Vieasures of Central tendency Measures of Location or Central Tendency or Meastre-oF-tocation gives an idea about the part of the distribution The main objective of Measure of Central Tendency are: © To condense data ina single value. © To facilitate comparison between data Requisites of a Good Measure of Central Tendency (i) [should be rigidly defined. Gi) It should be simple to understand and easy to calculate (iii) Ie should be based upon all values of given data (iv) It should be capable of further mathematical treatment, (¥) It should have sampling stability, (vi) It should be not be unduly affected by extreme values. ‘There are different types of averages, each has its own advantages and disadvantag Measure of Central Tendency Location (positional average) one Average Partition values, Mode Avithmetie Geamettie i Mean Mean Median Quartles_—_Deciles Percentiles, 1, Arithmetic Mean: Case I: Discrete data Arithmetic Mean (A.M.) ofa set of observa ns is their sum divided by the number of observations Let X41, Xp, Xp are 7 observations, then Xp tXy + wt Xn AM=% ff Example I: Let there are five numbers 5, 4, 3, 7, 6, then arithmetic mean is o_S+44+34746 _ 25 AM.= & == 5 centralCase Hs Dit ete or ungrouped frequency distribution Incase of frequency distribution X;[fisi = 1,2,..n, where fj is the frequency of the vatiableX, te aM AX, + foXo + fl, BEI 1 py — fhthtotfa Thi fi nae where N= Phy fj Example 2: Let observations and their frequencies are given as 5 7 8 9 10 3 1 2 3 2 Then arithmetic mean is calculated as AM.= 7 -3X245%34 7X148x249x3410%2 91 ues 24341424342 13 Case III: Grouped or continuous frequency distribution uped or continuous frequency distribution, X is taken as the midpoint of the class, Example 3a: Distribution of marks of students in a class are given as below: Marks: 0-10 10-20 20-30 30-40 40-50 No. of Students 5 3 9 2 6 Find the arithmetic mean. Solution: ee ee, Marks No. of Mid Point (x), Weo-10 3 10-20 3 Is 0-30 9 25 30-40 2 38 » 40-50 ‘Therefore, arithmetic mean is ytMerits of Arithmetic Mean: 1. Itis rigidly defined It is easy to understand and easy to calculate. . It is based upon all values of the given data. + ~ Itis capable of further mathematical treatment. - Itis not much affected by sampling fluctuations. Demerits of Arithmetic Mean: 1. Tt cannot be calculated if any observation is missing. - It cannot be calculated for the data with open-end classes. 2 3. Itis affected by extreme values, 4. It cannot be located graphically. 5 . It may be number which is not present in the data. 6. It cannot be calculated for the data representing qualitative characteristic. Solved Examples: Problem I: The sum of deviations of a certain numbers of observations measured from 4 is 72 and the sum of deviations of observations measured from 7 is -3. Find the number of observations their mean. and Solution: Let n be the required number of observations, We have given that 7 o(X; — 4) = 72, therefore LyX; — 4n = 72 @ and Ef.o(Xi — 7) = —3, therefore Diag Xj — 7n = (id Subtracting the above two equations we get, n = 25 and hence [¥q X; = 172, ‘Therefore, required mean is given byProblem 2: ‘The mean weight of 98 students is found to be SO Ibs. It is later discovered that the frequency of the class interval (30- 40) was wrongly taken as 8 instead of 10, Calculate the correct mean Solution: We have given that Incorrect mean (X) = 0 Kg and ¥ fi = 98 Therefore, Incorrect mean (R) = eerrect E Lik whi 50 = ct EFM 98 Therefore, Incorrect ¥ fiX; = 98 x 50 = 4900 Now, Correct 5 fiX; = Incorrect ¥ fiX; — (8 x 35) + (10 x 35) = 4900 — 280 + 350 = 4970 Also the Correct ¥ f; = 98 +2 = 100 Therefore, the correct mean = Gorvect Ef1X _ 4970 X= correct Ef, 100 ¥ = 49.701 lb Problem 3: The average marks of three batches of students having 70, 50 and 30 students respectively are 50, 55 and 45. Find the average marks of all the 150 students, taken together. Solution: We have given that [ Baten T 1 rl ‘Average Marks | R= 45 No. of Students. ny = 30 Let X be the average marks ofall 150 students taken togel ‘Therefore, _ ky + nake tsk, _ 70 x 50 +50 x 55 +30 x 45, 7600 ny mg try 70450430 0 or, 50.67 marks,Median Median of a distribution is the value of the variable which divides it into two equal parts, that is the alue such that the number of observations above it is equal to the number of observations below it. ‘The median is a positional average. Application: Median is only average to be used while dealing with qualitative data which cannot be measured quantitatively but can he anged in ascending or descending order of magnitude. Case lL: Diser ele data Arrange all the obs vations in ascending or descending order. © Ifthe number of observations arc edd, then n+] - Median = (“5*) th observation =Yne1 z Example 10: Let the observations are 3, 2, 5, 1,9, 8,7 Th range these obsi ations like 1,2,3,5,7.8,9 or 9,8,.7,8, 3,2, 1 —> Ascending order —> Descending order Here n = 7, so 7+1 Median - (“S) th observation = Xau1= Ifthe number of observations are even, then Median :Example 11: Let the observations are 3, 2. ‘Then arrange these observations like or Here n = 8, so Xn +Xn, Median 2241 _ Xa + Xs 2 2 5+6 Median =——— =5: ian = Case Il: Discrete frequency distribution Steps for calculating median © Find N/2, where N = tea fi Locate cumulative frequency (c.f) just greater than N/2 * The value of X corresponding to that cf: is the median Example 12: Find median for the following data x f Cf. 1 3 3 3 2 iS 4 14 6 6 20 9 «—24—2 10 5 27 12 9 36 15 4 40 | Solution: Here N = 40 | + N/2=20 + Cumulative frequency (¢,/.) just greater than 20 is 22 © The value of X corresponding to 2 ‘Therefore, Median = 9se HI: Continuous frequency distribution (grouped data) Steps for calculating median * Find N/ © Locate cumulative frequency (c,/) just greater than N/2 © The class corresponding to that cf is the median class * Then use following formula to calculate median Payy (N Median = In += ( ~G) a where, tn = lower limit of the median class fm = frequency of the median class h = class width Cy = ¢.f-of the class preceeding to the median class Example 13: Find the median wage (salary) per day of workers from the following data ages in dollar 20-30 30-40 40-50 50-60 60-70 No. of Workers 3 5 20 10 2 Solutio, Wages No. of Workers A (in dollar) wy 20-30 3 3 30-40 # 8 40-50 20 28 30-60 10 9 60-70 2 40 SS eran AOI Here N= 40, therefore © N/2=20 | 28 * Cumulative frequency (c,f:) just greater than 20 * The class corresponding to 28 is (40 - 50). This is the median class, © Thus the 10 Median = 40 + 5 (20~8) = 46 20 |Merits: J. Itis rigidly defined. 2. Itis easy to calculate and understand 3. Whe n be located merely by inspection, tis not all affected by extreme values Wcan be calculated for the distributions with open: Demerits: \ 2. Itis not based on all observations. nd classes. ase of even number of observations median cannot be determined exactly. 3. Lis not suitable for al ebraic treatment, 4. As compared to mean, it is affected much by fluctuations of sampling. Quartiles: Three points which divide the whole distribution into four equal parts are called quartiles. Re ql Q2 Q3 Calculation of Quartile: Case I: Discrete data ‘The First quartile Q, is the value such that 25% of the ranked data are smaller and 75% are lever: whenn isodd a= ; X, +X, |, when iseven ating a4 «The Second quartile Q> is the Median. hhe First quartile Qs is the value such that 75% of the ranked data ate smaller and 25% are « The Fi Jaret: when isodd } when iseven 41 te 14: Find the first quartile, second quartile (median) and third quartile of the following data Example 14: of scores: i 2s 1s 5 7 14 36 53 30 a2 Solution: First arrange the data in ascending order: ° 5 7 # Mos * 2 30g 2 53 2 Q) (Median)Example 18: Find the first quartile, median and third quantile. if 63 is added to Example 14 Solution: First, arrange the data in ascending order: 3 7 py 15 Rs 30 4% 33 2, Q> (Median) (12414) =13 1 (22425 7 22+25 dey, +X) 436442 31% + Mo)=5 2 ‘When the subscript of X'is not a whole number (whether » is odd or even), then we use the follo steps Stepl: Calculate Q, and Qs using the formulae =X noi and Qs = Xs net) 4 4 Step 2: Select the ranked po: sitions immediately below and above the number calculated, For example, for 10 values, Q,= X nay = 4 X75. So select second and third ranked v ralues. 4 4 With these values, do the followin; Multiply the larger ranked value by the decimal fraction of the original result (0.75 in the example), Multiply the smaller ranked value by 1 minus the decimal fraction of the original result (0.35 for the example. because 1-0, 75 is 0.25), Add the two products to determine the quartile value H case: If the two ranked values selected he the Same number, the quartile is that number and above two multiplication and addition can be skipped. 6SExample 16: Find the first and third qu artiles of the following data values 99 4389 gg Solution: First, 3 35 440 arrange the data in ascending order 29 34 . 3939 ,* 144 ay $2 QO, Q3 (Median) Or Calculate Xeniy 10) = Xa.75 4 4 ‘The second and third ranked values are 31 and 35. 1. Multiply the larger ranked value by the decimal fraction of the original result, we get 35x0.75 = 26.25 2. Multiply the smaller ranked value by | minus the decimal fraction of the original result, we get 31x0,25=7.75 3. Therefore final Q, =26.25+7.75=34 # Calculate Os = X3un41) =Xsa0st) =Xa2s 4 + ‘The eighth and ninth ranked values are 44 and 44, 1, Multiply the larger ranked value by the decimal fraction of the original result, we 44%0,25=11 Multiply the smaller ranked value by | minus the decimal fraction of the original result, we get 440.75 =33 3. Therefore final Q; =11+33=44 OR Since eighth and ninth ranked values a ime, that is 44, Therefore Qy = 44 Example 17: \d the first and third quartiles of the following data values: 5239 443 gg Solution: First, arrange the data in ascending order: 3103S 390 4p 4B 4 4 i 2 2, (Median oy 354g © Calculate = Xena) = Key 4 4 cond and third ranked values are 35 and 39, 1, Multiply the larger ranked value by the decimal traction of the original resul. 390,5 = 19.50 2. Multiply the smaller ranked value by 1 minus the 350.5 =17.50 We get decimal fraction of the original result, we 3, Therefore final @ =19.50417.50=37© Calculate QO; = Vx, Vx 4 4 ghth ranked values are 44 and 44, The seventh and e' 1. Multiply the larger ranked value by the decimal fraction of the of 44x0, al result, we get Multiply the smaller ranked value by | minus the decimal fraction of the original result, we get 44x 0. 2 3. Therefore final Qy = 22+ OR Since eighth and ninth ranked values are same, that is 44, Therefore Q; = 44 Case II: Discrete frequency distribution (ungrouped data) Steps for calculating quartiles . Find, j = 1,23 where N =, fi * See cumulative frequency (c.f) just greater than # The value of X corresponding to that ef is the Q),j = 1,2,3 Example 18: Calculate all quartiles for the given data x f Gf. 1 3 3 2 2 5 2 ee 4 6 20 Ss <— 24-2 8 2 3 9 3 36 10 4 40 Solution: Here N = © Cumulative frequency (c.f) just greater than 10, 20 and 30 are 14, 22, 31 respectively. © ‘The value of X corresponding to 14, 22, 31 3, 5, 7 respectively Therefore, a Q2 = Sand Qs =7Case 11; Continuous frequeney distribution (grouped data) Steps for calculating quartiles 7 Find ® j= 1 where N = 5? 1 } ini fi © See cumulative frequency (c.f) just greater than en quiattile class * The class corresponding to that cf is the © Then use following formula to calculate quartiles where, 1, = lower limit of the that quartile class fq = frequency of the that quartile class h = class width Cp = ¢.f.of the class preceeding to the quartile class, Example 19: Calculate first and third quartiles of wage (salary) per day of workers from the following data Wages in dollar 20-30 30-40 40-50 50-60 60-70 70-80 No. of Workers 3 8 17 1 a 5 Solution: Be Wages No. of Workers CH (in dollar) n es 20-30 5 5 30-40, 8 13 40 - 50 17 30 60-70 4 45 3 48 48, therefore Here N= 2 Weigand =36 7 Cumulative frequency (c/.) just greater than 12 is 13, so first quartile class is 30 - 40, Cumulative frequency (cf) just greater than 36 is 41, so third quartile c 50 - 60, 2 Thus the 10 = 22-5) = 38.75 Q, = 30+ 2 ~ 5) = 3.7 50 +74 (36 ~ 30) = andDeciles: The nine points which divide the whole distribution into ten equal parts are called decites eciles: The nine points whic $$ $+} ++ — 8 D. D9 t 1 2 Method of Calculating Deciles Case I: Discrete frequency distribution (ungrouped data) Steps for calculating deciles =12,...,9 where N = DP, f; 3 kN * See cumulative frequeney (c.f ) just greater than Sot The value of X corresponding to that cf’ is the Dy, ke = 1, 2, Case II: Continuous frequency distribution (grouped data) Steps for calculating quartiles kN . Find [k= 1,2,... kw * See cumulative frequency (c.f) just greater than a * The class corresponding to that cf. is the given decile class Then use following formula to calculate deciles hk De =a tee a). k= 12, where, 'a = lower limit of the that decile class fa = frequency of the that decile class lass width Ca = c.f. of the cl The ninety. called percentiles. lass preceeding to the decile class nine points which divide the whole dis bution into hundred equal parts are a Pi p99 P2 Method of Calculating Percentiles Case I; Discrete frequency distrily ion (ungrouped data) Steps for calculating deciles Pind!) ©. Find 5.! = 1,2, 99 where N =P", See cumulative frequency (Gf) just greater than < ‘The value of X corresponding io that cf is the Pj. = 1.Deciles: The nine points which divide the whole distribution into ten equal pa 's are called deciles, eciles: The nine points whic A+++ +-+-+— 8 D D9 t 1 2 Method of Calculating Deciles Case I: Diserete frequency distribution (ungrouped data) Steps for caleulating deciles = 12,...,9 whereN =D", f; kw © See cumulative frequeney (c.f) just greater than a The value of X corresponding to that cf’ is the Dy, ke = 1, 2, Case II: Continuous frequeney distribution (grouped data) Steps for calculating kN . Find Sk = 1,2,... ky * See cumulative frequency (c.f) just greater than a * The class corresponding to that cis the given decile class Then use following formula to calculate deciles h (kN eater a(S a). k = 1,2, where, 14 = lower limit of the that decile class Ja = frequency of the that decile class lass width Ca = c.f. of the cl The ninety. called percentiles. lass preceeding to the decile class nine points which divide the whole dis bution into hundred equal parts are a PA p99 P2 Method of Calculating Percentiles Case I: Discrete frequency distrib ion (ungrouped data) Steps for calculating deciles 8 Pind 5 1,2, 099 where N= SI, See cumulative frequency (6) just greater than ae * The value of X corresporCase Ht Continuous fr weney distribution (grouped data) Steps for calculating quartiles 7 100" 199 where y aa lv * See cumulative frequency (cf, just greater than * The class corresponding to that ef. is the given percentile class * Then use following formula to calculate deciles h (iN P=ly+ “(a- Cy) f= 1,2, 99 fy \100 where, 1, = lower limit of the that percentile class fy = frequency of the that percentile class h = class width Cy = ¢.f-of the class preceeding to the percentile class, Example: In Example 19, calculate Dy,Dy, Py and Pg Mode ‘Mode is the value which occurs most frequently in a set of observations. Case 1: Diserete frequency distribution In case of discrete frequency distribution, mode is the value, which corresponds to. maximum frequency. \ Example 20: Find mode for the following data x f \ H 7 \ 2 3 i 4 5. 6 15, % 7 & 3 Here maximum frequency is 25 and value corresponding to 25 is 4, therefore move is 4. Note: In any one or more of the following cases # Ifthe maximum frequency is repeated | + Ifthe maximum frequency oceurs in the very beginning or at the end of the distribution, | © I there are irregularities in the distribution, | ‘The value of mode is determined by the method of grouping, | Example 21; Find mode of the following distribution | x 2 3 4 Ss 6 7 8 7 10 i LPs | f3 8 IS 2 35 4 32 2% mM 4 4 6Solution: arities in the distribution may be observed. So mode will be calculated by grouping Gln (WG OPO ollausind bay ; Grosbioavir Frepatuat Clu Q) 26 thera bea ally ba comity 3 46 awnings fir fre, f ‘ a j pe brepeg, . 98 ° Gibawent¥t is Yaudd . 107 &) tembining bepoontes : woo thes by “het ofl ; 80 dist Go 8 os ek 9 a 2 Fern ealig, fhe "0 65 auekyp yor tale ao MW 4 i Be stese saith, cline sree, brecend Qlomn anti” Analysis Table ee enaatemin brepemsy Gl Max. 12 6 789 it a the No cororhont a er 7 x Solum i> : ar ond . . bbe valve, =e * oF comabtnadee wy x Grelves wy x oka x qin (sy to ree-eaer Drecsermine Total nes aa t ve ne Therefore, Mode ~ 6 ClO TL Case I: Continuous frequency distribution (grouped data) uel obreyvasirr Adentity the snodal class corresponding to the maximuun trequency © Coleulate the mode using the following formula eed weer es af deamen tthe Mode = 14 xh ee 4h WK Kad vali) where, be Hee venues 1 lower Jimit of the modal class if a ns ass width mort / Jui? frequency of the modal cass | Jy: requency of the preceeding class | fy: frequency of the successive elass | _— Y TG wkat) Aliaseh InaExample 2 Calculate mode of the following data Class Interval (cp *reqeney | ss 10-20 8 20-30 7 30-40, 12 50-60 20 60-70 10 70 - 80 10 ae Solution: Here maximum frequency is 28, therefore modal clas Now, 40 - 50. 10 x (28 - 12) Mode = 40 +s 2x 28-12-20 40 + 6.67 = 46.67 Remarks: 1 In any one or more of the eases as defined for discrete case, modal class is determined by the method of grouping If the method of grouping gives the modal class which does not correspond to the maximum frequency or in some cases 2f, — fy — fz = 0, in such cases mode is obtained by using formula | | hin = fe) ; P2fn ~ fol + Vin — fl Rematl’, Fer a symmebica! diduluhin mean, meolan | 1. It is easy to understand and easy to calculate. omnck mode coimecda ; Itis not affected by extreme values or sampling fluctuations. Ib Ue dobrlubin 6 ae asiymoretrial Mode = 1+ 3. It can be located just by inspection in many cases. 4, It is always present within data. ba 5. Itis applicable for both quantitative and qualitative data, igh te eal Demerits: 1, It is not rigidly defined. 2. It is not based upon all values of the given data, 3. It is not capable of further mathematical treatment,Measures of Dispersion Measure of Variabilit ariability of ci ly or Measure mean, of Dispersion indicates how spread out the data around the Larger variation Same Center The following are the measure of dispersions: 1. Range 2. Quartile Deviation 3. Mean Deviation 4 ‘Standard Deviation Characteristic for ideal measure of dispersion (i) Mtshould be rigidly defined. ii) It should be simple to understand and easy to calculate, (iii) I should be based upon all values of given data. reatment. (iv) It should be capable of further mathematical t (v) It should have sampling stability. 1. Range: The range is the difference between the largest and smallest values of the distribution. Let X4,Xp).Xq are the set of m observations, then Range = max(Xp,Xp.+Xq) — MINX Xa Xn) Example 1: Let there are five numbers 25, 34, 13, 27, 36, then Range = 36-13 = 23 4 One of the simplest measures of variability to calculate. 4 Depends only on extreme values and provides no information about how the remaining, data is distributed.4. Standard Deviation Case I: Diserete data Let X1, Xp, «.,Xq are n observations, then Standard Deviation generally denoted by @ is given by Standard Deviation (S.D.) = 0 = where X is the arithmetic mean, The square of Standard Deviation is called Variance. Therefore the formula for calculating Variance is 2 Variance = 6? ==S1.,(X; — X)? The above formulation can also be written as = Sha? nF) Variance = Example 6: Find Standard Deviation and Variance from the given data 5, 4, 3, 7, 6 Solution: Here p_St+4+3+74+6 25 AM, = ¥ = 5 5 Therefore standard deviation is given by oa [Cr+ G- 58+ 8-540-546-5710 = eee weevw' = Variance = 07 = 2 S.D. Hence the variance is Case II: Discrete frequency distribution bution X[fi i = 1,2, ..,m, where fj is the frequency of the variable Xj, then In case of frequency dis standard deviation is Die ii — 4? S.D. where N = Diy fi and Variance =? = 25%. f(X;— )? The above formulation can also be written as {Zier fiX? — NX?) Variancegsample 7: Find Standard Deviation for the given data 305° #7 8 9 10 fi 2 3 1 2 3 2 Solution: Hare Arithmetic Mean is X = 7. Te 3 2 32 5 3 12 1 1 9 8 2 2 9 3 12 10 2 18 Ts — ¥)? =76 Therefore S.D.=0= ms fi(Xe— RP? = [2.x 76 = VSB = 242 Case 111: Grouped or continuous frequency distribution In case of grouped or continuous frequency distribution, X is taken as the midpoint of the corresponding class. Example 8: Distribution of marks of students in a class is given as below: Marks: 0-10 10-20 20-30 30-40 40 - 50 No. of Students 5 3 9 2 6 Find the standard deviation and hence variance. No. of Mid Point ma Students (f;) (x) £iX (%- x)? FX — XP 0-10 5 5 25 400 2000 10-20 4 1s 60 100 400 20-30 8 25 200 0 0 30-40. 2 35 70 100 200 40-50 6 45 210 400 2400 EFiCX; — X)* =5000Therefore 1 200 = 14.14 eo ft 00 = 4200 SDs 0 = ISI fk 2? = [x so00 and Variance = 200 : i iation. * Itis based on all the observations better measure of dispersion than mean devi ssitantl ing the (X;— ). * The artificiality created in the case of mean deviation is removed by squaring the (X; — X). * Its capable of further mathematical treatment. Population Standard Deviation and Population Variance Let X,, X2, ..., Xv are the population values and mean of the population is 8, then population Standard Deviation and population variance is given by Standard Deviation (S.D.) and poi — 8)? mand Sample Variance Variance * Sample Standard Devi Let X1,X2400Xn are n observations taken from a population and sample size is small, then we replace the denominator n by (n — 1) in formula of standard deviation and variance. Therefore the Sample Standard Deviation generally denoted by s is given by Sample Standard Deviation (S. D.) a Hy Soe) Sample Variance (X — X)? ph ost \ This s? is also called sample mean square, But when sample size is large, then no need to do this modification as + 1as n > 00.aple 9: Consider the following data set 47 9 12 4 e Cateulate Mean and variance of the above dat : ) If2 is subtracted from each value, then mean andl variance oF new series series i) 113 is multiplied to each value, then mean standard deviation and variance oF ness SeFieS and variance of new series. n mean iv) If cach value is multiplied by 3 and then inereased by 5th Coefficient of Dispersi Sometimes we want to compare Which are measured in different units, we do not only calculate the measure of " " it of nt. calculate coefficient of dispersipn. which are pure numbers independent of unit of measureme _ ie ¢ PANKKBR CDV HC [Dar aree Pre Be Beoorea EHTS PPE YM ‘The Coefficient of Dispersion (C.D.) based on different measure of dispersion are as below: the variability of two series which differ widely in their averages or ispersion but we 1. C.D, based on range Where 4 is the largest value and B is the smallest value in the series. 2. C.D. based on quartile deviation Qs a) FOR OQ) CD. where Q; and Qs are first and third quartiles respectively. deviation 3. C.D. based on me: Mean Deviation (M.D.) Average from which it is calculated 4. C.D. based on standard deviation Standard Deviation o Mean OX C.D. Coefficient of Variation 100 times the coefficient of dispersion based on standard deviation (S.D.) is called Coefficient of Variation, that is Coetficient of Variation (C.V.) = 4x 100 degser OM t rato & le Romank! The senter han more. canvotints ern the atheyExample 10: In the game of Soccer goal scored by (wo teams A and B in ten matches are given below: Match No.: | 2 3 4 5 6 4 8 9 10 Team A 2 0 4 3 | 0 5 2 1 2 Team Bo: 5 0 1 4 1 1 4 3 2 6 Test which team is more consistent in the game?Measure of Skewness - from sy When a distribution is ‘The term skewness refers to lack of symmetry oF departure from symmetry. When 8 GX" TN © hot symmetrical (or is asymmetrical), itis called skewed distribution. The measur indi iff i istributed in particular indicate the difference between the manners in which the observations are distril r hewnise calté distribution compared with a symmetrical (or normal) distribution. ‘The concept of skewness & i ion of the normal importance from the fact that statistical theory is often based upon the assumption of the distribution. Nature of Skewness: Skewness can be positive or negative or zero. * When the values of mean, median and mode are equal, there is no skewness. * When mean > median > mode, skewness will be positive * When mean < median < mode, skewness will be negative. Normal eves) Pen vere Tesemtana: Pate etn rmprowrsapencty _treeve ere ont There ig different measuresof skewness, which are discussed below. 1. Karl Pearson’s Coefficient of Skewness This method is most frequently used for measuring skewness. The formula for measuring coefficient based on mean, mode and standard deviation is given by Mean ~ Mode * skp = See SD Where skp denotes the Pearsonian coefficient of skewness. ned, then the above formula can be modified by using the empirical relation as, 4 mode is ill det 3( Mean — Median) skp = ; Be TE skp =0, thecymmetsical dictribution, Zhen fhe olesbrcbu ties is aymeel neo! # Ifmean is greater than mode, implies sk,, >O. then positive skewness. C Alno fer Mteax> Ifmean is less than mode, implies sk,, <0, then negative skewness. (Alo S77 Met & The Karl Pearson's coefficient of skewness lies between -3 t0+3.Measure of Kurtosis ‘The relative flatness of the top is called kurtosis or convexity of curve. jhe coefficients fy and yy are us an ; re used fo measure the kurtosis, These coefficients are defined as below: Me By a 72> Py-3 ‘The quantity yy = By —3 is called excess of kurtosis. I yy<3. then curve is called Platykurtic. pene © Hope * Wf yy=3. then curve is called Mesokustic, © If y)>3. then curve is called Leptokurtie. Example 6: Refer to example 2 and caleulate .71./%) and 72 coefficients. Also study the nature of the distribution, Solution: In example 2, we have calculated the following first four central moments as dy =0. fty =1.8896, 425 = 7.491 and s1y =126,5874 fore (ZAM _ 56.1508 ANY _ 5611508 _ § 347 (1.8896) 6.7469 Hs _ 126.5874 _ 4B (1.88967 7 = Py -3= 32.4527 ce 7 >0, therefore distribution is leptokurtic,