 12 complaints altogether on any three given days? The number of fish that a person catches per hour at Woods Canyon Lake is a random variable having the Poisson distribution with 2 = 1.6. What are the probabilities that a person fishing there will catch (a) four fish in 2 hours; (b) atleast two fish in 3 hours; (€)_ at most three fish in 4 hours? If the number of minutes it takes a service station attendant to balance a tire is a random variable having an exponential distribution with the parameter & what are the probabilities that the attendant will take (a) less than 8 minutes to balance two tres; (b) atleast 12 minutes to balance three tires? If the number of minutes that a doctor spends with a patient is a random variable having an exponential distribution with the parameter 4 = 9, what are the probabilities that it will take the doctor at least 20 minutes to treat (a) one patient, (b) two patients; (¢) three patients? If Xis the number of 7's obtained when rolling a pair of dice three times, find the probability that ¥ = X? will exceed 2 If X has the exponential distribution given by probability that x > 1 x) = 05 eS", > 0, find the SEC. 7.6 If,d, the diameter ofa circle is selected at random from the density function (-4).0<4- (ini Semiy tn lhe cr ot Siler ltd fld) = 7.10. The lo, calledi current REFERENCES The use of the p TonNson, R.A. Saddle River A generalization Which fx) 4 apply separat aLPout, RE. ed. New York More detailed a many advanes Hooc, R. V., an Macmillan Pu Rovssas, G. G. Publishing Co Winks, 8S. Mat 272 Chapter 8 Sampling Distributions | Itisofinteresttonote that the formulas we obtained for var(X)in'Theorems 8.1 are Non .7. The and 8.6 differ only by the finite population correction factor ~—".* Indeed, when varial exists Nis large compared to m, the difference between the two formulas for var() is = usualy negligible and the formula oy = -7= is often used as an approximation acd ‘when we are sampling from a large finite population. A general rule of thumb is to use this approximation when the sample does not constitute more than 5 percent of becom the popalation. | ex condit EXERCISES: = 8.1. Use the corollary of Theorem 4.15 to show that if Xy, X2,...,Xq constitute af random sample from an infinite population, then | cov(x, -¥.¥) =0 forr =1,2,...00 88 Consid 82, Use Theorem 4.14 and its corollary to show that if X31,Xi2,..., Xiao X21-% the un Xan, are independent random variables, with the frst m constituting « random sample from an infinite population with the mean 1 and the variance o? and the other m2 constituting a randem sample from an infinite population with the mean 2 and the variance ¢3, then (a) EX, — Xp) = oy - we: (b) vary — Fy = 4 Use thy mm theoren 83, With reference to Exercise 82, show that if the wo samples eome from normal 2.1 populations, then X; ~ Xz isa random variable having anormal distribution with =a the mean and the vance E+ 2, (in: roe a in he roa of sae Theorem 8.4.) i BA, If X),No,...» %y are independent random variables having identical Berno distributions with the parameter ®, then X is the proportion of successes in andif trials, which we denote by 6. Verify that (a) EO) =; ‘sa -0) (0) van) = C= aa 85, Ifthe first ny random variables of Exercise 8.2 have Bernoulli distributions with ofhed the parameter 0) and the other nz random variables have Bernoulli distributions al with the parameter d, show that, inthe notation of Exercise 84, paar: | (0) E(61~ 6) = 6) ox: 810. Use thee (©) vari; — 6p) = the seque mi 7m SLL Explain 6, Looking at binomial random variables as on page 258, that is, as sums of pare identically distributed independent Bernoulli random variables, and using the 812. Explaint central limit theorem, prove Theorem 68, ai tenas ‘Since there are many problems in which we are interested in the standard deviation rather than the nee ‘avane th em “nit population orecton ator” oe fest = insted of =". Tis (@) then oes not matter, ofcourse, as long asthe usage is clearly understood. () the v (Bin Theorems 8.1 =) Indeed, when gnulas for var(X) is ‘san approximation {rule of thumb isto ore than 5 percent of, Xq constitute a ges Xin X20 first m constituting a tye and the variance oF infinite population with spe come fom normal eva aisbuon wih ‘occed as in the proof of aving identical Bernoulli portion of successes in mnoulli distributions with ine Bernoulli distributions ercise 84, 258, that is, as sums of py variables, and using the fad deviation rather than the N=? insead of Vv=r N= This Section 8.3 The Distribution of the Mean: Finite Populations 273 87. The following is a sufficient condition for the contra limit theorem: Ifthe random variables X1, X2,...,X» are independent and uniformly bounded (that is, there exists a positive constant k such that the probability is zero that any one of the random variables X; will take on a value greater than k or less than —k), then if the variance of y= Xit Xt Xe becomes infinite when n co, the distribution of the standardized mean of the X; approaches the standard normal distribution. Show that this sufficient condition holds for a sequence of independent random variables X; having the respective probability distributions 1 © tory fo =|? 1 torn = 88. Consider the sequence of independent random variables Xy, X2, Xs,... having the uniform densities 1 far oo. Use this condition to show that the central limit theorem holds for the sequence of random variables of Exercise 8.7. 8.10, Use the condition of Exercise 89 to show that the central limit theorem holds for the sequence of random variables of Exercise 88, ALL. Explain why, when we sample with replacement from a finite population, the results of Theorem 8.1 apply rather than those of Theorem 8.6. 8.12. Explain the results of Exercise 5.28 in the light of Theorem 86. 8.13, Ifa random sample of size n is selected from the finite population that consists of the integers | show that +1 N (a) the mean of X is “+ vig (N+IKN =n) (by the variance of Kis “+00 284 Chapter Sampling Distributions fe) 8? =2(x;- XP = 21 836. Show th (the joint density of ¥ and Ss given by definite ne.ne= Le? Leyte the eis oe ye 8.37. Verify t for s? > O and 00 < ¥ < 00, so X and S? are independent. ae 8.29, (Proof of the independence of X and $*) If X1, X2. X, constitute a random 1 degre: sample from a normal population with the mean and the variance 02 838. Verity t (@) find the conditional density of Xy given Xz = x2, X3 = 44... Xe = tas and an F dis then set X =n — X>—---— Xy and use the transformation technique to 839, 1FX has find the conditional density of X given X3 = 19,X3 = +++ Xp = ia (@) find the joint density of X. X3,Xa,-...Xy by multiplying the conditional ae density of X obtained in part (a) by the joint density of X2, X, Xn, and, . show that serosa . 841. Verity th for 00 is the s are limiting this di there is zero probe To be more between the value: (Hint: Make the substitution 1 += = 7 8.33. Use Stirling's formula of Exercise 1.6 to Show that when » —> oo, the distribution approaches, the standard normal distribution 8.34, By what name did we refer to the 1 distribution with v = 1 degree of freedom? 8.35. Use the transformation technique based on Theorem 7.2 to rework the proof of Theorem 8.14, (Hint: Let f= * and w =.) X, constitute a random =, S xq, and formation technique to tiplying the conditional yot Xz, X3,....Xq.and given 4.8? are independent; it tution with n — 1 degrees ceferences at the end of 2.to rework the proof of a7 ith v degrees of freedom edom oo, the distribution 1 degree of freedom? to rework the proof of Section 8.7 Order Statistics 285 8.36. Show that for vp > 2 the mean of the F distribution is ‘making use of the definition of F in Theorem 8.14 and the fact that for a random variable V having 1 837, Verify that if X has an F distribution with vy and v2 degrees of freedom and v2 + 00, the distribution of Y = v.X approaches the chi-square distribution with vy degrees of freedom, 8.38, Verify that if T has a distribution with v deg an F distribution with vy = 1 and v 1 of freedom, then X = 7? has v degrees of fret Use the result of Exercise 8.39 to show that 840, Sian BAL. Verify that if ¥ has a beta distribution with ny wa-%) has an F distribution with vy and vp degrees of freedom. 842, Show that the F distribution with 4 and 4 degrees of freedom is given by oft ft for f>o a= 10 elsewhere and use this density to find the probability that for independent random samples of size n = 5 from normal populations with the same varian will ake on a value less than } or greater than 2 ORDER STATISTICS. Consider a random sample of size n from an infinite population with a continuous density, and suppose that we arrange the values of X1, X2,..., and X, according to size. If we look upon the smallest of the x's as a value of the random variable ¥,, the next largest as a value of the random variable Y, the next largest after that as a value of the random variable Y, ..., and the largest as a value of the random variable ¥», we refer to these ¥"s as order statisties. In particular, ¥ isthe first order statistic, ¥p is the second order statistic, Ys is the third order statisti, and so on. (We are limiting this discussion to infinite populations with continuous densities so that there is zero probability that any two of the x’s will be alike.) ‘To be more explicit, consider the case where n = 2 and the relationship between the values of the X's and the Y's is mex andy. x when < men and 922m when x2

    You might also like