ee a 20 te INTRODUCTION TO STATISTICAL THEORY PE 2.4 FREQUENCY DISTRIBUTION ™ ec: ‘The organization of a set of data in a table showing the distribution of the data into classes of groups together with the number of observations in each class ot group is called a Frequency | '™ Distribution, The numberof observations fling ina particular casi refered toa the class frequency} 7 ‘of simply frequency an is denoted by f Data presented in the form of a frequency distribution are also called grouped data while the data in the original form are referred to as ungronped data. The data are said to be arranged in-an array when arranged in ascending or descending order of magnitude. The sh purpose of a frequency distribution is to produce a meaningful pattem for the overall distribution of the data from which conclusions can be drawn. A faitly common frequency pattern is the rising to a peak and then declining. In terms of its construction, each class or group has lower and upper limits, lower and ‘upper boundaries, an interval and a middle value. 24.1 Class The clas-limits are defined as the numbers or the values of the variables which describe the classes: the smaller number isthe fower class limit and the larger number is the upper class limit. Class-limits should be well defined and there should be ino overlapping. In other words, the limits shouldbe inclusive, ie. the values coresponding exactly to the lower limit or the upper limit be included in that class, The class-limits are therefore selected in such a way that they have the same number of significant places as the recorded values. Suppose the data are recorded to the nearest integers. Then an appropriate method for defining the class limits without overlapping, for example, may be 10 - 14, 15 19, 20 ~ 24, etc. The class limits may be defined as 10.0 - 14.9, 15.0 - 199, 20.0 - 24.9, et. when the data are recorded to nearest tenth ofan integer. Sometimes, a class has either no lower class limit or ‘no upper class-limit. Such a class is called an open-end class. The open-end classes, if possible, should be avoided as they are a hindrance in performing certain calculations. A class indicated as 10 ~ 15 will include 10 but not 15, ie. 10¢ X<15, 24.2. Class-boundaries. The clas-boundaries are the precise numbers which separate one class ii from another. The selection of these numbers removes the difficulty, if any in knowing the clas to which a particular value should be assigned. A class-boundary is located midway between the upper limit of a class and the lower limit of the next higher class, ¢.g. 9.5 - 14.5, 14.5 ~ 19.5, 19.5 - 24.5, or 9.95 - 14.95, 1495 ~ 19.95, ete. The class-boundaries are thus always defined more precisely than the level of measurements being used so that the possibility of any observation'falling exactly on the boundary is avoided. That is why the class boundaries carry one more decimal place than the class limits or the observed values. The upper class boundary of a class coincides with the lower boundary of the next class. 24,3 Class Mark. A class mark, also called class midpoint, is that number which divides each class into two parts. In practice, itis obtained by diving either the sum of the lower and upper limits ofa class, or the sum of the lower and upper boundaries of the class by 2 but in a few cases, it does not hold, particularly in modem practice of age grouping. For purposes of calculations, the frequency in a particular class is assumed to have the same value as the class-mark or midpoint. This assumption may introduce an error, called the grouping error but statistical experience has shown that such errors usually tend to counterbalance over the entire distribution. The grouping error may also be minimized by selecting a class (gxoup) in such a way tat its midpoint corresponds to the mean of the observed values vi falling in that class 244 Class Width or Interval. ‘The class-width or interval ofa class is equal to the difference between the class boundaries. It may also be obtained by finding the ‘difference either between two suecessivg lower clas limits, or between two successive class marks. The lower limit ofa class should not be subtracted from its upper limit to get the class interval. An equal clas interval, usually denoted by ‘or c, facilitates the calculations of statistical constants such as the mean, the standard deviation Scanned with CamScanner~~ sLTHEORY PRESENTATION OF DATA moments, ete, That is why in practice, itis desirable to have equal class-intervals, ut in some “a economic and medical data, it is wise to use unequal class-intervals on account of greater concentra to classe or J yeasurements in certain classes, Such Class intervals usually become uniform When logarithms o} 1 Frequency parks are taken, It should be noted that some people use the terms “class” and “elass-interva ieee interchangeably andthe width ofthe class is referred to as the size ot length ofthe class-interval. tion area ' : ‘ The data are 24S Constructing a Grouped Frequency Distribution, ‘The following are some basic rules that ‘nitude, The J should be Kept in mind when constructing a grouped frequency distribution: tution of the i) Decide on the number of classes into which the data are to be grouped. There are no hard ya peak and ‘and fast rules for deciding on the number of classes which actually depends on the size of » Tower and data, Statistical experience tells us that no less than 5 and no more than 20 classes are generally used. Use of too many classes will defeat the purpose of condensation and too few ables which will result in too much loss of information. H.A. Sturges has proposed an empirical rule for upper class determining the number of classes into which a set of observations should be grouped. The 3 the limits rule is 2e included k= L433 10g, = where k denotes the numberof classes and Nis the total number of observations. For example, a iS ir if there are 100 observations, then by applying Sturges’ rule, we should have . ete. when K=143,3(2,0000)=7.6, i.e. 8 classes Beun ‘Thus eight classes are required but this rule is rarely used in practice. = 15 will ii) Determine the range of variation in the data, ic. the difference between the largest and the simallest values inthe data one class, iii) Divide the range of variation by the number of classes to determine the approximate width 5 t0 which ot size ofthe equal class-interval. In case of fractional results, the next higher whole number is limit of a usually taken as the size or width of class-interval. If equal class-intervals are inconvenient or 5- 1495, may be undesirable, then classes of unequal size are used, But in practice, intervals that are level of multiple of Sor 10, are commonly used as people can understand them more readily. eae iv) Decide where to locate the class-limit of the lowest class and then the lower class boundary. = tae ‘The lowest class usually starts with the smallest data value or a number less than it. It is better ted if itis a multiple of clas-interval. Find the upper class boundary by adding the width of the ides each class-interval to the lower class-boundary and write down the upper class limits.too. The ‘mits ofa open-end classes, ie. classes with the lowermost or uppermost class boundary unknown, not hold, should be avoided if possible. xy ¥) Determine the remaining class-imits and class boundaries by adding the class-interval o repeatedly. The lowest class should be placed at the top and the rest should follow according sabe to size. In some cases, the highest class is placed at the top. values vi) Distribute the data into the appropriate classes. This is best done by using a “Tally- Column” where values are tabulated against appropriate classes by merely making short bars ference or tally marks to represent them, It is customary for convenience in counting to place the first en four bars vertically and the fifth one diagonally and to leave a space. The number of tallies is eal then, writen in the fequency column. The tally column is usually omitted"in the final ‘asdby presentation of the frequency distribution. But in'case of small number of values, the actual ane values should be shown against each class to mitigate chances of error. Scanned with CamScannerINTRODUCTION TO STATISTICAL THEORY | FRESE Frit) Finally, total the frequeney column to see that all the data have been accounted for. anarra These rules ate applied to group raw data which are assumed to be continuous, In case of discrete data which carry only integral values, the concept of a class boundary is unrealistic ay there can be no i points where the adjoining classes meet In spite of this logical difficulty, when the diserere dane are sufficiently large, they are treated for convenience of calculations as continuous and hence are grouped in the same way as the continuous data Example 22 Make a grouped frequency distribution from the following data, relating to the weight recorded to the nearest grams of 60 apples picked out at random from a consignment. 106 107 76-82 «109 107«1S- 93.187, «9S—«123-—1S M1 92 86 70 126 8130 129139119 HS 128 100 18 84 «99 «113 204 M4136 128 HIS 98 110 78 18S 162 178 140 152 173 146158194 148 90 1071811317584 10411080882 By scanning the data, we find that the largest weight is 204 grams and the smallest weight is 68 ‘grams so that the range is 204 ~ 68 = 136 grams. Suppose we decide to take 7 classes of equal size. Then size or width of the equal clas interval would be ES ivar. But we take += 20, the next integra value higher than 19.47 to facilitate the numerical work. E Let us decide to locate the lower limit of the lowest class at 65. With this choice, the class limits will be 65 ~ 84, 85 - 104, 105 ~ 124, ..., the class boundaries become 64.5 ~ 84.5, 84.5~ 104.5, 104.5 ~ occupat 124.5, ..:, and the class-marks are 74,5, 94.5, 1145.... The grouped frequency distribution is then constructed as follows: i) By listing the actual values: , FREQUENCY DISTRIBUTION OF WEIGHTS OF 60 APPLES Weight Entries Frequeney 65-84 | 76, 82, 70, 68, 84,78, 75, 80,82 9 85-104 | 93,95, 92, 86, 100, 9, 90, 98, 90, 104 10 106, 107, 109, 107, 115, 123, 111, 119, 115, 113, 111, 123, 115, 110, 107, 110, 118 125-144. | 125,126, 130,129,139, 128, 141, 136, 140,131 10 105 ~ 124 1 145 — 164 | 162, 152, 146, 158, 148, 5 1 165-184 | 178,173, 181, 184 4 isl4o- 185-204 | 187, 186,204, 185, 194 5 4 Total i 60 group a Scanned with CamScanner‘AL THEORY J for. ase of discrete ere can be mo rete da are 3 are grouped slting to the 23 8 1s 94 82 weight is 68 lass interval ‘aciltate the class limits 45, 104.5 - jon is then PRESENTATION OF DATA. This table is sometimes known as an entry fable, The values against each class may be a anaray. ii) By using a Tally-Column: } : FREQUENCY DISTRIBUTION OF WEIGHTS OF 60 APPLES T us Mid-points Classes Class- (eight) | boundaries | CHS = Frequency osu | os-eas | as | Ml 9 85-104 | 845-1045 94s | MMI 10 1os—124 | 104.5 - 1245 nas | MMMM 7 125-144 | 124s-144s | 1345 | MIM 10 145-164 | 1445-1645 | 1545 | HL 164-184 | 164.5 ~ 1845 1745 | Ill 185-204 | 184.5 - 204.5 1945 | Total Sluanu Example 2.3 Given below are the mean annual death rates per 1,000 at ages 20 ~ 65 in each of 88 occupational groups. Construct a grouped frequency distribution. TS: 82 G2 iB th 4B $4 S499 109 ios 14 v7 its 126 50 102 92 10 99 73 73 84 ios 11 io Md 6S 125 78 65 87 93 124 WA 91 97 93 62 103 66 74 86 27 94 17 128 87 S$ 86 96 119 04 78 76 121 46. 140 81 114 106 116 104 81 46 66 128 68 71 66 8&8 88 107 108 60 79 73 93 93 89 Ial 39 60 69 90 8&8 94 M4 109 (BASE. Lahore, 1971) ‘A scan ofthe data shows thatthe largest value is 14.0 and the smallest value is 3.9 so that the range 8140-39-10 ‘As the data are recorded to one decimal place, we may therefore locate the lower limit of the first ‘group at 3.5. Let us choose a class interval of 1.0. Then the class limits are specified as 3.5 - 4.4, Scanned with CamScannerINTRODUCTION TO STATISTICAL THEORY 15-54, 55-644, ... With this choice, the class-boundaties are 3.45~4.45,4.45 5.45, 5.45 - 645, 18.00 i db ot nie with ie given vas red The following table shows the required frequency distribution: FREQUENCY DISTRIBUTION OF MEAN DEATH RATES ¥ Death Rates | Class-boundaries || Tally Frequency 35-44 345-445 {I 1 45-54 445-545 | Ill 4 55-64 545-645 | 5 65-74 645-745 | THT I 8 : follows 75-84 745-845 | HIT 12 , 85-94 845-945” | MUIIHI 9 2 95-104 | 945-1045 | MUI . wos-114 | tods— 114s | 10 t MS 124 145-1245 | HL if, B 125-134 1245-1345 4 small, s N distibut 135-144 | 1345-1445 |] 1 Total 88 Example 2.4 Construct a frequency distbution forthe data below. Indicate the class boundaries and class limits clearly. 41.78 | 2932 | 3147 | 3535 | 3282 | 39.42 6.65 | 28.31 | 4463 | 22.78 | 44.44 | 48.12 8171 | 33.47 |'5035 | 29.19 | 51.26, | 50.32 2684 | 1895 | 48:19 | 43.72 | 43.89 | 47.15 6020 | 44.43 | 41.17 | 370°] 2235 | 29.17 By scanning the data, we find that the largest value is 81.71 and the smallest value is 18.95 so tht the range is 81.71 ~ 18.95 = 62.76. Suppose we decide to take 5 classes of equal size. Then size or width of the equal cass interval 6 S wold be S78 55. But we take h= 13.0, the next integral value higher than 12.55 to take OE discrete or numerical work. —_ Scanned with CamScannerIEORY 645, >undaries 15 so that + imterval to take the PRI [As the data are recorded to two decimal places, we may locate he class Limits wil be 18,00 ~ 30.99, 31,00 = 43.99, vo become 17,995 ~ 30,995, 30.995 - 43.995, The grouped frequency distribution 18.00. With this choi VIATIONOFDATA the lower limit of the fw the class be} then constra. follows: Gases] Gasman [Tay [| 1800-3099 | 17,995~30.995 | MH Ill 8 31.00-43.99 | 3099543995 | I IH 10 44,00- 56.99 | 43.995-56.995 | If Ill 9 57.00- 69.99 | $6.995-69.995 | |] 2 70.00 82.99 | ‘69.995 -82.995 | | _| Total oo | a | Example 25 A survey of $0 retail establishments had assistants, excluding proprietors, as follows: 2 388, 2 426 4 1 5,2 6 ‘ Arrange the values as a frequency distribution. By scanning the data, we find that the number of assistants is a discrete variable and the range 's small, so the data can be conveniently sorted by taking the values of classes as 0, 1, 2, ete. The frequency, distribution is then constructed as shown below: FREQUENCY DISTRIBUTION OF ASSISTANTS IN 50 RETAIL ESTABLISHMENTS Number of Tally Number of Assistants (x) Establishments () 0 Ml 3 1 {i 4 2 WI 6 3 Wu 7 4 Wi 10 5 WI 6 6 W 5 7 H 5 8 I 3 9 | 1 Total 50 Such a frequency distribution jn which each class consists of a single value is sometimes called a discrete or ungrouped freduency distribution Scanned with CamScannerSTICAL THEORY oe 4.6 Cumulative Frequency Distribution, The total flequency ofa variable from its one end ta ii); jn value (usually upper class boundary in grouped data), called the base, is known as the cumulative Fjwency, less than or more than the base ofthe variable. A table that shows the cumulative frequencies, led a cumulative frequency distribution. The cumulative frequency of the lst class is the sum ofall frequencies in the distribution, If the cumulation process is from the lowest value to the highest, itis referred to as “a less than” type cumulative frequency distribution. For example, let us consider a ) frequency distribution having classes, each of width h, Let us denote the midpoint ofthe ith class by +, L with frequency f; such that)’ /, =n. Now the lower class-boundary of the first group is x, ~/2 and ¢ oh 7 the upper class boundaries are 1x, +H/2, (7=1, Zink). The cumulative frequency distribution is then : obtained by adding each successive frequency to the cumulative total of frequencies for the preceding i classes as shown below: i Class-boundary | Cumulative Frequency 25 st less than x, ~h/2 0 A Jess than xy 4/2 fh in grou! less than x +/2 hth the Sten display LP y+ G less than x, +h/2 hthth _— : the trali example less than x, +/2 hen Itshould be noted that a less shan type cumulative frequency distribution starts with the lower class boundary of the first group indicating that there is no frequency below 1, ~h/2 ‘When the fequences ae cumulated from the highest value othe lowest value, itis called amore FAH pos than” type cumulative frequency. : of the in If the class frequencies against various classes are divided by the total frequency, we get the 1" relative frequencies which alvays add to one. The cass frequencies may also be expressed a5 fie stem percentages, the toal of which would be 100, A percentage cumulative distribution is useful to read off fF a ea the percentage of values falling between certain specified values aap Example 2.6 Construct (i) a “less than” type cumulative distribution, and (ii) a “more than” type : ‘cumulative distribution from the frequency distribution of weights of 60 apples of Example 2.2. as fol lice {) _A“less than” type cumulative frequency distribution is shown below: 7 ‘Weight (grams) | Cumulative Frequency (F) a | Tess than 64.5 0 : i Less than 84.5 9 o | Less than 104.5 9 2 | Less than 124.5 36 A: Less than 14,5 46 (or leadit Less than 164.5, 31 hich hat | Less than 184.5 55 order inv | Less than 204.5 0 Scanned with CamScannerOBOE EARN ME Ale AL THEORY oneendwoa fj) A*more than” type cumulative frequency distribution i given below 1¢ cumulative Weight (grams) | Cumulative Frequency (F) frequencies, More than 64.5 @ ve sum of all highest, it is More than 84,5 51 5 consider a More than 104.5 41 Chass by 2 More than 124.5, m4 er More than 144.5 4 : More than 164.5 9 tion is then More than 184.5 5 » preceding More than 204.5 0 1.5 STEM-AND-LEAF DISPLAY AA clear disadvantage of using a frequency able is that the identity of individual observations is lost in grouping process. To overcome this drawback, John Tukey (1977) introduced a technique known as the Stem-and-Leaf Display. This technique offers a quick and novel way for simultaneously sorting and displaying data sets where each number in the data set is divided into two parts, a Stem’and a Leaf A stems the leading digi(s) of each number and is used in sorting, while a leaf isthe rest of the number or the trailing digt(s) and shown in display. A vertical line separates the leaf (or leaves) from the stem. For example, the number 243 could be split two ways: leading digit | trailing digits OR leading digit | trailing digits ‘wer class 2 a 4 3 stem leaf. stem | leat ‘amore J) au possible stems ae arranged in order from the smallest tothe largest and placed on the lft hand side ofthe line “eet the The stem-and-leaf display isa useful step for listing the data in an array, leaves are associated with sssed_ as the stem to know the numbers. The stem-and-leaf table provides a useful description of the data set and read off can easily be converted to a frequency table. It is a common practice to arrange the trailing digits in each row from smallest to highest. 1” type Example 2.7 The ages of 30 patients admitted to certain hospital during a particular week were as follows 48, 31, 54, 37, 18, 64, 61, 43, 40, 71, 51, 12, 52, 65, 53 42, 39, 62, 74, 48, 29, 67, 30, 49, 68, 35, 57, 26, 27, 58 Construct a stem-and-leaf display from the data and list the data in an array. ‘A scan of the data indicates that the observations range (in age) from 12 to 74, We use the first (or leading) digit as the stem and the second (or trailing) digit as the leaf. The. first observation is 48, hich has a stem of 4 and a leaf of 8, the second a stem of'3 and a leaf of 1, etc. Placing the Ieaves in the order in which they appear in the data, we get the stem-and-leaf display as shown on next page: Scanned with CamScannerINTRODUCTION TO STATISTICAL THEORY PRE Stem (leading digit) Leaf (trailing digit) Pc ery 82 na isan | 2 967 other 3 17905 4 830289 5 5 412378 pn 6 415278 best 7 14 figur being To get the array, we associate the leaves in order of size withthe stems as shown below: sake 12, 18, 26, 27, 29, 30, 31, 35, 37, 39, 40, 42, 43, 48, 48, 49, ae : infor 51, 52,. 53, 54, 57, $8, 61, 62, 64, 65, 67, 68, 71, 74 anda Example 2.8 Construct a stem-and-leaf display for the data of annual death rates given in Example 23 : fon Using the decimal part in each number as the /eaf and the rest of the digits as the stem, we get the i following stem-and-leaf display (leaves are ordered): Stem Leaf ii) 3 9 4 66 o 5 045 it) 6 00225566689 7 13334456778889 8 1124667788899 i”) 9 012333344467799 10 011233446678899 y a 144669 n 0145688 4 0 2.6 GRAPHICAL REPRESENTATION * While Tabulation, we know, is a good method of condensing and representing statistical data in a readily i) understandable form, but many people have no taste for figures. They would prefer a way of representation where figures could be avoided, This purpose is achieved by the presentation of statistical data in a visual form. The visual display of statistical data in the form of points, lines, areas and other ii) geometrical forms and symbols, is in the most general terms known as Graphical Representation Statistical data can be studied with this method without going through figures, presented in the form of iy tables. iv) Scanned with CamScannerL THEORY 28 given in We get the a readily way of statistical ind other entation. form of PRESENTATION OF DATA. Such visual representation can’ be divided into two main groups, graphs and dla described in the sections that follow. The basic difference between a graph and a diagram 1s is a representation of data by a continious curve, usually shown on a graph paper while a diagr cher one, two or three--dimensional form of visual representation. 2.7 DIAGRAMS Diagrammatic representation is best suited to spatial series and data split into different categories, Whenever a comparison of the same type of data at different places is to be made, diagrams will be the« best way to do that. Diagrammatic representation has several advantages over tabular representation of figures. Beautifully and neatly constructed diagrams are more attractive than simple figures, Diagrams, being a visual display, leave more effective and long lasting impression on the mind of a reader. They’ rake unwieldy data intelligible at a glance. Comparison is made easier with diagrams. Diagrams have some disadvantages too. Diagrams are less accurate than tables; cost money and time and the amount of information conveyed is limited. However, this method of representation is excessively used in business and administration, Different types of diagrams or charts commonly used for displaying statistical data are described below: i) Linear or One-Dimensional Diagrams. They consist of Simple Bars, Multiple Bars and Component Bar charts. Here the values are represented only by one dimension, generally the length of the bar. ii) Areal or Two-Dimensional Diagrams. They consist of Rectangles, Sub-divided Rectangles and Squares, the areas of which are proportional to the values of the given quantities. This device is used to represent data having moderately large variations. iil) Cubic or Three-Dimensional Diagrams. They are in the form of Cubes and cylinders, whose volumes are proportional to the values they represent, These diagrams are used when the variation among the values of the data to be portrayed isso large that even the square roots of the values concerned ful to reduce the variation appreciably. iv) Pie-Diagrams. They are in the form of Circles and Sectors. Here the areas of circles or sectors are in proportion tothe values they represent or compare. : VW) Pictograms. They consist of pictures or small symbolic figures representing the statistical data. A pictogram is an effective way of visual comparisons. For example, we can compare the armed strength of various countries by drawing pictures of the number of soldiers, where each pictorial soldier may denote, say, 1,000 solders, In a similar way, the production of ‘wheat can be compared by means of the pictures of wheat bags of a specified size. It is essential to repeat the pictures a number of times to represent the differences in magnitudes. While drawing diagrams, the following points should be kept in mind: i) _Anappropriate scale consistent with the size of paper available and the size of the data to be represented, should be chosen and indicated either atthe side‘or at the bottom of the diagram. This scale must start at zero. ii) A diagram like a table, must have a title, which should be-brief and self-explanatory. A key, footnote or source will also be necessary. iii) A diagram should be shaded, coloured or cross-hatched to show the different parts, ifany. iv) Lettering should be shown horizontally, Scanned with CamScannerINTRODUCTION TO STATISTICAL THEORY 7.1. Simple Bar Chart, A simple bar chart consists of horizontal or vertical bars of equal widths Fengths proportional to the values they represent. AS the basis of comparison is linear or one- a rensional, the widths of these bars have no significance but ate taken to make the chart fook attractive. fre space separating the bars should wot esceed the width of the bar and should not be less than half of its width. The bary should neither be exceedingly long and narrow nor short and broad The vertical bar chart is an effective way for presenting a time series and qualitatively classified data whereas horizomal bars are useful for geographical or spunal distadutions. The data when do not relate to time, should be arranged in ascending of descending oner before charting PRESEt Example 2.9 Draw a simple bar diagram to represent the tunover of a company for 6 yeas, Years: 1980 1981 1982 1983 1984 1985 Turnover (Rupees): 38,000 45,000. 48,000 52.500 55,000, 58,000 ‘The bar chart is drawn below: Bar diagram showing the Tarnover 000. ‘of a company for 5 years 60,000 27. is divided 50,000 by each ba ° 10 increase 40,000 ‘cumulation 4 bars. 30,000 Exa 20,000 10,000. 1980 1981 1982 1983 1984 1985 Year 2.7.2 Multiple Bar Chart, A multiple bar chart shows two or more characteristics corresponding f. The g to the values of a common variable in the form of grouped bars, whose leingths are proportional to the F dawn below values of the characteristics, and each of which is shaded or coloured differently to aid identification This is a good device for the comparison of two or three kinds of information, For example, imports, exports and productions of a country can be compared fiom year to year by grouping the three bars together. Example 2.10 Draw multiple bar charts to show the aea and production of cotton in the Punjab from the following data: Year ‘Area (000 ares) | Production (000 bales) 1965 - 66 2866 1588 1970-71 3233 29 | 1975-76 3420 1937 | (Source: Statistical Wing, Agriculture Deptt. Lahore) Scanned with CamScannerLTHEORY qual widths ‘eat oF one. ‘Kattadtve, than half of vertical bar ¥ horizonial % should be cars, =sponding nal to the tification » imports, hree bars ‘e.Punjab |PIRESENTATION OF DATA Seo ccea ee are eee eee sue drawn below: ‘The maltiple bar charts sve dra meer ve awk aca nacre . a TD roovcron wens 2.73 Component Bar Chart, is divided imto two or Example 2.11 Drawa ‘Component bar chart for the following data, (Population in Lakiis) Division Both Genders [Male | Female] Peshawar e Gi aara\peal Rawalpindi 40 a 19 Sargodha © 2 | 28 [Lahore 65 35 30 The appropriate conponent bar chact after ‘ranging the population fi drawn below: ‘COMPONENT win CHART SHOWING * PUPULATIC.‘OF 4 DIVISION ‘tes in ascending order is Scanned with CamScannerINTRODUCTION TO STATISTICAL THEORY y ——__—_$—_— $$$ enc seas snd TISTICAL THEORY 2.74 Rectangles and Sub-divided Rectangles. The area of a rectangle is equal to the product of| jength and breadth. To represent a quantity by a rectangle, both length and breadth of the rectangle are 7,sed. Sub-divided rectangles are drawn for the data where the quantities along with their components are to be compared. These diagrams are generally drawn to compare the budgets of various families. In the construction of sub-civided rectangles, we are required to i) + change each component into the percentage ofthe corresponding total, ii) draw one rectangle for cach total, taking equal lengths (100 units) and breadths proportional to the totals, iii) divide every rectangle so drawn ini parts equal in number to the number of components. Each part shaded or coloured will represent percentage size of one component. Example 2.12 Compare the budgets of families A and B with a suitable diagram. ems of Expenditure | Family A _| Family B Food, mu 60 Clothing 4 4 House Rent 4 16 Education 3 6 Litigation 2 10 ‘Conventional Needs 1 6 Miscellaneous 2 8 Total 40 120 The necessary computations required for the drawing of sub-divided rectangles are given below and the diagram is shown on page 33: Family A Family B Items of Expenditure | Actual | Percentage | Actual |. Percentage Expenses | Expenses | Expenses | Expenses Food uM 60.0 60 50.0 Clothing 4 10.0 4 17 House Rent 4 10.0 16 133 Education 3 15 6 5.0 Litigation 2 5.0 10 83 Conventional Needs 1 25 6 5.0 Miscellaneous 2 5.0 8 67 Total 40 100.0 120 100.0 Scanned with CamScanner2AL THEOR| «the product te rectangle an} ‘omponents a} ‘amilis, Inte sroportional £ components given below PRESENTATION OF DATA NEEDS “| uricarion ta EDUCATION. HOUSE RENT 60 40} 20 3 i j RSA Rs.120 2.1.5 Pietograms, A pictogram is a popular device for portraying the statistical data by means of F pictures or small symbols. It is said that « picture ix worth ten thousand words. It is customary to represent a unit value of the data by a standard symbol or a picture and the whole quantity by an appropriate number of repetitious of symbol concemed. This means the larger quantities should be represented by a larger manber of symbols and not by larger symbols. quantity smaller than the unit is represented by a part of the picture or symbol used. The symbols or pictures to be used, must be simple and clear A pictogram is virtually a bar chart constructed in pictorial way as the number of symbols or pictures corresponds to the length ofa bar. Example 2.13 The following table shows the number of employees in a certain Textile Mills Represent the data by means of a pictogram. Year _| No. of Employees 1950 2,004 1955 2,990 1960 4240 1965 5,380 Representing 1,000 employees by one picture, the pictogram is drawn below: ver, prcrocran ol KK Kmoies oa RI Scanned with CamScannerINTRODUCTION TO STATISTICAL THEORY } PRES 2-16 Pie Diagrams. A pie-diagram, alsd known as sector diagram, is graphic device consisting of a citcle divided into sectors or pie-shaped pieces whose areas are proportional to the various parts ints can br which |the whole quantity is divided. The sectors are shaded or coloured differently to show the } from relationship of parts to the whole. If space permits, the descriptive titles of the constituent parts should be | repres placed horizontally on each sector, otherwise a key becomes necessary. It is a convenient way of displaying the component patts in proportion to the tlal and therefore is used as an alternative to a component bar chart. It is an effective way’of showing percentage parts when the whole quantity 1 taken a5 100. It is also used when the basic categories are not quantifiable as with expenditure, classified into food, clothing, fuel and light, etc, The arrangement of the sectors must be made uniform in comparing pie charts, To construct a pie chart, draw a circle of any convenient radius. As a circle consists of 360°, the whole quantity to be displayed is equated to 360. The proportion that each component part or category bears to the whole quantity will be the corresponding proportion of 360°. These corresponding Proportions, i.e. angles, are calculated by the formula Angle SOMPOREHL 55 ‘whole quantity “Then divide the circle into diferent sectors by constructing angles atthe centre by means of a protractr and draw the corresponding radi. Example 2.14 Represent the total expenditure and expenditures on various items of a family bya pie diagram. sltems: Food Clothing House Rent Fuel and Light Misc, Expentiture:(inRs) 50.30 20 15 35 The corresponding angles needed to draw the chart are computed below. , ems Expenditure (in Rs.) Food 50 Clothing 30 House Rent 20 Angles of the Seciors (in Degrees) Fuel and Light 1S 120 Miscellaneous 35 2 Total 150 360 “The pie diagram consisting of a circle divided into five sectors defined by angles 120°, 72°, 48%, F a 36° and 84°, is drawn below:”" PIE DIAGRAM 28 Gl | a | to repre Variable 4 C or mon advanta Scanned with CamScannerAL THEORY. ice consisting} cus part in to show the arts should & sient way of erative 104 nity 1s taken Iassifed ino omparing pi a protractor: family bya 72°, 48°, canbe shown above the ntmal base fine and Fosse below the base ine, Since the bars are n from the zero line show losses. we start fiom the top. For an illustration, the following represented Particulars | foc i) Matenals ii) Wages iii) Polishing, ete. Total cost Proceeds, Profit( + ) or loss ( PROFIT AND L¢ can OSS CHART 19601970 100) WAGES 80} POUSHING oo} 40} PROFIT 1 > voss (ET ese ome ‘pie chart may also be used for this purpose. 2.8 GRAPHS As already stated, diagrams are usefil for rep nt a statistical series spread over a period of time, or a frequency distribution or two related variables in visual form, For such representations, graphs are employed. Graphs present the data in a simple, clear and effective manne!, facilitate comparison between two ‘or more than two statistical series, and help us in appreciating their significance readily. Another advantage Of graphs is that they provide an overall picture of a statistical series. Graphs are also sometimes used to make predictions and forecasls. Certain partition values can also be located sraphically. But graphs are less accurate as they do not give minute details. Moreover, they cost considerable expenditure and time. Scanned with CamScannerINTRODUCTION TO STATISTICAL THEORY Z.,ruction of Graphs. In the construction of @ graph, the first step isto lake a starting point, known re ogi, in the left-hand bottom comer of the graph paper. Two straight lines perpendicular 1o each Paver are drawn through the origin. ‘The horizontal line is called the X-axis or abscissa and the vertical line is labeled as Y-axis or ordinate, The two lines together are known as co-ordinate axes Some suitable scales are selected along -axis and Y-axis. Independent variable is taken along X-axis and dependent variable along Y-axis. Points are plotted and joined to get the requited graph. While constructing a graph, the following points should be kept in mind: i A scale and the form of representati to be selected in such a way that the true impression of the data to be represented is given by.the graph ii) Every graph niust have a clear and comprehensive title at top. Where necessary, sub-titles shoilld be added. iii) The source of the data must be given. A key and footnotes should be provided when necessary, iv) The independent variable should always be placed on the horizontal axis. v) The vertical scale should always begin with zero, otherwise the graph will give a false impression. If, however, the first item of the data is quite large, a scale-break should be shown between zero and next member. vi) The horizontal axis does not have to begin with zero unless of course, the independent variable or the lower limit of the first class interval is zero. vii) The axes of the graph should be properly labelled. Labels should clearly state both the variable and the units, eg. “Distance” and “Kilometer”, “Sales” and “Rupees”, etc Curves if more than one, must be clearly distinguished either by different colours: or by differentiated lines (solid, dashed, dot-dashed). ix) The graph should not be loaded with too many curves. Graphs can be divided into two main categories, namely: 2) Graphs of Time-Series or Graphs of Historical Data, and b) Graphs of Frequency Distributions. The important graphs of frequency distributions are Histogram, Frequency Polygon, Frequency Curve and the Cumulative Frequency Curve ot Osive. 2.8.1 Graph of Time Series-Historigram. A curve showing changes in the value of one or more items from one period of time to the next is known as the graph of a time series. This curve is also callec a Historigram. Thus a historigram displays the variations in time series dealing with prices, production impoits. population, etc. To construct a historigram; time is taken along X-axis and the values of tht variables along Y-axis. Points are plotted and are then connected by means of straight line segments to ge the “Historigra Example 2.15 ‘The following table gives the number of cars produced in Germany during the years 1929-1936. Draw a suitable graphs, i.c, Historigram of the series. Years: 1929 1930 1931 1932 19331934 19351936 No. of Cars: oR m4 68 50 9 12 45 302 Scanned with CamScannerL THEORY ‘fo FRESENTA LUN UF DATA — int, known The historigram is drawn for the data by taking years on horizontal axis ular to cach > vertical axis as below: vertical line me suitable | dependent nga graph, and the number g} impression « substtes ded when re a fake be shown Tependent zvanable BOA aa YEAR, 2.8.2 Histogram, A histogram consists of a set of adjacent rectangles whose bases are marked off by class boundaries (not class limits) on the X-axis and whose heights are proportional tothe frequencies associated with respective classes. The area of each rectangle represents the respective class frequencies, This is one of the most important graphical representation ofa frequency distribution, When the class- intervals are equal, the rectangles all have the same width and their heights directly represent the class ions are | frequencies, that is they are numerically proportional to the frequencies in the respective classes. The “uve or | following figure shows the histogram forthe ftequency distribution of Example 23, 1929] 1930] 1933] 1934] 1935] 1936] 1928] sor by more ‘MISTOGRAM FOR FREQUENCY DISTRIBUTION oe ‘OF ANNUAL DEATH RATES >ealled as] Juction, of the Sto get | 2 prequency ing the cuss souNoARIES Scanned with CamScannerINTRODUCTION TO STATISTICAL THEORY /| intervals are not all equal, the height of the rectangle over an unequal class-interval, is 10 be adjusted because it ws arcu and not height that measures frequency. This means that the height of a rectangle must be proportionally decreased if the length of the corresponding elass-sterval increases. For example, if the length of a class-interval becomes double, then the height of the rectangle is to be halved so that the area, being the fundamental property of the rectangle of a histogram, remains unchanged. This sort of rescaling is necessary so that the correct pattern of the distribution is to be conveyed. When the frequencies in a frequency distribution are given against the class-marks x, of equal class-intervals of width fa histogram is constructed by drawing vertical lines (dotted) whose heights correspond to the respective classfrequencies a the class-marks marked off on the axis of X and erecting a series of adjacent rectangles with widths equal to x,£/2 (ie. half ofthe width is token on either side of x). {tis important to note that in the construction of a histogram, we assume that within any one class, the values of the variable are evenly spread out between the class-boundaries. A histogram which must not be confused with the historigram (graph ofa time series) is useful in forming a rough idea of the 2 overall patter and shape ofthe frequency distribution. which i: Example 2.16 Construct a Histogram for the-following frequency distribution relating to the ages fF the core (to nearest birthday) of telephone operators. intervalt same tet ‘Age (Years) 18-19 | 20-24 | 25-29 | 30-34 | 35-44 | 45-59 rectangl reach tht No.of Operators | 9 | 188 | 160 | 123 | 84 | 15 Saeed : the poly As the class-intervals are unequal, the height of each rectangle cannot be made equal to the frequency. Fin the 1 The eight ofa rectangle is therefore calculated by dividing the frequency (the area) by the corresponding J Example class interval (the width). The necessary calculations andthe histogram follow: | Class- Class- | Frequency | — Proportional boundaries Interval (i) Heights | 175-195 2 9 9+2=45 195-245 5 188 188 = 5 =37.6 45-295 5 160 160 =$ 32.0 w5-M5 5 123 13-546 | 345-4455 10 4 84-10-84 | 445-595 15 15 15~15=1.0 Af position Scanned with CamScannerPRESENTATION 7 DAT ‘ ts £ s Be ge z 28.3 Frequency Polygon. A frequency polyge.t is a graphic form of a frequency distribution. which is constructed by plotting the points (x;, fj) where. is the class-mark of the ith class and f; is ‘the corresponding frequency, and then connecting shem by straigh' line segments provided the clzss- intervals are equal. In case of vneual class-intervals, heights of unequal classes are adjusted by using the same technique that was use! for histogram. It can also b2 obtained by joining the tops of the successive rectangles in the histogram by means of straight line segments. The graph drawn in this way does not reach the herizontal axis. But a polygor, ac we know, isa closed figure having many sides. itis therefore customary to edd “extra” clsss anarks et both ends cf the distribution with zero class frequencies so that the polygon docs form a closed figure with the horizontal axis. This should be done even ifthe curve ends in the minus part of the graph. The frequency polygon for the frequency distribution of weights in Example 2.2 is given below: y FREQUENCY PCLYGON FOR | FREQUENCY DISTRIBUTION | OF WEIGHTS 20 15 20 5 9 x spre MID POINTS ‘A fequency polvgon whicis can be used! fcr comparing two or more data sets, gives roughly the d kurtosis ofthe curve (these terms ate defined later). Scanned with CamScannerINTRODUCTION TO STATISTICAL THEORY, Fs Frequency Curve, When a frequency polygon or a histogram constructed over class vas made sufficiently small for a large number of observations, is smoothed, it approaches a Finuous curve, called a frequency curve. The concept of a frequency curve is of great importance in Fatistics. Mathematically, the curve is represented by the relation y= f(x) and has an important property concerning its area, The following graph represents histogram and frequency curve for the frequency distribution of the mean annual death rates of Example 2.3. Y FREQUENCY CURVE MID POINTS 2.85 Cumulative Frequency Polygon or Ogive. A cumulative frequency polygon, popularly known as Ogive (shymes with “alive” and pronounced o'jiv) is a graph obtained by plotting the cumulated frequencies of a distribution against the upper or lower class boundaries depending upon whether ihe cumulation is ofthe “less than” or “more than” type, and the points are joined by straight line segments. Because of its likeness to an architectural moulding called an ogee, a cumulative frequency polygon is called an Ogive. An Ogive, when the cumulation is of fess-than type, is constructed by plotting the points x;+h/2, F;) where x; +//2 is the upper class-boundary of the ith class and Fis the cumulative frequency for the ith class, and connecting the successive points by straight line segments. The polygon should start from zero atthe lower boundary of the first interval, i. the point (x; ~h/2, 0) is plotted and joined, and to have a polygon, the last point is also joined with the last upper class- boundary. In case of unequal classes, we merely join the unequally spaced points. ‘CUMULATIVE FREQUENCY POLYGON (OGIVE) yy FORFREQUENCY DISTRIBUTION OF sense 4 3 3 go 40 i $ 30 § 2 : 8 ° Pony Ee ew = PEPEREES ee tren cise nouns Scanned with CamScanner