UNIT'V - ESTIMATION AND HYPOTHESES| \ Chapter 6 ESTIMATION AND HYPOTHESES (A REDE es TCA INTERFERENCE \TION AND TESTING OF HYPOTHESES) C FATRODUCTION _ to study @ group of large number wooed of items or ind devel times in every day life, viduals we requ - ee ete in aera arora isl Wefne the terms ike population, ann sme plays ety. iriportank roe cea Aa jg otro replacement (SRSWOR) eter ‘le random sampling with replacement (SRSWR), simple random DOM SAMPLE FROM A PROBABILITY DISTRIBUTION = Forside 2 popuetion with probability | \ eit Continuous] ity law f(x €) [pamé. in case of discrete distribution and p.A pefinition : A sample Ky. X2. o.) Xp from the , vert and ental sted population with probity lw fx) scaled 2 random sample it they are Peon cena fr ect random variables with probability 'aw f(x 8) : ing pa ve reuse numeral ves rg parame 1 for any other inference is genera (a STATISTIC AND PARAMETER Using the random sample X, Xp, Mq we draw concusion abot pet sition can be studied if the parameter @ is known. In ov st a ey of parameter @, We use sampled observations for this purpose, There are various WAYS ‘of summarizing the Se ‘ont, The summarized quantity is called asa statistic. We define it precisely as follows * rition : If Xy Xo = %n i @ random sample from a probablity distribution F(x @), then T = T (ia, Xo sample values which does not involve unknown parameter 8 6 called as a statistic (or estimator. Some typical statistics are given below : xi T= TO tke) = is jn case the probability ily @ random sample, For drawing The unknown probability distebution However er words study of probability cistibutlon reduces xq) @ function of (Sample mean: T = X isastatistic (i) Sample variance : isa statistic iy similarly, it ean be show that sample proportion (p) is aso a statistic Note: {@) Verbally, statistic is = summarized quantity of sample valves such 95 0 (On the other hand similar quantities ‘uhich correspond to population are called as Pal the constant @ involved ints 280 called as parame Note that, the popul vrariance ete are the functions of ®. Thus By the term function of @ proportion, correlation coefficient, an, variance, rameters. If f(% 8) isa pd. then parameter we mean either 6 oF lation mean, (6a)sma ple and the later 8 cong, able. It varies from san, el PROBABILITY AND STATISTICS (DBATU) ‘ga random V2 that the former is ‘a random ¥3" {@) Anoteworthy dfference in statistic and parameter iis 052 Since statistics is a function of random variables XX Xn be the 62M@ 2 that sample. iy aistabutor ma ity astrbution £3) Since statisti is random variable it possesses some PPOPA "4, not possess pro parent distribution f(x, @. However, parameter being constat SARD ERROR > (GA SAMPLING DISTRIBUTION OF STATISTIC sity distribution. The generat they Further statistical inference is based on statistic, therefore We is developed in the subsequent discussion. Definition # If X,. Xp a, Xn. random sample from f( €) th called asit sampling distribution and standard deviation of Tis ‘ sem = VEIT it In the problem of testing of hypothesis standard error of statistic use quantity. We list below some typical statistics alongwith standard errors 5 9 W T2k, SEM ‘ na where, o? is the population variance. {il) Suppose there are two samples of size n and nz resPe and the second is drawn from a population with variance ohindependentl. statistic Ty Xp gy i yen the ane test procedure tums oUt (0 be impor in tively Fist is draw from 2? population with variance gt Let, T= &-%, then sem = \ varti-%) 7 2 \ | 24% Werf m *m (il) 1fp is the sample proportion then SE (p) = V2 where, P: Population proportion and Q = 1 P. fi (iv) If py and pp are proportions obtained using two samples one from first population with population Proportion , land the other sample from second population with P2 as population proportion. Let, T= pi-P2 then, it ean be shown that 6.5 TESTING OF HYPOTHESES. 6.5.1 Introduction {Quite often we need to test some claims regarding populations onthe basis of sample. For example (2) A particular scooter gives average of $0 km per litre (2) Proportion of unemployed persons is same in two different states. {3) Average life of an article praduced by company Ais larger than that of company 8. (4) Marketing agencies advertise the goods on radio, television or in newspaper etc, with a view to increase the ss In this situation one may be interested to test whether such an advertisement campaign was really effective | increasing the sales. Thus mainly, we ae interested in testing certain clams, These claims are tobe stated n terms of population parameters statistical distribution.hs = 2 _——_— — ems aa ypothesis : Definition (63) ESTIMATION AND HYPOTHESES ment or claim or assert saree assertion abou tthe statis A pe words hypothesis fs an assy sical distibum imption to be bution or parameter of statistical distribu istibution {a) Average life of @ bulb is 1000 Nested, Nous. can be writen as Hy: 1 = 1000, where u represents popuiation f hypoth ce wero pes of Myatheses in eve it These hypothes TY problem of tests ? ove vers® \ypotheses are referred to. of significance. If as null ‘one is rejected the other is tA eee hypoth other is to be accepted cai Ae wh ce ess nd aemae hypotes as we we n "ose en unde spon a according to RA Fisher null hypothesis s on ei oiu * 100, Here the hypothesis sates th ion that itis true, Null hypothesis is denoted by He gba 2 B2 OF Ho: waa = 0. This hypothesis states that ee there is no difference in population mean and 100. contd jie test sme iflerence wil be cbserved in a is no difference between two population means: While Ho Hh P< 05, H, , 3 Hy a < Ba ase >a, etc are cal as one sided hypothesis. On ther hand th : the other hand the hypothesis of the type Hi : Pi #05, Hy: 14 Gy Hy: He Ho ete are called as two sided hypothesis. 65.3 Test of Hypothesis inorder to study the above stated clsims (hypothesed) imoractcabe to study each and every tem i the population. Moturaly one has to use sampling methods, n ths siwaton Temmple of appropriate siz Is selected, suitable statistic such . pr i, ezample mean, sample variance sample POPOriON © ved to take decision about accepting of r6}eCtN9 hypothesis. How tpeary out the test willbe clear inthe subsequent discussion. 63.4 Type Land Type I Errors. Since, decision of acceptance oF reject ie subject to two kinds of errors. For instance. 0 the ingpection of a lot of manufactured items, the inspector wil wuitable size and accordingly take decison whether to accept or reject the wr inthis process, two eos are POSSOIS Yo rejection of a good lot and if ba ot In testing of Rypothesis test pers are called as type and type rros: Type Terror : Rejecting Ho when itis true ype H error: Accepting Howhen t is false. these errors can be put in tabular form to remember eas Table 6.1 _ ne see ‘Actual ‘Situation His tue Type lero cet decison | ‘Type ll error His aise Cconectdecision |ESTIMATION AND HYPOTHE PROBABILITY AND STATISTICS (DBATU) (64) 6.5.5 Critical Region Let, Xp ws Xn be a random sample taken for testing Hp While deciding whether (due to sample observations) is divided into two non-overlapping regions: In other into two subsets W and WE in the test procedure. One of the regions is acceptance region of Hg (or rejection region oF ia value of statistic which is us acceptance region of Hy). A region in which Ho is rejected is treated as critica! region I i ted by Wor C. ed testing Ho falls in critical region, Hy i rejected and accepted otherwise. Critical region ies Fa woe or fe: emir spe > words the sample space is Patong ry and te other is rection rion of 4, Fig. 6.1 6.5.6 Test ‘Arule which leads to the decision of acceptance of Hy or rejection of Ho is called a test | Tere Static: A funtion of sample obsentons whichis used 0st Hois called a te S305 Level of Significance : A probability of rejecting Ho when itis trues called as level of significance. It is denoted by Level of significance can be interpreted 2s proportion of cases in which Hp is rected though It is true. It's a probability committing type I error. Level of significance cannot be made zero, However, we can fix it in advance a5 ‘the cases it is taken as 5%. 6.5.7 One Sided and Two Sided Tests ‘As seen earlier the alternative hypothesis Hy hypothesis H; is one-sided is termed as one sided test or one tailed test whi hypothesis is two sided is called as two sided (tailed) tests. The details of these tests. 6.5.8 Observed Level of Significance (p-value) Use of probability value (p-value) or observed level of significance is bect fixing the level of significance uniformly same for all data sets, sometimes it s computed for the data set used in testing f hypothesis If Z isthe test statistic and its value under Hy for 2 given data set is Zo then for the above two tale tet observed p-value will be P (| Z| > | Za) In the above illustration p-value = P (12 |> |~ 6 |) = 0. Thus probability of type! ‘error in this case is 0 If Zy would have been 1.5 then P-value = P (|Z > 1.5) = 0.0868. p-vatue is the smallest level of significance at which Hp would be rejected. p-value is more informative as compared to level of significance. The details are beyond the scope of the book. Some software packages use p-value instead of level of significance. POINTS TO REMEMBER. ] «Population is collection of items or individuals having some common characteristic. Any part of the populater selected according to some rule to study population is called sample. 106, Xr ns Xi a random sample from a probabilty distribution fx 8) then T= T(x, x2.) @ function of sample values which does not involve unknown parameter 8 is called a statistic. For example, sample mean, sample proportion, sample variance are statistics. If Xy, Xa esse Mo is @ random sample from ffx, 8) the probability distribution of statistic Téx, x2, .-., Xe) is called as is sampling distribution and standard deviation of Ti called its standard error (SE) 101 (ie. 1%) oF 0.05 (Le. 55). In most g either one sided or two sided. The test used to test Hp when alternate the test applied to test Hp when alternate ‘are discussed in the next chapter, ‘oming papular in tests of hypothesis. Instead ofjs of “no difference” Is called as nuit gnosis ant ih tested oF posable ejecton hypothe or according to Pot. RA. Faber rol rypothesis is the tt x inder the. tne ye: Ho: H = 200 OF ba = assumption that it i ynple = 1 = Ha ete. Alternati that itis trus Trams jected. Its denoted by H, For exampl hypothesis is a hypothesis, which is to be 2ccePl ee exis of the tye Ha =H > ig Hy: samen 100 then He. > 200 or Hap < 3000° Ha we 4300 His bz etc. are called two sided alternative a i < are called one sided hypotheses on oll 4 error in testing null hypothesis Hy is rejecting Hy ena a error is call . when it is true. eaity OF PS eae led level of significance (Los) and ti ing nul hypothesis Hes accepting He when tifa itis denoted by a. Generally, ding whether Hp is true ‘ RE false: oo 140 which Ho is rojo ed sample space is divided into two non-overlapping regions. The reo) cttical region or rejection region. Its denoted by W oF < a= 005 oF 0.01. Type qqvesee jo ws p value of test statistic Z then for 12th? n for two tailed test p-value = We geld observed lve of sigrfeance, esos Pe (EXERCISES (A] «_ eaplin the terms: (population, (i) sample, (i) statistic, (iv) hypothesis, (y) null hypothesis, (vi). alternative hypothesis, ip critical region, (vill) type terror, iid ype error, 6) level of significance, (x pevelue. state the utility of sampling in statistical inference Distinguish between a parameter and a statistic. papain the term sampling distribution of statistic with istration, papain the term standard error with illustrations state the standard errors of the following : () sample mean, (i) sample propo ‘ample means, (v) difference between two sample proportions. +. faplain SRSWR and SRSWOR methods of sampl EXERCISE 6 (8) jents are true or false. tion, (il) difference between State the whether the following statem Statistic is a constant. Parameter is a constant. Standard error is human error in using ™ Random selection of units must be done while drawing a sample for statistical inference. ‘at which Ho would be rejected P-value is the smallest level of significance 5. True. ethod of sampling. L fale 2 True 3. False 4 Truewained for any Sample s fic can De distrib, thy Ting distribution of the 12S! sme cases em rege ss tet sa In testing of hypotheses, whenever exact sampling distibutlO” © yin SOME Opies size, sie tests ae called "small omple tests or * eax SOMBIE TS tun which Shows that up oes cmne os assumed to follow T=ED ig standard norms! certain conations, for lagen the asibuton of Z= SEC cenerally on this result, the corresponding tests are called as ‘Large S2™P! 1 whenever the test Procedure sy, for 0230, the large Sample tye | applicable Tests of Hypotheses ct le tests F Large sample tests small ann wo carrying out these tests involves, In this chapter, we will discuss large sample two tailed tests only. The procs following steps. (2) Set null hypothesis Hp and the alternative hypothesis Hy. (2) Choose appropriate test statistic Z @) Compute numerical value of Z for given sample sow (4) Compare Z with the critical value of Z (22) 3s given in the table 6.2 Table6.2 Tr Zor two Critical valve of test statistic ‘Level of significance (a) mee 10% 14 3% 1% 1% 258 Otherwise we accept Hy. If|Z | 2 Zaya, we take the decision to reject Hp at level of significance & ; a i ds act upon the decision taken in the above (6) State the conclusion about the claim made in the problem. In other wor step. In this chapter, we study large sample tests mainly regarding hypotheses about population mean and proportion. Case (1) : Two tailed test : Suppose x, X, Xn is a sample taken from 2 population with mean p and known variance 0. The sample is selected fa testing whether population mean is equal to a specified value pp against itis not equal to Wo. In this case, the null hypothesis will be Ho : 1 = tp and the alternative hypothesis will be Hi: # Ho A test statistic! based on a sample mean X. 2 Clearly € (X) =p and Var x) = Hence, we get Under He zt NG)( wt [AND STATISTICS (DBATU) (Pgs, under Hp for large n. we 5 ate at mean O and variance 1 assume distribution of 2 to oni this c0ses |Z] 2 1.96 ‘ol aL 5% level Ae 88 of Z is more than 1.96 of le of ee aod accepted otherwise, th if Hy is tested at 1% ley aires Ho vel of significa gf gl ZI? 258 ‘Thus Hy is rejected at 1% feat te critical significance i a ale of Zis more tha joe 2.58 oF less than ~ 258 significance, an ~ 1.96, His +g) Incase 0? is unknown, we repla we ce it by its sample m vean square, ge = Fa] O¥ nF) i” ESTIMATION AND HYPOTHESES roa = 0.025, aeneone 2 a2 Rejection region for Hy Acceptance region fot Hy Fig. 62 198 LL Rejection region for Ho ste sample i rv fom a normal popula : 7 on then this tes is exact sample test. is applicable for smal values of" jet gis KNOWN. ene a SOLVED EXAMPLES Z yn of 0.1 kilogram. e6.1 An automatic machine pomel {5 designed to fill tins with 2 kilograms of oil with a standard deviation Test statistic ote that X = 1.94, Ho = 2 (Under He), 6 = 0.1, n= 100 t l= 194-2 0.06 oad 100 01 of 100 tins was examined ond the average weight was found tobe 1.94 kilograms, Can we say thatthe machine =-6 1z1 (cleloted) = 6 > 196. Hence, we reject Hy at 5% level of significance. Since it's rejected at 5% level of sinfcance, ve test it at 1% level of significance, Z| = 6 > 258. Therefore, Hois rejected at 2% level of significance also. Conclusion : Machine may not be working propel Interpretation: 4a) Intesting of hypotheses if the hypothes's Ho is accepted, it doesnot mathematically Prove oF disprove any cairn. Only we can conclude whether the sample data Qe 2° evidence to reject Hp or accept Hy (2) Here the level of significance used is 5% which means ifthe abi viet situations wrong rejection of His Posse {@) In Example 6:1, population mean is # © 2 and 52 considerable or negligible cannot be decided just testing of hypotheses. Whether the existing dif answered by test procedure. (a) Test procedure gives a common yard stick for comparison PM used and corresponding parameter. / Inthe above example the difference between X and pis very small itis which seems considerably large. Thus ‘the visible difference may appeal étests of hypotheses such a subjectivity 5 eliminated. rove procedure is used in 100 situations. en in at the imple mean is X = 194. Whether the difference between them i Ty inspection or intuition. An objective way oY 4 tool is given b) Hey ie due to chance (je. sampling) of aly Ho rot true, i rpose, by standarizing the difference between statist 0.06, however ater standardizing. t becomes =| + to be small or large due to subjectivity. By reaESTIMATION AND H¥Ponig [PROBABILITY AND STATISTICS (OBATU) 8) Case (il) One - Tailed Test cedure ie same a 350, poth the types, one yo Here Ho: = Ho and Hy: W> bo (or < Ps) ing test PIO In this case ertial region (CR) is one tailed it depends uRON ME TNT 1. combining TRH: HO» Ho then CR. is Z > Zq and for Hy pcr ck convenient rule for practical purpose as CR. |Z1 > Za. ms. Can we Assume that a ight of 125 970" ‘am Example 6.2: A sample of 900 ball bearings is found to have overa9® MOO) ams Fissure tat the POPULATION raga ‘against that iti fess than is coming from @ population with mean 13 grams deviation is 2 gram. Solution : tet = Weight of ball beating We formulate first the null hypothesis and alternative hypothesis. Hy: Wo = 13 Versus Hy <3, We are given thatX = 125,0=1,n=900 X- Test statistic z= ot = 258. 95x30=-15 ome 16a), So we reject Ho at 5% lee Here CR. Is Z < -Zq Note that Z = -15 < ~Zq = ~1.64, for | 2) = 15> 20> significance. Moreover itis rejected at 1% level of significance also. Conclusion : A sample may not be regarded as coming from population with mean 13. [6.8 TESTING EQ JALI So ONS POPULATION MEANS. ulation means. Here ye In statistical investigations sometimes a problem arises, whether two samples have same po discuss how to test such claims Suppose, there are two diferent independent samples selected from two different populatio with mean ws and known standard deviation 0}, Let mean of first sample be Xy. Similar known standard deviation o>. Nui <2 ory Bs > Ha ns, Let first sample be of sce ri. Its taken from a population second sample is of size n; with sample mean a, from population with mean B2 and hypothesis will be Ho: Hs = H2 and alternative hypothesis willbe Hy ms #2 OF Hye Let, T= 1 -% therefore E(x -%2) = Him ba: Var (i =) = Var) + Var 6% )~ 2 Cov (2) and 2g -2,4_6 mm Covariance term will vanish because the samples are independent. Test statisticap STATISTICS (OBATU) (69) ESTIMATION AND HYPOTHESES: oe so approximation we get for large n; and 10 in hand ny (each greater than 30), the limiti s iting distribution of Z to be under Ho ont got! of sig agence et , fh we py CRis Z<—Zqand for Hy: wy > py CRIs Z > Zo. ie eptational BUPOSE, ONE MAY Us for convenience CR as|Z|> Za yp CRIs given by |Z|> Zar at a % " \ if ifcance otherwise accept Hy If hoes of significance. In ther words if} Z (calculated) | > 1.96 reject Ho at is rejected at 5% level of significance, it may be tested at 1% evel of le test, it holds for small values of M1 and ny ote: ay the samples are drawn from normal populations the testis exact sampl and si where, 2 sample estimates Si 2 2 and oy are unknown, we replace them by corresponding 1 gia Jae e me i ands} = 63iAs alec i oo Se an le Fe bulbs from company A gave an average life 1225 hours ‘with a standard deviation 42 lbs fram company B gave an ‘average life 1265 hours with @ standard deviation 60 hours. row soy thatthe two companies are producing bulbs af same overage lie? solution : Let 1 be population mean of life of bulbs due to company A and 2 be population mean of life Gym me of bulbs due to company B. Ho : Hh = Ba versus Ha: Hi #2 = 60, my = 400, m2 = 200. Here % = 1225, ¥2 = 1265, estimate of 0; = 42 estimate of 6 1225 - 1265. -400 | te fee et ~ 47382 © ~BAt6. 400 * 200 since | Z| = 8.4496 > 196 we reject Ho at 5% level of significance, similarly [Z| > 2.58 we reject Ho at 1% level 0 significance tbe same. Conclusion: The by two companies may 0 ‘Example 6.4 : A rand from county A gave mean height 170 cr. from country B gave ‘mean height 178 cm. Can you say that persons in country B are taller population standard deviations 73s 6 cm and 8 cm respectively ? Solution : Here, We want to test equality of average height of persons: jn two countries : 7 ith person in country A Yi = Height of th person in country B Versus Hi: Ba < 8 average ife of bulbs produced fom sample of 400 persons ‘Another sample of 80 perso” than those of A given that theESTIMATION AND We rroummaosunmseuny 9 Test statistic : ze - 800 Note that; = 170, 0 = 6& nm, = 400, %)= 178 278 ™ 170-178, 8. = - 19.4029 a 2s ee 7 as 1400 * 800 Ris given by2 <2 (or|2| > Za) 4 level of significance, n, Here, Z = ~ 194029 < ~ 1.64 (or|2| = 194029 > 1.64) therefore we et asia language we say ae < ° ° , also : Z < - 233 (or [Z| > 2.33), we reject Hp at 1% level of saiance aoe eth significant. difference between the average heights of persons from county A 2 Conclusion : Persons in country 8 may be taller than those of county A VALUE (Pa [9 TESTING POPULATION PROPORTION (P) EQ QuaL To A SPECIFE SPEGHIED servations of specifi 4 Sometimes observations are taken on qualitative characteristic, In this case, number ofo ee counted and its proportion is determined. i For example : Proportion of unemployed persons, proportion of neo “ne we doy test in order to test whether proportion of certain type of observations is €4! Pee cision P = Proportion of certain type of coals n = Sample size. X = Number of items of p= % = Proportion of certain 9 of females, propertion ual to a SI F items in Por certain type in a sample of size 0 1 of items in sample. Py = Specified value of P. Here Hy: P = Po versus H, :P# Pp In this situation the observations are lassie into two different cases. Assuming thatthe probability that an Sbsenot belongs to the first class is P, we find the probability distribution of X. Cleary, X follows binomial distribution wit parameters n and P symbolically, we write X +B (n, P). Hence E 00) = nP and Var (X) = NPQ. ‘Where Q = 1~P. 1 oP rn co=e(%) =b em =F oP ant varie) = var(®) = 2 vary = 280 = 92 Using normal approximation, we get limiting distribution of =E p-P. ae = NOD 2 eo VB asN(0, 0). n D Under Hp: P = Po we get Z fs distributed as N (0, 1) for large n. n Case (i): If Hy: P# Pp then CR is given by |Z| > Zay2 at a% level of significance, Case (ii): IfHy:P < Py then CRis Z <-Zg and for Hy: P > Py, CR.is Z > Za (or combining both types of C.R. together for practic purpose we get |Z] > Za).ESTIMATION AND HYPOTHESES: Sed We expect 50% heads ie. p = 0.5, Hence Hy : p= 05 = 2525-05 _ 2.025 5x05 ~ 0.025 _ 400 Since |Z| = 1 < 196 accept Hy at 5% level of significa jsion : The coin may be regarded as fair “ ple 6.6: manufacturer of ball - bearings guarantees th ge 25 defective. Con we Say the product meets guaran 2% of items are defective. A sample of 1000 ball bearings Squton: The hypothesis regarding this ob MT ecive ball bearings = 25 Problem will be Hy P = 0.02 versus Hg: P > 0.02. Here n = 1000, x = number of Hence, -t,3 P= 5 * Toop = 0025 2 = of }PoQo 7 0.025 ~ 0.02 =~ Topeoss = 294 1000 since Z = 1.1294 < 1.64, we accept Ho at 5% level of significance. Conclusion : We may say that product meets guarantee. (GOTESTING EQUALITY GF TWO POPULATION PROPORTIONS (P; = Pd) ees Here, we discuss the cases when the two samples are taken from two distinct materials or populations. Suppose, a sample ‘sdrawn from each of the populations. However the test statistic is based on both the samples. Suppose these samples give proportions of specific items as py and p, respectively. One may be interested in knowing that the population froportions from which these samples are chosen are same. In other words we want to know whether difference Retween two sample proportions is negligible and it has arisen merely due to sampling variations. let n, = Size of sample drawn from the first population. nh = Size of sample drawn from the second population. ‘Number of items of specific type in first sample. Number of items of specific type in second sample. x p, = & = Proportion of specific items in a frst sample. p; = % = Proportion of specific items in a second sample.{TON AND HYPOTHE. mousumuosinmesouny wi en PROBABILITY AND STATISTICS (DBATU) (sa st option . speci lation. iA Proparton of vein second POP P, = Propartion The hypothesis for such problems will be Hy = Py = Pp versus Hy: Pi ® Pz = eset Using normal approximation for large n, and ny. Under Hp: Pi ttcan be shown that 1B =F 2 ze +1N(0,1) for large nim: var tp. P? =n ne Since P is unknown we use its estimate. Here we combine (pool) these $27 ples and find proportion of specific type of te is byP items. Proportion in a pooled sample is taken as estimate of P if Ho is true. Iris given by . 4 aR 5 Rt | Mmm JP Pe “mem ~ men ; and ns Note that B isa weighted arithmetic mean of py and with weights as sample sizes M. =P willbe the test statistic 5 ze = f a(4 + Z N(,1) forn, and n, large. Case (I) If Hy Py > Pa then CR.is given by [Z] > Za; at a% level of significance. ent Case (il): Hy: Py < Po then CR. is Z < =Zq and for Hy Py > P,CR.is Z > Ze (or combining both types oF Cs together for practical purpose we get [Z| > Zo). sind fo be smokers. In a sample of Example 6.7 : In a random sample of 800 persons from rural area 200 were fot 1000 persons from urban area 350 were found to be smokers. Test whether prop populations. i he Solution : Let Py and P; be population proportions of smokers in rural and urban area respectively. Hence Ho : PL=P, versus Hy:Pi #2. Note that :n = 800, X; = 200, = 1000, X; = 350 cartion of smokers is same for both 350 Hence, pr = 20 2025 and pr = apg = 035 Pooled estimate of 0.25 - 0.35 0.3056 x 0.6944 s&s a) muro 800 * 1000, _ 2010. * 0.021851 Since [Z| = 2.2882 > 1.96, reject Haat 5% level of significance. However, we can test t at 1% level of significance. Note that: |Z| = 2.2882 < 2.58 hence we accept Hy at 1% level of significance, Conclusion : Proportion of smokers in rural and that of in urban area may regarded same at 1 % level of significance. = ~ 457643OO —-___——— np STATISTICS (OBATU) , production of day shit, gave 52 ‘tion of night shit revealed 45 items defective. Can W® 68 ts Atop oa MS selected from p items selected from a! lected p that of second shift? és 6 Home of 70 E ie i A rogortion of defective ites in the frst shi Shift is te ns ind Pr FePresent proportion of defective icnste aie’: ietie a versus Hus Pu Pe TO = 1000, %1 = 52, mz = 700, x) = 45, we jefore PL Zt wi from day sift and right sit respectively. Formulation of eh * Yoo = 0052 We 228 Pre hy = 79g = 006428, Xt m+n, = eS * Too +700 .0s71 Pi-p2 2,3 PQ @ * 2) 0.052 - 0.06428 - 1 1 0.0571 x 0.9429 (ae +700) 0.01228. = pone 7 71079 since Z= 3.0739 + pga 6A, accept at 3 ve of tienes F120) eZ=- Ul onclus Ci fts may be same — + Proportion of defective articles in Wo shi f ae SUMMARY OF LARGE SAMPLE TESTS Ton = ate ‘erative hypettesis is Hy P# Po 0 zr 1 | ee 2 | 258 * | 8 mby:P>Py | ZN o 0 nr Gp P 288 1% Py? Py \o 16h | ob Hy Py? Pa feePs an 8 ath P

258 % | 5% ae i) Hy > Ho (i) 27184 “m 7>232 1% (i) Hy BS Bo (i) Z<- 1.64 oh 2Z<-232 1% 4 (Hy Be He (21> 198 5% [2|> 258 1% ze Hy b> Be (i 2> 1.64 5% Z>232 1% (Hy: By

You might also like

• Testing of Hypothesis Notes

  

  No ratings yet

  Testing of Hypothesis Notes

  10 pages
• Engineering Hypothesis Testing Guide

  

  No ratings yet

  Engineering Hypothesis Testing Guide

  9 pages
• Chapter Five Hypothesis Testing

  

  No ratings yet

  Chapter Five Hypothesis Testing

  50 pages
• Sawpl: Nef - The 7.V

  

  No ratings yet

  Sawpl: Nef - The 7.V

  24 pages
• 9 Nov Maths Notes

  

  No ratings yet

  9 Nov Maths Notes

  12 pages
• Faculty of Information Science & Technology (FIST) : PSM 0325 Introduction To Probability and Statistics

  

  No ratings yet

  Faculty of Information Science & Technology (FIST) : PSM 0325 Introduction To Probability and Statistics

  6 pages
• Z Test

  

  No ratings yet

  Z Test

  25 pages
• Sample Theory and Test of Significance

  

  No ratings yet

  Sample Theory and Test of Significance

  11 pages
• Tests of Hypotheses

  

  No ratings yet

  Tests of Hypotheses

  24 pages
• Statistics 5.testing of Significance

  

  No ratings yet

  Statistics 5.testing of Significance

  15 pages
• Hypothesis Testing Notes

  

  No ratings yet

  Hypothesis Testing Notes

  18 pages
• Statistics 2

  

  No ratings yet

  Statistics 2

  25 pages
• Chapter 6. Hypothesis Testing

  

  No ratings yet

  Chapter 6. Hypothesis Testing

  49 pages
• Unit 6

  

  No ratings yet

  Unit 6

  51 pages
• 6.4 The Normal Distribution

  

  No ratings yet

  6.4 The Normal Distribution

  23 pages
• Lecture Notes ON Parametric & Non-Parametric Tests FOR Social Scientists/ Participants OF Research Metodology Workshop Bbau, Lucknow

  

  No ratings yet

  Lecture Notes ON Parametric & Non-Parametric Tests FOR Social Scientists/ Participants OF Research Metodology Workshop Bbau, Lucknow

  19 pages
• Hypothesis Testing

  

  0% (1)

  Hypothesis Testing

  139 pages
• Hypothesis Notesz

  

  No ratings yet

  Hypothesis Notesz

  6 pages
• Testing of Hypothesis-Terms-LargeSampleTestfor Mean

  

  No ratings yet

  Testing of Hypothesis-Terms-LargeSampleTestfor Mean

  30 pages
• Sample Tests

  

  No ratings yet

  Sample Tests

  66 pages
• Chi Square Test

  

  No ratings yet

  Chi Square Test

  10 pages
• Hypothesis Testing

  

  No ratings yet

  Hypothesis Testing

  12 pages
• Research pt.2

  

  No ratings yet

  Research pt.2

  36 pages
• RM Study Material - Unit 4

  

  No ratings yet

  RM Study Material - Unit 4

  66 pages
• Point Estimation of Process Parameters

  

  No ratings yet

  Point Estimation of Process Parameters

  64 pages
• Notes of U-4 and U-5

  

  No ratings yet

  Notes of U-4 and U-5

  35 pages
• Lecture 4

  

  No ratings yet

  Lecture 4

  28 pages
• Statistical Inferences Guide

  

  No ratings yet

  Statistical Inferences Guide

  37 pages
• 5 - Test of Hypothesis (Part - 1)

  

  No ratings yet

  5 - Test of Hypothesis (Part - 1)

  46 pages
• Chapter 3

  

  No ratings yet

  Chapter 3

  35 pages
• Complete Collection To Be Studied, It Contains All Subjects of Interest

  

  No ratings yet

  Complete Collection To Be Studied, It Contains All Subjects of Interest

  3 pages
• Premili Definitions

  

  No ratings yet

  Premili Definitions

  3 pages
• Hypothesis Testing

  

  No ratings yet

  Hypothesis Testing

  26 pages
• 5 Hypothesis Tests

  

  No ratings yet

  5 Hypothesis Tests

  14 pages
• Quantitative Reasoning Examples Using Fundamental Statistical Concepts - 1

  

  No ratings yet

  Quantitative Reasoning Examples Using Fundamental Statistical Concepts - 1

  5 pages
• Ch8 Statistics Ver1

  

  No ratings yet

  Ch8 Statistics Ver1

  21 pages
• STAT-101-Chapter 8,9,10,11,12 BY ATHA

  

  No ratings yet

  STAT-101-Chapter 8,9,10,11,12 BY ATHA

  13 pages
• Lecture7 Hypothesistest PDF

  

  No ratings yet

  Lecture7 Hypothesistest PDF

  22 pages
• Statistics & Hypothesis Testing

  

  No ratings yet

  Statistics & Hypothesis Testing

  37 pages
• Module 6 Testing of Hypothesis

  

  No ratings yet

  Module 6 Testing of Hypothesis

  49 pages
• Stat2602 Chapter5

  

  No ratings yet

  Stat2602 Chapter5

  28 pages
• Ed Inference1

  

  No ratings yet

  Ed Inference1

  20 pages
• Testing of Single Mean

  

  No ratings yet

  Testing of Single Mean

  15 pages
• Topic06. Analysis of Differences

  

  No ratings yet

  Topic06. Analysis of Differences

  63 pages
• Basic Concepts in Hypothesis Testing (Rosalind L P Phang)

  

  No ratings yet

  Basic Concepts in Hypothesis Testing (Rosalind L P Phang)

  7 pages
• 7.hypothesis Testing and Sample Size Determination

  

  No ratings yet

  7.hypothesis Testing and Sample Size Determination

  60 pages
• COM 201 - Inferential Statistics - 18032022-1

  

  No ratings yet

  COM 201 - Inferential Statistics - 18032022-1

  58 pages
• Business Statistics by S P Gupta 1

  

  No ratings yet

  Business Statistics by S P Gupta 1

  18 pages
• Statistical Inference Basics

  

  No ratings yet

  Statistical Inference Basics

  63 pages
• Module - 6 Statistic and Probability

  

  No ratings yet

  Module - 6 Statistic and Probability

  7 pages
• Large Sample Test

  

  No ratings yet

  Large Sample Test

  6 pages
• Chapter3 Notes

  

  No ratings yet

  Chapter3 Notes

  24 pages
• StatisticalInference-Tests of Hypotheses-1

  

  No ratings yet

  StatisticalInference-Tests of Hypotheses-1

  25 pages
• Hypothesis Test

  

  83% (6)

  Hypothesis Test

  15 pages
• Z Test

  

  No ratings yet

  Z Test

  14 pages
• Updated Day2

  

  No ratings yet

  Updated Day2

  107 pages