Scribd
Upload
37 views114 pages

Statistical Inference Workbook

book of inferential statistics

Uploaded by

enosrusike24
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
37 views114 pages

Statistical Inference Workbook

book of inferential statistics

Uploaded by

enosrusike24
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
WOOSMHOM LNSGNLS ONSYSANI Contents s of Statistics | 1 Worksheet 1 Bxereises e What This Unit is All About hse (d} a census and a survey i statistics [e} Statistic (0) parametric statistics and non-parametric statisties (2) a parameter and a statistic le shows the number of fouls committed (X) in the World frequencies (f) 2. The follow cup of 1998 (France 98) and Use the data to answer 4 (a) Find the mean (b) What is a mode. In this mode of X (c) Plot X against its frequenc (a) Find the median of X and is range (e) Calculate the standard deviation of X (£) Is the distribution of A. symmer 3, Distinguish between the four scales of measurement 4. Give some of the advantages of parametric statistics and ¢ appropriate to use them as opposed to 7 b._._.__2 1.2 Worksheet 1 Solutions: a 1. Sow Mout swan Rads x14 x45 0 446%2 64344042 BL 20 20a (b) Mode is the number occurring the most, that is, with the highest In this case the move is 4 fouls committed. fe) Attempt it (4) The median iy the contral value when X is arranged in asce ling order or descending order. In this case the median of X 4 fouls committed Range of X is given by Xmax Xmin =9= 25 (6) First find the variance of X given by 1 yale 24,-n2°] Var(X) 5 [ete - 4] = 5 [Eh \ a Fx bo 4G x 220 x 405] = 1.3132 % ins the standard deviation « (i X is given by s = VT31S2 = 1.1459 From the plot in part (c), itis evid vat that the distribution of is not symmetric Note: You 3. Distinguishing should be wad ordinal scale terval scale rat si scale e module for expla ation of these scal Advantages of parametric Worksheet sh. tet (90,22, 92,20, 24,40, 34,.36.92 99,28. 30) be a reatnation from cera plato. Using this lati ay fl te na (op fal the tansdand devo ted prdser a ateun nol hal layne fen vin Bt 1) precept ur ts at rata Skewes ob first a seve co fica nat nu ek Tw A “st Br Gt 57 49 28 42 8 AS )Canetonet a stenr an Bea ign for this at ve ata an fad gr Gil Ho adie tis lata a (ey Whit pueavrtiay of the Ssappoese X-has a peduity sist ite lensity firey rc ood (05) Find the evict af ken fe} Charenton te stain of Additional Exerc 2. Lot V be a random aia le with ft oxra2 FV a sterwise (a) Sketch fr (b) Calculate the coefficient of skewues (6) Calculate the coefficient of knrtosie (4) Comment on part (a) nnd pint (9) ‘ Wosedt 1 Wen Limit theorems for non-para het leks specify th Tho hype eae y ] i: i ACA, Worksheet 1 Second quartile Qo=median position = p(n) a Fost 9) + 03)32 Third quartile: Q, Q = Hr Or IK Tati Q=%50, 0 Mi m0. Pearson's second coicient 1 1 +1= 40a =65 hans n= 9 % so] = 30+ 1= 31 1 sa] = 38+ 0 H, Q=33 Maz = 40 SOO Sa 50808 10 89 Unit 2 Estimation: Point Estimation of Population (Model) Parameters 2.1 Worksheet 2: Exercises What This Unit is All About This Unit mvolves exercises and solutions hased on Unit 2 of Module STD108. 1. Given the deusity fanetion ) Devermine what fan n do estimate 8 if you are ving the method of parame me it Larned out that 2. Given ) Bind the maximums tikel tment mr at m5. You should use te Gif you are using the method! of moments for estan 3. Given the probability mass fusetion given b {aio 2 1b) Whiclt method sul you wse to comme up with the estimator? Wocksheet 2 (b) Use Metiad of Moments(MM) to find the estimator 8 a le realisation of X as {c} Suppose you are given a possibl : (05, 1,12,07,15, 02, 11, 075,09, 12} Fina two couimators of @ using the two methods above, fe to determine the effect of stirring rate on the amouny, 8. A study was mad = The study yielded the ‘of ampurity mr paunt produced by a chemical process following dats ‘o(X) [117 116 180 S571 oe Pps 200) (a) Plot tte data above and comment. (b) Giveg that the model Yat AX = LQ um (6) Show that, using Least Squares method fy = ¥ ~ AX (4) Fit the mode! 9. Let X,,Xa,... be a random sample from the probability mass funetion fivea by ay={ HI-PF, 2 = 0,12 y= {Fo eats Find an estimator of p wsing (0) Methods of Momsents(MM), (b) Maximum Likelihood & imator(MLE) (6) Given a realisation of (24,5,8,2,2,3} find an estimate of p: 10. Let Xi, q bea random sample fiom a poisson density f say { 5 T=O12, otberwise {a} Name two estimation proceds Zl | likelihood estintation. Procedures that vow know besides maxims! 16. Cie | ‘ihood Estimator of ¢ et () Suppose you are given a realist fd Find the maxinn Hlethood eaten ty NOME 7.65079) al | () Shove that the estinator a | | (¢) ove is unbiased and consistent cad | nd estimator of @ Wb YX, ss sap ‘ Praranioter A. Fiat i LE Given Gi Walistion SEN {2.935.323 2423.33} Caine the wsuh above fi 19) state ot & Workhest 2 used in parameter estimation y, 1 Method can be 18. Explain bw Jucgement ina gave aw example 1 | sample from 2 normal distribatio y with be a toncdon parameters #1 andl o-USing Meth, foe Estimate the P Methioc! of Moments 20, Fan pont estumatons for yaad ‘he mictho of masini __ Worksheet 2 2.2 Worksheet 2: Solutions I \ x | LSI ae ls ies Worksheet 2 ——— 0 sven de = x 100.9) {oe hee amnesia noestinat N02 gg (2) Using tae mea lg are Lew an) = Per tert ee Inbar tm a = RMA AD ; p fn blar ta.A) =O tvs differentiating and equating jy In L Dee eter , tee ts ! = : Soa, = nF doe | 1 / Thus the maximum likelihood estimator of Ais 4 = — (6) Formethod of moments, equate E(X" | Since te) ~ exp) with UX) = we estimator i given by P= ENO Baye 100d fun Sees 6) egies ta Var it {x ] 1 Sala! Worksheet 2 ue for detais of calculations (a) Refer you to regression mode f Suis 3 AS 085 THO Ta) x] 2 e108 137 2 a 7 93120 13 M4 245 26.0 38G y[0% 24 3247 | Draw a plot and your comment show'd be, the plot shows ay wpmeard wend 951612 73863799 Thus ¥; = 227386%, si Brom the plot, the line almost passes through the origin. Yes, eters could bee est model a? = mean square e1'or | voici . O12 aaeesso 0778 6. Given tan ={ (Uw 20.23, | b otherwise ) Ung hood estimate mod. the blood fo thon and its natural log are Le ) Tren Gs oie pret In Lan, -.tmp) *C)] Same (0 se) AMecetaing and eowaing Stary yy) O gives 20 Worksheet 2 int 1 , az eer z= ap (ony the eax lkeleoc esti Now as n > 20, lim 852) 0 int estimator of “Ths p isa consist eaton (2.3, 344.2 5 3.6518) 8 POSSI ting (c) Using this realist ms #4 _om 1. Given that X1-Xay---,X» ita zandon sample from Uniform city with the density function: plana b, 090, 067<8 ‘with 6 unknown, (a) ‘The estimator of the parameter @ using judgemental method j smmaz21,--. a) (b) Using Method of Moments(MM), * ea [je - el 20\, = & i 128 3 2 : 8 ee Thus b= 28 (©) Suppose you are siven a possible realisation o {05, 1, 12,07, Th ‘ on 15,02, 11,075.09, 19) Sure eA) impurity%(9) Worksheet 2 7 1h) Giver that the nuded 5 Least Squares sath * Takug, partial senivataves wr " Pi , y ow Soa = 3 OM y rm 1180 190 106 s91) — 4,100) his US 9 35738 a1 since b- A. 106 ~ 0.35758 < 19 3 80 Thus the ite miodel is j, = 88056 + 0.2976: ttt worksheet 2_ : com the probability mas § sample f tye a random S20P po. tat XoXo ° a exe 02 awe fo oe (ama a Using, Meo asta) Lp B10) (Using Maximum Likebood Pst (MLE). ue.) = [[pu- 7 Fina Ziman = ome nth =p) = np + l-peap 2 b= ptap > t=pies) be re (6) Given a rea isation of {[Link].2.2, 3} Au estim nate of p = ji 10. Given Sen) “ rege : otherwise, 1 e Worksheet 2 (a) Methods of moments(MM) and the Least squates method LSM) lihood estimation method, the hkehihood fane [he ote “ wlft3] eas Since E(G) = # this mean that 4 is an unbiased estimate o veniy = ver(23x) ell Worksheet 2 0, therefore 2 i mor off Ay m+ 0. ination of X as (1046.77 11. Given that P w-p oe aul it ») = Tmo ’ Be eee ee coe oe h \ Ths Sh ypc ney = 844 = 10158902) = 1673 ma 0(9.5)* = 82.5 Sun = Sod net = 385 - 10055)" = 8 ° Sey _ 1673 Se 2.030303 = A= S= 5 fp = gus ~ 123- 2.0309(5.5) = 1.13905 Thus j= L134 + 2.09032 vat Try fon p>0, r>0, A>D0 Recall that for a eeal number r > 0, the gamma function of r is given by voy [eens ot fares = Meat and the properties of the gamma function : rays Tiles v T1601, ray postive el 1 pdr teen ‘Het = Sow tay {9) —Kewied to show that &(3) <¢ - Fro [tee [re Unit 3 Desirable Propertie Estimators of Point 3.1 Worksheet 3: Exercises What This Unit is All About Tus Ui sercises and solutions hased on Unit Sof Module STD 1 bet (XS 1 of w indepenclent random variants wih mean ps and ans Lat Ynx Ses LS ) Find the mean aud variance of X i innit Theotemn(CLT 1) Test 1 exannnstion mau th stawlard deviation of a = 6. Use a appropriate appr havea : 2. Let [Link],_,Xy he identical and independeutly distributed il) rand les with ean ys an a L Worksheet 3 2 at 2 isa consistent estimator 2 {b) Var(X) = Hence deduce thé oy «from & dstributio lon sample from with : Ca Kos Me be a a : Te al anee 3 Shoe ta at (ky 2X) FIX + 8%a) i an unbiased and cong cate fs more efficient? fc) Which ofthe above estitnators 4. (a) Define the fllowing eras us b) Let Xs... Ny be & random sample from a population with va Ls 2 82 Sia 8 o Otineeatoceerne : ( Find the maximum likebiheod estimator of @ and show that i sw é ase, consistent and sufficient é Sal Mien 1% be a random sample from the Bernoulli dsb Sie Raameterp- Find the maximum lieinood eats of? ‘How tha 5s an unbiased and consistent estimator of G Let x, Xo... Xs bea rand he it iigg'*%* B84 random sample fem the distribution on the 7 He feom the distrib Ae et a BAF 9 ond b= Xqy = 2 10. Let Se me HOt B where Key w mas( Xo) ; (4) Show that both estimators are unbiased estimators of 8 | (©) Whi of the estimators is more efcient? EI O cae, : tent estiniator of @ 2 - : - Worksheet 3 an 7. bet [Link]..,Xq bo random sample foe 4 populetion with mean a sel valance Sen dehned hy 2S, 1) Hence dee that = SO" LX, ~ X14 an unbiased extnaton ote an, fr Datioe an Ly k=1y x, aa 8 x,-Xy ote 8. Given the tel Veet en where 05 = MQ, (4) Fina the feast snare estimate (9) Show that the estinnator of unfinsed (4) Find Varta) ea {d) Find the distribution of (6) Give the flowing data tion [a3 [ow [oe [ar [2 [a5 | 9907 05 | a85_ Too [10 ence (elo }aata2 [artes feota. Mertas e030 Fit the mel above eval Additional Exercises TO. Let Ny.X3.... Ny be identical and independent'y distributed{i i deandom ariables with nyean j2) aud variance a? Let x ad S pat = Worksheet 3 4) Show that X isan biased estimator of 2) and coumient © Hence deduce that X is a consistent (by Find 6 fe) Prove that Var(X) ator of <4) Show that $? 36a consistent estimator of 27 11. Explain the importance of the Invariance Property, 1 12, How do you use the factorisation theorem to show that & static siufcient estimator of a parameter 15. Let Ny... .Xq bea random sample from < normal distribution with m by and vance «) Explain what you understand by the phrase ‘a sufficient statist (b) Prove that X = 130%, X, isa sulicient statistic for 1, using 4 the definition ii. the factorisation theorem (©) Show that X is a consistent estimator of Worksheot y Phevvlone it FW 3 Solutions : Worksheet 3__— w tk i] =e io[St0-0- am py? = n(X | = ny = BX | « ats Eves wort] Thus £ Recall that 315° fallows a chisquare dstition with m4 and variance 2(n ~ 1). and note that St Worksheet 3 3 0asn 40 For consistency lim Var(X) se! y} m> yl i Pid > yx 2 TSP Ss ~ We Pu ey (0,1) as # > 00. sariable and V be a chi-square ves of freedom. If Z and V are indepen then ty ~t int) +> ulPLXy > yl by inne nd Ya is given by so the joint density of Yi Pte cangs. eta leuing # = 9-H fe) of ayy = 6r(1 7) Thus firy= Grr}, Genet N.B Solution ean be obtained via Jacobian technique i our ha (0) = FRE — IPE ~ Bet? 2) where FEA = So It has a Beta distribution ie Beta(2,2) B(R) = of rr(1 = rd =o f (7 = Nae = oft tal iE =! 2 Ue) = Using th "the cumulative distribution f™ By insy Letting Letting & Which is ar Worksheet 4 — sis also normal formation of a normal f tinear trans ‘Thus wo N(0,1 £~ Mi @ L ‘Thus ( ~ ¥) isan un biased estimator of (ay — yx wre z whee U~ NOL, y 0 yy and T is tdisteib (6) Assume of ce 7 Vel) —verty ty udepentmee ee Unit 5 Interval Intervals Estimation: Confidence jet 5: Exercises 5.1 Workshe What This Unit is All About “This Unit involves exercises and solutions based on Unit § of Module STD103 1. Suppose that the realisation (24, ~L6, 4 45, ~ 15, 2 0.8 33: 47) fea random sample drawn ffom a normal distribution with mean 4 and variance 0%, Construct a 90% confidence interval for if (a) @= 121 (6) o is unknown 2 Atandon sample of 10 voters wete elected and 8 were found to supp ‘an annexation suit. Find m (a) A 99% confidence interval for the tru A. coats tl rte ee proportion of the vung (the marin ange thesane ite with OS a eve ta te ppt A food inspector examining oi], measured perce id bottles had the following observations: Qe (23.15, 20.22, 44.20.19, 31 12,20. 19) Pind a 90% confidenc i Find 90% confidence interval forthe variance of th fo" nce of the percentage of 4. Two independent randon from each of the norinal popula ek" eed Perlaon The dace gas tis in the following t= a. ed a (a) Calculate 53, the pooled estimator of 92 (b) Find 9 98% confidence interval for yy 8 of wool of standard fem before anil after washing. “The length of each aa and after (Y) washing are given in the following tab Ta 1205 2115 16 2008 aE ¥|20 075 28 15 19 25 16 31 28 Fin a 98% confidence interval for the difference in Uke length before and after stretching? Assume Ue data to he normally disteduted 6. Suppose that two independent random samples from the normal popula. tions N(u4, 2) and (j0,,03) gave the following statistics: ‘aniple size sample variance ‘Sample 2 a Saupe 2 9 Fina 99% confidence interval for the ratin 3} 7. The following, table yives th leuts before and after attendin 1 course in Statistic ([Studewe [1 ame oreo one Before 6/65 W037 15 80 WD 70 85 75 65 Altera | 66 44 60 40 75 5) 75 8 0 (a) Is this a matched pair or not? Explain Me ~ by (6) Using the confidence int tw effect ou the scares obta * Supine a mauufactre of cap ws wants fo cpa te pve we i ate a St ectores, the compat Tandonly choses 1 900 houseliolds the northeast (NE) an | 900 Holds in the southeast aud deteranine wether eack Houseiet! Ht “super within 5 years, The results obtaiued are sam eastern Uui ” Wkshet 5 semple size 5 Sample T (NEY 1 000 = aise) | 1000 «ay Constsuct a 95% confidence interval for the difference inthe twp portions p ~ Pe (b) Using the confidence interval, test the hypothesis that the propotiay are equal 9. The variance of scores in a standardised mathematics test for high sho seunes was 130m year 200, 4 sample for 20 high school semors who took that test this year gave a variance of 170 (a) Construct a 95% confidence interval of 0? (b) Explain how this confidence interval can be used to test the assertion that Ho: a? = 150 versus Ms 0? #150 Hence is this assertion true? 10. Two independent samples of measurements of diameters of ball beatae ‘sing two different calipers were taken resulting in the following messi 26 07 OW 035] 0.40 035 030 032 035) [eh aasgaregnsog oar Caliper 2] 027 029 31 (@) Calculate sb (b) Construct a 90% cont the te clipe Confidence interval for the diference in the neat : M4 1 Let Xi at in be roudon sa ran and variance 160, with wles from a woral distribution x, = Suppose we ae given that (0) Find a 95% (b) Find a 95% ees terval for ge of she form (0,%!) lowing results Seles from uoemal populations produced ‘onfidence in interval for x ‘confidence in - (Ss Samer sof 200: , for {a) Explain what you understand by the pitase independ estimate of the (c) Fin a 95% confidence interval for (b) Calculate the pool ‘Mt samples (4) Using the results above test the hypothesis that My a= agaanst an 13, The monthly net meote (24) and munthly expenes (y) of@ random sample fof 10 families fron at low income bracket sere recorded as shown ia the following, table faye [1]2] 3] 4] 510] ToT) Tncome (x4) 1 840°] 190) 210 | 290 | 320) 4207 300 | 305 [320 Lp ppewses (y.) | 220 | 180 | 205 | 265 | 29% | 395 | 380 | 285 | 215 | 195 | Tu dieneuce dy =e, ~ thas epaesents the fain’ monthly income, ste Avouuae X, 2 Nye) and Ye My shat), where d > O ay mai a) Explain how the confidence interval terknique can be used in the test ig of hypothesis (b) Pind a 95' (6) Find also an apprapriate 95% (Wu) ‘confidence suterval for d ronfidence anterval for dof the form Additional Exercises from a normal distribution with wn M4, Let Ny. Neg he vanedont sap ation with Fanon maar rand variance 40. Suppose we ate given that Sia) Xe = 700 (8) Find « 95% confidence interval foe 4 real for jf the form (0.4) (0) Phu « 95% confidence ite xf 400 tosses of a coi 15. Suppose thar 175 heads and 225 tails resulted from 400 « probability of a head (2) Find 90% confidence interval for the proba (6) Does the coin appeat to be a fair one” station with wuknown mean 14 and 1 6. A sample of readings from a norisal poP' mi ‘unknown variance 0? gave the follow! aadings taken from the sam* population gave 4 second snl a 19072, 522536 mples to give estimates of ye and o° (a) Combine the two san al for je (b) Construct an appropriate 90% confidence intery yn Harate last year, was discover 17, Ia sample of 400 shops taken fro vetted pri 136 of them were selling their goods above the 8: {a) estimate the percentage of all shops selling above the gazetted prices, (b) Calculate a 959 confidence limits for this estimate and explain briely what i means (c) What sample size would have to be taken in order to be 95% sure to estimate the percentage to be within 4:2% 18. Let Xs, ...Xeo be random samples from a normal distribution with wa ‘known mean y and variance 64, Suppose we are given that 0%? Xy = 382 (a) Find a 95% confidence interval for y (b) Find a 95% confidence interval for y of the form (0,2) Workshest 5 Worksheet 5: Solutions Ss [ Sie s = Vai = 2172065 (a) Ifo? is known, «90% confidence interval for is given by Fie; % = 108+ 2067 121 198 4 1.65421 vio 1.98 4 0.631349 MMB? , 26113 (b) If 0? is unknown, « 90% confidence interval fo- is given by Simumary Statisties too, p= 0.83. ¢= 047 | (a) A 90% confidence interval for the true proper ton P | 053 40.12 (0.4014 . 0.6986) d Worksheet 5 in order to be 95% confident that yum sample site baer eve value of the proportion is (8) the a the samp riven by = VE = VORB [A 90% confidence interval for the variance is given by ane feone) (saan 1 MeO dye) ~ sr 4, Summary statistics Ho = 75 = 8 f= wa (6) The pooled estimator of ois given by PE A08D 3909 Worksheet § ae A. 95% confidence interval for the ratio Zf is given by: ( ESTEE Reels 5) (Feat Frenstiiii| (sa oan) (0.5420 , 3.0833) NB: Fyans(24.20) = peach 1, = 94292 T. Given the following data: Suudent [12 3 48 6 7 8 9 2 Belore 6 [65 40 63 75 G2 35 80 50 70 5 Atera [66 44 60 76 62 a0 75 50 75 70 d=a-b]1 4 3 105 5 0 5 ea (a) This is @ matched pair sample because we are assessing the effect of the course on the same students before aid after attending the course. This gives us a par of seores for each student, that 15, par (a,6) for the 12 students, (b) A 95% confidence interval for the diference in the two means ja ts given by Summary statis: = alle (E5] 1 g?] “sr = oa Tine 7 [me Lao] y 00605 = 3 1718 | 9586 confidence interval for a= te uy is given by de tyin— 158 1-6 13993 4 gant} 1.3099 4 2.993171 viz 13883 201000 (206811, 3.3477) . Ho ma =0 ee an Hewe>o 7m) Brom te aie we gc ie be 0.0062 = 08 i Ps given by uote ¢ o.utcamespei6 (asym =O or tad : (Sx) J-" Vi =v Y= 130 4.95% confidence interval for the arian’ : somo) 19070) yet ee) = ar) sgn (yao feta ‘Worksheet 5 (b) Ifo? . wore Since 150 € (98.1763 , 362:5140) we fail to rect Ho, Th an _ 150 belongs to the confidence inter 1, we fail to reject Hy 10, Summary statistics: i = 02863 sf = 00012 z = 00018 a a (m= Ast + (na — 195 (0.0012) + 710.0018) _ 9 gox5 ia (b) A 90% confidence interval for 44 ein stgin em aah Tr ] i = (0.2868 — 0.825) + togs(L4ivO.0015y/ + + = (02 25) tous MiVOODIBY 5 + 0.0387 + 0.0311 (0.0728 , 0.0046) 11, Summary data: 1 = 5(200) = 10 7 = 160 known (a) A 95% confidence interval for jis given by (60 10 196 / 965 10 1.968 (4.450, 15.544) —— a Worksheet 5 a tm Dstt (ra = DSH ) +92 8129) is 28a Thus = 28404 (c) A. 95% confidence witersal for see ~ 0, = 006 4 1586 16496 15 8) @) (e) Using the results above, test the hypothesis that nd a 95% confidence interval for jo ~ 51 ee \ Ay mtn Mo ia-m FO Now 0 € {16436 , 1.526) we fail vo reject Ho. Therefore we conclude | that pi = ss at 5% level of significance. 13, Given the data Faun i Tats pert rsy Taeome (a3) 2807 190°, 210 NT! 320 FAT 390 305 Expenses (y) 220 | 180 | 205 330 | 285, (Savings d= ey | 20, 10 ~ MUd,204) given a? 200, wea, 2 a From the data (0) 6 Hsing the confidence inter the 10001 ~ 0)% ann some the interval contains the jy te™¥Al oF the ORME, soy Me hypothesised wate Paameter (eve Me then we fait gg Haak Al 0 Lexy {h) A 95% contidlence interval fo, iS en hy tet, a4 1 = 6% 1o6yx = 4648765385 = (68U6 , 23.3654) («) A 95% confidence interval for d of te foom (0,2) Given P(-Z,, < Z - Bn.) = 0.95 1S given by Lower Limit = (-2,)v% = 0 ss (53) d+ 3.000 “ (54) Upper Limit Using Equation 5.3 we get M46 -(-2,V%~0 2 M6+2,vil=0 2 = AS. as 7 v20 Now P(-3.536 < 2 < Zqq) = 0.95 eer SS &(Z,,) = 0.95 ~ 0005 = 0.9005 = 2, = 1.05 5.4 we get 146+ 1.05v20 21.9790 elo of the for" ©: €95% confidence interval for # Unit 6 Hypothesis Testing 6.1 Worksheet 6: Exercises What This Unit is All About 1 Unit 6 ef Module STDIO3 Tris Unit involves exercises and solutions based on sunple obtained frem a bernoulli iste 1. Suppose Xi... Xiw is a random s bution with parameter & where 0 oct) Its desired to test Hy:@=02 versus Hy 0 = 06 using the decision 7: reject Me if ¥ = Dy Xe > 3 (a) Find «and 8 for this test (b) Find the power for this test 2. Given the density non {8 05% ° 6 otherwise IF you ave testing the hypothesis He @ = 3 versus Hy B= 2by ncans of a single observed value of X {a) What would the sizeof Type 1 and Type It eras be if you choose the Inverval z-< Las the eritical region” {b) What would the size of Type [and Type I eters be if yu choose inverval 1 < 2 < 2 as the critical region? 41. A manufacturer of cetain brand of automotive batteries claims that the ama life of these Bateres is 43 months A cousemer protection age? aa don sample of 36 such batteries and fou thatthe ntean ie fo ‘ihe sample 6 4275 months witl a standard deviation of & months (a) Test at 59% signiicance level if the mea life of these batteries is 1. more than 45 months ee Worksheet ii less than 45 months : iil is not equal to 5 months bp) Construct a 9 nflence interval of BF A chemical roces has produce, ve per day. ‘The daily yields i tom —. : Onmes for the five days are ° (a) Test the assertion that th he population variance is 800 level of nj Tost whether ye ~ 800, using 5, Two independent random samples of sizes ny <4 and » \ from each of the normal populations. The data isn the follomins eth Sample [1738-5 Sample [M4779 6 (a) Calculate $2, the pooled estimator of 0 (b) Test at 5% signafican wliether the two populations are sane 6. A producer of frozen nvestigated by the Bureau of Fair sad Each package of fish thal this producer matkets earties the claim tha Af fish, A conmplaint has b uo. The bueat acquires 100 packages of fist iarketedl by this ex Yoox, = 1150 test at oy = 0.01 Hy 2 12 versus Hy oy < 12 on ; Hew < 12 versus My > 22 a : Any <5 were selected om 0s my = 6 and Be eependent raucous samples of sizes = BARI Mach of the normal populations. The data i so 30 33 Soper [zT aa te 30? eee day Calculate 5}, the pooled estimator ol o” a 1% significance level that 12 > Woishet 6 Additional Exercises 8. Suppose X isa random variable from a poisson distribution with parameter u. This desired to test: My p= 2 versus Hye using the decision T* reject Hy if X > 2 (a) Define Type | error(a) and Type Hl estor((3) for she test (b) Find @ and 2 for this test. 9. To test whether a coin is fair, the following decision rule is adopted. Toss a coin 120 times. If the number of heads is between 50 and 70 inclusive, accept the hypothesis that the coin is fair, otherwise reject it (a) Find the probability of rejecting the hypothesis when itis correct. (b) How should the decision be modified if P(Type I error) < 0.01? (c) With the original decision rule, find P(Type Il error) if the coin is biased and the probability that a head is obtained is in-fact 0.6. 10. Suppose X;,...Xy is a random sample obtained from an exponential dis Uuibution with parameter 0 It is desired to test. Hy: versus Hy using the decision 7" reject Hg if X > 1 (a) define type L error(a) and type IL error() for the test (b) Pind a and for this test (©) Find the power for this test, 11, Two independent randon samples of sizes my from each of the normal populations. The data is ‘ in the following table (a) Calculate $3, the pooled estimator of o (b) Test at 5% significance level whether the two populations are sam - Worksheet 6 Bern(0) > PIX = 2) = 0511 gyre 4 on Deion Rule 7 reject Hof ¥ = 3°, x 6 a= PUT ype | error = Plreect Halts tue) P (Sox, >39-02) ¥ ~sinti.02) 1-PY <3) 1 0879 = 0.121 Bin(10.06) 1.055 ) = 0.058 Thus B= P(Type erro) = iven by 1-9 = 094 b) The power for th A iven the density , for 0 a= 4 otterne fe=1o (8) a= p(Type fener ca pireiect Half ee =9) : nh} 1 5 a= file je i Workshest © esas te sim P(Type Ul error) 6 = Placcept Holdy is false) P(X > 110 =2) Me f3e-F Thus J = P(Type Lerror) = § Lertor) a = Pleeject HolMy is true) = Plc X<2\0=3) "pte? 1_t -[-G-i ig P(Type I error) = + 0 = P(Type It error) B = Placcep Halts se) = PX S12) = fla [2f =) fb 2=[5]-5 ‘Thus 6 = P(Type H error In this case the sizes are unchanged, 8. Summary statistics: n= 36, #= 4575, 5 4 (4) {Since ois unknown and n is Asean, ee a standard nor eee Ma = 48 again Hs > 45 (igh sided) om: Reject Hyif Z > a that is Z € [1.6449, 20) § Worksheet 6 a : 4. Summary statistics fl [ t= 1S aam6 2 1 co ILS 1 . ‘ zh a 1) 0 s = Ve = Ji = 8.9443 | (3) Hypotheses: yo? = 800 against Ho? £ 300 (obese) hal Fegon: Rejet fy ence ede + x8 > xdel) = 111 06 x3 < xds(4) = 0484 pe one 4160) X30) gg Dassion: Since x2 < xByn(4) = 0.484 we eect Ho, Conclusion, Sine rected we once tat ther i sufcet canal to 810 | (b) Since of in unknown and n ic small we make ue of a - tant) = 2.78 | ton Suni ae | 800 Vio7s Jay Decision. Since T 1 is not an element of (-20, -2 78] U "ay DB) we ta to jet My Conclusion: Stace Hy i= not rejected we conclude that there is ullicient statistical evidence ta conclude that she population mean 1s equal to 800, Gynmary statist \ a | Hypotl Hi, eee : Test, Statist Din: See 7 = 18 Conclusion: Sine Worksheet 6 Is 916) Is ooes7i43 wish Hy double sed! i 214 oF 7 > laus(T) = 24 2asgju not a eleweut of (-26. 2.14] Hy jected we conclude that there i & mnake use ofthe standard normal Ths of = 04203, \ Worksheet 6 (b) Since 6} and 03 are unknown and Perret "1 and ny are small we make use of Hypotheses: hy sa, Crical Rew He against Hy p> py (one sided) on Reyeet Hy a T > tal + my ~ 2) ‘Tost Statistic oo S58 208 08967 aps3c0%04 VOR Decision: Since T= 2.2681 < 2.82 we fail to reject Ho Gonclusion: Since flys wot rejected we conclude that there ws Ho smficient statistical evidence to coucinde thal sta > #4 Unit 7 Sample Past Exam Papers 7.1 Past Exam Papers What This Unit is All About ‘This Unit sample past exam papers and solutions, Past exam papers included © Exam paper June 2002 «= Solutions June 2002 ‘+ Exam paper May 2003. # Solutions May 2003, 85 Aa. ZIMBABWE OPE BSe Mathematics SITY UNIV Patt STATISTICAL INFERENCE 1 a shows vid attempt ALL questions in section A and 3 questions in section B Marks | pare! ; — | — —____— ——— a | j SECTION A (40 or | ' them 10 AB. ' ws may attempt ALL questions b eareful to unmet AL | | {ay el of th following ® r lest Ai. (o) Explain briefly wh rr | rl {i} power of au “| {i efficiene | iv) eritical value "| | , | naan sions that follow yo lata anne sins | (h) Consider she Fellow | 4 a mo 4 ua o6 a7 39 ot sea ane a tinge cor UE (i) Construct a ste" aan (ui) Using the stem on the distributio eAot ru estnator cet cand cons unbiased a al x,yisam : _Sro103 fc) Which of the above estimators is more efficient? Let ¥ic¥p._-o¥q be a random sample from a distribution with location parameter y. a) Give one example of a location parameter f by) Derive the least squares estimator for kj ) What is the limiting distribution of this estimator ] SECTION B (60 matks) aves may attempt THREE questious being careful to number then B4 to BT. Ba. A point whose cordinatesare X, ¥ with respect to rectangular axes is chosen at random whee O<2-<1and00, 2 x = (0.784) = 8.136 A 90% confidence interval for pis given by TATE Zap = 0784+ L6H, eae 1000 0.784 + 0.0214 (0.7626; 0.8034) > P(07626 < 5 < 0.8034) = 090 = P(30501 <= < 3.2016 [3.7 (ii) IC the value for + is tobe in exror by less than 0.005 then the value for $ must be in error by less than 0.00123. ed page 5 of BS. srows n= (Zt) (a) (641.5199 aso X has a probability distesbution with densty fo) Suppose X has ap 4) = 2 Were @, O>0 $00) = Ble- 0 bu Be = 1)! . ) merit EIN-)! Coefficient of hurtosis (Pe = sroi0s boat = Ba) and (2?) = now Bas zits we (8) iosas * (Gg 135 ~ ~24963 Since fy = ~24963 < 0 it means thatthe distribution is plathurtic | (in) Since @¢ > 0 i means the distibution is skewed to the right and Oe (b) Let 1, a... %e bea radon sample fom exponential disteibution with meat {hen the probability distribution Function of X is given b Bo. a =| Un, ee differentiating and equating 2 tn E(x a 5M L(t, 9,2) T] ing = arent nina aS) 15d) = O gives ‘Thus the maxima likelihood estimator of 4s 92 Therefore A= 2-1 x, = 1 = 0.305, (ii) Advantages of maxinnam likelihood estimation are-(any one of * invariance property «estimates are always consistent alley estimate varia ) () A hypothesis is ‘or model for data generat Terror: when Ho s reject ny ‘decision Conve’ { Decison hypothesis eject Ho accept Ho D6 atts chain is jst Pthrowing a sis = § Sixes obtained in six thew bet ¥ bea Then X~ py about t a beliet or suspicion Hw Cocect Deesion Type Tetor rable sepre anda Binl6.3) he parameter(s) of a population ng umber of page B OFT {i) Note that ET) varity) = 4 Since T, = 23h wa N(0, 4) (ii) lot l sd 4 Now T Tire They (Gy mnedhan as the amide value descending order (wy x~U@t Por ordered data statistics Vur(xy = 49? 4 hon data is anrauged an evcler ascending 6 Xp is the medians 16 F page 10 of 11 w) ot _gyrta ytd po = gayest = Gn(-w O< mT ung y= me | fry) =6ult-y), OST Nove thar fly) = fgBhyt MUL = g)Pt = Betal2.2) where FE = 6 So R has a Beta distribution ie Beta(2,2) END OF SOLUTIONS pare 11 of 1 rps Vidy oe ZIMBABWE OPEN UNIVERSITY | Se Mathemati STATISTICAL INFERENCE | 03 hours Hates should attempt ALL questions an seetion 4 at 15 queso in section B. Mas] fe allocates! as ind Explain what y Dro. thustrate the graphs of & | ) Teptokntic 2 (0) inesokrt ) paykit od is B. The cetationship bere S (a) Bind the Least sanares tu b} Pind BU) aud (©) Pind Varta srbis _ i ‘Ad. (a) Copy and complete the following table 7 Decision/ Hypothesis [| Hy true | 7h true reject He scm ‘aceaplH me + bb) Give definitions of each of the following terms (6) Type Land Type Ul error 2} (un) Significance level ) 2 vu) Power ofa test |AB. Suppose two independent random samples from two noriral populations N (0) and Xiju.o}) yield the following summary statistics mpleT | 50] 32) 4 Sample 2 | 100 | 30 | 2 a) Pinel a 95% confidence interval for js — '5 (b) Comment on whether there is a difference in means (uy SEC JON B (60 marks) Candidates may attempt THREE questions being careful to mumber them BO to BS. BG. Let X;,Xp,..,Xy be a random sample from the probability mass function given bi. pip)", r= 0,1 no {F ‘evi (a) Find an estimator p of p using: (4) Maximum Likelihood Estimator(MLE) '5 (i) Methods of Moments(MM) ial } 0) Given a realisation of (1,1,0,1.0}. find an estimate of p a c) Show that p18 a consistent estimator of p 4 |) Prove that P= £2. X, is ascent statistic for p 5] (e) Hence, dedice a sufficient estimator for p. y BT. Let Xs, Xo... Xu be identical and inde Mer te Aan lie ented and independeaty dsb (i) eandamy sare page 2013 ee a pow that BX) isan unbiased estimator ofp ) p Pe tw show that 5 is a consistent estimator of of that Var(X) = 2 Hence deduce that $2) and comment on your findings from the disteibutio [Link] density fu bution function Fy Ib) Let Xi, Xa, Xs be a i O41). Let XapeXur.X 7 (0) Name the dist fi) Hence find the 0 Hint: The 3 fly) {a} State the relationstip be (B) Let X4,Xa,.-- Xm, aed YY bles with means ea 5 2 be unbiased spl ZIMBABWE OPEN UN IVERSITY sand Statistics Part L 1 SOLUTIONS BSc Mathemat Si) STATISTICAL INFEREN( May 2003 ‘Candidates wore requested to attempt ALL questions in section A and 3 questions in section 8 SECTION A (40 mark Ibered then AL Candidates showld have attempted ALL questions being careful to have to AS | AL. (a) Unbiased estimator: Let X1,...,.Xq bea random sample from a population with parameter @ An estimator 8 of 8 is said to be unbiased if a Ba =8 ive the expectation of the estimator is equal to the paameter 8, | (b) Relative effcioney: Let Xy,--.,Xq be a randoun sample from a distribution de pending on an unknown parameter @ and let 6 aud dy be two estimators of @ with respective variances of(n) and of(n). Then the relative efficiency ey, is defined | ty if ny)" _ oft ue = | ain) ain) | (6) Consisten- estimator: Let X,.... 2% be a random samp fom a population with parameter 9 Then an extmator = @(X)-. ts) a) eed eee lirmg Pll] >) =0 depending on 8. Then a statistic T = T(X,._... X,) is anid to be saffeient for 8 AR. Let Oia be a coefclent of kurtosis. Mlustati Mlustavtions ate given in igure 1 page of As. o ) VS ya a Lett stows Tha Sy Since E(3) = B+ isan unbiased estimator of il Var(A) Since Var(B) -» 0 8 n+ 0-= (18a consistent estimator a} Decision Hypothesis | wwe [fh te 4. (@) wt Hy Type Teor [correct deision i (0). () Type error: when Ho i ected when infact tis conrct i Type Trevor: when My is etaived when in fact fale i (i) Significance level: this s the probability of Type T ersor 2 (ii) Power ofa test: this isthe probability of correctly rejecting Ha 2 AS. (a) A.95% confidence interval for 1 ~ By Is given by a 8 x-mas rood + 5 2206198 (1.3802 , 26198) fs Ee ——E—Eee ace 3009 i SECTION f (60 marks) ates should have attempted THREE ques 0B h them ons being, careful 0 have nusubered (9) the tet @ Taking, In and differentiating gives ta Lip) = ntinp + (n-ne) (1 ~P) a nant ig In Lip) a 1p ae 1 np tables B(X) (ji) method of moment. Usin spl equating population moment i (y= 4 over var(s) = Var(t)= 3520 Ly n-») PUL=P) asm > 00 ml br Thus i a consistent estimator ofp (4) Recall that T= 2 Xs ~ Bin(n,p) Herm =8 = tan 1 goes not depend on p 2 T is asuficient statistic for p (e) A suficient estimator of pis given by = £3" (®) | +0 as n -+ 0 This implies that X is consistent M4 c (a) Recall that ol ae distribution wit BB. (a) (1) Let Y= Xa = min(Xi. Xn) Fr(y) PIY oh 1 Platt X, > y) 1 PIX > Xe > ah rt~—CiCSCisC_C 1 1 [1 P(X Sy) fh PIX Sv) [Fey A) = 3 and n = 3 we obtain: fely) = tea Gyl-y) fore yet Thus f(y) = 6y(1 = v) (is) Note that f(y) = edb yt HCL ~ yyPt ~ Betel2, rare = © where So Y has a Beta distribution ie Beta (iii) The mean and the variance are given by WY sf wayll — vidy =6 of“ 2 - G75 0 y- BWV ent | > Van(h) = El FJ po, ta) tee and ¥ beep esnune randoms ales wth ad dees | sftjcagonvapectvely ‘Then te andor vail i um luas au F distribution with (>, 1) degrees of freedom axe 8 of 9 stpiaa Now using the relationship i part (a Uf 1 _ fim ~1)s8) PV =) mde) [2]. [4 i loi} * Leal 7) : } aid | [Brim UFip55(0.035) foarag (aie oasa 6ay4o | (276 + 3801 = loarae 3.3569] (i) of = of > J =1. Since Le oanae 3.3568) = variances are equal, Daze 9 of 9

You might also like