UNIVARIATE ANALYSIS Introducton w Single variable’ Distributions and Variables - Numerical Summaries of Level and Spread - Scaling and Standardizing — Inequality - Smoothing ‘Time Sens. 1. What isa case and variable? A Dataset consists of cases. Cases are nothing but the objects in the collection or are units of analysis, the things about which information is collected Each ease hes one or autributes or qualities. called variables which are characteristics of cases. Example: The following dataset contains 6 cases and 3 variables: Variables Store_| Sales _| Customers 1 14 9 2 a | 3 Gices, 3 22 49 4 2a 20 5 29 26 6 40 2. What is sample - A population is the enfice A sample is the 5 less than the total for ET jobs in The top 50 search results for al vertisements f ‘Netherlands ‘Netherlands on May 1. 2020 FT jobs in the lereraduate students in the 300 undergraduate students from three Dutch universities who Nétherlsnds, ‘volunleer tr your psychology research study:3. Whatis random sampling? Random sampling isa part of the sampling technique m winch euch sample has an equal probability of being chosen. A sample chosen randomly is meant to be an unbiased representation of the total populstion. 4. Explain the household Survey in London? themselves IL 1s normally conducted by Ollice for Nateanal collected from them includes household, family and individ classities the drinkers in to the tallowmg euteguries, Hardly drink ax all 27dnnk a hile 3, drink a moderate arnaunt4 drmk quite a lot 5. drink heavily Also, National Statstics of Si onomic Chssification (NS-SEC) or Social Class based on Occupation (SC) or Social Class Classification or Socio-economic Groups uses five classes, 1, Managenal and Professional Occupations 2. Intermediate Occupations. 3. Small employers. and own account workers: 4. Lower supervisory ars! technical occupations. 5, Sem-routine occupations, of different reasons why it case may have missing «lata wn a parlacula missing data is some householders not ready to express the data. When numbers are used to represent categories thal have no inherent or scale. When numbers are used to convey full arithmetic properties, this is\ The varuhles used by social scientists are measured op§pommal se: or ordinal scales: [to as continubus variables). S. How to reduee the number of digits? ‘There are two different mechanisms are avail 3 fine 1s the Rounding the variable values to the nearest number. Example 199 99 may be rounded to, Seeand method is: comp Srop Example 899 9 an ime 6. Draw the bar chart and pie ch: D.. the above survey dataset. ‘A Bar chuet ts thé tisplay har is re the: representing each category of wartable such thal the length of the rofeases in the category. For instances, a bar ebart ofthe drinking wn in the following figure.hanily drink ar all drink a fittke drink que a fot Amoune of alcohol respondent din} Source: Coben $ of Faure 1 Figure 1.2 Bar chart showing drinking betravidur from GH$(2005) specimen data QO cases). 1 hardly d-ink at all DI drink aline El drink a mokrate amount IB crink quite alot I or drink heavily? Source: Colwnn $:of figure If. Figure 1.3 Pie chart showing drinking behaviour from GHS specimen data (20 cases).7. Whatis Histogram and explain? Chane rhat are somewhar similar to bar charts can be used to display interval level variables grouped into categories and these are called histograms They are constructed in exactly the same ‘way as bar charts except, of course, thal the ordering of the categories is fixed, and care has to be taken lo show exactly how the dala were grouped. A sample General Household Survey of ag shown in the following figure. Hiecogram 1.000: Frequency Tistograms-allovw inspection of four important aspects of any distribution: ‘Spreud= Whafaurefypncal values m the distribution? How widely dispersed are the values? much from one another? Shape Outliers: Is the disuibution flat or peaked? metric Are there any particularly unusual valves? ypical share of the frequency distribution of polygon or histogram is shown in the following,9. ‘What are all the characteristics of normal distrib Characteristics of Normal Distribution probability representations eponmveny skewes, Foun acon Frou 23 ‘Tpcat saapes Empirieal Rule: Ina normal fei Of the observations are confined within + Bythin + two standard deviahons, and jglion, 95% cone standard el almost 99.7% o| Bell-shaped C extrermbey, This r |-shaped Curve Mean am ‘This data representation is shaped by mean and standard deviation. Equal C jencies. The mean, median, and mode of this data are equal Symmetric: The distribution curve is centrally symmetric. Therefore, half of the he lef pT the center, and the remaming values appear on the nght ‘artosis: Skewness is the the symmetry. The skewness for a normal is zero. Kurtosis studies the tail of the represented dara. For a normal fie kurtosis is 3. The total value of the standard deviation, ie. the complete area of the ww lunetion, ts one Also, the entire meun 1s 727,10. Explain the frequency distribution of data sets. A frequency distribution is a collection of observations produced by sorting observations into classes and showing their frequency of eccurrence in each class. A liequency distribution belps us lo detect any pattern in the data (assuming a pattern ‘by superimposing some order on the inevitable variability among observations. * — Ttean be represented as graph or table 1 Explain the Honsehald level Incame duta in UK (General Household Survey also collects dats af the level af the household In this survey, the data includes Weekly Household Income, Having Cars, Persons in Household.1. North East 2. North West 3. Yorks and Humber 4. East Midlands 5, West Midlands 6. Eastern 7. London 8. South East 9. South West Wales 11. Scotland Figure 1.8 Specimen data from the 2005 GHS household Houschold Weekly household id income (pence) c i Weunueune 16523.35, SS384.62 155166.7 We ie eeeeat Mlilenls Coucenment Oifice Regions [ Yorks and Hesmber Neath Esa Frequency Sewer CHS 2005 Rawhald dota Figure 1k? Bar chart of number of households sefeact 15 2005. Office Region in te GH! % $09 1,000 1,500 2,009 2.500 000 Grr wericly howell insome (pone) Source’ GHS 7005 rexsehois data. Figure 1.10 Histogram of weekly household income from the GHS 2005. 10Di nosaror van D taro vaa Ei Zanor vans @Sormor an from the GHS 2005. mm 4 678 9 oN No. of persons in household Source: CHS 2005 household date. Figure 1.12. Bar chart of number of people in each household in the GHS 2005. a12. What ix SPSS SPSS (Statistical Package for the Social Sciences) isa statistical analysis and used to bring out the good sraphical displays. It is also frequently used by researchers in market research companies, local authonties, health authontes and goverment departments, 13. How to aecess the SPSS through Data Editor? ‘When you first start using the program. don't be overwhelmed by the number of different me? and options that are available, Rather than trying to discover and understand all SPSS provides, it is better to start by focusing on mastering justa few procedures. screen shown in the following figure. | etsplay sta Pla Information, Cache Bata. e Dara Editor window + The Output window + The Syntax window jhen you first open SPSS, the Data Editor window will be displayed. This will be empty until either open an existing dam file ot type in your own data - in the same way that you would enter data into a spreadsheet ke Excel You can open an existing data file usmg the File menu and then selectmg ‘Open’ Fed12. What ix SPSS SPSS (Statistical Package for the Social Sciences) isa statistical analysis and used to bring out the good sraphical displays. It is also frequently used by researchers in market research companies, local authonties, health authontes and goverment departments, 13. How to aecess the SPSS through Data Editor? ‘When you first start using the program. don't be overwhelmed by the number of different me? and options that are available, Rather than trying to discover and understand all SPSS provides, it is better to start by focusing on mastering justa few procedures. screen shown in the following figure. | etsplay sta Pla Information, Cache Bata. e Dara Editor window + The Output window + The Syntax window jhen you first open SPSS, the Data Editor window will be displayed. This will be empty until either open an existing dam file ot type in your own data - in the same way that you would enter data into a spreadsheet ke Excel You can open an existing data file usmg the File menu and then selectmg ‘Open’ Fedand 'Data!. You are then able to browse the directories on your computer until you find the dista file thar you need. The data in the SPSS Dama Editor are displayed in rows and columas. Each row provides the information about a single case m the dataset, As we have seen im this chapter this could be an individual person or a bousehold. Each column comprises the information about a spe variable, and the name of the variable appears at the top of each column The menus aerass th of the Data Editor allow you to access a range of procedures so that you can analyses yo modify your data and produce lubles, pie charts, histograms and olher graphical display then the Dialog boxes that SPSS provides to cheose exactly the variables that y with. The second is to type instructions into the SPSS Syntax window. SPSS keywords amd commands thal need to be entered very precisely und m the & In onder tn produce a pie chart of drinking betuviar, s lo thet showen in figure 1.3, trom the menus along the top. choose ‘Graphs’ and then ‘Pie’. ] Ee eee aie Is 50801001 _™ t . 450601001090101 49 362 7 Figure 1.15 Using menus to preduce a pie chart. A-SPSS ‘dalogue box will now appear. By default, this specefies thal the dala in the chart Tepresent summaries for groups of cases and this 1s what you want, Next click on the ‘Define’ button. BDaan Chat Are —-- © Surmsies for geups ol eases * Surrmosins of eparate variables = | Vakues of individual canes wil palpar af al oat. gu A second dialogue box for pie charts. 1Figure 1.18 The second dialogue box for pie charts with selected. *k on the button in the top right comer labelled ‘OK’, SPSS will automatically Jiewer window for vou. whieh will display your first pie char. This is shown in 15Drawing Bar chart In the next dialogue box, you can specify’ for which variable you want a bar chart displayed and also choose whether the bars represent the number of eases (N of cases) or the percentane of cases ((% of cases) m each category of the vanable Select the variable genbith from the wartable hist and move it te the “Category Axis window by clicking en the arrow button. Funally click on the button and the bar chart far the variable genhlch will appear in the SPSS output window. OQ Figure 1.24 The tar chart dsplayed in dre SPSS ourput window, 16‘The SPSS package itself provides good introduction to-all the main aspeets af the program via the Turorials. In order to view a tutorial, choose Tutorial’ from the Help menu (see below). Sta.stes Coach vat mc Command Sirbax Peference Figure 1.14 Accessing the SPSS tutorials. Defining missing values Retore producing graphical displays of single varahleSn S$ I normally have te tell the computer which values of the variable correspon; r " To do this, in the variable view of the data editor in SPSS click on the, nhith' of the column headed "Missmg and use the dialogue box thal apps -& and -6 and then click the OK button. i fs thee (Mer tom ‘zed wkd 3s hor. 644 event ef nevted sorecrnl Gk poe 4a a eed 0.0) te pon) wet ences goed : fam toot net nes Teste be ge pte Figure 1.20 Opening the SPSS dialogue box to show the missing values. resNext, from the menus along the top, choose "Graphs! and then "Rar? From the dinlogue box that appears, choose the first option ‘Simple’ and specify that data in the chart are summaries for groups of cases. Nextelick on the Define’ button Figure 1.21 Creating a bar chart in SPSS.14_Compare the (Zeneral Hausehold Survey? The histograms of the working hours distributions of men and women in the 2005 General Household Survey are shown in figures + The male batch is at a higher level than the female bateh = The two chstributions are somewhat similarly spresd out Ameen of cna fers CSS. tn tach se Figure 2.2 Weekly hours worked by women in the 2005 GHS. 19+The female batch 1s brmodal suggesting thene are two rather different underlying populations, +The male batch is unimodal IS. What is the difference between the sumple and papulati A population is the entire group that you want to draw conclusions about. A sample 1s the specidie® ptoup that you will collect data from. The size of the sample is always less than the total si @ the population. 16. What area Residuals? A residual can be defined as the difference beoween a data point and the abs&y average, value, For example, we had chosen 40 hours a week working hours, using, data from the General Household Survey in recorded in the survey ws working 45 hoursa week would have a resid of expressing this is to say that the residual is the observed data value and in this case 45-40= 5. Data = Fit + Residual 17. What is mode? Give example. ‘The mode retlects the value of the most freqy mnglgghre or the mode is the value [nat appears most often in a set of data. The mode‘of robability distribution is the value x at DADAOUONGONAINEaGoaD Sauress The Now York Times Almanac (2052). 20In this case there casts two modes, 4and 8 4 oceurs 7 times und 8 occurs 6 times, ‘Anown as bimodal. Bimodal describes any distribution with two obvious peaks, eg oem ireene eure 3.1 PRE of president rem, 18. What is median? Give example. ‘The median reflects the middle value when observations argimtdered from legtt to mast 19. How to compute or find median? A. INSTRUCTIONS ‘Order scores trom least to most [Frat the rdf poston by ack 2.Figure 2.30 Mens working hours ranked to show the Men's working hours (ranked 0 * ” 0 as o a3 Median value 48 30 sa 55 s o 70 %9 Shurea: Small random sample of men" weeking hore fea CHS 2008 dae. ¢ adding all scores and then dividing by the number of scores. sumof all scores Mean number of scores 2SUM OF FROM MEAN DEVIATIONS, 64a) 240 2ao 240 240 240 240 aaa 040 50} 1.60 169 =150 =160 =160 169 160 21, What is the difference hetween the po) ae jan => A mple me Sample Mean (X) The balance pois or Seb le, found by dividing the sum for the values he sample ‘balance point for a population, found by dividing the sum for all w the number of scores in the populztion, <) The total number of scores in the population POPULATION MEAN rx 2322. What he purpase and nuture of mean? ‘The mean serves as the balance point for its distribution because of a special property: The sum of all scores, expressed ity positive and negative deviations tiem the mean, always equals 2e10, ‘The mean reflects the values of all seures, not just those that are middle ranked (us wi median), or those that occur most frequently. 23. How to Interpret the Differences between Mean and Medi Ideally, when a distribution is skewed. report both the mean and the median. Appreciable differences between the values of the mean and med skewed distribution. Ifthe mean exceeds the median, 2 it does fir the, fant death rates, the tmuGfying distin bution is positively skewed because of one or more scores with relatively large val ‘has the very high infant death rates for a number of countnes, especially Siermgfeone On the other hand, if the median exceeds the mean, jerlying distribution is negatively skewed because of ane ar more scores with relatively Small vallaes FIGURE 3.2 Mode. median. ard mean in posizively and negarively skewed distributions. Special Stams ofthe Meun 24As has been seen, the mean sometimes finks ta desenbe the typical or midife-ranked value ofa distribution. Therefore, it should be used in conjunction with another average, such as the median. Tnthe long run, however, the mean is the single most preferred average for quantitative data. le Distrihntion Ix Not Skewed When a distribution of scores is not too skewed, the values of the mode, median, and meatf ate similar, and any of them can be used lo describe the central tendency a the destribuy 24. What is mid-spread and quartiles? The range of the middle 50 per cent of the distribution is a commott because it concentrates on the middle cases. It is quite stable from ints which divide the distribution into quarters are called the quartiles ‘founths’) The lower quartile is usually denoted QI.and the upper quartile is of course the median.) The distance between QL and QO is called the the “interquartile range’), ar the dQ for short Figure 2.3C Men's working hours ranked and showing the upper and lower quarries P{The middle quantile: -spread (sometimes ‘methods available for serve as valid measures of variability of the system, first one ‘the Inter Quartite Range (IQR), second one is the variance and the third one és the Standard eviation. rolesare reserved for the variance and particularly for its square roal, the skindard deviation, because these measures serve as key components for other important statistical measures. The 25vanance and standard devistion nccupy the same exalted position among measures of variability as does the mean among measures of central tendency. 26. Explain the method of computation of IQR with example ? Table 4.5 CALCULATION OF THE t0R A. INSTRUCTIONS 1 Order scores from least to most. 2 To determine how far to penetrate the set of ordered scores, begin at either then add 11a the total number of scores and divide by 4, I necessary, round resuttto the nearest whole number. 3. Beginning with the largest score, count the requisite number off in step 2) into the ordered scores to find the location of the third 4 The third quarsle equals the value of the score at this location, 5 Beginning with the smallest score, again court the requisite the ordered scores to find the location of the first quart. & The fist quartile equals the value ofthe score at this 7 The OR equals the third quartile minus the first, |. EXAMPLE 1 7,9,9, 10,11, 11,13 2741) 37,9, 9,10, 11, 19,13 27. What is star The st siaion essentially calculates a typical value of these distances from the mean. It confenti jeted s, and defined as: deviations from the mean are squared. summed and divided by the sample size (well, N - | actually, for technical reasons), and then the square root is taken to return to the original units. The order in which the calculations are pertormed 1s very important As always, calculations within 28brackets are performed First, then mul bpheation and diviston, then addition {meludimg, summation), and subtraction. Without the square root, the measure is called the variance, The layout for a worksheet to calculate the standard deviation of the hours worked by this small sample of' men is shown in table. Figure 24 Worksheet for standard deviation of men’s weekly working hours. y-¥ 8 ~6 36 40 “ul Py a “4 196, 48. -3 Q o 16 16 55 4 16 55 ‘ 16 90 » 841 9 361 n of squared residuals = 2472 ohh mean calculated at the and their squared values are placed values 1 shox at the toot af column 3, andl from rt the Sunrce: As gure 2.34. The eriginal data aa a728. What are Camulative Frequency Distribution ? Given an example. A frequency distribution showing the total number of observations in each class and all lower- ranked classes. Weight of different persons, and their frequency. and cumulative frequeney, cumulative percent i shown in the following table, QO WEIGHT 240-249 BO-28 220-29 210-219 re CF —the cumulative frequency—is the count of all scores less than or equal to the score of inferest, F is the flequency for the score of interest. and N is the number of scores in the distribution. 28Tt 1 10 9 6 0 9 9 5 2 9 8 4 2 7 60 3 3 5 3 2 1 2 15 1 1 1 5 Percentile ranks (PR) for a distribution of 1! 30. What are an extreme? ‘The tap and bottom of JH. What are deciles? |, he percentiles are the quantiles at depth N’100. and so on. an the Individual and aygregate level data with suitable example. ‘mien or’individdual level “data” are the specific entries, ‘The data are micra data itis possible to extract a small sample and examine the actual working hours of specific individuals within the General Household Survey dataset 2reregate means that some analysis has already been carried out, and that the data are summarized in some way rather than being provided in a raw form. For example, in contrast to the General Household Survey, data from the Annual Survey of Hours and Eamings in Britain (which replaced the New Farnings Survey m 2044) are nat generally avantable at the inehividual level 35. What are all the duties of Data Analysts. Duta analysts have to lear to be cnucal of ihe measures avunlable to them, but in a manner As well ax asking "Are there any errors m this measure” we also have ( anything better available” and. if not, How can I improve what I've got” Sf. Expl ‘the concept of Adding ar subtracting, a constant. One way of Tucusing attention on a particular feature af a dataset is 10 WBA yx from every data value. For exaruple,inaset of data on weekly family incGahé it would be possible to subtract the median ftom each of the data values, thus drawing attention io which families had incomes below orahave a hypothene! typical firmly, 37. Faplain the concept of Multiplying or dividing A. ‘We could change each dara point by moltinlvinggeMtiyine i i. (constant. A common example of bra the reecpresnon a ene surency fine Pt or earl on der to cone pounds to US dollars, the pounds are ates ye the§gurtent exchange rate. Multiplying or dividing each of the values has a more adding or subtracting. The result of multiplying or dividing by a const @ variable by a factor, evenly stretching, or shrinking the axis Tike a piece of is, Tel us see what happens if data Fram the General Household § We who classify themselves aS ‘alcohol consumption. as moderate or heavy which has been standardized in this way is forced to have a mean or median of O and a deviation or mid-spread of 1 isis achieved by standardizing each score. One common way of standardizing is to first subwact jean from each data value, and then divide the result by the standard deviation This process is summarized by the following formula, where the original variable 'Y" becomes the standardized variable 'Z': 30Z = (Y,—Y)/St. Dev. For example, the first value of 31 in the reading rest becomes: ‘This is shown in the fallow Figure 3.5 Scores of reading and mathematics tests at age 16. (31 — 25.377 or 0.8 1 2 3 4 5 Raw reading Raw maths Standardized Standardized mposice score score score reading score sore of attainment 31 7 08 061 LAL 3 1 212 31 3 1:98 30 18 0.84 28 O18 0.55 31 -025 0.55 29 -068 -016 23 0.61 0.98 23 06 1.02 25 0.02 19 we 139 32 Ls 2.69 3 132 212 29 -0.68 0.16 30 ast 127 eafy the value of the extremes and the number of cases fir it te be reproduced accurately 1 Child Development Study respandents at age 16. ‘normal chstribution or umafarm distribution, or example, #f the disteabution flat (a uniform distribution), this would be possible. We would only need to amd it would be possible to say exactly what proportion of the cases fell above and below # certsm In order to summarize the shape of a distribution suceinetly, it would need to be simple enough to. beable to specif how it should be drawn in a very few'statements. For example, ifthe distribution aLwere completely flat (a uniform disinbunon), this would be possible We would only need w- specify the value of the extremes and the number of cases for it to be reproduced accurately, and iL would be possible to say exuely what proportion of tbe cases fell above and below a certain. level. and which contains fixed proportions of the distribution st different distances from the center. The two curves in following figure look different - (a) has a smaller spread than (b) - but an fact they only chilfer by asculing, factor. OQ Number of fa) The Gaussian distribution. ‘igure 3.9 Fixed proportions under the Gaussian curve with mean O and standard fiation 1. a2® 68 per cent of the cases lie within one standard deviation unit uf the mean » 95 per cent of the cases lie within two standard deviations units of the mean » 99:7 per cent of the cases lie within three standard deviation units of the mean Example for standard deviation. m0. Sed. Dev. = 0.079332 N= 5,746 600: Frequency PN EH ih 1.260 1.300 1490 1.500 1.66 1.700 1.800 1.900 2.000 AD height in merres at 16 years fe i HH 40. D Standardizing distributions with respect to an appropriate base. dardiiving techmques considered up to now, the same numerical adjustment fe tg each of the values in a batch of data. Sometimes, however, it can be useful to eplual adjustment to each dala value, winch may involve a diffrent nurnber a tease s the figures stund, the mast dominant feature of the dataw is a ruther uninteresting ome: the ge in the value of the pound. While the median and mid-spreads of the money incomes each ‘year have increased substantially in this period, real incomes and differentials almost certainly 33have not. How could we present the dats m order to focus on the trend m real income differentials, over time? A dataset which can be viewed from several angles is shown in figure 3.13: the value of the lower quartile, the median and the upper quartile of male and female earnings in the period between 199) and 2000. The data are drawn from the New Earnings Survey that collects information earings ina fixed period each year from the employers of a large sample of emplayees. Figure 3.13 Male and female earnings 1990-2000; grass earnings in pounds week for full-time workers on adult rates whose pay was not by absence. Male earnings Year M 1990 258.2 1991 277.5 1992 2959 1993 304.6 309. 1994 312.8 320.1 1995 323.2 3325, 199% 3349 3473 1997 349.7 364.7 1998 362.8 379.1 1999 3743 3982 2000 389.7 476 Source: Exccact fc i aL and A26.2. Figure 3.14 ‘RR, sex. Female earnings: a M Qu 77 100 133 136 7 100 139 100 136 76 100 140 100 137 76 100 119 100 137 76 100 140 ‘3 100 37 76 100 140 596 B 100 138 75 100 140 1997 B 100 137 5 100 140 1998 B 100 138 75 100 140 999 73 100 138 75 100 140 ‘000 3B 100 138 75 190 M1 Source: Re-expression of data in figure 3.13. a4have not. How could we present the dats m order to focus on the trend m real income differentials, over time? A dataset which can be viewed from several angles is shown in figure 3.13: the value of the lower quartile, the median and the upper quartile of male and female earnings in the period between 199) and 2000. The data are drawn from the New Earnings Survey that collects information earings ina fixed period each year from the employers of a large sample of emplayees. Figure 3.13 Male and female earnings 1990-2000; grass earnings in pounds week for full-time workers on adult rates whose pay was not by absence. Male earnings Year M 1990 258.2 1991 277.5 1992 2959 1993 304.6 309. 1994 312.8 320.1 1995 323.2 3325, 199% 3349 3473 1997 349.7 364.7 1998 362.8 379.1 1999 3743 3982 2000 389.7 476 Source: Exccact fc i aL and A26.2. Figure 3.14 ‘RR, sex. Female earnings: a M Qu 77 100 133 136 7 100 139 100 136 76 100 140 100 137 76 100 119 100 137 76 100 140 ‘3 100 37 76 100 140 596 B 100 138 75 100 140 1997 B 100 137 5 100 140 1998 B 100 138 75 100 140 999 73 100 138 75 100 140 ‘000 3B 100 138 75 190 M1 Source: Re-expression of data in figure 3.13. a4One appreach would be to treat the distribution of incomes for euch sex in wach year as a separate distribution, and express each of the quartiles relative to the median, The result of doing this is ven in fivure 3.14. ‘The figure of 75 for the QL for men in 1990, for example. was obtained by dividing £193 by £258 and multiplying the result by 100. All of the results have been rounded to the nearest pound (£). In this relative to medians income computation, it is the male and fernale imcome almost equal to other ane, “This is one method for standardivabon wath a base, Q Z App s 35