2018 Farah mherat Hamza

1 2018 Farah mherat HamzaINTRODUCTIONINTO BIOSTATISTICSW hatisBiostatistics?Itisawordof2parts:1-B io=frombiology(whichindicatesthatthissci enceisconcernedwithlivingthings mainlyhumans-).2 Statistics= ,inotherwords, ,medicalsciences,healthsciences,educatio nalresearchandenvironmentalsciences. How isstatisticstheartof prediction?Hereishow:Supposethatyouareap hysicianoraresearcherandyouhaveaquestion inyourhead,andyouwonderwhetherornotthere isarelationshipbetween2factors(2variable s),forexamplelet ssaythatyouareadermatologistandyounotice dthattheskincancerproportionshadbeenincr easingamongwomeninthelast10years,andyoun oticedatthesameperiodoftime-inthelast10y ears-thatalsomorewomenwereusingsunblock, inyourheadyouwillwonderifthereisarelatio nshipbetweenthesetwothings!

2 Eachoneofthese2thingsiscalledavariable;b ecauseawomancanorcannotbewithskincancer, ,sothefirstthingtodoinordertofindananswe rforthisquestioninyourhead(whetherornott hereisarelationship),isthatyougotothelit erature(whichisthestudiesthathavebeenpub lishedbeforeyoubyotherresearchers),tryto findananswerintheliterature,ifyoudidn tfindanyanswer,thenyouarethepersonwhowea relookingfortoanswerthisquestion;because noonehaddonethisresearchquestionbeforeyo uinJordanforexample!-hurray,congrats!-, ,let ssupposethatthereare4millionwomeninJorda n,isitpossibletogotoeachoneofthemandaske achoneifshehasskincancerornot?!andifsheu sesunblockornot?!Thiswilltakeaverylongpe riodoftime,andthat sgoingtobesohardandrequiresagreateffort, anditcanbesodifficulttoimplement, ,youhavejusttocollectasamplefromthepopul ation,butthissampleisnotcollectedhaphaza rdly,itiscollectedusingcertainrulesandth ereareseveraltechniquesinsampling, |PageOk,youcollectedthesample,let ssay1000womenfromthose4millionwomen,youa skeachoneofthese1000thetwoquestions,andt heiranswersrepresentyourdata,butwaitamin ute!

3 Thesedataintheirprimarysimpleformaremean inglessanddon tmakesense!Instead,youexplain,organize, , ,whichiscalleddescriptivestatistics(whic hisbasicallytheprocessoforganizing,summa rizing,anddescribingthesamplecharacteris tics,usingthelowestnumberofwordsandnumbe rs, ,standarddeviation,upperandlowerlimit,qu antiles,percentiles,..etc.).Sofar,haveyo uansweredtheresearchquestionbydescriptiv estatisticsonly?!Notyet,youhavejustdescr ibedthedata,forexample,yousaythattheperc entageofskincanceramongwomeninJordanis12 %,andtheusersofsunblocksamongthemare17%. But,youdidn tseewhetherthereisarelationshipbetweenth e2factors,notonthesamplelevel, ,herecomesthesecondandmostimportantstepo fbiostatisticswhichiscalledtheinferentia lstatistics, ,Iusethemagicofbiostatisticstocomeupwith conclusionsandresultsregardingthebiggerp opulation(the4millionwomenintheexampleab ove).

4 Youhavetokeepinmindthatyoucanneverbe100% sureaboutyourconclusionregardingthebigge rpopulation,butyoucanstillbeashighas95%c onfident, ,attheendofyourstudy,youcansaysomethingl ikethis:(Weare95%confidentthatthereisast atisticallysignificantassociationbetween usingsunblockanddevelopingskincancer),th isisaveryimportantfinding;becausebasedon ityoucanprovideyourhealtheducation,yourr ecommendations, ,youaddedaveryimportantpiecetoourknowled ge, ,howcanwemovefromasamplecomposedjustofco upleofthousandstoapopulationwith4million ?usinginferentialstatistics. Whatarethetypesofinferentialstatistics?C hi-square,T-test, , | ,organization,summarization, ,more , : ,ortobemadeavailable, :Studyingtheself-esteemandacademicachiev ementamongcollegestudents.

5 Population:allstudentwhoareenrolledinany collegelevel. Sample:students ,itis theselectionofa |PageInferential statistics Representativeness: thekey characteristic of thesampleisclose tothepopulation. (howmuchdoesthesample representthepopulation). :1.(Referencepopulation):towhomaretheres ultsgoingtobeapplied?2.(Studypopulation) :Whatisthegroupofpeoplefromwhichwewantto drawasample?3.(SampleSize):Howmanypeople doweneedinoursample?4.(SamplingMethod):H owwillthesepeoplebeselected?Thepopulatio nwhichIwanttomakeaconclusion/inferenceto itattheendofmyresearchstudyiscalledthere ferencepopulation;becauseIamgeneralizing myfindingstothem, ( ), ,wehave12governorates,distributedin3regi ons(North,Middle,South),andineachgoverno ratethereareseveralcities,villages, ,inordertoapplymystudy, ,onefromthemiddle, , ,itdoesn tmakesensethatIgotoallthewomenineachgove rnorateorcitywhichIhadselectedinordertoa skthemmystudyquestions!

6 Yes,itisnarrower,butstilltoobroad!Ihavet oshrinkitdownmore,soIwillthinkaboutgoing toprimaryhealthcarecentersineachcitythat Ihadselected,andseethewomenthere, (alsocalledaccessiblepopulation).4|PageW e want the sample to be small enough so it s manageable and large enough so it s representativeness Neglect a significant or a considerable part of the original population I will choose my sample from the sampling frame which is smaller than study population Samplingframeisalistingofalltheunitsthat composethestudypopulation,oritisthelisti ngofallelementsofapopulation, ,alistofallmedicalstudentsattheuniversit yofJordan,2014-2016. ,butIdon twantallthesethousandstobeinmysample, ,Iselectoneoftheseveralsamplingtechnique s,andfromthosethousands, (alsocalledsamplemembers).

7 Notethatsamplesizeisnotchosenrandomly,th erearespecificwaystoselectmysamplesizeto answeranytypeofresearchquestion. Element:Thesinglememberofthepopulation(p opulationelementorpopulation memberareusedinterchangeably).element=sa mplemember=subject(thethreewordshavethes amemeaninginquantitative research).But,in (qualitativeresearchtheyarecalledpartici pants).TypesofSamplingMethods (samplingtechniques)Byusingsamplingtechn iques,wecanenterthesamplingframeandselec tthecertain numberofsubjectsneededforthe :1-Probabilitysamplingtechniques(methods ):Thetechniquesinwhichweguaranteeanequal chanceforeverybodyinthesampling |PageEx: all students who are enrolled in any college levelStudents at JUMedicine school and law school -Involvestherandomselectionprocesstosele ct asamplefrommembersorelementsof apopulations, ofthestudy population should have anequalor at least aknown chanceof samplingframe(Listingofall studyunits).

8 Therearemany typesofprobabilitysampling, andwewilldiscussfive inthis sheet,whichare: SimpleRandomSampling. Systematicsampling. Stratifiedsampling. Cluster sampling. Multistage 'thaveenoughtimeand , Wewilltalkaboutfourtypesofnonprobability sampling,whichare: Convenience sampling. Snowball sampling. Quotasampling. |PageIt used most in the case of having a specific criteria of the people you need in your research (ex: people with a rare disease ProbabilitySamplingMethods1 SimpleRandomSampling:Thisisthesimplestof canIensure thattheselectionprocessofthe samplewas100%random?Therearesomewebsites andbooksthatgiveyousomethinglikethis(see thetablebelow),basicallylistsofrandomnum bersandtellyouwhichnumberstochoose(these aretotallyrandomandtheyhavenosystemorord er).)

9 Forexample,Icametoaclasswhichcontains100 studentsandIwanttochoose10onlyformysampl e,soImakealistwhichcontainsthestudents , |PageTheproblemisthatsometimesthe10rando mlychosenstudentscanbefromthesamecategor yorkind, ,orallcouldbeJordanians, ,evensimplerandomsamplingisnotidealenoug h, (orSystematicRandomSampling): Individualsarechosenat regularintervalsfromthesamplingframe. ,Iwanttochoose10studentsfrom100randomly, butIdon twantthemalltobefromthefront,orfromthemi ddle, ,inordertoachievebetterrandomness,Itaket hefirst10andIclosemyeyesandpickoneofthem randomly,let ssupposeIpickedthestudentnumber7,sowhatI havetodoisthatIskip10studentsandpickthes tudentnumber17,skip10,pickthestudentnumb er27,skip10,pickthestudentnumber37,47,57 ,67,77,87.

10 Samplingfraction= Intervalsize= 3-StratifiedSampling:Sometimesyoursampli ngframeiswaytoomorecomplexthanjustasecti oninaclassroom,itcouldbeawholeuniversity ,soweneedmoresophisticatedprobabilitysam plingtechniques, |PageMaybe the sample you got doesn t representativeness Ifwehavesomestudyunitswithdifferentchara cteristicswhichwewanttoincludeinthestudy ,thenthesamplingframeneedstobedividedint ostrataaccordingtothesecharacteristics. Ensuresthatproportionsofindividualswithc ertaincharacteristicsinthesamplewillbeth esameasthoseinthewholestudypopulation. ,theUniversityofJordanhas50thousandstude nts,andIneedonly1000studentformysample,b utIdon'thavelistswithallthe50thousandstu dentsintheuniversity,orIcouldhave, 'tIdividetheuniversityinto3strataforexam ple(AstratumfortheMedicalcolleges,andone fortheScientificcolleges,andanotheronefo rHumanitiesColleges).