Biostatistics and Research Methodology Unit - 2 Abhay Pratap Singh (1903920500001 )as a X Correladion sralysia established foe fact Phot many Fey ‘too variables © an col: dlosely relied. D Now , we may be Interested in estimating (predicting) tee value 6) one variable Given pee vetur §} ametey > Regrenion amalyars re vetes average relectowip Wu toe variables qua this makes posible estima on/ predictian,, Oven Cx) pas Ini genset variable] Explanatory ust do predtet tee varia dl? § tenes}. Variable Rant) Dependent Yarioble/ Explained Voriable7 variable - Tt ia predicted by using en plan tony j Prnalyarh uses _i4 calle) Simple linear Regrevion Prrallgsis _ Regrersion 7 ‘on Xs ~ a gy: eau] Doig wince? Gaonet al yer ? “6 6 +" Slope” J regrenion Wre ¢ a con taut, : > Reprarents change 1h X variable fon g unit change | X variable. z Cy Ye)* Rest Bp -Cl-%) 20 Gest ft 272 Na + bX nS pure oad ax = 9 2X bEXE 4 Regresio 2p J xX ony atts iin ap ion re fi ‘ZX = Na+ bey ‘a bi ered eye geye beyTE canekl- Gra ant aya ag 262 = AX2 +b x'St 4s. Gag 1446 — Sots 47%4 + 1626 os -it1 = 7356 ee EE 4 3s . Vou gf & ae SIE %r a4 b 2.072) ba = S9~ 66-S7 Ue fese vt (so 2 6xq + bx2 a —& qe 7 FS? ape] aeYe -2:26,4 FAX yea . 5 Y= 126 4°37 #12 L2G + $004 Re gr-eviien eq: JX ony: [ee aby ] 2x> Na+ bey Ex = 92) 4 Gey ys (1~ Cxa+ by so) —o XG (262 = 4x 594 bx a3) —©O 1239~ 3544 + 24018 —@® - s9r~ 8f44q 4 4say —~@O eo = 333 = seth. : - $333 ~ fossa] b F lest lo a Wa 6x4 + bx S35 Pastry vale g 4 ‘a= 6x9 + O41 yEg ‘64 = 21 t6.S06 a ' to o8 6a = 2.474 Qe 24 =[o4us ] . Y= Gait oaay D 27 Raw Oven 2 X= o-4tay egy * = 06412 + o BEX FO 424 0499 = 809 . X% I oven Q» fiom the Following data , obtein dhe two Reyrenia equation a ee bye 4 tex ; : yon ‘ ir e Regranion e¢ va jig + 64x Exy = aex HEX?6 XY B= S XG 4h x0 ——O 2wt= Boxa + bh x 220 ——@® 2 = Seq/t 1804 — 204 = 3fq + ob angTE 794 (oun 9 ah | Sane medtrod or Le en¥] EX = Na + bey | exy = VE7 + eye ‘: I ~ Mutfiple reg euion: Conridening tre vales gf thy available multiple Inde pended variabler and Pred The vatur font dependent variables. . y * ¥D Studer na Men ky Live clay | Bock! 3 + + s 3 3 5 fs 2 5 Mn = Y= 40x ‘Ton xy Fe Y= ay oxo bx) 2 Fs ae bit bl 7) The variable conaiden g fe the model should be Pelee Gnd Religble. not 2 Tha modle should be font dA non have normal dishibeion a 4 The punpose J ug cons faut ‘id fo denote A dSe = (tnean : Variable wastWW yorable’s vale im cou hen all hy tndeperdeut vortablee vals tunh -fo ZERO, Ye Xr 4% = qt bX, ¥ bX te u 27> Nat bié¥, 4 bex, —o 21k1 = 98%, + biéx,? 4 br éxx, —@ ie ZY. = 942%. + bre%ix, t brex2 —o| Boe Les | | em Ty] vee Te oe |e | h 98 8 4 24 82 n 64] 3 {le 2 Ig LS ee ee oh 5 49 5 9 a 3 = 2) as | 2s i | s |s | so [Se | 2s fle] 4 Pye lo fs la fo fe fe bm fe & —+——la> fgg «| F2 «| 330 | 63 | 84 ‘pete tole te Persie it 4z Sa t1FH + 206, —@ (43 > Ita + 6361 + F2br Sn [66 s 2catyrh + OF b&Rolve fee 29" 0 §@ I (4° = Sq t IFbx + 20 br s (443 = Ita 4+ 63by + #26, G00. BSo/ + 289 by + 3466. —@ C2 FS = GSa t+ isha t 3 Gob, —6) —38 = -26 ~2b ex - by o @ t aehaTa ye SS 5 26 by 5 26 ss tah | Solve %q: @ 4 @ ti (40% Sq + 176, t Bob, 16 = 200+ aby 4 O4b oo = 209 4 Gag, 4 80 be —@ : ~[286 = 2oque, PE by 4 04. &) Oe ~b= ~4br4 4b, 6s Fort 4b, #4 ® ane} pe EMAL Synaral Sx\ 6 = 46, , 46, ) a | |SS = 26bi + 2Xby ~ 30 5 20 by ot a —_— ff o = bbe by 2 _s Le —— = 0-o2 q 0+ O23 on (s) YH by A vale pet wer 6 > 4bat4b, 6 = dyoaszt 4b ij ae eres Poe hy: 4b. = 6 -3.332 Hr = 2699 br= 2-600 + a % ® ae tig 4o = Sa + IFbz + 20h, go = Sat [#X6.8334+ 2X 6-669 fom Sq 4 14-162 + 13.24 Sq=2 40 -— 27-S¢! So 212/899 we HES I9 S| DES o. qe 2893.5You Xa Uh + ae 92.82) 10-033 yy) ony one Reads Me) Xe =. tie “far Or OES OEE T= 2.824 0-O23K 4 4 0667 x6 / Ys 2:S24 3332 4 3-325 [We ene) Standard Error 5] Reqverion @ Se Pert Ge) 3 “Ton XiY¥Y = 946% Roy Yoo Y= a toy et KOH Y 2 [OLOGY Y x £ Xz [OU OCKS yy $ oS (Ch a a {We Borelsxg ob 2 459% $2 2Syx= 2 (1 N"Prob abily | ov ‘Chance’ 6 yery com monly used ww Day to Day conver gotf7 On Gt statistical Sense , the deam " Probability “y nat connected cirth belvefs_or ony frm she Thinking « S It is based ch calcalcfng NumBticad Values, > Probability % q number which am ge hemloey) Zero» fon gn event which cannot ocean One fe aa Bat cen-teuin fo acer 2 Posbubility 10g mertiremedicel tool used fo atu Rom domnen » T+ deals with fie chance (the likeli boo) 4 Me ayn event Cccwuring. o eee? eI 7)2)3,4,8,6,& fleas Aad tare P: Probability 4 Occurrence (Suecen) sent > Pobabily gf Non Gee ureon j Relat ft Mead Pra} Ie a, co (fai lane) Nee gear p= + e/ ra] 1 = Feeg a hy ds 34 SG Do pttegeeehey 4 ay FP Probecbi ity at art A Prob - > GH te Sts Gun We 246 a) m a Jeu deom = PL Aah +. *r soi hucHia free QO fteam J Medical Refeutonale ane ois J meds in a village from o ba9 contalnng 20 Steipt q Ocols aud lo Strip gf Chock, Graton ack oucton, heck i tar prebablity of geting Crocuy 9 P= No. J faveroble Core (a) ToteP) be J Bgually “kely cow (my pe (suecen) €~ (fatwa) G9: to nH = to +20 = 8 te = oe a = Lr 2k, L 2 5Probability Bishibedion > Thaoricel Listing 6] cuttcamer 4 Pobabllity | (obtaind fom yaacte maeticetf model) > Dn Empire, isting g oletco mes 4 tuiv ebsenved relecHve frequencies, 9 fy subdectve [tng 6] cudeome ameciated wit Heit Bubject#e oy tondyivel” prababilPHiei yepreancting sts degree f conviction J fae deemion maken. Binomial dss: oa Binorneel Distribation ee 2H Binomial dicibution a FM Siecle grata bi by xhiladion fra gues only dus powible veruks w 99 expen men, eitten Succoy or failure. Robabitity Dishibechos Zs [mel Hinamlap Negative is Gian emicd) [* \sses + ois. Du. : va 2 gy. if we fom g coin fier could be aly choc possible aattoree: Heaas o+ “ils. TDcalled 9 > Pp aingle sycceu] failus dest i ea Benes Gennsulli tried ov Bernceelli Go penimact and 9 Aerien 4 cuftonme ja celled 4 Bonnout proce foltson Distribution 9 Ju adatistics, @ Parsston dishiketion 14 Probability duhibection fiat tA used fo Shou hour _ma ey times amy event ja Tikedy Jo occur _oven_a_ specified "In cts words, it ia g count distribution, "ob may be expected th carton where he chance 4 ahy Individuerd — Rvewts such gy, ne, é qecidents on oad or Pid try Wiskedcey om book. Hormel Dis hi bution » f norm! —disbechion "og Gelle shaped reguency | GS be OM Cenuy,Se a Mos4 q ae dela vet cluster cround “Je mear. > Th furfoey a ceetr pat 1A from JRC prvay foe lou Neck it W fo ocety . ws dittribetipn feud fe 2 Nowe) — distibedion ane fymetric , unimodal and OAY™ ptetic and The mean, median aud made ane all egunl. Why wht HTH THR OTTHS gut ) a 30 6gehlUS OG i) oe + © ahead payed ——— a h : 2S Pare Oph Qh x | ; woe mt 2 ap ee She) Me Te (a) = lop Pe Probability fA Sueeet I Arm gels mia (E be Probability Fei tae 1m single “bial n-¥= ilu ne no gf aj gould re to succenx m 4 ial h inl)ap eee Fx (a8)! Powe 4 ce Gygxrve Ylo2 © .2K2 Qe A com ia thvenen § dimes what in te Probase (ity Ql obtalung C4 Head () 1 Head P(x) ace keg ht x pr hy ee OY (nV! pe 6 ya ¢ Pon) = Ss * b)> Gy Ka) = Srdusvere xbag PRAT xd 6 QW oret (Ray = Se LL) *. (s} j | Pay est eee 1% 4590p 9 42S cana | } cleoge sgO- A coin a thrown 6 time what 4 fle proba bility J obtaining Sor maare Hed. » pet 1 get reo ta | raou Pony 5, x q* x ps flay = ~Sxseaxtmme dy Lf Seay yt 2 a/ * bx sg Fl z >T teiy—t Oo + vols oite fisvenfoissomty DSishribection Probabb lity Sat [Bowed g yar system g Caton) Loganitums ) Me fhe mean gf jae Paiarcw Diserivechs m= meee n> Ne of tries P= Probability a] Succey 4. i‘ "each tricep Al~ Ro q Syccew ih 4 porticeday Hit panied - O Q stofe hes (000 Pharma Companies h Peres (D factey ber fire Jeng yer OP doen wal comp tH WHE i be Pebabiy gf S compara cil hae dive, | Ve WFO? _ NE 2oe6 Mon p 2 pel | JocsersMe gece y LL (2+7-193)° jece * 2 a Les Rs) _(o- nosy? + OF (2 #183) 7 a F385 = OHBSS 432 = Seedy sy2y1 (Ess ] - 4336. fe [26 036 few AA oh) T= L- 6-036 D chanel Qe Jon g te PEM Tab pleduced by dy grenvlecha pacha) qd -% dab ef choses Rondon peaaetily G wild ty de feotee, by usmg Poisson's — dy frig on ta detective, find the Reveby dear iy g sont Re WELZ Abe me wxp Pe loy 3a me Lo Yo eee mer re 6 _— (anos? ¥- (26 fae —NNowwatl Bist pesto | 2 The oe in Bell shaped 2 mean jy lecete? at conte, S| Distribuction . 2 tse, ila Se Jsdvibediion orten ds inde Brifely buf even fouches fre % OKIA- i. Moan 2 Median > ‘Moe- aye hay Disbibuien J area under standard Wormal cunue 1 broadly Pollors 2 she + 60-244. . +27 95-44% y pr]e: 99434 iY x tie Symmetvicee) Sone from Meon tuOa aeyme ar mean welylt a Student im Pharmacy latitude to be 6S keg with 4 variance q 1b ka | | Vow many sfudeth a4 fea inatitete of foo Studert Le, would You ee pect fo be oem F2Kg, vaviguee XY =65 ky ean) =) 1enks X= HEIs o> Jig = 4g ~ ¥oF 2 Fer. ece < o + + ( — ol 04839 - Total student = goo foo xX a6 4oL { 160 | [gy— Q> Assum fae mean fage of Atudet fo Prealeg y of be Foy. with a Standard deviction J Sz ow. ray atudeat in Ste thstik be J 65 Student would you -eepect fo be Davinaol valet > from able =] 064 772 Titel ame 68000 [e —_6:4972 00220 sone ceil: 0-022 OY bo =< 1: | 6 (Dstrded | —_fopalation ; Sample F Census § popeledtion ing distinct Shp g aigi du sex a nation oF 4 Shoup au, Whether “Wet grup compvi § people wit a common ghanactenistic. Tndig_ oo / te pa Ragisteres ( e , Panmacrst ] \ Phaamactt J 1a Indi)” NEE tn Sasi fe Tudra. 2D In Statistica, a popelaction a fx fool of individual fiom whieh Statistic) Sam ple 14 draw h for q Study. ~ Thur, any” selection J individual grouped, ctogettes by @ common feafure can be said to be pepulasa, Re peed, Pinte pelcor : Ly poteeticeeg byisfen ce Population Sradeodt Infinete r puladt ey Populate ey [os omyCam ple 2 sample in a Starts cay » significa Powtion 4 a population not gy enttye population 2 for dha” Reason, 9 atetistical analysis da. Sample mus} Re por} fue appre ximmade £0 Staadand deviotion , or Sfaudend evot, q its results 2. | th entire Pepi tect an, 3 Only ay Omad yars 4 ay ere Population would have no Standand ennroy. ne mame) || | ( Lew: then Zo Sample Sir)” (30° or» move fy Major Small Sample Test 30 Sample Size | Students lange Sample Fest 2- fet| teach jae Foe ea | fro babili or Renders | [Ror Peeboortitts | Sampling Method Non « Random Sampling Metio) Sampling fechniquas ; e Soonging iA simply foe preeeu | lecnnmg about He fopalation on fhe barra f sample drawn trom if 5 The proco G] Sowgling tevelve 2 Glemeds, a Selecting “fhe Sample & Colleccag tre thfor mati ay Making qn inference abeth 4Le Ropu Garertials_g_Sonphg > Represerbativency, Adequacy Todependence Homagendly \4 —— i Non~ Probability o Pro babitity o Rem Jom hod Nov — Random Samp Ing Samp" g Melho eta ie - _ =f Sinaple Row Jom — oe , s : Sepia > Judgere} Samp ay Lottery Metrod > Quota Bap ag | Table 5] Rerydors Ny, 1 ————————— | | Ristrietes Rem dow, | Cam pling oY Stratified Sampling Ly (onvenrence Lamply d ? Syttemartic Sampling Cus tey Samp hing fe | aT dose not depend tpon Tee existence g deeile) Rreson abe} about he univons pe © Fhecfuentsbiased > I provide estimerfer which we essenaly 4m) gud have Aignift card Preciaioy oT fis sampling relative efficiency of Varta sample f design cov be evaluedded Limitatio, hag + Very hig level skill d xc periance repured 7 Time taker, Cost a daripling biga Sample Ravdom sampling Gath Guoy unit hy 3 a) ce Squall o vt? bean | Selected tm sie sample faery a 4 g -9 Ply Penson bras g die aweytigaton dose nef influence fhe selection. Chanactty sity & Tem awe selected Tidenpendenly % Scamp bmg widlorct replace mere} 4 widt Replace pout +Ristricted Rauidom Soup lng When — sampke aw selected Rardamlky Rat — under Conte Restriction Ht ik denmed ay Reshicted fa Jon Jonpy u] "TE wvolte dae personal attertion of fue MOF Gedy) while selecting q Aample. 9 tt on net purely Rem Jom. 6 Steertified Somp hg 9 Sgstematic Som Pring 2 Chuten Sampling Non - Rem dom Sampling TH ie one pr which all fae ifemy J The univoye do not have equal chancey J bemg Aelected 9 The selec tien 9 | prec weet Jeary Rotiall y subject | = Th Waveytigector Selec + samples on dee Lai 4 Comenieence or Ky Jv moved rethhe, fhe basin g Probabitity : aie> The Safistica) — ex000+ rete, to Suu difference between 1° fre collected deka (sample deta) and tht actyad volute] facts (Popaledion deca) » mn other worts, It is de difference Ye Getimeded vous and the ce tered) veluy, E> D Pham - 2, | agp & Prem 246%] 46, Mm s3e%l asx Pheennaeert PhO» 240, | 5, te (Ganong % Sampln Creer) , Grex Riaced Grom - . Onbiaud Grren, Sawmplrng firvo3) > T+ refene fo fle Jifference Yeo othe sample estimafe tind dhe acteal verte q she chana centsticy, J dhs Pepal ation.Biased Errors Biased Rovers _ fn Gnex thet atisex Om decoutd 4] “Some Baur eas imbolancer ov Are part gf ter ° Investig actos 7 Iwformetts ot 2 Tartu meets of counting, mearinin) a fin erron thee dose not take, place on deoynf | any bias with anybody be ® occurs Recidentali | May be due to a chance on due to oy ' } arthmetic — evroy A Inout 4 Unban) Rhron. | = Such — erroz, eyay oathomacticalty with oud ony mofive. Tht maghitude sampling eur0r Gn be Redued | Ly Hing 9 langen sumple. Sy "4 boy —_ y i | “Tse erut0 occtu) jh are es 1 FE Cot ding ob tabuleting » stetrcep deity,on Thi au more seriou then Samplmg ennory becaurr . sam plme evtor can be minimized by taking 4 largen sample, Howenwen Increow jn Tae sample Size ct Non - Saw pling erion art Whely fo mercer Some qf tee Maxdr reason dred lead qe tee Bon Somphe g eNOS ant. Tnadequate — decfa specticection with fee objective 8) Sun very ty > Duplicecion Joa subject m sth Aunvey ” baelk 3 fain d mua tigeders “Geren committed why le abu lecting te day > Lack q SUP ErvIsTEn q pamany tafe.| te . Hy porbwsia dehy ty 424 some hypo thor in bout Parent Populetion from which fi comple 14 dau, Nall Py portiean,, > Thene iA no sight cout Fi fference between foo populections, specitie J my dee yperioarets due te any | Lyperimentep ay Sampling NNO}. D> The null hypothena is denoted by Ho Alte native Hypetussia Than Altenative bopetrens , the Simple Obsenvecti ony oun eeuny nfl ence J by ome tandom cou , Tt uw dene deg by fe [He ov Hy SEM (Slardard Greer meas) a | rieAiffreee . [so ~b0| SEM teu x grovel ) Preceding Hype tena Tarbes @ set J Hypotain 8 @ Set up 4 Aigniticance Lee] ot | 14, Sy ® Setting q Test Cri denjon Ze test Cle Studeg a desk ® ang computedtion Fr fest aon Malorng Heer dims, = Reyect H Ho is Tre Comect Decision [yn Bl) | Ho 1A false Type -t) Error (2) T comect Deny, |! ie A |Rocedun for Hypetian Testy [ba gt Confiden ep Smngurficore Level “Level Soy. Sy Gor Ly 95 x. bLR 59-97.typotte Testing en fanonedric 4254 In a kind of fron -fanomeic “est at bnppe team dest which dese not rogue any Give genevedtadion toa pepulachen dishi bechon , Generecting record BB whre, wh ment by . a4 Peyanding WH ™ J Hsbinct pangmeters, +e prearz] original Popul tia ne a conte | lng Z- test (236 Samples) (Te tesz gue ¥ Ja sample, known) > b- 45+ (280 Sample) few Hen So (compenmg KR the 2 sample bere + fesp. (ta dipendeect) > pained st - ferk (beper sen) potest Ceempany st a sd) oh way Ccomponng 72 Sompts) ~ pwr way ANOVANow fea metric Test Good new t Tet x? 4es4 (The dis tibechieh gq the. Variable P Prdensen Bee bag eg bemg anahyra) 6 fat goon. M Shapiro Wilk “ hypo tretica? test) * kapevs Jes, | ~ Hos men. Lemey hay Tesy Test + {fest tee Todi pendence = ne | { Row and cotumny gf varigl ly Fis he, | “3 Exacd deg) bemg dest ane mdcpendet) a Test for _Homeyenay) yh fest ‘| Variable bermg crm lyzed Wilcoyes Rak Test | me FiShrbuted 6 Pans withney py a telly) ihney U Test 4 DP krusted « walis I Tesh 7 Priedmanniy Test > Levene Yes}, ' » Studedt’s + dest ww used when’ sample size tA Kk jer than So amd Populedton Standand Jeviedtion (o-) iA unknown, > The variable t+ distibadion vonge fom o 4 to pe [rect ct) ioral Jistbohe ; Aadtspibedion ( = ° ; SE Sobibata” hor @ bet) shaped cere and symmetry + aimiler to normal) disturdection | The shape gf tA t> aspibetion” tovies wath He thange rh degree f free don. = ey ten peakes eh named elder ot (Gta) ¢ More _peoked wm the Fils| Stugewps fesp ——— | Selves (Depew dev) On paired (Tadepersed C9) G00 0 “a PSR im HRP R (©) Joo rg a © : my BRER Pe my© ANOVA (Galpin gj varome) > fANovA ww developed by [RA Fisker) 19 1926 B Palys <] variance , or ANOVA, 14 4 sheng statiseHca) fethnigue ahat Ww used Fe shows de difherence Yeo qwo or More mean ov components through Aignificance desta. i a TH ako shew ug way to make multiple ) fom pani som q severed) Popalecion mean. The ANOVA test in pentomed by companig choo types 5 variate, , Hu variection, betwee ft Jomple pmeary | ey well a te variccficy with each ft somple. | eS vovieier fw tg Meat ae Se ae F s BMA 2 Tain wotthin TAL datibutio __ pes ae novi : A Aissureption g ane Nowra Our Gach populetty Tr > Tk popu ection from which que samples ant drqwl have Pe equap varlance. i Gach sample ww drawh randomly fhey au inde pou der He = Hee Yn = pe My re 2 My Hat WA Bey Fm mo Me ANOVA [co ropanng eg = Sample ) One Way "hao way (one Fac fon) Three way [ie fase) a (Tree faster’) Thanks