You are on page 1of 42

Excel For Statistical Data Analysis

MENU
1. Introduction 2. Entering Data 3. Descriptive Statistics 4. Normal Distribution 5. Confidence Interval for t e !ean ". #est of $%pot esis Concerning t e &opulation !ean '. Difference (et)een !ean of #)o &opulations *. +N,-+. +nal%sis of -ariances /. 0oodness1of12it #est for Discrete 3andom -ariables 14. #est of Independence. Contingenc% #ables 11. #est $%pot esis Concerning t e -ariance of #)o &opulations 12. 5inear Correlation and 3egression +nal%sis 13. !oving +verage and E6ponential Smoot ing 14. +pplications and Numerical E6amples 15. E15abs to 2ull% 7nderstand Statistical Concepts 1". Interesting and 7seful Sites Companion Sites:

#opics in Statistical Data +nal%sis #ime Series +nal%sis and (usiness 2orecasting Computers and Computational Statistics 8uestionnaire Design and Surve%s Sampling &robabilistic !odeling S%stems Simulation &robabilit% and Statistics 3esources # e (usiness Statistics ,nline Course

Introduction
# is site provides illustrative e6perience in t e use of E6cel for data summar%9 presentation9 and for ot er basic statistical anal%sis. I believe t e popular use of E6cel is on t e areas ) ere E6cel reall% can e6cel. # is includes organi:ing data9 i.e. basic data management9 tabulation and grap ics. 2or real statistical anal%sis on must learn using t e professional commercial statistical pac;ages suc as S+S9 and S&SS. Microsoft Excel 2000 <version /= provides a set of data anal%sis tools called t e Analysis ToolPak ) ic %ou can use to save steps ) en %ou develop comple6 statistical anal%ses. >ou provide t e data and parameters for eac anal%sis? t e tool uses t e appropriate statistical macro functions and t en displa%s t e results in an output table. Some tools generate c arts in addition to output tables.

If t e Data +nal%sis command is selectable on t e #ools menu9 t en t e +nal%sis #ool&a; is installed on %our s%stem. $o)ever9 if t e Data +nal%sis command is not on t e #ools menu9 %ou need to install t e +nal%sis #ool&a; b% doing t e follo)ing. Step : ,n t e #ools menu9 clic; +dd1Ins.... If +nal%sis #ool&a; is not listed in t e +dd1Ins dialog bo69 clic; (ro)se and locate t e drive9 folder name9 and file name for t e +nal%sis #ool&a; +dd1in @ +nal%s32.6ll @ usuall% located in t e &rogram 2ilesA!icrosoft ,fficeA,fficeA5ibrar%A+nal%sis folder. ,nce %ou find t e file9 select it and clic; ,B. Step 2: If %ou donCt find t e +nal%s32.6ll file9 t en %ou must install it. 1. Insert %our !icrosoft ,ffice 2444 Dis; 1 into t e CD 3,! drive. 2. Select 3un from t e Dindo)s Start menu. 3. (ro)se and select t e drive for %our CD. Select Setup.e6e9 clic; ,pen9 and clic; ,B. 4. Clic; t e +dd or 3emove 2eatures button. 5. Clic; t e E ne6t to !icrosoft E6cel for Dindo)s. ". Clic; t e E ne6t to +dd1ins. '. Clic; t e do)n arro) ne6t to +nal%sis #ool&a;. *. Select 3un from !% Computer. /. Select t e 7pdate No) button. 14. E6cel )ill no) update %our s%stem to include +nal%sis #ool&a;. 11. 5aunc E6cel. 12. ,n t e #ools menu9 clic; +dd1Ins... 1 and select t e +nal%sis #ool&a; c ec; bo6. Step !: # e +nal%sis #ool&a; +dd1In is no) installed and Data +nal%sis... )ill no) be selectable on t e #ools menu. !icrosoft E6cel is a po)erful spreads eet pac;age available for !icrosoft Dindo)s and t e +pple !acintos . Spreads eet soft)are is used to store information in columns and ro)s ) ic can t en be organi:ed andFor processed. Spreads eets are designed to )or; )ell )it numbers but often include te6t. E6cel organi:es %our )or; into )or;boo;s? eac )or;boo; can contain man% )or;s eets? )or;s eets are used to list and anal%:e data . E6cel is available on all public1access &Cs <i.e.9 t ose9 e.g.9 in t e 5ibrar% and &C 5abs=. It can be opened eit er b% selecting Start 1 &rograms 1 !icrosoft E6cel or b% clic;ing on t e E6cel S ort Cut ) ic is eit er on %our des;top9 or on an% &C9 or on t e ,ffice #ool bar. "penin# a Doc$ment:

Clic; on 2ile1,pen <CtrlE,= to openFretrieve an e6isting )or;boo;? c ange t e director% area or drive to loo; for files in ot er locations #o create a ne) )or;boo;9 clic; on 2ile1Ne)1(lan; Document.

Sa%in# an& Closin# a Doc$ment:

#o save %our document )it its current filename9 location and file format eit er clic; on 2ile 1 Save. If %ou are saving for t e first time9 clic; 2ile1Save? c ooseFt%pe a name for %our document? t en clic; ,B. +lso use 2ile1Save if %ou )ant to save to a different filenameFlocation. D en %ou ave finis ed )or;ing on a document %ou s ould close it. 0o to t e 2ile menu and clic; on Close. If %ou ave made an% c anges since t e file )as last saved9 %ou )ill be as;ed if %ou )is to save t em. T'e Excel screen

(ork)ooks an& *orks'eets: D en %ou start E6cel9 a blan; )or;s eet is displa%ed ) ic consists of a multiple grid of cells )it numbered ro)s do)n t e page and alp abeticall%1titled columns across t e page. Eac cell is referenced b% its coordinates <e.g.9 +3 is used to refer to t e cell in column + and ro) 3? (14.(24 is used to refer to t e range of cells in column ( and ro)s 14 t roug 24=. >our )or; is stored in an E6cel file called a )or;boo;. Eac )or;boo; ma% contain several )or;s eets andFor c arts 1 t e current )or;s eet is called t e active s eet. #o vie) a different )or;s eet in a )or;boo; clic; t e appropriate S eet #ab. >ou can access and e6ecute commands directl% from t e main menu or %ou can point to one of t e toolbar buttons <t e displa% bo6 t at appears belo) t e button9 ) en %ou place t e cursor over it9 indicates t e nameFaction of t e button= and clic; once. Mo%in# Aro$n& t'e (orks'eet: It is important to be able to move around t e )or;s eet effectivel% because %ou can onl% enter or c ange data at t e position of t e cursor. >ou can move t e cursor b% using t e arro) ;e%s or b% moving t e mouse to t e reGuired cell and clic;ing. ,nce selected t e cell becomes t e active cell and is identified b% a t ic; border? onl% one cell can be active at a time.

#o move from one )or;s eet to anot er clic; t e s eet tabs. <If %our )or;boo; contains man% s eets9 rig t1clic; t e tab scrolling buttons t en clic; t e s eet %ou )ant.= # e name of t e active s eet is s o)n in bold. Mo%in# +et*een Cells: $ere is a ;e%board s ortcuts to move t e active cell.

$ome 1 moves to t e first column in t e current ro) CtrlE$ome 1 moves to t e top left corner of t e document End t en $ome 1 moves to t e last cell in t e document

#o move bet)een cells on a )or;s eet9 clic; an% cell or use t e arro) ;e%s. #o see a different area of t e s eet9 use t e scroll bars and clic; on t e arro)s or t e area aboveFbelo) t e scroll bo6 in eit er t e vertical or ori:ontal scroll bars. Note t at t e si:e of a scroll bo6 indicates t e proportional amount of t e used area of t e s eet t at is visible in t e )indo). # e position of a scroll bo6 indicates t e relative location of t e visible area )it in t e )or;s eet.

Entering Data
+ ne) )or;s eet is a grid of ro*s and col$mns. # e ro)s are labeled )it numbers9 and t e columns are labeled )it letters. Eac intersection of a ro) and a column is a cell. Eac cell as an a&&ress9 ) ic is t e column letter and t e ro) number. # e arro) on t e )or;s eet to t e rig t points to cell +19 ) ic is currentl% 'i#'li#'te&9 indicating t at it is an acti%e cell. + cell must be active to enter information into it. #o ig lig t <select= a cell9 clic; on it. #o select more t an one cell.

Clic; on a cell <e.g. +1=9 t en old t e s ift ;e% ) ile %ou clic; on anot er <e.g. D4= to select all cells bet)een and including +1 and D4. Clic; on a cell <e.g. +1= and drag t e mouse across t e desired range9 unclic;ing on anot er cell <e.g. D4= to select all cells bet)een and including +1 and D4. #o select several cells ) ic are not adHacent9 press IcontrolI and clic; on t e cells %ou )ant to select. Clic; a number or letter labeling a ro) or column to select t at entire ro) or column.

,ne )or;s eet can ave up to 25" columns and "5953" ro)s9 so itCll be a ) ile before %ou run out of space. Eac cell can contain a la)el9 %al$e9 lo#ical %al$e9 or form$la.

5abels can contain an% combination of letters9 numbers9 or s%mbols. -alues are numbers. ,nl% values <numbers= can be used in calculations. + value can also be a date or a time 5ogical values are ItrueI or Ifalse.I

2ormulas automaticall% do calculations on t e values in ot er specified cells and displa% t e result in t e cell in ) ic t e formula is entered <for e6ample9 %ou can specif% t at cell D3 is to contain t e sum of t e numbers in (3 and C3? t e number displa%ed in D3 )ill t en be a funtion of t e numbers entered into (3 and C3=.

#o enter information into a cell9 select t e cell and begin t%ping. Note t at as %ou t%pe information into t e cell9 t e information %ou enter also displa%s in t e formula bar. >ou can also enter information into t e formula bar9 and t e information )ill appear in t e selected cell. D en %ou ave finis ed entering t e label or value.

&ress IEnterI to move to t e ne6t cell belo) <in t is case9 +2= &ress I#abI to move to t e ne6t cell to t e rig t <in t is case9 (1= Clic; in an% cell to select it

Enterin# ,a)els

7nless t e information %ou enter is formatted as a value or a formula9 E6cel )ill interpret it as a label9 and defaults to align t e te6t on t e left side of t e cell. If %ou are creating a long )or;s eet and %ou )ill be repeating t e same label information in man% different cells9 %ou can use t e A$toComplete function. # is function )ill loo; at ot er entries in t e same column and attempt to matc a previous entr% )it %our current entr%. 2or e6ample9 if %ou ave alread% t%ped IDesle%anI in anot er cell and %ou t%pe IDI in a ne) cell9 E6cel )ill automaticall% enter IDesle%an.I If %ou intended to t%pe IDesle%anI into t e cell9 %our tas; is done9 and %ou can move on to t e ne6t cell. If %ou intended to t%pe somet ing else9 e.g. IDilliams9I into t e cell9 Hust continue t%ping to enter t e term. #o turn on t e +utoComplete funtion9 clic; on I#oolsI in t e menu bar9 t en select I,ptions9I t en select IEdit9I and clic; to put a c ec; in t e bo6 beside IEnable +utoComplete for cell values.I

+not er )a% to Guic;l% enter repeated labels is to use t e Pick ,ist feature. 3ig t clic; on a cell9 t en select I&ic; 2rom 5ist.I # is )ill give %ou a menu of all ot er entries in cells in t at column. Clic; on an item in t e menu to enter it into t e currentl% selected cell. Enterin# -al$es + value is a number9 date9 or time9 plus a fe) s%mbols if necessar% to furt er define t e numbers Jsuc as. . 9 E 1 < = K L F M. N$m)ers are assumed to be positive? to enter a negative number9 use a minus sign I1I or enclose t e number in parent eses I<=I. Dates are stored as !!FDDF>>>>9 but %ou do not ave to enter it precisel% in t at format. If %ou enter IHan /I or IHan1/I9 E6cel )ill recogni:e it at Nanuar% / of t e current %ear9 and store it as 1F/F2442. Enter t e four1digit %ear for a %ear ot er t an t e current %ear <e.g. IHan /9 1///I=. #o enter t e current da%Cs date9 press IcontrolI and I?I at t e same time. Times default to a 24 our cloc;. 7se IaI or IpI to indicate IamI or IpmI if %ou use a 12 our cloc; <e.g. I*.34 pI is interpreted as *.34 &!=. #o enter t e current time9 press IcontrolI and I.I <s ift1semicolon= at t e same time.

+n entr% interpreted as a value <number9 date9 or time= is aligned to t e rig t side of t e cell9 to reformat a value. .o$n&in# N$m)ers t'at Meet Specifie& Criteria: #o appl% colors to ma6imum andFor minimum values. 1. Select a cell in t e region9 and press CtrlES iftEO <in E6cel 24439 press t is or CtrlE+= to select t e Current 3egion. 2. 2rom t e 2ormat menu9 select Conditional 2ormatting. 3. In Condition 19 select 2ormula Is9 and t%pe P!+Q<L2.L2= PL21. 4. Clic; 2ormat9 select t e 2ont tab9 select a color9 and t en clic; ,B. 5. In Condition 29 select 2ormula Is9 and t%pe P!IN<L2.L2= PL21. ". 3epeat step 49 select a different color t an %ou selected for Condition 19 and t en clic; ,B. Note: (e sure to distinguis bet)een absolute reference and relative reference ) en entering t e formulas. .o$n&in# N$m)ers t'at Meet Specifie& Criteria

Pro)lem: 3ounding all t e numbers in column + to :ero decimal places9 e6cept for t ose t at ave I5I in t e first decimal place. Sol$tion: 7se t e I29 !,D9 and 3,7ND functions in t e follo)ing formula. PI2<!,D<+291=P4.59+293,7ND<+294== To Copy an& Paste All Cells in a S'eet 1. Select t e cells in t e s eet b% pressing CtrlE+ <in E6cel 24439 select a cell in a blan; area before pressing CtrlE+9 or from a selected cell in a Current 3egionF5ist range9 press CtrlE+E+=. ". Clic; Select +ll at t e top1left intersection of ro)s and columns. 2. &ress CtrlEC. 3. &ress CtrlE&age Do)n to select anot er s eet9 t en select cell +1. 4. &ress Enter. To Copy t'e Entire S'eet Cop%ing t e entire s eet means cop%ing t e cells9 t e page setup parameters9 and t e defined range Names. "ption : 1. !ove t e mouse pointer to a s eet tab. 2. &ress Ctrl9 and old t e mouse to drag t e s eet to a different location. 3. 3elease t e mouse button and t e Ctrl ;e%. "ption 2: 1. 3ig t1clic; t e appropriate s eet tab. 2. 2rom t e s ortcut menu9 select !ove or Cop%. # e !ove or Cop% dialog bo6 enables one to cop% t e s eet eit er to a different location in t e current )or;boo; or to a different )or;boo;. (e sure to mar; t e Create a cop% c ec;bo6. "ption !: 1. 2rom t e Dindo) menu9 select +rrange. 2. Select #iled to tile all open )or;boo;s in t e )indo). 3. 7se ,ption 1 <dragging t e s eet ) ile pressing Ctrl= to cop% or move a s eet. Sortin# )y Col$mns # e default setting for sorting in +scending or Descending order is b% ro). #o sort b% columns. 1. 2rom t e Data menu9 select Sort9 and t en ,ptions. 2. Select t e Sort left to rig t option button and clic; ,B.

3. In t e Sort b% option of t e Sort dialog bo69 select t e ro) number b% ) ic t e columns )ill be sorted and clic; ,B.

Descriptive Statistics
# e Data +nal%sis #ool&a; as a Descriptive Statistics tool t at provides %ou )it an eas% )a% to calculate summar% statistics for a set of sample data. Summar% statistics includes !ean9 Standard Error9 !edian9 !ode9 Standard Deviation9 -ariance9 Burtosis9 S;e)ness9 3ange9 !inimum9 !a6imum9 Sum9 and Count. # is tool eliminates t e need to t%pe indivividual functions to find eac of t ese results. E6cel includes elaborate and customisable toolbars9 for e6ample t e IstandardI toolbar s o)n ere.

Some of t e icons are useful mat ematical computation.


is t e I+utosumI icon9 ) ic enters t e formula IPsum<=I to add up a range of cells. is t e I2unctionDi:ardI icon9 ) ic gives %ou access to all t e functions available. is t e I0rap Di:ardI icon9 giving access to all grap t%pes available9 as s o)n in t is displa%.

E6cel can be used to generate measures of location and variabilit% for a variable. Suppose )e )is to find descriptive statistics for a sample data. 29 49 "9 and *. Step 1. Select t e #ools Opull1do)n menu9 if %ou see &ata analysis/ click on t'is option/ ot'er*ise/ click on a&&0in11 option to install anal%sis tool pa;. Step 2. Clic; on t e data anal%sis option. Step 3. C oose Descripti%e Statistics from +nal%sis #ools list. Step 4. D en t e dialog bo6 appears.

Enter +1.+4 in t e inp$t ran#e bo69 A is a value in col$mn A and ro* 9 in t is case t is value is 2. 7sing t e same tec niGue enter ot er -+57ES until %ou reac t e last one. If a sample consists of 24 numbers9 %ou can select for e6ample +19 +29 +39 etc. as t e input range. Step 5. Select an o$tp$t ran#e9 in t is case (1. Clic; on summar% statistics to see t e results. Select "2. D en %ou clic; "29 %ou )ill see t e result in t e selected range. +s %ou )ill see9 t e mean of t e sample is 59 t e median is 59 t e standard deviation is 2.5*1/*/9 t e sample variance is "."""""'9t e range is " and so on. Eac of t ese factors mig t be important in %our calculation of different statistical procedures.

Normal Distribution
Consider t e problem of finding t e probabilit% of getting less t an a certain value under an% normal probabilit% distribution. +s an illustrative e6ample9 let us suppose t e S+# scores nation)ide are normall% distributed )it a mean and standard deviation of 544 and 1449 respectivel%. +ns)er t e follo)ing Guestions based on t e given information. +. D at is t e probabilit% t at a randoml% selected student score )ill be less t an "44 pointsR (. D at is t e probabilit% t at a randoml% selected student score )ill e6ceed "44 pointsR C. D at is t e probabilit% t at a randoml% selected student score )ill be bet)een 444 and "44R $int. 7sing E6cel %ou can find t e probabilit% of getting a value appro6imatel% less t an or eGual to a given value. In a problem9 ) en t e mean and t e standard deviation of t e population are given9 %ou ave to use common sense to find different probabilities based on t e Guestion since %ou ;no) t e area under a normal curve is 1. Sol$tion: In t e )or; s eet9 select t e cell ) ere %ou )ant t e ans)er to appear. Suppose9 %ou c ose cell number one9 +1. 2rom t e menus9 select Iinsert pull1do)nI. Steps 20! 2rom t e menus9 select insert9 t en clic; on t e 2unction option. Step 31 +fter clic;ing on t e 2unction option9 t e &aste 2unction dialog appears from 2unction Categor%. C oose Statistical t en NORMDIST from t e Function Name bo6? Clic; OK Step 41 +fter clic;ing on ,B9 t e N,3!DIS# distribution bo6 appears. i. Enter "44 in Q <t e value bo6=?

ii. Enter 544 in t e !ean bo6? iii. Enter 144 in t e Standard deviation bo6? iv. #%pe ItrueI in t e cumulative bo69 t en clic; ,B. +s %ou see t e value 4.*41344'4 appears in +19 indicating t e probabilit% t at a randoml% selected studentCs score is belo) "44 points. 7sing common sense )e can ans)er part IbI b% subtracting 4.*41344'4 from 1. So t e part IbI ans)er is 11 4.*4134'4 or 4.15*"53. # is is t e probabilit% t at a randoml% selected studentCs score is greater t an "44 points. #o ans)er part IcI9 use t e same tec niGues to find t e probabilities or area in t e left sides of values "44 and 444. Since t ese areas or probabilities overlap eac ot er to ans)er t e Guestion %ou s ould subtract t e smaller probabilit% from t e larger probabilit%. # e ans)er eGuals 4.*41344'4 1 4.15*"552" t at is9 4."*2"/. # e screen s ot s ould loo; li;e follo)ing. 5n%erse Case Calculating t e value of a random variable often called t e I6I value >ou can use N".M5N- from t e function bo6 to calculate a value for t e random variable 1 if t e probabilit% to t e left side of t is variable is given. +ctuall%9 %ou s ould use t is function to calculate different percentiles. In t is problem one could as; ) at is t e score of a student ) ose percentile is /4R # is means appro6imatel% /4K of students scores are less t an t is number. ,n t e ot er and if )e )ere as;ed to do t is problem b% and9 )e )ould ave ad to calculate t e 6 value using t e normal distribution formula 6 P m E :d. No) letCs use E6cel to calculate &/4. In t e &aste function9 dialog clic; on statistical9 t en clic; on N".M5N-. # e screen s ot )ould loo; li;e t e follo)ing. D en %ou see N".M5N- t e dialog bo6 appears. i. Enter 4./4 for t e probabilit% <t is means t at appro6imatel% /4K of studentsC score is less t an t e value )e are loo;ing for= ii. Enter 544 for t e mean <t is is t e mean of t e normal distribution in our case= iii. Enter 144 for t e standard deviation <t is is t e standard deviation of t e normal distribution in our case= +t t e end of t is screen %ou )ill see t e formula result ) ic is appro6imatel% "2* points. # is means t e top 14K of t e students scored better t an "2*.

Confidence Interval for the Mean


Suppose )e )is for estimating a confidence interval for t e mean of a population. Depending on t e si:e of %our sample si:e %ou ma% use one of t e follo)ing cases. ,ar#e Sample Si6e 7n is lar#er t'an/ say !08: # e general formula for developing a confidence interval for a population means is.

In t is formula is t e mean of t e sample? S is t e interval coefficient9 ) ic can be found from t e normal distribution table <for e6ample t e interval coefficient for a /5K confidence level is 1./"=. S is t e standard deviation of t e sample and n is t e sample si:e. No) )e )ould li;e to s o) o) E6cel is used to develop a certain confidence interval of a population mean based on a sample information. +s %ou see in order to evaluate t is formula %ou need It e mean of t e sampleI and t e margin of error E6cel )ill automaticall% calculate t ese Guantities for %ou. # e onl% t ings %ou ave to do are. add t e margin of error to t e mean of t e sample9 ? 2ind t e upper limit of t e interval and subtract t e margin of error from t e mean to t e lo)er limit of t e interval. #o demonstrate o) E6cel finds t ese Guantities )e )ill use t e data set9 ) ic contains t e ourl% income of 3" )or;1stud% students ere9 at t e 7niversit% of (altimore. # ese numbers appear in cells +1 to +3" on an E6cel )or; s eet. +fter entering t e data9 )e follo)ed t e descriptive statistic procedure to calculate t e un;no)n Guantities. # e onl% additional step is to clic; on t e confidence interval in t e descriptive statistics dialog bo6 and enter t e given confidence level9 in t is case /5K. $ere is9 t e above procedures in step1b%1step. Step 1. Enter data in cells +1 to +3" <on t e spreads eet= Step 2. 2rom t e menus select Tools Step 3. Clic; on Data Analysis t en c oose t e Descripti%e Statistics option t en clic; "2. ,n t e descriptive statistics dialog9 clic; on Summar% Statistic. +fter %ou ave done t at9 clic; on t e confidence interval level and t%pe /5K 1 or in ot er problems ) atever confidence interval %ou desire. In t e ,utput 3ange bo6 enter (1 or ) at ever location %ou desire. No) clic; on "2. # e screen s ot )ould loo; li;e t e follo)ing.

+s %ou see9 t e spreads eet s o)s t at t e mean of t e sample is

P "./42'''''*

and t e absolute value of t e margin of error P 4.231"'*14/. # is mean is based on t is sample information. + /5K confidence interval for t e ourl% income of t e 7( )or;1stud% students as an upper limit of "./42'''''* E 4.231"'*14/ and a lo)er limit of "./42'''''* 1 4.231"'*14/. ,n t e ot er and9 )e can sa% t at of all t e intervals formed t is )a% /5K contains t e mean of t e population. ,r9 for practical purposes9 )e can be /5K confident t at t e mean of t e population is bet)een "./42'''''* 1 4.231"'*14/ and "./42'''''* E 4.231"'*14/. De can be at least /5K confident t at interval JL"."* and L'.13M contains t e average ourl% income of a )or;1stud% student. Smal Sample Si6e 7say less t'an !08 If t e sample n is less t an 34 or )e must use t e small sample procedure to develop a confidence interval for t e mean of a

population. # e general formula for developing confidence intervals for t e population mean based on small a sample is.

In t is formula is t e mean of t e sample. is t e interval coefficient providing an area of in t e upper tail of a t distribution )it n11 degrees of freedom ) ic can be found from a t distribution table <for e6ample t e interval coefficient for a /4K confidence level is 1.*33 if t e sample is 14=. S is t e standard deviation of t e sample and n is t e sample si:e. No) %ou )ould li;e to see o) E6cel is used to develop a certain confidence interval of a population mean based on t is small sample information. +s %ou see9 to evaluate t is formula %ou need margin of error )a% it did for large samples. It e mean of t e sampleI and t e

E6cel )ill automaticall% calculate t ese Guantities t e

+gain9 t e onl% t ings %ou ave to do are. add t e margin of error to t e mean of t e sample9 9 find t e upper limit of t e interval and to subtract t e margin of error from t e mean to find t e lo)er limit of t e interval. #o demonstrate o) E6cel finds t ese Guantities )e )ill use t e data set9 ) ic contains t e ourl% incomes of 14 )or;1stud% students ere9 at t e 7niversit% of (altimore. # ese numbers appear in cells +1 to +14 on an E6cel )or; s eet. +fter entering t e data )e follo) t e descriptive statistic procedure to calculate t e un;no)n Guantities <e6actl% t e )a% )e found Guantities for large sample=. $ere %ou are )it t e procedures in step1b%1step form. Step 1. Enter data in cells +1 to +14 on t e spreads eet Step 2. 2rom t e menus select Tools Step 3. Clic; on Data Analysis t en c oose t e Descripti%e Statistics option. Clic; "2 on t e descriptive statistics dialog9 clic; on Summar% Statistic9 clic; on t e confidence interval level and t%pe in /4K or in ot er problems ) ic ever confidence interval %ou desire. In t e ,utput 3ange bo69 enter (1 or ) atever location %ou desire. No) clic; on "2. # e screen s ot )ill loo; li;e t e follo)ing.

No)9 li;e t e calculation of t e confidence interval for t e large sample9 calculate t e confidence interval of t e population based on t is small sample information. # e confidence interval is. ".* T 4.41442"142 or L".3/UPPPVL'.21. De can be at least /4K confidant t at t e interval JL".3/ and L'.21M contains t e true mean of t e population.

Test of Hypothesis Concerning the Population Mean


+gain9 )e must distinguis t)o cases )it respect to t e si:e of %our sample ,ar#e Sample Si6e 7say/ o%er !08: In t is section %ou )is to ;no) o) E6cel can be used to conduct a %pot esis test about a population mean. De )ill use t e ourl% incomes of different )or;1stud% students t an t ose introduced earlier in t e

confidence interval section. Data are entered in cells +1 to +3". # e obHective is to test t e follo)ing N$ll and Alternati%e %pot esis.

# e null %pot esis indicates t at t e average ourl% income of a )or;1stud% student is eGual to L' per our? o)ever9 t e alternative %pot esis indicates t at t e average ourl% income is not eGual to L' per our. I )ill repeat t e steps ta;en in descriptive statistics and at t e ver% end )ill s o) o) to find t e value of t e test statistics in t is case9 :9 using a cell formula. Step 1. Enter data in cells +1 to +3" <on t e spreads eet= Step 2. 2rom t e menus select Tools Step 3. Clic; on Data Analysis t en c oose t e Descripti%e Statistics option9 clic; "2. ,n t e descriptive statistics dialog9 clic; on Summar% Statistic. Select t e "$tp$t .an#e bo69 enter (1 or ) ic ever location %ou desire. No) clic; "2. <#o calculate t e value of t e test statistics searc for t e mean of t e sample t en t e standard error. In t is output9 t ese values are in cells C3 and C4.= Step 4. Select cell D1 and enter t e cell formula P <C3 1 '=FC4. # e screen s ot s ould loo; li;e t e follo)ing.

# e value in cell D1 is t e value of t e test statistics. Since t is value falls in acceptance range of 11./" to 1./" <from t e normal distribution table=9 )e fail to reHect t e null %pot esis. Small Sample Si6e 7say/ less t'an !08: 7sing steps ta;en t e large sample si:e case9 E6cel can be used to conduct a %pot esis for small1sample case. 5etCs use t e ourl% income of 14 )or;1stud% students at 7( to conduct t e follo)ing %pot esis.

# e null %pot esis indicates t at average ourl% income of a )or;1stud% student is eGual to L' per our .# e alternative %pot esis indicates t at average ourl% income

is not eGual to L' per our. I )ill repeat t e steps ta;en in descriptive statistics and at t e ver% end )ill s o) o) to find t e value of t e test statistics in t is case ItI using a cell formula. Step 1. Enter data in cells +1 to +14 <on t e spreads eet= Step 2. 2rom t e menus select Tools Step 3. Clic; on Data Analysis t en c oose t e Descripti%e Statistics option. Clic; "2. ,n t e descriptive statistics dialog9 clic; on Summar% Statistic. Select t e ,utput 3ange bo6es9 enter (1 or ) atever location %ou c ose. +gain9 clic; on "2. <#o calculate t e value of t e test statistics searc for t e mean of t e sample t en t e standard error9 in t is output t ese values are in cells C3 and C4.= Step 4. Select cell D1 and enter t e cell formula P <C3 1 '=FC4. # e screen s ot )ould loo; li;e t e follo)ing.

Since t e value of test statistic t P 14.""*/" falls in acceptance range 12.2"2 to E2.2"2 <from t table9 ) ere P 4.425 and t e degrees of freedom is /=9 )e fail to reHect t e null %pot esis.

Difference

et!een Mean of T!o Populations

In t is section )e )ill s o) o) E6cel is used to conduct a %pot esis test about t e difference bet)een t)o population means assuming t at populations ave eGual variances. # e data in t is case are ta;en from various offices ere at t e 7niversit% of (altimore. I collected t e ourl% income data of 3" randoml% selected )or;1stud% students and 3" student assistants. # e ourl% income range for )or;1stud% students )as L" 1 L* ) ile t e ourl% income range for student assistants )as L"1L/. # e main obHective in t is %pot esis testing is to see ) et er t ere is a significant difference bet)een t e means of t e t)o populations. # e NU,, and t e A,TE.NAT5-E %pot esis is t at t e means are eGual and t e means are not eGual9 respectivel%.

3eferring to t e spreads eet9 I c ose A and A2 as label centers. # e )or;1stud% studentsC ourl% income for a sample si:e 3" are s o)n in cells A2:A!99 and t e student assistantsC ourl% income for a sample si:e 3" is s o)n in cells +2:+!9 Data for (ork St$&y St$&ent: "9 "9 "9 "9 "9 "9 "9 ".59 ".59 ".59 ".59 ".59 ".59 '9 '9 '9 '9 '9 '9 '9 '.59 '.59 '.59 '.59 '.59 '.59 *9 *9 *9 *9 *9 *9 *9 *9 *. Data for St$&ent Assistant: "9 "9 "9 "9 "9 ".59 ".59 ".59 ".59 ".59 '9 '9 '9 '9 '9 '.59 '.59 '.59 '.59 '.59 '.59 *9 *9 *9 *9 *9 *9 *9 *.59 *.59 *.59 *.59 *.59 /9 /9 /9 /. 7se t e Descriptive Statistics procedure to calculate t e variances of t e t)o samples. # e E6cel procedure for testing t e difference bet)een t e t)o population means )ill reGuire information on t e variances of t e t)o populations. Since t e variances of t e t)o populations are un;no)ns t e% s ould be replaced )it sample variances. # e descriptive for bot samples s o) t at t e variance of first sample is s12 P 014443:2 ;9 ) ile t e variance of t e second sample s22 P01<:<93;. *ork0st$&y st$&ent !ean Standard Error !edian !ode Standard Deviation Sample -ariance Burtosis S;e)ness 3ange !inimum !a6imum Sum Count '.45'142*" 4.125/''5' ' * 4.'452/335 4.5554"21* 11.3**'455* 14.4/3'43'5 2 " * 24' 35 st$&ent assistant !ean Standard Error !edian !ode Standard Deviation Sample -ariance Burtosis S;e)ness 3ange !inimum !a6imum Sum Count '.4'142/ 4.1""454 '.5 * 4./*4'5* 4./"/'4* 11.1/2*25 14.413*1/ 3 " / 2"1.5 35

#o conduct t e desired test %pot esis )it E6cel t e follo)ing steps can be ta;en. Step 1 2rom t e menus select Tools t en clic; on t e Data Analysis option. Step 21 D en t e Data Analysis dialog bo6 appears. C'oose 60Test: T*o Sample for means t en clic; ,B Step !1 D en t e 60Test: T*o Sample for means dialog bo6 appears.

Enter A :A!: in t e %aria)le ran#e )ox 7*ork0st$&y st$&ents= 'o$rly income8 Enter + :+!: in t e %aria)le 2 ran#e )ox 7st$&ent assistants= 'o$rly income8 Enter 4 in t e Hypothesis Mean Difference bo6 <if %ou desire to test a mean difference ot er t an 49 enter t at value= Enter t e variance of t e first sample in t e -aria)le -ariance )ox Enter t e variance of t e second sample in t e -aria)le 2 -ariance )ox and select 5abels Enter 4.45 or9 ) atever le%el of si#nificance %ou desire9 in t e Alp'a )ox Select a suitable "$tp$t .an#e for t e results9 I c ose C <9 t en clic; ,B. # e value of test statistic 6>0 1<;34;23 appears in our case in cell D24. # e reHection rule for t is test is : U 11./" or : V 1./" from t e normal distribution table. In t e E6cel output t ese values for a t)o1tail test are :U11./5//"14*2 and :VE1./5//"14*2. Since t e value of t e test statistic :P11./*45*24 is less t an 11./5//"14*2 )e reHect t e null %pot esis. De can also dra) t is conclusion b% comparing t e p1value for a t)o tail 1test and t e alp a value. Since p1value 01039 <0; ! is less t an aP4.45 )e reHect t e null %pot esis. ,verall )e can sa%9 based on t e sample results9 t e t)o populationsC means are different.

Small Samples: n ". n2 are less t'an !0 In t is section )e )ill s o) o) E6cel is used to conduct a %pot esis test about t e difference bet)een t)o population means. 1 0iven t at t e populations ave eGual variances ) en t)o small independent samples are ta;en from bot populations. Similar to t e above case9 t e data in t is case are ta;en from various offices ere at t e 7niversit% of (altimore. I collected ourl% income data of 11 randoml% selected )or;1stud% students and 11 randoml% selected student assistants. # e ourl% income range for bot groups )as similar range9 L" 1 L* and L"1L/. # e main obHective in t is %pot esis testing is similar too9 to see ) et er t ere is a significant difference bet)een t e means of t e t)o populations. # e NU,, and t e +5#E3N+#I-E %pot esis are t at t e means are eGual and t e% are not eGual9 respectivel%. )or;1stud% student student assistant " * '.5 ".5 ' " '.5 * " " / *.5 ' ".5 ' '.5 " *

".5 '

/ '.5V

3eferring to t e spreads eet9 )e c ose A and A2 as label centers. # e )or;1stud% studentsC ourl% income for a sample si:e 11 are s o)n in cells A2:A 29 and t e student assistantsC ourl% income for a sample si:e 11 is s o)n in cells +2:+ 2. 7nli;e previous case9 %ou do not ave to calculate t e variances of t e t)o samples9 E6cel )ill automaticall% calculate t ese Guantities and use t em in t e calculation of t e value of t e test statistic. Similar to t e previous case9 but a bit different in step W 29 to conduct t e desired test %pot esis )it E6cel t e follo)ing steps can be ta;en. Step 1 2rom t e menus select Tools t en clic; on t e Data Analysis option. Step 21 D en t e Data Analysis dialog bo6 appears. C oose t0Test: T*o Sample Ass$min# E?$al -ariances t en clic; ,B Step ! D en t e t0Test: T*o Sample Ass$min# E?$al -ariances &ialo# )ox appears. Enter +1.+12 in t e %aria)le ran#e )ox <)or;1stud% student ourl% income= Enter (1.(12 in t e %aria)le 2 ran#e )ox <student assistant ourl% income= Enter 4 in t e @ypot'esis Mean Difference bo6<if %ou desire to test a mean difference ot er t an :ero9 enter t at value= t en select 5abels Enter 4.45 or9 ) atever level of significance %ou desire9 in t e Alpha bo6 Select a suitable Output Range for t e results9 I c ose C19 t en clic; ,B. # e value of t e test statistic t>0 1!:222<;2; appears9 in our case9 in cell D14. # e reHection rule for t is test is tU12.4*" or tVE2.4*" from t e t &istri)$tion ta)le ) ere t e t value is based on a t distribution )it n11n212 degrees of freedom and ) ere t e area of t e upper one tail is 4.425 < t at is eGual to alp aF2=. In t e E6cel output t e values for a t)o1tail test are tU12.4*5/"24'* and tVE2.4*5/"24'*. Since t e value of t e test statistic tP11.3"222/*2*9 is in an acceptance range of tU12.4*5/"24'* and tVE2.4*5/"24'*9 )e fail to reHect t e null %pot esis. De can also dra) t is conclusion b% comparing t e p1value for a t)o1tail test and t e alp a value. Since t e p0%al$e 01 ;;29 29; is #reater t'an a>0104 a#ain9 )e fail to reHect t e null %pot esis. ,verall )e can sa%9 based on sample results9 t e t)o populationsC means are eGual.

)or;1stud% student student assistant !ean -ariance ,bservations &ooled -ariance Df t Stat &<#UPt= one tail t Critical one tail &<#UPt=t)o tail t Critical t)o tail "./4/4/4/4/ 4.5/4/4/4/1 11 4.**1*1*1*2 24 11.3"222/*2* 4.4/4135"3/ 1.'24'1*444 4.1**2'12'* 2.4*5/"24'* '.454545455 1.1'2'2'2'3 11

$%pot esi:ed !ean Difference 4

"N#$"% "nalysis of $ariances


In t is section t e obHective is to see ) et er or not means of t ree or more populations based on random samples ta;en from populations are eGual or not. +ssuming independents samples are ta;en from normall% distributed populations )it eGual variances9 E6cel )ould do t is anal%sis if %ou c oose one )a% anova from t e menus. De can also c oose +nova. t)o )a% factor )it or )it out replication option and see ) et er t ere is significant difference bet)een means ) en different factors are involved. Sin#le0Factor AN"-A Test In t is case )e )ere interested to see ) et er t ere a significant difference among ourl% )ages of student assistants in t ree different service departments ere at t e 7niversit% of (altimore. Si6 student assistants )ere randoml% )ere selected from t e t ree departments and t eir ourl% )ages )ere recorded as follo)ing. +3C 14.44 *.44 '.54 *.44 '.44 CSI ".54 '.44 '.44 '.54 '.44 #CC /.44 '.44 '.44 '.44 ".54

Enter data in an E6cel )or; s eet starting )it cell +2 and ending )it cell C*. # e follo)ing steps s ould be ta;en to find t e proper output for interpretation. Step 1 2rom t e menus select #ools and clic; on Data +nal%sis option. Step 21 D en data anal%sis dialog appears9 c oose +nova single1factor option? enter +2.C* in t e input range bo6. Select labels in first ro).

Step!1Select an% cell as output<in ere )e selected +11=. Clic; ,B. # e general form of +nova table loo;s li;e follo)ing. Source of -ariation (et)een 0roups Dit in 0roups #otal Suppose t e test is done at level of significance a P 4.459 )e reHect t e null %pot esis. # is means t ere is a significant difference bet)een means of ourl% incomes of student assistants in t ese departments. T'e T*o0*ay AN"-A (it'o$t .eplication In t is section9 t e stud% involves si6 students ) o )ere offered different ourl% )ages in t ree different department services ere at t e 7niversit% of (altimore. # e obHective is to see ) et er t e ourl% incomes are t e same. # erefore9 )e can consider t e follo)ing. 2actor. Department #reatment. $ourl% pa%ments in t e t ree departments (loc;s. Eac student is a bloc; since eac student as )or;ed in t e t ree different departments Student +3C CSI #CC 1 2 3 4 5 " 14.44 '.54 '.44 *.44 '.44 *.44 /.44 *.44 '.44 ".44 ".44 ".44 ".54 ".54 *.44 '.44 *.44 ".44 SS SS#3 SSE df B11 nt1B !S !S#3 !SE 2 !S#F!SE &1value 4.44"'25 2 crit 3."*231""'4

T'e #eneral form of Ano%a ta)le *o$l& look like: Source of -ariation Sum of SGuares Degrees of freedom !ean SGuares 2

#reatment (loc;s Error #otal

SS# SS( SSE SS#

B11 b11 <B11=<b11= nt11

!S# !S( !S(

2P!S#F!SE

To fin& t'e Excel o$tp$t for t'e a)o%e &ata t'e follo*in# steps can )e taken: Step 1 2rom t e menus select #ools and clic; on Data +nal%sis option. Step21 D en data anal%sis bo6 appears. select +nova t)o1factor )it out replication t en Enter +2. D* in t e input range. Select labels in first ro). Step!1 Select an output range <in ere )e selected +11= t en ,B. S7!!+3> 1 2 3 4 5 " C,7N# S7! +-E3+0E -+3I+NCE 3 3 3 3 3 3 24.5 21 1/.5 21.5 23 22 *.1""""' ' ".5 '.1""""' '."""""' '.333333 2.5*3333 1 4.25 4.5*3333 2.333333 1.333333

+3C CSI #CC AN"-A Source of -ariation 3o)s Columns Error SS

" " "

54 43 3*.5

*.333333 '.1""""' ".41"""'

1.4""""' 4."""""' 4.241""'

df !S 5 4./*455" 2 5.5/'222

&1value

2 crit

4./42''* 11.1/444

1./'24"' 4.1"*'/2 3.325*3' 11.25"/* 4.442'52 4.142*1"

4./'2222 14 4.4/'222

#otal

21.4"/44 1'

N,#E. 2P!S#F!SE P4./*455"F4.4/'222 P 1./'24"' 2 P 3.33 from table <5 numerator D2 and 14 denominator D2= Since 1./'24"'U3.33 )e fail to reHect t e null. Concl$sion: # ere is not sufficient evidence to conclude t at ourl% rates differ for t e t ree departments. T*o0(ay AN"-A *it' .eplication 3eferring to t e student assistant and t e )or; stud% ourl% )ages ere at t e universit% of (altimore t e follo)ing data s o)s t e ourl% )ages for t e t)o categories in t ree different departments. +3C ".54 Dor; Stud% ".*4 '.14 '.44 Student +ssistant '.54 *.44 2actors Factor A: Student Hob categor% <in ere t)o different Hob categories e6ists= Factor +: Departments <in ere )e ave t ree departments= 3eplication. # e number of students in eac e6perimental condition. In t is case t ere are t ree replications. Interaction. +3C ".54 Dor; Stud% ".*4 '.14 '.44 Student +ssistant '.54 *.44 S7!!+3> +3C CSI CSI ".14 ".44 ".54 ".*4 '.44 "."4 #CC #CC "./4 '.24 '.14 '.54 '.44 '.14 #otal CSI ".14 ".44 ".54 ".*4 '.44 "."4 #CC "./4 '.24 '.14 '.54 '.44 '.14

Count Sum +verage -ariance Count Sum +verage -ariance #otal 3 22./

3 24.4 ".* 4.4/

3 1/ ".2 4.1 3 24 ".* 4

3 21 '.1 4 3 22 '.2 4.1

/ "4.2 "."/ 4.1/ / "4./ '.21 4.1*

'."3333 4.14333

#otal Count Sum +verage -ariance +N,-+ Source of -ariation Sample<2actor += Columns<2actor (= Interaction Dit in #otal SS 1.22'22 1.*4333 4.3*111 4.'/333 4.245 df 1 2 2 !S 1.2 4./ 4.2 2 1*." 13./ 2.** &1value 4.44141"55' 4.444'41//* 4.4/5443443 2 crit 4.'4'221 3.**52/ 3.**52/ " 43.3 '.21""' 4.2*5"' " 3/ ".5 4.2 " 43 '.1 4

12 4.1 1'

Concl$sion: !ean ourl% income differ b% Hob categor%. !ean ourl% income differ b% department. Interaction is not significant.

&oodness'of'(it Test for Discrete )andom $ariables


# e C@50SAUA.E distribution can be used in a %pot esis test involving a population variance. $o)ever9 in t is section )e )ould li;e to test and see o) close a sample results are to t e e6pected results. Example: T'e M$ltinomial .an&om -aria)le In t is e6ample t e obHective is to see ) et er or not based on a randoml% selected sample information t e standards set for a population is met. # ere are so man%

practical e6amples t at can be used in t is situation. 2or e6ample it is assumed t e guidelines for iring people )it different et nic bac;ground for t e 7S government is set at '4K<D$I#E=9 24K<+frican +merican= and 14K<ot ers=9 respectivel%. + randoml% selected sample of 1444 7S emplo%ees s o)s t e follo)ing results t at is summari:ed in a table. E#$NIC (+CB03,7ND D$I#E +23IC+N +merican ,#$E3S EQ&EC#ED N7!(E3 ,2 ,(SE3-ED 23,! E!&5,>EES S+!&5E '44 P'4K,2 1444 244 P24K,2 1444 144 P14K,2 1444 '54 1'4 *4

+s %ou see t e observed sample numbers for groups t)o and t ree are lo)er t an t eir e6pected values unli;e group one ) ic as a ig er e6pected value. Is t is a clear sign of discrimination )it respect to et nic bac;groundR Dell depends on o) muc lo)er t e e6pected values are. # e lo)er amount mig t not statisticall% be significant. #o see ) et er t ese differences are significant )e can use E6cel and find t e value of t e C$I1S87+3E. If t is value falls )it in t e acceptance region )e can assume t at t e guidelines are met ot er)ise t e% are not. No) lets enter t ese numbers into E6cel spread1 s eet. De used cells ('1(/ for t e e6pected proportions9 C'1C/ for t e observed values and D'1D/ for t e e6pected freGuenc%. #o calculate t e e6pected freGuenc% for a categor%9 %ou can multipl% t e proportion of t at categor% b% t e sample si:e <in ere 1444=. # e formula for t e first cell of t e e6pected value column9 D' is 1444O('. #o find ot er entries in t e e6pected value column9 use t e cop% and t e paste menu as s o)n in t e follo)ing picture. # ese are important values for t e c i1sGuare test. # e observed range in t is case is C'. C/ ) ile t e e6pected range is D'. D/. # e null and t e alternative %pot esis for t is test are as follo)s. $4 . &D P 4.'49 &+P4.24 and &, P4.14 $+. # e population proportions are not &D P 4.'49 &+P 4.24 and &, P 4.14 No) lets use E6cel to calculate t e p1value in a C@50SAUA.E test. Step 1Select a cell in t e )or; s eet9 t e location ) ic %ou li;e t e p value of t e C@50SAUA.E to appear. De c ose cell D12. Step 21 2rom t e menus9 select insert t en clic; on t e F$nction option9 Paste F$nction dialog bo6 appears. Step !13efer to function categor% bo6 and c oose statistical9 from function name bo6 select C@5TEST and clic; on "2. Step 31D en t e C@5TEST dialog appears. Enter C'. C/ in t e act$al0ran#e bo6 t en enter D'. D/ in t e expecte&0ran#e bo69 and finall% clic; on "2. # e p1value )ill appear in t e selected cell9 D12.

+s %ou see t e p value is 4.4423/2 ) ic is less t an t e value of t e level of significance <in t is case t e level of significance9 aP 4.14=. $ence t e null %pot esis s ould be reHected. # is means based on t e sample information t e guidelines are not met. Notice if %ou t%pe IPC$I#ES#<C'.C/9D'.D/=I in t e formula bar t e p1value )ill s o) up in t e designated cell. N"TE: E6cel can actuall% find t e value of t e C$I1S87+3E. #o find t is value first select an empt% cell on t e spread s eet t en in t e formula bar t%pe IPC$IIN-<D1292=.I D12 designates t e p1-alue found previousl% and 2 is t e degrees of freedom <number of ro)s minus one=. # e C$I1S87+3E value in t is case is 12.4'121. If )e refer to t e C$I1S87+3E table )e )ill see t at t e cut off is 4."451' since 12.4'121V4."451' )e reHect t e null. # e follo)ing screen s ot s o)s %ou o) to t e C$I1S87+3E value.

Test of Independence% Contingency Tables


# e C$I1S87+3E distribution is also used to test and see ) et er t)o variables are independent or not. 2or e6ample based on sample data %ou mig t )ant to see ) et er smo;ing and gender are independent events for a certain population. # e variables of interest in t is case are smo;ing and t e gender of an individual. +not er e6ample in t is situation could involve t e age range of an individual and is or er smo;ing abit. Similar to case one data ma% appear in a table but unli;e t e case one t is table ma% contains several columns in addition to ro)s. # e initial table contains t e observed values. #o find e6pected values for t is table )e set up anot er table similar to t is one. #o find t e value of eac cell in t e ne) table )e s ould multipl% t e sum of t e cell column b% t e sum of t e cell ro) and divide t e results b% t e grand total. # e grand total is t e total number of observations in a stud%. No) based on t e follo)ing table test ) et er or not t e smo;ing abit and gender of t e population t at t e follo)ing sample ta;en from are independent. ,n t e ot er and is t at true t at males in t is population smo;e more t an femalesR >ou could use formula bar to calculate t e e6pected values for t e e6pected range. 2or e6ample to find t e e6pected value for t e cell C5 ) ic is replaced in c11 %ou could clic; on t e formula bar and enter C"OD5FD" t en enter in cell C11. Step 1 ")ser%e& .an#e )3:c4 Smo;ing and gender %es male female total Step21 Expecte& .an#e ) 0:c 2*.4"442 '1.5355* 31 45 '" no "/ 122 1/1 total 144 1"' 2"'

4'.5355* 11/.4"44 So t e observed range is b4.c5 and t e e6pected range is b14.c11. Step !1 Clic; on fx<paste function= Step 31 D en &aste 2unction dialog bo6 appears9 clic; on Statistical in function categor% and C$I#ES# in t e function name t en clic; ,B. D en t e C$I#ES# bo6 appears9 enter b4.c5 for t e actual range9 t en b14.c11 for t e e6pected range. Step 41 Clic; on ,B <t e p1value appears=. 4.4''3/5 Concl$sion: Since p1value is greater t an t e level of significance <4.45=9 fails to reHect t e null. # is means smo;ing and gender are independent events. (ased on sample information one can not assure females smo;e more t an males or t e ot er )a% around. Step :1 #o find t e c i1sGuare value9 use C$IN- function9 ) en C inv bo6 appears enter 4.4''3/5 for probabilit% part9 t en 1 for t e degrees of freedom. Degrees of freedomP<number of columns11=Q<number of ro)s11= C$I1S87+3EP4.544*4'

Test Hypothesis Concerning the $ariance of T!o Populations


In t is section )e )ould li;e to e6amine ) et er or not t e variances of t)o populations are eGual. D enever independent simple random samples of eGual or different si:es suc as n1 and n2 are ta;en from t)o normal distributions )it eGual variances9 t e sampling distribution of s12Fs22 as 2 distribution )it n11 1 degrees of freedom for t e numerator and n2 1 1 degrees of freedom for t e denominator. In t e ratio s12Fs22 t e numerator s12 and t e denominator s22 are variances of t e first and t e second sample9 respectivel%. # e follo)ing figure s o)s t e grap of an 2 distribution )it 14 degrees of freedom for bot t e numerator and t e denominator. 7nli;e t e normal distribution as %ou see t e 2 distribution is not s%mmetric. # e s ape of an 2 distribution is positivel% s;e)ed and depends on t e degrees of freedom for t e numerator and t e denominator. # e value of 2 is al)a%s positive. No) let see ) et er or not t e variances of ourl% income of student1assistant and )or;1stud% students based on samples ta;en from populations previousl% are eGual. +ssume t at t e %pot esis test in t is case is conducted at a P 4.14. # e null and t e alternative are.

.eBection .$le: 3eHect t e null %pot esis if 2U 24.4/5 or 2V 24.45 ) ere 29 t e value of t e test statistic is eGual to s12Fs229 )it 14 degrees of freedom for bot t e numerator and t e denominator. De can find t e value of 2.45 from t e 2 distribution table. If s12Fs229 )e do not need to ;no) t e value of 24.4/5 ot er)ise9 24./5 P 1F 24.45 for eGual sample si:es.

+ surve% of eleven student1assistant and eleven )or;1stud% students s o)s t e follo)ing descriptive statistics. ,ur obHective is to find t e value of s12Fs229 ) ere s12 is t e value of t e variance of student assistant sample and s22 is t e value of t e variance of t e )or; stud% students sample. +s %ou see t ese values are in cells 2* and D* of t e descriptive statistic output.

#o calculate t e value of s12Fs229 select a cell suc as +1" and enter cell formula P 2*FD* and enter. # is is t e value of 2 in our problem. Since t is value9 2P1./*4"153*59 falls in acceptance area )e fail to reHect t e null %pot esis. $ence9 t e sample results do support t e conclusion t at student assistants ourl% income variance is eGual to t e )or; stud% students ourl% income variance. # e follo)ing screen s oot s o)s o) to find t e 2 value. De can follo) t e same format for one tail test<s=.

*inear Correlation and )egression "nalysis


In t is section t e obHective is to see ) et er t ere is a correlation bet)een t)o variables and to find a model t at predicts one variable in terms of t e ot er variable. # ere are so man% e6amples t at )e could mention but )e )ill mention t e popular ones in t e )orld of business. 7suall% independent variable is presented b% t e letter 6 and t e dependent variable is presented b% t e letter %. + business man )ould li;e to see ) et er t ere is a relations ip bet)een t e number of cases of sold and t e temperature in a ot summer da% based on information ta;en from t e past. $e also )ould li;e to estimate t e number cases of soda ) ic )ill be sold in a particular ot summer da% in a ball game. $e clearl% recorded temperatures and number of cases of soda sold on t ose particular da%s. # e follo)ing table s o)s t e recorded data from Nune 1 t roug Nune 13. # e )eat erman predicts a /42 degree temperature for Nune 14. # e businessman )ould li;e to meet all demands for t e cases of sodas ordered b% customers on Nune 14. DAC 11Nun 21Nun 31Nun Cases of #emperature Soda 5' 5/ "5 5" 5* "3

41Nun 51Nun "1Nun '1Nun *1Nun /1Nun 141 Nun 111 Nun 121 Nun 131 Nun

"' '5 *1 *" ** ** *4 *2 *4 *3

"" '3 '* *5 *5 *' *4 ** *4 */

No) lets use E6cel to find t e linear correlation coefficient and t e regression line eGuation. # e linear correlation coefficient is a Guantit% bet)een 11 and E1. # is Guantit% is denoted b% .. # e closer . to D t e stronger positive <direct= correlation and similarl% t e closer . to 11 t e stronger negative <inverse= correlation e6ists bet)een t e t)o variables. # e general form of t e regression line is % P m6 E b. In t is formula9 m is t e slope of t e line and b is t e %1intercept. >ou can find t ese Guantities from t e E6cel output. In t is situation t e variable % <t e dependent variable= is t e number of cases of soda and t e 6 <independent variable= is t e temperature. #o find t e E6cel output t e follo)ing steps can be ta;en. Step 1 2rom t e menus c oose #ools and clic; on Data +nal%sis. Step 21 D en Data +nal%sis dialog bo6 appears9 clic; on correlation. Step !1 D en correlation dialog bo6 appears9 enter (1.C14 in t e input range bo6. Clic; on 5abels in first ro) and enter a1" in t e output range bo6. Clic; on ,B. Cases of Soda #emperature Cases of Soda #emperature 1 4./""5/*'' 1

+s %ou see t e correlation bet)een t e number of cases of soda demanded and t e temperature is a ver% strong positive correlation. # is means as t e temperature increases t e demand for cases of soda is also increasing. # e linear correlation coefficient is 4./""5/*5'' ) ic is ver% close to E1. No* lets follo* same steps )$t a )it &ifferent to fin& t'e re#ression e?$ation1

Step 1 2rom t e menus c oose Tools and clic; on Data Analysis Step 2. D en Data Analysis dialog bo6 appears9 clic; on regression. Step !1 D en Regression dialog bo6 appears9 enter b1.b14 in t e %1range bo6 and c1.c14 in t e 61range bo6. Clic; on labels. Step 31 Enter a1/ in t e output range box. Note. # e regression eGuation in general s ould loo; li;e >Pm Q E b. In t is eGuation m is t e slope of t e regression line and b is its %1intercept. SUMMARY OUTPUT 3egression Statistics !ultiple 3 3 SGuare Standard Error ,bservations AN"-A df SS 3egression 1 3esidual #otal !S 2 Significance 2 4./""5/*5'' 4./34312*4/ 2./1/3*31/1 13

+dHusted 3 SGuare 4./2*34124"

1333.4'//*/ 1333.4'//*/ 15".4"434/' '.5*511E14*

11 /3.'54'*434 *522'/*213 12 142'.234'"/ Standard Error 7pper /5K

Coefficients

t Stat

&1value

5o)er /5K

Intercept

/.1'*44'"'

5.445'42*3 1."*53545* 4.124444*4 1 21.1"44 " ' 1 2.*4'//'5" 1 4.'244/''" 1.433/4 3 *

#emperatur 4.*'/242'1 12.54*4111 '.5*511E1 4.4'42**/2 e 1 " 4*

# e relations ip bet)een t e number of cans of soda and t e temperature is. > P 4.*'/242'11 Q E /.1'*44'"' # e number of cans of soda P 4.*'/242'11O<#emperature= E /.1'*44'"'. 3eferring to t is e6pression )e can appro6imatel% predict t e number of cases of soda needed on Nune 14. # e )eat er forecast for t is is /4 degrees9 ence t e number of cans of

soda needed is eGual to? # e number of cases of sodaP4.*'/242'11O</4= E /.1'*44'"' P /1.*2 or about /2 cases.

Moving "verage and E+ponential Smoothing


!oving +verage !odels. 7se t e +dd #rendline option to anal%:e a moving average forecasting model in E6cel. >ou must first create a grap of t e time series %ou )ant to anal%:e. Select t e range t at contains %our data and ma;e a scatter plot of t e data. ,nce t e c art is created9 follo) t ese steps. 1. Clic; on t e c art to select it9 and clic; on an% point on t e line to select t e data series. D en %ou clic; on t e c art to select it9 a ne) option9 C art9 s added to t e menu bar. 2. 2rom t e C art menu9 select +dd #rendline. # e follo)ing is t e moving average of order 4 for )ee;l% sales.

E6ponential Smoot ing !odels. # e simplest )a% to anal%:e a timer series using an E6ponential Smoot ing model in E6cel is to use t e data anal%sis tool. # is tool )or;s almost e6actl% li;e t e one for !oving +verage9 e6cept t at %ou )ill need to input t e value of a instead of t e number of periods9 ;. ,nce %ou ave entered t e data range and t e damping factor9 11 9 and indicated ) at output %ou )ant and a location9 t e anal%sis is t e same as t e one for t e !oving +verage model.

"pplications and Numerical E+amples


Descripti%e Statistics: Suppose %ou ave t e follo)ing9 n P 149 data. 1.29 1.59 2."9 3.*9 2.49 1./9 3.59 2.59 2.49 3.4 1. #%pe %our n data points into t e cells +1 t roug +n. 2. Clic; on t e I#oolsI menu. <+t t e bottom of t e I#oolsI menu )ill be a submenu IData +nal%sis...I9 if t e +nal%sis #ool &ac; as been properl% installed.=

3. Clic;ing on IData +nal%sis...I )ill lead to a menu from ) ic IDescriptive StatisticsI is to be selected. 4. Select IDescriptive StatisticsI b% pointing at it and clic;ing t)ice9 or b% ig lig ting it and clic;ing on t e I,;a%I button. 5. Dit in t e Descriptive Statistics submenu9 a. for t e Iinput rangeI enter I+1.DnI9 assuming %ou t%ped t e data into cells +1 to +n. b. clic; on t e Ioutput rangeI button and enter t e output range IC1.C1"I. c. clic; on t e Summar% Statistics bo6 d. finall%9 clic; on I,;a%.I T'e Central Ten&ency: # e data can be sorted in ascending order. 1.29 1.59 1./9 2.49 2.49 2.59 2."9 3.49 3.59 3.* # e mean9 median and mode are computed as follo)s. <1.2 1.5 2." 3.* 2.4 1./ 3.5 2.5 2.4 3.4= F 14 P 2.4* <2.4 E 2.5= F 2 P 2.45 # e mode is 2.49 since it is t e onl% value t at occurs t)ice. # e midrange is <1.2E 3.*= F 2 P 2.5. Note t at t e mean9 median and mode of t is set of data are ver% close to eac ot er. # is suggests t at t e data is ver% s%mmetricall% distributed. -ariance: # e variance of a set of data is t e average of t e cumulative measure of t e sGuares of t e difference of all t e data values from t e mean. # e sample variance1based estimation for t e population variance are computed differentl%. # e sample variance is simpl% t e arit metic mean of t e sGuares of t e difference bet)een eac data value in t e sample and t e mean of t e sample. ,n t e ot er and9 t e formula for an estimate for t e variance in t e population is similar to t e formula for t e sample variance9 e6cept t at t e denominator in t e fraction is <n1 1= instead of n. $o)ever9 %ou s ould not )orr% about t is difference if t e sample si:e is large9 sa% over 34. Compute an estimate for the variance of the population9 given t e follo)ing sorted data. 1.29 1.59 1./9 2.49 2.49 2.59 2."9 3.49 3.59 3.* mean P 2.4* as computed earlier. +n estimate for t e population variance is. s2 P 1 F <1411= J <1.2 1 2.4*=2 E <1.5 1 2.4*=2 E <1./ 1 2.4*=2 E <2.4 12.4*=2 E <2.4 1 2.4*=2 E <2.5 1 2.4*=2 E <2." 1 2.4*=2 E <3.4 1 2.4*=2 E <3.5 12.4*=2 E <3.* 1 2.4*=2 M P <1 F /= <1."3*4 E 4./"44 E 4.33"4 E 4.44"4 E 4.44"4 E 4.4444 E 4.4144 E 4.2'44 E 1.4444 E 1.'424= P 4.""*4

# erefore9 t'e stan&ar& &e%iation is s P < 4.""*4 =1F2 P 4.*1'"

Pro)a)ility an& Expecte& -al$es: Ne)s)ee; reported t at Iaverage ta;eI for ban; robberies )as L39244 but *5 percent of t e robbers )ere caug t. +ssuming "4 percent of t ose caug t lose t eir entire ta;e and 44 percent lose alf9 grap t e probabilit% mass function using EQCE5. Calculate t e e6pected ta;e from a ban; robber%. Does it pa% to be a ban; robberR #o construct t e probabilit% function for ban; robberies9 first define t e random variable 69 ban; robber% ta;e. If t e robber is not caug t9 6 P L39244. If t e robber is caug t and manages to ;eep alf9 6 P L19"22. If t e robber is caug t and loses it all9 t en 6 P 4. # e associated probabilities for t ese 6 values are 4.15 P <1 1 4.*5=9 4.34 P <4.*5=<4.4=9 and 4.51 P <4.*5=<4."=. +fter entering t e 6 values in cells +19 +2 and +3 and after entering t e associated probabilities in (19 (29 and (39 t e follo)ing steps lead to t e probabilit% mass function. 1. Clic; on C artDi:ard. # e IC artDi:ard Step 1 of 4I screen )ill appear. 2. $ig lig t IColumnI at IC artDi:ard Step 1 of 4I and clic; INe6t.I 3. +t IC artDi:ard Step 2 of 4 C art Source Data9I enter IP(1.(3I for IData range9I and clic; IcolumnI button for ISeries in.I + grap )ill appear. Clic; on IseriesI to)ard t e top of t e screen to get a ne) page. 4. +t t e bottom of t e ISeriesI page9 is a rectangle for ICategor% <Q= a6is labels.I Clic; on t is rectangle and t en ig lig t +1.+3. 5. +t IStep 3 of 4I? move on b% clic;ing on INe6t9I and at IStep 4 of 4I9 clic; on I2inis .I # e e6pected value of a robber% is L1943*.4*. E<Q= P <4=<4.51=E<1"22=<4.34= E <3244=<4.15= P 4 E 551.4* E 4*"."4 P 143*.4* # e e6pected return on a ban; robber% is positive. ,n average9 ban; robbers get L1943*.4* per eist. If criminals ma;e t eir decisions strictl% on t is e6pected value9 t en it pa%s to rob ban;s. + decision rule based onl% on an e6pected value9 o)ever9 ignores t e ris;s or variabilit% in t e returns. In addition9 our e6pected value calculations do not include t e cost of Hail time9 ) ic could be vie)ed b% criminals as substantial.

Discrete E Contin$o$s .an&om -aria)les: +inomial Distri)$tion Application: + multiple c oice test as four unrelated Guestions. Eac Guestion as five possible c oices but onl% one is correct. # us9 a person ) o guesses randoml% as a probabilit% of 4.2 of guessing correctl%. Dra) a tree diagram s o)ing t e different )a%s in ) ic a test ta;er could get 49 19 29 3 and 4 correct ans)ers. S;etc t e probabilit% mass function for t is test. D at is t e probabilit% a person ) o guesses )ill get t)o or more correctR Sol$tion: 5etting > stand for a correct ans)er and N a )rong ans)er9 ) ere t e probabilit% of > is 4.2 and t e probabilit% of N is 4.* for eac of t e four Guestions9

t e probabilit% tree diagram is s o)n in t e te6tboo; on page 1*2. # is probabilit% tree diagram s o)s t e Ibranc esI t at must be follo)ed to s o) t e calculations captured in t e binomial mass function for n P 4 and P 4.2. 2or e6ample9 t e tree diagram s o)s t e si6 different branc s%stems t at %ield t)o correct and t)o )rong ans)ers <) ic corresponds to 4XF<2X2X= P ". # e binomial mass function s o)s t e probabilit% of t)o correct ans)ers as &<6 P 2 Y n P 49 p P 4.2= P "<.2=2<.*=2 P "<4.425"= P 4.153" P &<2= D ic is obtained from e6cel b% using t e I(IN,!DIS#I Command9 ) ere t e first entr% is 69 t e second is n9 and t e t ird is mass <4= or cumulative <1=? t at is9 entering P(IN,!DIS#<29494.294= IN +N> EQCE5 CE55 >IE5DS 4.153" +ND P(IN,!DIS#<39494.294= >IE5DS &<6P3YnP49 p P 4.2= P 4.425" P(IN,!DIS#<49494.294= >IE5DS &<6P4YnP49 p P 4.2= P 4.441" P11(IN,!DIS#<19494.291= >IE5DS &<6 2 Y n P 49 p P 4.2= P 4.1*4* Normal Example: If t e time reGuired to complete an e6amination b% t ose )it a certain learning disabilit% is believed to be distributed normall%9 )it mean of "5 minutes and a standard deviation of 15 minutes9 t en ) en can t e e6am be terminated so t at // percent of t ose )it t e disabilit% can finis R Sol$tion: (ecause t e average and standard deviation are ;no)n9 ) at needs to be establis ed is t e amount of time9 above t e mean time9 suc t at // percent of t e distribution is lo)er. # is is a distance t at is measured in standard deviations as given b% t e S value corresponding to t e 4.// probabilit% found in t e bod% of +ppendi6 (9 #able 59as s o)n in t e te6tboo; ,3 t e commands entered into an% cell of E6cel to find t is S value is PN,3!IN-<4.//9491= for 2.32"342. # e closest cumulative probabilit% t at can be found is 4.//419 in t e ro) labeled 2.3 and column eaded b% .439 S P 2.339 ) ic is onl% an appro6imation for t e more e6act 2.32"342 found in E6cel. 7sing t is more e6act value t e calculation )it mean and standard deviation in t e follo)ing formula )ould be SP<Q1=F # at is9 S P < 6 1 "5=F15 # us9 6 P "5 E 15<2.32"34= P //./ minutes. +lternativel%9 instead of standardi:ing )it t e S distribution using E6cel )e can simpl% )or; directl% )it t e normal distribution )it a mean of "5 and standard deviation of 15 and enter IPN,3!IN-<4.//9"5915=I. In general to obtain t e 6 value for ) ic alp a percent of a normal random variableCs values are lo)er9 t e follo)ing IN,3!IN-I command ma% be used9 ) ere t e first entr% is 9 t e second is 9 and t e t ird is . Anot'er Example: In t e earl% 1/*4s9 t e #oro Compan% of !inneapolis9 !innesota9 advertised t at it )ould refund t e purc ase price of a sno) blo)er if t e follo)ing )interCs sno)fall )as less t an 21 percent of t e local average. If t e

average sno)fall is 45.25 inc es9 )it a standard deviation of 12.2 inc es9 ) at is t e li;eli ood t at #oro )ill ave to ma;e refundsR Sol$tion: Dit in limits9 sno)fall is a continuous random variable t at can be e6pected to var% s%mmetricall% around its mean9 )it values closer to t e mean occurring most often. # us9 it seems reasonable to assume t at sno)fall <6= is appro6imatel% normall% distributed )it a mean of 45.25 inc es and standard deviation of 12.2 inc es. Nine and one alf inc es is 21 percent of t e mean sno)fall of 45.25 inc es and9 )it a standard deviation of 12.2 inc es9 t e number of standard deviations bet)een 45.25 inc es and /.5 inc es is S. S P < 6 1 = F s P </.54 1 45.25=F12.2 P 12./3 7sing +ppendi6 (9 #able 59 t e te6tboo; demonstrates t e determination of &<6 /.54= P &<: 12./3= P 4.1'9 t e probabilit% of sno)fall less t an /.5 inc es. 7sing E6cel9 t is normal probabilit% is obtained )it t e IN,3!DIS#I command9 ) ere t e first entr% is 69 t e second is mean 9 t e t ird is standard deviation s9 and t e fourt is C7!75+#I-E <1=. Entering PN,3!DIS#</.5945.25912.291=9 0ives &< 6 /.54= P 4.441"/3. Samplin# Distri)$tion an& t'e Central ,imit T'eorem : + ba;er% sells an average of 24 loaves of bread per da%. Sales <6= are normall% distributed )it a standard deviation of 4. If a random sample of si:e n P 1 <da%= is selected9 ) at is t e probabilit% t is 6 value )ill e6ceed 2*R If a random sample of si:e n P 4 <da%s= is selected9 ) at is t eprobabilit% t at 6bar 2*R D % does t e ans)er in part 1 differ from t at in part 2R Sol$tions: 1. # e sampling distribution of t e sample mean 6bar is normal )it a mean of 24 and a standard error of t e mean of 4. # us9 using E6cel9 4.15*"" P11 N,3!DIS#<2*9249491=. 2. # e sampling distribution of t e sample mean 6bar is normal )it a mean of 24 and a standard error of t e mean of 2 using E6cel9 4.422'5 P11N,3!DIS#<2*9249291=.

.e#ression Analysis: # e ig )a% deat s per 144 million ve icle miles and ig )a% speed limits for 14 countries9 are given belo). <Deat 9 Speed= P <3.49 55=9 <3.39 55=9 <3.49 55=9 <3.59 '4=9 <4.19 55=9 <4.39 "4=9 <4.'9 55=9 <4./9 "4=9 <5.19 "4=9 and <".19 '5=.

2rom t is )e can see t at five countries )it t e same speed limit ave ver% different positions on t e safet% list. 2or e6ample9 (ritain ... )it a speed limit of '4 is demonstrabl% safer t an Napan9 at 55. Can )e argue t at9 speed as little to do )it safet%. 7se regression anal%sis to ans)er t is Guestion. Sol$tion: Enter t e ten paired % and 6 data into cells +2 to +11 and (2 to (119 )it t e Ideat I rate label in +1 and IspeedI limits label in (19 t e follo)ing steps produce t e regression output. C oose I3egressionI from IData +nal%sisI in t e I#oolsI menu. # e 3egression dialog bo6 )ill )ill appear. Note. 7se t e mouse to move bet)een t e bo6es and buttons. Clic; on t e desired bo6 or button. # e large rectangular bo6es reGuire a range from t e )or;s eet. + range ma% be t%ped in or selected b% ig lig ting t e cells )it t e mouse after clic;ing on t e bo6. If t e dialog bo6 bloc;s t e data9 it can be moved on t e screen b% clic;ing on t e title bar and dragging. 2or t e IInput > 3ange9I enter +1 to +119 and for t e IInput Q 3angeI enter (1 to (11. (ecause t e > and Q ranges include t e IDeat I and ISpeedI labels in +1 and (19 select t e I5abelsI bo6 )it a clic;. Clic; t e I,utput 3angeI button and t%pe reference cell9 ) ic in t is demonstration is +13. #o get t e predicted values of > <Deat rates= and residuals select t e I3esidualsI bo6 )it a clic;. >our screen displa% s ould s o) a #able9 clic;ing I,BI )ill give t e IS7!!+3> ,7#&7#9I I+N,-+I +ND 3ESID7+5 ,7#&7#I # e first section of t e EQCE5 printout gives IS7!!+3> ,7#&7#.I # e I!ultiple 3I is t e sGuare root of t e I3 SGuare?I t e computation and interpretation of ) ic )e ave alread% discussed. # e IStandard ErrorI of estimate <) ic )ill be discussed in t e ne6t c apter= is s P 4.*"4239 ) ic is t e sGuare root of I3esidual SSI P 5./'511 divided b% its degrees of freedom9 df P *9 as given in t e I+N,-+I section. De )ill also discuss t e adHusted 31sGuare of 4.21325 in t e follo)ing c apters. 7nder t e I+N,-+I section are t e estimated regression coefficients and related statistics t at )ill be discussed in detail in t e ne6t c apter. 2or no) it is sufficient to recogni:e t at t e calculated coefficient values for t e slope and % intercept are provided <b P 4.4'55" and a P 14.2/333=. Ne6t to t ese coefficient estimates is information on t e variabilit% in t e distribution of t e least1sGuares estimators from ) ic t ese specific estimates )ere dra)n. t e column titled IStd. ErrorI contains t e standard deviations <standard errors= of t e intercept and slope distributions? t e It1 ratioI and IpI columns give t e calculated values of t e t statistics and associated p1 values. +s s o)n in C apter 139 t e t statistic of 1.*545* and p1value of 4.144''9 for

e6ample9 indicates t at t e sample slope <4.4'55"= is sufficientl% different from :ero9 at even t e 4.14 t)o1tail #%pe I error level9 to conclude t at t ere is a significant relations ip bet)een deat s and speed limits in t e population. # is conclusion is contrar% to assertion t at Ispeed as little to do )it safet%.I SUMMA.C "UTPUT: !ultiple 3 P 4.54*339 3 SGuare P 4.344"'9 +dHusted 3 SGuare P 4.213259 Standard Error P 4.*"4239 ,bservations P 14 +N,-+ df SS !S 3egression 1 2.5"**/ 2.5"**/ 3esidual * 5./'511 4.'4"*/ #otal / *.54444 Coeffs. Intercept Speed 2 3.43/45 &1value 4.144''

Estimate Std. Error # Stat &1value 5o)er /5K 7pper /5K 14.2/333 2.45/"3 14.11/2" 4./4*41 15./"52" 5.3'*"4 4.4'55" 4.444'4 1.*545* 4.144'' 14.41*3/ 4.1"/54

.esi&$al "$tp$t: Pre&icte& .esi&$als 3.*"222 14.*"222 3.*"222 14.5"222 3.*"222 14.4"222 4.//55" 11.4/55" 3.*"222 4.23''* 4.24444 4.4"444 3.*"222 4.*3''* 4.24444 4.""444 4.24444 4.*"444 5.3'333 4.'2""'

E'*abs to (ully ,nderstand Statistical Concepts


T'e -al$e of Performin# Experiment: If t e learning environment is focused on bac;ground information9 ;no)ledge of terms and ne) concepts9 t e learner is li;el% to learn t at basic information successfull%. $o)ever9 t is basic ;no)ledge ma% not be sufficient to enable t e learner to carr% out successfull% t e on1t e1Hob tas;s t at reGuire more t an basic ;no)ledge. # us9 t e probalilit% of ma;ing real errors in t e business environment is ig . ,n t e ot er and9 if t e learning environment allo)s t e learner to e6perience and learn from failures )it in a variet% of situations similar to ) at t e% )ould e6perience in t e Ireal )orldI of t eir Hob9 t e probalilit% of aving similar failures in t eir business environment is lo). # is is t e realm of simulations1a safe place to fail. # e appearance of statistical soft)are is one of t e most important events in t e process of decision ma;ing under uncertaint%. Statistical soft)are s%stems are used to construct e6amples9 to understand t e e6isting concepts9 and to find ne) statistical properties. ,n t e ot er and9 ne) developments in t e process of decision ma;ing under uncertaint% often motivate developments of ne) approac es and revision of t e e6isting soft)are s%stems. Statistical soft)are s%stems rel% on a cooperation of statisticians9 and soft)are developers.

(eside t e statistical soft)are9 Fa%a Applets9 "nline statistical comp$tation9 and t e use of a scientific calc$lator is reGuired for t e course. + Scientific Calculator is t e one9 ) ic as capabilit% to give %ou9 sa%9 t e result of sGuare root of 5. +n% calculator t at goes be%ond t e 4 operations is fine for t is course. # ese calculators allo) %ou to perform simple calculations %ou need in t is course9 for e6ample9 enabling %ou to ta;e sGuare root9 to raise e to t e po)er of sa%9 4.3". and so on. # ese t%pes of calculators are called general Scientific Calculators. # ere are also more specific and advanced calculators for mat ematical computations in ot er areas suc as 2inance9 +ccounting9 Civil Engineering9 and even Statistics. # e last one9 for e6ample9 computes mean9 variance9 s;e)ness9 and ;urtosis of a sample b% simpl% entering all data one1b%1one and t en pressing an% of t e mean9 variance9 s;e)ness9 and ;urtosis ;e%s. Dit out a computer one cannot perform an% realistic statistical data anal%sis. Students ) o are signing up for t e course are e6pected to ;no) t e basics of E6cel9 and ot er popular Sprea&s'eet. +s a starting point9 %ou need visiting t e E6cel Deb site created for t is course.

You might also like