6.3 BASIC TERMINOLOGY OF PROBABILITY In this section we shall explain the vai probability theory. tious terms which are used in the st 7 ne study of some examples of random experiment a ; re ti Fee dice) and drawing a card from a deck of $2 hae goin, rolling a die (singular of Sample Space. The sample space S, associated with a random experiment, is the set of all possible outcomes of the experiment. Each Sas called a sample por ach outcome (or element) of a sample Remark. The sample space associated with an experime) e same role as the xperiment play plays the sau Example 3. A die is rolled and the number that turns up is observed. Determine the sample space of this experiment. Solution. The only possible outcomes are 1, 2, 3, 4, 5 or 6. Thus the sample space is S = {1, 2, 3, 4, 5, 6}. Example 4. Two coins are tossed and the result for each coin is observed. Determine the sample space. Solution. If “head” is abbreviated as Hand “tail” is abbreviated as T, then the possible outcomes are : Hon the first coin and H on the second coin Hon the first coin and T on the second coin T on the first coin and H on the second coin T on the first coin and T on the second coin Thus the sample space is S = (HH, HT, TH, TT) where the sample point HH is used to indicate Hon the first coin coin, and Hon the second Note. The experiment of tossing two coins gives the same sample space aS does the ®xperiment of tossing a single coin twice in succession ae Ss Three coins are tossed simultancously and the Sot - Determine the sample space. the a One possible outcome is Hon the first coin. We can indicate this sample poin' x result for each coin is coin, Hon the second coin and Hon t by HHH. Another outcome is H on6A Fundamentals of Business Stagg, the first coin, H on the second coin sample space is and Ton the third coin, HHT, and so on, 1) + The S = {HHH, HHT, HTH, HTT, THH, THT, TTH, TTT, sample space of an experiment is called an event. Events ary Event. Any subset of the 1¢ capital letters such as A, B, Cand so on. usually denoted by singl For example, if a die is rolled, then the set {2, 4, 6}, which is a subset of the sample space {1, 2, 3, 4, 5, 6} is an event. This event may be described as “an even numbe; appears”. Some other events are A = {2}, B= {1,3,5},, C= 01,2,3, 4 Simple Event. An event that consists of a single outcome of an experiment is calleq a simple event, For example, for the experiment of rolling a die, the event (2} is q simple event. Compound Event. An event that consists of more than one sample point is called q compound event. For example, for the experiment of rolling a die, the event [2, 4, 6)is a compound event. Sure or Certain Event. The sample space S, which is a subset of itself, is called the sure or certain event Impossible Event. An event which does not contain any outcome of an experimentis called an impossible event. The impossible event is denoted by 9. For example, for the experiment of rolling a die, the event {7} is an impossible event. Remark. Sometimes it is convenient to represent the sample space Sand an event A for an experiment by a Venn Diagram, as in Fig. 6.1. The sample points in S are represented by the points inside the rectangle, and the sample points in A are represented by the points inside the circle. Complement of an Event. The complement of an event A, denoted by A°or A orA, is the event consisting of all sample points in Swhich are not in A. The shaded region in Fig 6.2 represents the event A. Pe Ne Fig. 6.1 Fig. 6.2 Union of Two Events. The union of two events A and B denoted by AU B, is the event consisting of all sample points that are either in A or in B or in both Aand B.I'* represented by the shaded portion in Fig. 6.3 (a). iceceaiiaesprobability Concepts intersection of Two Events. The intersect; 2 pice a cmt Won of two events A and B, denoted by A P ” yA mple poi : represented by the shaded region 1 a a Ste common to both A and B. It is Remark. If A and Bare two events of q sample space s (9 only A occurs Pace S, then the events (ii at least one of them occurs (0 only Boceurs (i2) none of them occurs | and AC) BC respectively Occurrence or Happening of an Event, eee espectively the experiment is a sample point of A, For exampl Az (2.4, 6} occurs if the outcome is 2, 4 or 6 Tee : Acan happen is 3, which is called the number of cases o- out happening of A. rable Number of Cases. Th Favour a ee re ‘¢ number of outcomes favourable to the happening ofan even : ‘¢ number of outcomes which entail the occurrence of the event A. Thus the number of outcomes favourable to the happening of an event Ais simply the number of sample points in A, For example, if two dice are rolled, and we define the event A as “the sum of the numbers on the two faces is 8”. Then A can be represented by A = {(2, 6), (6, 2), (3, 5), 65, 3), (4, 4)}. Obviously, the number of outcomes favourable to the happening of A is 5 because this represents the number of different ways in which A can happen. Equally Likely Events. Two or more events are said to be equally likely if each has the same chance of occurrence. For example, in rolling a well-balanced die, all the six faces are equally likely to occur. Mutually Exclusive Events. Two or more events of a sample space are said to be mutually exclusive if and only if one of them can occur at a time, ie., if the occurrence of one event precludes the occurrence of all the other events. For example, in tossing a coin, the events “head” and “tail” are mutually exclusive. Exhaustive Events. Two or more events of a sample space are said to be exhaustive iftheir union is the sample space. Thus the events A,, Ay, . . ., A, are exhaustive if A,UA,U...UA, = S, where Sis the sample space. For example, in tossing a coin, the events “head” and “tail” are exhaustive. are represented by the sets A RB SATA BA UE occur if the outcome of ‘comes favourable to the ‘s APPROACHES TO PROBABILITY “re are three ways of defining the notion of probability. These are : 8 : The classical or the mathematical approach ‘ ; a empirical or the statistical approach. lth he axiomatic approach. probe of these approaches derstand the important and useful concept in the sPlity. The classical ical definition of probability are both used e axiomatic approach, introduced in “dy and applications6.6 Fundamentals of Business Statics, the twentiet the logical structure of probabilit study of the subject th century, consists of a set of axioms. The axiomatic approach exhipy theory and it opens the way for intense mathemati: cal 6.4.1 The Classical Definition of Probability the oldest definition of probability. It was used extensivg, e Pascal in connection with games of chance * The classical definition i by earlier workers such as Blais Definition. Let a random experiment result in n equally likely, mutually exchusive “vents and let m (out of n) be the number of outcomes favourable and exhaustive simple e' ‘ to the happening of an event A, then the probability of happening of A, denoted by P(A), is defined as number of outcomes favourable to A a) = total number of outcomes n(A) Equivalently, P(A) = (Ss): where n (A) is the number of outcomes in A and n (S) is the number of outcomes in § It follows immediately from the definition that for any event A, we have Os drawing a card of heart B : drawing of card of queen we have Pia) = P= + p 1 Then 52 Bo PNB = since we are looking for the probability of A U B, we use the addition theorem and find that 1 P(AUB) = P(A) + P(B)-PiAn & = Bo-s - ig = %. Example 18. In a single throw of two dice, find the probability of obtaining either a sum of 9 or a sum of 11. Solution. Let S be the sample space associated with the given experiment. Then n(S) = 36. Define the following events : A :sum is 9 = (3, 6), (6, 3), (4, 5), (5, 4)} B: sumis 11 = (5, 6), (6, 5)} n(A) = 4 and n(B) = 2 n(A 4 n(B) 2 z pa - Hae & and | POAT cn Sue . 36. Clearly, events A and B are mutually exclusive. Thus by the addition theorem for mutually exclusive events, 1 P(AUB) = P(Al+P(B = 5 a slo “ae 6 36swe Tr U.2u-U.U$ = 1-00.45 = 0.55. Example 29. From a sales force of 150 persons, one will be selected to attend a special sales meeting. If 52 of them are unmarried, 72 are college graduates, and 3/4 of the 52 that are unmarried are college graduates, find the probability that a salesperson selected at random will be neither single nor a college graduate. Solution. Let A be the event, “the salesperson is unmarried” and let B be the event, sthe salesperson is a college graduate”. Then we have 3 ol _ 72 4%52 39 P(A) = i50 P(B) = Ts P(ANB = “50 = is0 The probability that a salesperson selected at random will be neither single nor a college graduate is given by P(A B) = 1-P(AUB) = 1-{P(A) + P(B)- P(An Bh} [$2.72 39\ | 13 150150 150 30°Example 35. The probability that a management trainee will remain with a company is 0.8. The probability that an employee earns more than % 20,000 per month is 0.4. ‘The probability that an employee, who was a management trainee and remained with the company or who earns more than % 20,000 per month is 0.9. What is the probability that an employee earns more than % 20,000 per month given that he is a management trainee, who stayed with the company. [C.A. (Foundation), Nov 2000} Solution. Let us define events A and Bas follows : A: anemployee, who was a management trainee, will remain with the company B : anemployee earns more than % 20,000 per month Then we are given : P(A) = 0.8 P(B) = 0.4 P(AUB) = 0.9 P(AN Itis required to find P(B /A) = ~ p(ay have . By addition theorem of probability, we P(AN B) = P(A)+ P(B)- P(AUB) =0.8+0.4-0.9 = 0.3. P(AOB) _ 03 P(B/A) = P(A) ~ 08 = 0.375. looility that a person stopping at a petrol pump wil} ceed 0. ae Bei hee he will get his oil checked is 0.29 probability that he will get both checked is 0.07. ( What is the probability that a person stopping at this pump will have neithe, his tyres checked nor oil checked. aay Ret his aNd the : ili i ed, will also have hi Find the probability that a person who has oil checked, is tyre o ree [Dethi Univ. B.Com. (H) 2007, 2013) Solution. Define the events A and Bas follows : A : aperson stopping at the petrol pump will get his tyres checkeg B : aperson stopping at the petrol pump will get his oil checked Then we are given : P(A) = 0.12 P(B) = 0.29 P(AN B = 0.07 () The probability that a person stopping at the petrol pump will have neither his tyres checked nor oil checked is : P(AOB) = 1-P(AU B) =1-{P(A) + P(B)- P(An By} =1- {0.12 + 0.29 — 0.07} = 0.66 (ij The probability that a person who has his oil checked, will also have his tyres checked is : ~ P(ANB) _ 007 7 indies: P(8),,...7 039. ~_ 50:ple 45. Three marksmen can hit the target with probabilities 2, 3 and 4 respectively. They shoot simultaneously. What is the probability that thee abe exactly two hits? [Delhi Univ. B.Com. (H) 2004] Solution. Define the following events : A : first marksman hits the target B: : second marksman hits the target C : third marksman hits the target ‘The events A, Band C are clearly independent and P(A) = P= 3 PIQ= Nie ole wl wl = P(A) = P(B) = P(C) = The probability that there are exactly two hits is given by PAN Bn C)+ PAN BOQ + PAN BAG - = P(A) P(B) P(C) + P(A) P(B) P(Q + P(A) PB) PQ) Ho 2 cl 1h Bear 4. 223, A - 4 3 q + 2°93 4fe 48. A speaks truth in 60% of the cases and Bin 70% of the cases. In what aajtage of cases are they likely to contradict each other in stating the same fact? % [Dethi Univ. B.Com. (H) 1999, 2019] solution. Let E be the event “A speaks truth” and F be the event “B speaks truth” and Fare clearly independent, and ‘then E = 60 70 P(A) = Tog and PI = 365 i = 1-89 _ 40 A - 1-20 . 30 2 P= I-i99 = 700 amd PIF = 1-769 = i00 The event “they are likely to contradict each other in stating the same fact” can occur in the following mutually exclusive ways : () Aspeaks truth and B tells a lie (i Atells a lie and B speaks truth P (they contradict each other) = P(i) + P(i) = P(E F) + P(EN A) = P(B P(E) + P(é) P(A = oe ee = B38 = 8 Hence, in 46% of the cases, they are likely to contradict each other in stating the same fact.(ple 53. From a city population, the probability of selecting (i a male or a smoker £710, (i) a male smoker is 2/5 and (iti) a male, if a smoker is already selected is / on : » 3. Find the probability of selecting (a) a non-smoker, (b) a male, and (o) a smoker, 2, Be male is first selected. [Dethi Univ. B.A. (Econ. Hons.) 1986] solution. Define the following events : A: amale is selected B: a smoker is selected Zz 2 2 Wearegiven: P(AUB) = ig PIANB = § PIA/B= 3 (a) The probability of selecting a non-smoker is P(B) = 1-P(B).To find P(B), we have P(A/B) = ~pa) PI) = “p(aye) ~ 2/ 5 ieee P(B) = 1-P(B)= 1-3 = =. (0) The probability of selecting a male is P(A) = P(AU B)+ P(Ad B) - P(B) (Addition theorem) 10°: (q The probability of selecting a smoker if a male is first selected is P(ANB) 2/5 _ 4the above formula for finding F wet 1 — : D Bayes’ theorer Se Bayes’ Theorem. Let Ay Aas An ee tition of the sample space 9 6%» re a oa Porn in conjunction with only one Of occur i c §=1, 2,05 oe uitional probability of occurrence of A; ( by tually exclusive events such that they forma UA, = S. Let E be an event so that E cay FAN Ay vy Ay and P(E) 40. Then the J) given that B has occurred, is given p(A)P(E/A FQ hn TE TAs) +4 P(An)P(E/ An)’ = P(A) P(E/Ai)- Ag) ++ P(An)P(E/An P(A// 2) = p(A,)P(E/ Ai) + P(A2)P(E/ A2 : | Remark 1.A Seen approach using tree diagram, helpful in conducting the Bayes ‘emark 1. theorem calculations, is shown in Fig. 6.6. E . PIEIAY) tay). PLEIAY) P(Ag). P(EIAg) P(EIA, — few 6 P(An) - P(EIAn) Fig, 6.6. Tree diagram for Bayes’ Theorem Remark 2. In Bayes’ formula, the probabilities P (A,), P (Ag), ..., P (A,) which were already known before performing an experiment are referred to as a priori probabilities~ that is, “before - the ~ fact” probabilities. The probabilities P (A,/5, P (A,/2), .... P (A,/E) which are calculated in the light of sample information are referred to as gsposteriori probabilities, or “after-the-fact” probabilities. Thus, Bayes formula gjy€6 us a technique for computing a posteriori probabilities. Let usfiscuss the following examples to illustrate Bayes’ theorem. imple 62. A company 0% of the scooter and PI: are of standard quality quality. A scooter is pick is the chance that it ha: has two plants to manufacture scooters. Plant | manufacture lant II manufactures 30%. At Plant I, 80% of scooters produced and at Plant II 90% of Scooters produced are of standard ed up at random and is found to be of standard quality. What S come from Plant Il? [Delhi Univ. B.Com. (H) 2002, 2011] Solution. Let us define the following events : mel A, : Scooter manufactured by Plant I A, : Scooter manufactured by Plant II E : Scooter is of standard qualitycepts 6.45 pability CO” goat e events A, and A, are mutually exclusive a on ree sive and from the data given in the P(A,) = .70 P(A,) = 30 P(E/A,) = 80 PiR/ Aj - 90 ooter chosen is of stan s of standard qualit ality, we want to find th o find the probability t the sc given tha tpatit came from Plant Il ie., we wish to find P(A, / E). Using Bayes’ f 1 sing Bayes’ formula, we get P(A,/ = aA P(Ap) P(B/ A2) 1) P(E/Ai) + P(A2) P(E/ A2) (€30)(90)2 = 2 0.27 T7080 = 0.325 = (-70)(.80) +(30)(90) ~ 0.83 d that hat the information concerning various probabilities in the pote. 1 m8Y be note Move exe ‘may be summarised in a table as shown below Computation of Posterior Probabilities Prior Conditional Joint Posterior (revised) Probability Probability Probability probability PIA) P(B/A) Piaod eure al 3) (4) = 2)* 2) (5)=_@/P 8 0.80 | 0.90 | Wee nee \ | L ———_——_— ccample 63’An insurance company insured 2,000 scooter drivers) 4,000 car drivers and 6,90 truck drivers: The probability of accident is 0.01, 0.03 and 0. 15 respectively. One of the insured persons meets an accident. What jis the probability that he is scooter driver? [Delhi Univ. B.Com. (H) 1987, 2005 (Modified)] Solution. Define the following events " An insured person is 2 scooter driver 1 A, : An insured ured person isat person is a car driver ruck driver ‘A, : An ins Th EB: An insured perso” meets an accident en we are given the following probabilities pia) = 2000. 1 4000, 1 _ 6000. 1 Pie 4) = Bo e. Peas) = 12000 - 5 PUA) = 72000? ne moon P(E/ A) = 0-09 Pes As) = 0-15 scl that an ins' gon is a scooter driver given that he meets an is P(A, / B). Usi remula, we have —————tlFundamentals ol 6.46 Of Business Sg, P(A)P(E/A) P(A, / ) = B(Ay) P(E/Ai) + P(A2) P(E/ Aa) + P(A3) P(E/ 3) 6x00 au Teac runiie sea oar 3X 0.08 + 5 0.15 Example 64. By examining the chest X-ray, probability that TB is detected whe, person is actually suffering is 0.99. The probability that the doctor diagnose incorrecy, that a person has TB on the basis of X-ray is 0.001. In a certain city 1 in jog, persons suffers from TB. A person selected at random is diagnosed to have TB, Wha, x0.014 is the chance that he actually has TB? [Delhi Univ. B.Com. (H) 1986) Solution. Define the events : A, : Aperson has TB A, : Aperson does not have TB E : Aperson is diagnosed to have TB Then we have 1 P(A) = ZO99 7 0-001 P(A.) = 1-P(A,) = 0.999 P(F/A,) = 0.99 P(E/A,) = 0.001 We wish to find P(A, / 5), the probability that the person actually has 7.B. given that he is diagnosed to have T.P. By Bayes’ formula, Hae P(A;) P(E/ Aj) P(A\) P(E/A\) + P(A2) P(E/ Ap) P(A,/ = _ ___(0.001)(0.99) © (0.001)(0.99)+(0.999)(0.001) = 0-499.75 BINOMIAL DISTRIBUTION 2 shall discuss one of the most important discrete probability robability distribution. This distribution was first hematician about 1700 and, for this reason, the s a Bernoulli distribution. ematical model in which av formance of the experimen +t of assumptions whic! In this section w distributions, calles studied by J. Bernoulli, distribution is sometimes The binomial probability distri experiment is repeated several times, Ww! called a trial. This model is developed un are described as follows : rials is finite and small. possible mutually exclusive outcomes ineach trial, called su d the binomial p' a Swiss Mat! referred to ai ibution studies a mat! where each per! der a very specific se () The number of t (i). There are only two ess and failure. (ii) The probability (iy. The trials are independent. That is, outcome of any other trial. e binomial probability mo ‘Any random experiment for which th ailed 2 Bernoulli trial. A common example of a Bernoulli trial i of success remains constant from trial to trial 1 the outcome of any trial is independent the del is appropriate | 5 coin flipping: FY J |y 7.16 Fundamentals of Business sig, stig , P(X=7) =8C, (0.4)" (0.6)5-", re 0,1,... 5 () Required probability = P (none will graduate) = P(X= 0) = (0.6) = oor (i) Required probability = P (one will graduate) = P(X= 1) = °C, (0.4) 065 =5 * 0.4 x (0.6)* = 2 x (0.6)* = 0.2592 1 - P (none will graduate) = 1-0.07776 = 0.92224 quired probability = P(X = 5) = (0.4)° = 0.01024. mple 22. A bag contains 5 white, 7 red and 8 black balls. If 4 balls are dra, jy one with replacement, what is the probability that () none is white? (i) all ay, white? (ii only 2 are white? i (ii) Required probabilit (using par (iv) : WN One Solution. Let X denote the number of white balls drawn. Then the possible values of Xare 0, 1, 2, 3, 4. Now p = Plawhite ball is drawn)= 2 = 4. q=1-p=i- Using formula for Binomial distribution, the probability of r white balls is 1 4 q\' (3\io% Pik=n = 40, (3) (3) hn. 01142, 3,4, ty it) P(none is white) = P(X=0) = 4C, ayey -(3)' - x (a P(all are white) = P(X= 4) = *C, (2)'(3y = ay’ -s (ii P(only 2 are white) = P(X=2) = *C, Gy @y - Oxi6 "ig -BFundamentals of Busing, SS "Otis 7.24 7.8 FITTING OF BINOMIAL DISTRIBUTION ndom experiment consisting of n trials and satisfying the og ia e n Suppose a rai .d Ntimes. Then the frequency Of x succesga,' : Is - Bie, binomial distribution is repeate by the formula : eet ; fl = NP(X= x)= Nx P(x) = Nx CP x=0,1, 7 Substituting x = 0, 1, 2, ..., mwe get the expected or theoretical frequencies 7 \ the Binomial distribution = ; , NC, pq'!, N "C, pear’, ,N p” Na", These expected frequencies can easily be obtained if p, the probability of fell known. However, if p is not known, then first we find the mean of the given frequer,. ~ . LX distribution by using the formula X = “Sv and equate it to np, the mean of, 7 x binomial distribution. That is, np = X = p= {> and hence q= 1 ~ p. With the, p and q, the expected or theoretical Binomial frequencies can then be obtains q. (1). imple 38. Fit a binomial distribution to the following data : Kets oO 1 2 3 4 cata. ae 62 46 10 4 [Delhi Univ. B.Com. (# 2011) Solution. In terms of usual notations, we have n= 4 and N= Lf = 150. The mean of the given distribution is : go . LIK _ 0x28+1x624+2%464+3x1044%4 ee op = 150 200 _ 4 150 ~ 3 If ‘p’ is the parameter of the binomial distribution, then np = mean of the distribution = $ wl 4 > 45 pe 2 and = pe jederse 2 Cirle Be 3 Using the formula for binomial distribution, the expected binomial frequencies are Riven by: 4 ) Et e643 3,4 F(X) = NP(X= x = 150 x#0(2)"(7.25 ‘es(t)’ 2 - tex(4).(3)’ 7 3 = 0.3950 50.25 = 59 2 te,(3) 2)" ~ SP = 02000 44.43 = 44 3 *e,(3) (3) o ~ = 0.0987 14.80 = 15 + | se. QP= & - oo bac Hence the fitted binomial distribution is : x 0 1 2 3 4 fee 30 59 44 15, 27.31 pst per of accidents at a busy traffic intersection during a certain part of the we num ould count the number of accidents that occur during that period of time fy Ne cgagt is a success since i's what we are looking for), but how could we ete She number of NON cidents. Thus, in this case the number of possible caunt PE accesses + failures) is difficult, if not possible, to determine and hence gyeomes (SM ribution cannot be applied. Fortunately, the Poisson distribution can mia uch as above, without knowing the total possible outcomes probability Distributions te “eunibution was derived in 1837 by a French mathematician Simeon Denis poseon is 1840). The distribution is used to describe the behaviour of rare events pasen (numberof accidents on road andl has been called “the law of improbable uch a eerts nts sxperime! puteomes OCCU} which yield numerical values of a random variable X, the number of ring during a given time interval or in a specified region, are often sng gon experiments, The given time interval may be of any length, such 29 a caled Pojay a week, @ month oF even a year. The specified region could be a line eae ier ea or even a piece of material. Examples of Poisson experiments are ware numberof telephone calls per hour received by an office. ib) The number of typing errors on a given printed page. ig The number of bacteria in a given culture ig) Te number of cars arriving at toll gate during a fixed period of time ig The aval of customers at the checkout counter of a supermarket in a given time interval {f The number of deaths in a district in a given period, say, a year (g The number of defective material such as blades, pins, screws, etc., ina packing manufactured by a good concern. (t) The number of births resulting in twins during a year in a certain city. Arandom variable X representing the number of outcomes (or successes) occurring l¢Posson experiment is called @ Poisson random variable and its probability sorption 3s called the Poisson distribution. The derivation of the Poisson oo in ne beyond the scope of this text. However, we state the result (without following definition. isson taste tribution. Let X be a Poisson random variable representing the number Ptbabity thon successes) occurring in a given interval of time or space, then the at exactly r successes occur in a given interval be given by “ine P(x=y = 2 p= 0,1,2, 2s the ay rr? “Nal Ay erage number of outcomes (or successes) occurring in the given ny ray 2 é + tibutign, "dom variable with this distribution is said to have a Poisson "Sdescribe eee the Poisson distribution is a discrete probability distribution Single parameter 4. To apply the Poisson distribution, we needASTM t.Schi2).01 + 0.139,*,5.".0.325. \_ Baemte 42. Suppose that the number of claims for missing baggage average 6 p: day. Find the probability that on a given day, there will be (i) no claim, (i) exac: claims, and (ii) at least 2 claims. (use e = 0.0025). [Delhi Univ. B.Com. (H) 2001 Solution. Let X be the random variable denoting the number of claims for mis baggage on a given day. Then X follows Poisson distribution with 2 = 6. Thus the probability that there will be exactly r claims on a particular day is given by -66r Pik=) = 2) = 1- P(X< 2)=1-[P(X=0)+P(X= 1] =1-[e* + e©6] = 1- e8 x7 =1-0,0025*7 2 = 0.9825.4! ITING CASE OF THE BINOY, n fia) = ore RIBUTION AS A LIMI a binomial random variable and if the Probatj., endent trials, " 7.10 THE POISSON DIST! DISTRIBUTION Recall from Section 7.6 that if Xis r trial is p, then for nindep = "Cpa" at the distribution of X can be approximated j, of success on any Lor ix=4 p. It can be seen t where q= 1- Poisson random variable under t nt Ge ie (9 nis very large, (i9 pis very small, ie, p>0 y~ isconstant .—~ ple 46. A company knows on the basis of past experience that 3% of its bus g Poisson distribution, find the probability that in a sax: foduced are defective. Usin{ of 100, none is defective. (use e? = 0.04979). Solution. With the usual notations, we have p = 0.03, n=100 sothati=np=3 Let the rand. i iom variable X denote the number of defective bulbs. Then X follovs he following conditions : Poisson distribution such that -h purey = 22M aia tei P(X=0) = o!84 PROPERTES OF NORMAL DISTRIBUTION We list the following properties of normal distribution 1. The normal curve is bell-shaped and is symmetric about the vertical line through the mean. In other iOrds, the mean isa the center. See Fig. 8.1 : 2. The normal curve never touches the x-axis and extends to infinity on either side of the mean. In other Words, the normal curve approaches the x-axis asymptotically direction away from the mean. normal curve and above the as we proceed in either 3. Irrespective of the shape, the total area under the horizontal axis is always equal to I. 4. No portion of the normal curve lies below the x-axis 5. The normal distribution is unimodal, the only mode occurring anX= GHA has its mean, median and mode at the same point. "eae pedian =Madk 7. The area under the normal curve bounded by the two ordinates X= @ Oe na equals the probability that the random variable X assumes avalue between X =aandx=b omExafple 4. It is known that amounts of money spent on textbooks in a \_--Students of a particular university follow a normal distribution with mean %380 year standard deviation 500. () What is the probability that a randomly selected student will spend less thax % 4000 on text-books in a year? (i What is the probability that a randomly selected student will spend more th= % 3600 on textbooks in a year? (ii) What is the probability that a randomly selected student will spend betwee % 3000 and % 4000 on textbooks in a year. [Dethi Univ. B.Com. (H) 2013 Solution. Let X be the random variable denoting the amounts of money spent on '* books in a year by students. Then Xis normally distributed with mean y= 38002" s.d. o = 500. Converting X into standard normal variable Z by means of the transformation gu Xou _ X= 3800 o 500 » we have () P(X < 4000) = oz < 4000-3800 500nat visrbution Fy : - P(Z<0.4) 05+ PO~0.4) 795+ PCO4<7 2g = 05+ PO 1.2) = 1-P(z< 1.2) a Oh = 1-0.8849 = 0.1151 md ; m qi May ae number of units having weights more 'e~ a . x 2000 = 230 units (apP-)Fundamentals of Business Stati, tics me 1 Decision making under certainty < Decision making under uncertainty < Decision making under conflict g Under Certainty oosing one action from ai 1. Decision Makin: naker is faced with ch when the outcome oO} is said to be faced wit certainty entails t! action will yield. not discussed in this text, but here is a simp feach alternative action 7S completely knoy, h decision making with certainty, 7}, he selection of a course of action when y, If a decision ™ alternative the decis decision making under know the outcome each alternative Decision problems of this type are example of this type of problem. Example 1. A company makes two types of furni % 20 per chair and % 30 per table. Both products are processed on three Machines jy, M, and M,, It is known that each chair requires hours on Machine M,, 5 hours o: Machine M, and 2 hours on Machine M,, while each table requires 3 hours on Machine M,, 2 hours on Machine M, and 6 hours on Machine M,. During 1 week the company has 36 hours available on Machine M,, 50 hours available on Machine M, and 6 hours available on Machine M,. How many chairs and how many tables should the the week in order to maximize profit. tures, chairs and tables. Profits are company produce over Many problems involving decision making under certainty, solved using a technique known as linear programming. including this one, are 2. Decision Making Under Uncertainty If a decision maker is faced with choosing one action from among two or more alternative actions, and each of these has possible outcomes that depend on the chance occurrence of one of a set of mutually exclusive and collectively exhaustive states of nature, the decision maker is said to be faced with decision making under uncertainty a eecion making under uncertainty involves the selection of a course of action en we 1 wit is — paeomE 1 ol iow with certainty the results that each alternative action ye. BSEIeaeEe aie ASSIS that the outcome of whatever course of action-the decision ae cted only by chance and not by his opponent or competitor. e below some examples of i: isi ede under ypeértainty. i problems where business decisions must be ™ ea a 2 e PaEsEaDeH vender buys newspapers at % 3 each and sells them att * ch, Any papers not sold by the end of the day are completely worthless to Mil . but not too many, eee ra i oe newspapers to buy : enough to make a good me ni , to avoid loss due to i ¥ decision problem in the face of incertae The newspaper vender is faced Exa i laine 2 4 company wishes to know whether an increase in the amu in a toothpaste will increase thi "a sales. If the flavouring is increas?sion Theory 93 pecis increase a cis ao not sncrease appreciably, the company will lose money d oer decides bt linea the flavouring when in fact it should hare eed ential sales. 1e ii i “Hy company is faced with a decision problem in the fx Ea face of i] company wishes to et ‘0 drill fo: ina iple 4. An oil company wishes to know whether t pe tion. How likely is oil there? How much? How ae will fen ea 2 di Aa z i rill? (cal gxample 5.A saa firm wants to invest in securities. The problem of the firm is to determine ne = the market sectors and individual securities with the best pects? Where is the economy headed? How about interest rates? How should pros} . these factors affect the investment decisions? 3, Decision Making Under Conflict Decision making under conflict is similar to decision making under uncertainty in that we do not know with certainty the result each available alternative course of action will yield. However, the reason for this uncertainty is different in the case of decision making under conflict. In fact, in decision making under conflict, the outcome of our chosen course of action depends on decisions made by our competitors or Spponents. Decision problems of this type fall under the discipline known as gare gory. The discussion of game theory is beyond the scope of this text, however, an is type of decision problem is illustrated as follows. example 9 fe 6. A television manufacturer is currently facing a decision about the price rs will ofa TV it makes. If the company sets the price too high; potential custome: he other hand, the price is set too low, the competitors profits. This example concerning purchase competitor's TV. If, on t demonstrates decision making will also drop their prices, thereby reducing everyone's the decision of whether to change the price of a TV under conflict.M9 OF MUSINESS Stata combination is called the outcome or payoff An e may b ~eEithee negative Note. It may be noted that a payoff is a ‘conditional value’ because its value de on the event that occurs and the action which is taken. Pend Payoff Table (or Matrix). A payoff table is a table showing conditional payoffs res, from all possible action/state of nature combinations. The general form of q p,,8 Table for a decision problem involving m possible acts Ay, Ag, A, and n Stater nature S,, S), .... S, is shown in Table 9-1 me TABLE 9.1. Payoff Matrix wa “State of Natu >————— Action Ss, Ss, A, Oy Or A, On, On An Ont Ong decision maker has chosen is associated with Each of the set of possible actions the : c ‘ a row of the payoff matrix. Each state of nature is associated with a column, The numbers in the table are .ne outcomes (or conditional payoffs) of the decision problem, For example, the number Oijis the outcome resulting from the combination of the ith action with the j th state of nature. aking problem, we should first write down all the In order to analyze a decision-m: possible acts and their corresponding payoffs. This is demonstrated in the following example on inventories. Example 7. A retailer has space for four units of a particular perishable product per week. The cost of a unit is € 30, the selling price is ¥ 50 and the profit is € 20. Any units not sold at the end of the week are wasted. The retailer does not know the demand, but must nevertheless decide on the number of units of stock. Write down all the possible acts, states of nature and their corresponding payoffs. Solution. Tabie 9.2 shows the payoffs resulting from all possible combination. of actions (no. of units stocked) and events (no. of units demanded). TABLE 9.2. The Payoff Table Act Event (No. of units demanded) Z (No. of units stocked) 0 1 2 2 * 0 20 0 0 0 | 1 -% 30 20 20 20 20 | 2 -% 60 -10 40 40 a f -7.90 -40 10 60 aa00 |2 TREOTY 8 eo" ty Loss (or Regret}. The opportunity loss is defined opportu ayoft a decision-maker received fora chosen peter tt on-maker could have received for choosing the lect ate of nature that occurred, pee as the difference action and the maximum action Fielding the highest he stat e, suppose the retailer in Example 7 had stocked two units, but that four -manded. His actual profit from selling the two units which he had is @4q dl stocked four units he would have made % 80 profit. Thus his opportunity loss 0-40) oF 40. ; se it may be noted that opportunity loss can never be negative. The loss for the note I Mole act is zero. For all other acts the loss is positive. oe Pe nity Loss or Regret Table. An opportunity loss table is a table showing ee losses (or regrets) due to imperfect information. PP xt show how to construct the opportunity loss table from the ayoff table bi We next shi pay y epeating the following two-step procedure : step 1. Find the maximum payoff for a given state of nature. The opportunity loss associated with this payoff is zero. step 2. The opportunity loss associated with any other payoff for this state of nature is obtained by subtracting that payoff from the maximum payoff from that pay! state of nature. Example 8. Construct the opportunity loss table for Example 7. Solution. The payoff table for Example 7 is reproduced here for ready reference. TABLE 9.3, The Payoff Table *~ CC" we | Act Event (No. of units demanded) | te of units stocked) 0 1 2 3 4 0 o* 0 20°70 9 Ape 0 9 1 - 300) 20" 2o-Ze 20 20 20 2 ~60 -~10 20-4 ie) 40* 40 40 3 -90 -40 10 60* 60 4 - 120 -70 - 20 30 80" We derive the pportunity loss table with the following two-step procedure : Step 1. For each column of the payoff table (Table 9.3), mark the maximum payoff oi 4 star («). The opportunity loss associated with each column maximum is Step 2. The opportunity loss table is constructed, column by column, by subtracting €ach value in Table 9.3 from the starred value in its column. The resulting °Pportunity loss table is shown in Table 9.4.Fundamentals of By. sine. 9.6 TABLE 9.4. The Opportunity Loss Table Ace Event (No. of units demanded) (No. of units stocked) 0 | 2 2 = 0 0 20 40 4 ; 30 0 20 a 2 60 30 ° 3 90 60 30 4 120 90 60Pome 11. Given the following payoff matrix : Payoff (in @) a Act State of Nature Cold Weather Hot Weather | Sell Cold Drinks 50 1007S 7eqes Sell Hot Drinks 120 40 and given the probability of weather being hot is 0.8, set up the opportunity loss tab and compute the opportunity loss of each action. Select the best act. [C.A. Foundation, Nov. 1996 Solution. The opportunity loss table for the given payoff matrix is shown in Table 9.8. TABLE 9.8. The Opportunity Loss Table State of Nature Act Cold Weather Hot Weather ‘Sell Cold Drinks | 120-50= 70 100 - 100= 0 Sell Hot Drinks ____120-120= 0 100 - 40 = 60 Probability 1-0.8= 0.2 08 | Computation of Expected Opportunity Loss (EOL) Expected Opportunity Loss for the act ‘Sell Cold Drinks’ = 0.2 x 70 + 0.8 x 0= 14 act ‘Sell Hot Drinks’ = 0.2 x 0 + 0.8 x 60= 48 Expected Opportunity Loss for the . Since expgeted opportunity loss is minimum for the act ‘Sell Cold Drinks’, therefo® to EOL criterion, the act ‘Sell Cold Drinks’ is the optimum (best) act. i ee nae” ta ee asxrned Falve of Perfect Information (EVP!) to unddestand the concept of expected value of perfect information, also referred to ae BVPI, let us consider the preceding example again. ‘The optimal stock action by the fruit seller in the preceding example was to stock 12 coses each day in order to get the highest possible average daily profit under the ‘conditions given. This stock action was obtained under uncertainty using the expected sayoff criterion. However, as long as uncertainty exists, there is the possibility that the expected payoff criterion may lead to the wrong course of action. Now suppose that the fruit seller in Example 18 could remove all uncertainty from his problem by obtaining complete and accurate information about the future, called the perfect information. This does not mean that sales would not vary from 10 to 13 cases per day. Sales would still be 10 cases per day 15 percent of the time, 11 cases 25 percent of the time, 12 cases 40 percent of the time, and 13 cases 20 percent of the time. However, with perfect information, the retailer would know i advance how many cases were going to be called for each day. Under these circumstances, the fruit seller would stock today the exact number of Dizes buyera will want tamorrow: When sales were going to Be. ea seler would stock exactly 10 cases, thus realizing a profit of < 598 Baden cases, the fruit seller would stock 11 cases and realize @ profit of € 3300, an: ate Tkble9.17 shows the conditional profit table with perfect information ae ‘othe fruit seller’s problem.Sant 19. A magazine vendor has assigned the fopowjne probabilities ty dofnand for a weekly magazine : Gee by, Su Copies Demanded: 10 u 12 13 14 (n) th Probability 2 0.1 0.3 0.3 0.2 0.1 A copy of the magazine sells for € 10 and costs € 8. An unsold copy can be returteli 5. How many copies of the magazine should the vendor buy? Find his exe! profit also, [Dethi Univ. B.Com. (f) 2004,an THEO 9.17 isi : ne wren (Be number of copies of the magazine purchased (or_stocked) Pet jutior vce and the weekly demand of the magazine is an event, Le, state of week Since the cost price of each magazine is @ 8 and its selling price is 710, the ature: © vendor m a profit of €10— for 2.on each magazine sold. However, + Snot sold within the week, it can be returned for @ 5 and ve incur a loss of & 3 on each unsold copy. ae nage og the ™: if ae vendor i ner gional payoffs. Let the magazine vendor buy (or stock) m copies of the magazine at Pade demand of the magazine be r copies, Then the vendor's profit, ie., i P (m,n) can be obtained as explained below : the demand is more than the stock. z ve demand Se this CaS all the stocked copies are ‘sold out. Hence the vendor's profit is given. by t P(m, n) = %2m, n>m a) g.nm P(mn) = j00 Z hee, ifn = > Fundamentals of Business sap 9.18 Expected profit when stock is 11 copies = 3[0.1*17+0.9 x 22) =%21.5 Expected profit when stock is 12 copies = F[0.1 x 14+0.3 x 19+ 0.6 x 24] =%21.5 Expected profit when stock is 13 copies = %[0.1 x 11+ 0.3 x 16 +0.3 x 21 + 0.3 x 26] = 220 Expected profit when stock is 14 copies = 70.1 8+0.3 13+0.3« 18+0.2 «23 +0.1 * 28) =@i75 is 11 or 12 copies, therefore the Since the expected profit is maximum when stock or 12 copies. His expected profi optimuny$tock action for the magazine vendor is 11 is®2jon THEOTY p 0. «businessman has two indepe, 9.35 le i ‘Pendent inye: pte Ricks the capital to undertake both of then” bes capital of 30,000 and investment 19 tagre A boat igium and low demands with corresponding prob OD eee eee OO tid 0-3 0-4 and 0.3 for inves spec estment A are & 75,000, €55,000 and & 35,000 and op stnent B. Returns on ent Bate likely t0 be 1,00,000, 880,000 and 7o,06e intiemand respectively. What decision should the company ee ioretracting an appropriate decision true. [Delhi Univ, pee aa pie by sotution. The decision tree based on the given data is depicted in pale (SOL) Acts Events Pay off (2) Expected pajee my (Prob. « Payor 75000 - 30000 = 45000 18000 (UGeu» xo. soe se Rew) wont 55000-20000 = 25000 10000 ( 25e7D Xo" 4) 000-1000 “S23 36000 - 30000 Total [25000 EM-Tot =f 100000 - 50000 = 50000 15000 80000 50000= 30000 12000 Fig. 9.2. Decision ‘Tree i having constructed the decision true, we now use roll back technique 10 determine a odes Band les ‘ein by considering the rightmost column of nodes, namely SE the - ach chance node, we calculate its expected monetary value ( See ee thegg Pe°ted payoff of each branch by the probability Products. ‘thus of that branch an‘Fundamentals of Business sia, EMV (node B) = 0.4 x 45000 + 0.4 x 25000 + 0.2 = 5000 = 29000 EMV (node G = 0.3 x 50000 + 0.4 x 30000 + 0.3 x 20000 = 33000 These expected monetary values are placed above these nodes as shown in Fig g., 9.36 Next, moving one column to the left brings us to the decision node A. At de. node A, we compare the EMVof its branches and choose the alternative whose by. has the highest EMV. Thus " Atnode A: Investment A has EMV = 29000 Investment B has EMV = 33000 33000 > 29000, so choose the Investment B alternative. The EMV of 33000 is recorded over the node A and a double dash is inserted to indicate the rejected branch as shown in Fig. 9.2. The optimum decision is to choose investment B._——= 7 SOTOTTES 7524-3130 vivOAVe —veewd Us aa. ‘24. Find Karl Pearson's coefficient of correlation between th, e Example f playing habits of the people from the following information : Be ang i; [Age groups (im years) ‘No. of people * 0 Of Playe 15 and less than 20 200 {Pye 20 and less than 25 270 ay | 25 and less than 30 340 a | 30 and less than 35 360 be 35 and less than 40 400 aed 40 and less than 45 300 ee {Delhi Univ. B.Com. (P| 2012, B.Com. (H) 2014 (SOL ea solution ts required to find the coeficent of correlation berween the ager slaving habits. We must, therefore, express the number of players in Eat d percentage by applying : No. of players eS ei pleyels’ = “Nec of persia Let this be denoted by Yand mid-point of class interval by X. Calculation for Coefficient of Correlation ‘Age group |Mid-point | No. of | No.of | %of ¥s we) eilw finyears) | x | people | players | players ¥ is eniiess|f17'5 |; 200°"). 150 388x100 a) 8 9 | 2 |-5 than 20 = 75 20 and less| 22.5 | 270 162 |392 «100 -2 2 a-- [pans] eat than 25, 60 | Bee trealia7s | 340° |" 170°" |Sg57100° |= 0 eee than 30 = 50 Baraca! 226° | 360 |” 180° |5an*100 0 0 oF | 2 e | than 35 = 50 ee Vie’ | sco |i2ci0 |. |) -2 | ee than 40 a 45 a 40 andiess| 42.5 | 300 | 120 |420x100 2 -2 Olas than.45 = 40 a Toor lav? | mo ey [Ee 12 | S =19 | 45 |e— uur sptinear COTTE snp! con's coefficient of correlation between the 7 ars | Pet wa the playing habit 3 is cen BY ne UV ~(Xu)yvy 6 a a £(-24)-(-3),.4 r* T5u2-(xup. nYV? (py Yer 97 Vx 34 ~(4)2 144412 ~132 14-9. 204-16 ~ Vi0S. Jigs ~ 0.9395, * Wiis. pindicates a high degree of negative Correlation between the age and the playing which 4 habits.g table gives a bivariate frequency distr; \3 cyees of a company according to age (in years) and pay (in rupees) : - ~w wiuu Wwe nleIp of following examples, Example 30. The followin, ‘bution of 55 ‘Age Pay (in) (Years) 25,000 ~ 30,000 | 30,000 - 35,000 |35,000 - 40,000 40,000- 45005.) | 20-30 8 3 Ss 3 \ 30-40 2 5 2 2 | | 40-50 - 2 9 6 | 50-60 - - 5 6 [ethi Univ. B.Com. (#9 2013ie ww inear Correlation she Le sino 10.31 ‘ent of ton ‘rrelation gota! (es rep = |asano 30000. 400007 in €)_{ 30000|35000 40000] nanny Mid-Value = ¥__|27500|32500|57500| 4509 ae an [ye on =H ° 1 2 f fu | pe | i | Pees jee 8 S| Lo Ta] tse ii sm |ics oH Heel |i |e 0 oO | Lo | [30-40] 35 8 e s 5 : . : : ; [30-9 0 o! Ll a Ete aes s: | 17 | a7 wha | 0 ol Lag Bal 60] 55 2 = “ = > 50- fo -10 [| o | 16 | 28 [ap Ve = 34 A Eno |For os ' 36 |Z fur Jo fuv 8 2 a We 50 x 63 - 28x 34 = 50x72-28?.V5082-34 NE fw (EAYEL) 8 Uy pe (5 ANE? (E40) 2198 2198 - 0.763 5 = = 9879.29 Eso Taos * Ties * 2 3600-784 . E-- 19 sepeat per of tures ovsvins vt 18 TeDeated and so on.” “Pete, m, represents the sum late Spearman’s i i iple §1. Calculate Spx Coefficient of Tank correlation from the following Be sy’ 16 24 465.16. 16 Soest, 83 4g pi-19.---6~ 9. 20.4" iss age 13 i rane y [ethi Univ. B.Com. (p) 1997, m the highest value in both the € repeated, so ranking is done in tion. The following table shows the ranking fro ” Moreover, certain items in both the series ar sees. " srordance with suitable average r Calquiation of Rank Correlation CqeMficient eee Rank assigned D pb? | xX [Rank aioe Ye an ee oe | ; 19 3 -1 1 | = - 6 8.5 -0.5 0.25 | 16 8 : _ 1 | 4 6 9 . = 1 | 65 1 20 - i: | 6 8 3 19 4 1 | 16 s é 4 “2 16 i 9 16 6 a8 i 40 4 24 aa =05 @ 5 13 aie 2.5 48 3 13 ee ht ston T=10 is repe mes (i.e., m, = 3). In series y, the item i imes (i.¢., ™, = 9)- metatin ies X, the item 16 is repeated 3 ti ee d twice is also repeate “repeated twice (ie., m, = 2) and 6 is ion is given by Jation is given i } S Spearman's coefficient of rank a \_ (v3 eal lms -ms) 6[0? + lm 1)" (soleil al lee n(n?-1) __—_a 8 =l —y Fundamentals of Busines “fers fo-3) 0 ( 02) fea oo 6[41+2+0.5+0.5] _ | 6x44 a “990 990 ~ 15ee 18. From the data given below, find: () The two regression equations, (i) The coefficient of correlation between the marks in Economics and Statistics (ii) The most likely marks in Statistics when marks in Economics are 30. | Marks in Economics 2 25) 28. 98 92 31 36 - 20m. 3872341 i | Marks in Statistics gie7 43 46149. 41 2 36..-32° 31. 8047335 vay [Dethi Univ. B. Com. (# 1982] Solution. (i) Let X denote the marks in Economics and Y denote the marks in Statistics.~ hy imple Linear Regression si 11.19 [2s 28 = 35 « 32 121 31 9 36 : 4 a 2» 31 -3 “7 _ 36 |-24 38 30 6 o a 9 21 34 33 2 a i las 32 39 0 A ; % {19 1 0 rx xy 2 = 320 = 380 ae Xy? [yay = = 398 |s_93 x - 2X _ 320 Ne Rate Sowa! 82 meee ~ LY _ 360 _ 0 n 10 738 xy _ -93 b. REDS B88 2s O55 id by = L2Y _ -93 _ a Sy? BA898 See Sx? ~ 140 =~ 0.66 Regression Equation of X on Y is ; Regression Equation of ¥ on X is : X-X = by, (¥-¥) ¥-¥ = b,(x-X) or, —- X-~32 = 0.23 (Y- 38) or, ~ 0.66 (X- 32) te, X-32= -0.237+8.74 ie, -0.66X + 21.12 = = -0.23Y+ 40.74 ...(1) | = ~0.66X +5912. ( (ij To find the coefficient of correlation, we have 1? = bey . yy = (0.23) (0.66) = 0.1518 = r= + JOIST As both the regression coefficients are negative, r= - 0.3896 (ii) To find the most likely marks in Statistics (bent in Seen ay ’ We use regression line of Yon X, Thus substituting X= 30 i ao Y = — 0.66 x 30 + 59.12 = - 19.8 + 59.12 = 8 = +0.380611.8 DIFFERENCE BETWEEN CoRR 1ON AND REGRESSION 1, Correlation literally means the relationship between two o vary in sympathy so that the movements in one tend to bean bles Wick corresponding movement in the other (s). On the other hand. seemed by the stepping back or returning to the average value and is a math nme expressing the average relationship between the two variables Correlation analysis attempts “ a ly pts to determine the “degree of relationship” bet variables. On the other hand, regression analysis atte ss et ee ott analysis attempts to establish the “nature of the relationship” between variables, ic., to study the functional relations ween the vari ; ee between t ables and thereby provide a mechanism for prediction or forecastipg. pana or Correlation need not imply cause and effect relationship between the variables under study. However, regression analysis clearly indicates the cause and effect relationship between the variables. —— Jriere may be non-sense (or spurious) correlation between two variables which is ue to pure chance and has no practical relevance such as the height and blood di pressure of a group of people. However, there is no such thing like non-sense 4. regression. 5. Correlation analysis is confined only to the study of linear relationship between the variables and therefore, has limited applications. Regression analysis hasao Fundamentals of Business Stang to determine the average length of life of a certain brand of light bulbs, it w impossible to test every light bulb because the test destroys the light bulb, Exnee be costs can also be a prohibitive factor in studying the entire population, ‘tant The process of selecting a sample from a population is called sampling. The py to allow us to use information gained from the sample to make stater | ‘Men S chapter we of sampling 5 about the entire population from which the sample was taken, In thi. as well as some distributions that s esult consider some common sampling methods from sampling. 14.2 SAMPLING TERMINOLOGY Sampling has its own set of terms it uses. Here is a brief description of these. Population. A populations the collection of all individuals or items that are of intereg, in a study, The population of interest may be finite or infinite. population is one where the total number of elements ig Finite Population. A finite fixed and could be listed. Examples of finite populations include the number of employees of a given company the number of students enrolled in a university, or the potential consumers in target market. Infinite Population. An infinite population has an unlimited number of elements Examples of infinite populations include the number of widgets manufactured by a company that plans to be in business forever, or the grains of sand on the beaches of world. Census. A census is a complete enumeration of an entire population. When the population is very large, it is impractical or even impossible to enumerate the entire population and moreover such a procedure would be too costly in terms of both time and money. A sample, on the other hand, is a subset of a population. It requires only a partial enumeration of the population. Sample. A sample is a portion or subset taken from a population. Figure 14.1 depicts the relationship between a population and a sample from the population. Population SampleVER 143 Tential Statistics is = “en is the process of us Sampling and Sampling Distributions Inferential statistics, Infe} draw conclusions about the sing The major use of infere, Aracteristics of Population. MHal statistics is to u mple results something about a popuigaye inferences anogn 8 froin asamp to igs tained from asan? 8 Population are based on enn obtaine ms een tBeSample ment PEP fon goa res is drawn, that is, it must be a represe, the Population from whieh 1 Native sample. are drawn about representative these conclusions are likely 45 if the sample is iPresentative, inferences abour © are likely to be correct a population from a sample tee is not " gonclusions be erroneous. On the other hana, the population based on the sampl 2 uPle is selected in such a way © less likely to be inchuded thay an be ents in the population of interest e.g., name of individuale Pee 14.3 REASONS FOR SAMPLING Cerned with selecting a sample in order to obtain information about the entire population. To ‘study an entire Population, a statistician may examine all the members of the population, Such a complete examination of all the members is called a census We might ask why we should sample, and thus obtain only partial information, when a census could provide us with complete information. A number of reasons why a sample may be preferred to a census are given below, 1, Infinitely Large Population : The first reason for sampling is that a population may be too large for every element to be examined, When the population is very large, a census is physically impossible. In such 4 situation, the sample is the only option. For example, a manufacturing process that runs indefinitely under constant conditions can generate an infinite number of output units. A census of the entire output would thus be theoretically impossible, ‘ is cheaper than a census. It is obviously ; Reduced Cost : A sample is c! ; 7 hes coat i okies data for a selected subset of a population, rather than the nie ‘ample, the information collected by conducting the nee i of 100 customers rather than with a population of ‘interviews with a samp S i ensive, 1,00,000 customers is less exp 2 cone S, Time, Beleciins\ateomblas so can cave Gone. 60 iio ee in is the sample can ling can provide them os ig eee ie eanes is a matter of urgency, ee ay oe some Stetshe This is particularly useful in situations pam eee ulation data completely. that is almost as years 19 Brasess 1b¢ bee 1 sample provides information that is slmost fo) ROCUTACy JS omc ee cian a cea aaa Canna TOr "accurate as the results obtaine14.4 METHODS OF SAMPLING Although there are a number of different methods that might be used to create 2 sample, they can be divided into two general categories : Probability (or Random) Sampling and Non-probability (or Non-random) Sampling Probability Sampling. A probability sampling is a technique in which a sample is drawn in such a way that every element in the population has a known and non-zero probability of being included in the sample. The sample so drawn is known a. Probability sample (or random sample) There are four common methods of obtaining random samples. These include : simple random sampling, systematic sampling, stratified sampling and cluster sampling (se Figure 14.2) ( Sampling Probability Sampling Non-Probability Sampling Quota ‘Sampling Convenience Sampling Systematic | Stratified |B Cluster Sampling | ff Sampling |f Sampling Judgment Sampling Fig 14.2: Methods of Sampling al Non-Probability Sampling. In non-probability sampling personal Inowlede® opinion are used to identify the elements from the population that are to be inc! in the sample.