Regression and Causal Relationships.pdf
Regression and Causal Relationships.pdf PSY 0035 - Research Methods
Popular in Research Methods
Popular in Psychlogy
This 14 page Class Notes was uploaded by Alisa Kim on Saturday February 7, 2015. The Class Notes belongs to PSY 0035 - Research Methods at University of Pittsburgh taught by Barbara Kucinski in Winter2015. Since its upload, it has received 105 views. For similar materials see Research Methods in Psychlogy at University of Pittsburgh.
Reviews for Regression and Causal Relationships.pdf
Report this Material
What is Karma?
Karma is the currency of StudySoup.
You can buy or earn more Karma at anytime and redeem it for class notes, study guides, flashcards, and more!
Date Created: 02/07/15
12815 To err is human but to really ful things up you need a computer Paul R Ehrlich Using Regression to Begin to Understand Causal Relationships in Observational Studies 1i Linear Regression In correlation the two variables are treated as equals In regression one variable is considered the independent predictor variable X and the other the dependent outcome variable Y If you know something about X this knowledge helps you predict something about Y How regression works with 2 variables quot WM My 7 WW Low beg at M wrw tmm tum uuc Go 39 l MM Emcul WMKML DV Wt mot 4 VALLMm g S M 28 3 ADVNV O unnuwdmmmgl vogvmm rqcrhueme 0 quot pvalue significant when we U g g m i 4 o are confident that slope 20 IV Mildew 7 3 Y e smvidovrl sd some M w hu M E39SLRJY C vcprtmtm W1 2 vwmw Wm M AW 2 12815 What s Slope U03 Slope ofmeans that every l unit change in X yields a 2 unlt change in y re Slope of means that every t unfit change in x Yields a 75unit change in Y h Slope of means that every Mohange in 23 yields a 3unit change in Y Regression finds the line of best fit b Assumptions or the tine print Linear regression assumes that 1 The relationship between X and Y is linear 2 Y is distributed normally at each value of X 3 The variance of Y at every value of X is the same homogeneity of variances 4 The observations are independent Curvilinear relationshipsj Y Y X Y Y i 4 Manges e manna cm 6W 1 Nonconstant variance Constant variance 1 w va fumw 11 f o a VAX e wam MM bruth Lt mtmm mi 1r m mnwm 1 28 15 How f f r f regressnon works with 2 variables o w o E 39 o o 0 Ex I gsquotH x39 o o a f f quot IV How regression works with 2 variables a M mm b 3 b 75 J K Wrwv I N no PSlg pns 9g qw w Llc a r u 0 mm 90er a 0 9X 61 go39 3 gt O 7 O CI 0 3 Still raggm z 1 4 x o quotv we W N Q or mmammw IV With 3 or more variables regressron w ill identify whether the main W has an influence on the DV above arises beyond aka controlling for the other varia Reality 1 Reality 2 Anxiety W Anxiety Depression WM l Deprestiaa W Insomnia LW Insomnia ff Reality 3 Anxiety DeprescsgirJ rM Insomnia Does anxiety have an effect on insomnia above and beyond depression Third Variable Confounds aka Common Causes Anxiety 39139 Insomnia Depression Key 039 NotDepressed Depressed Regression Within a Group Overall Regression 39 ignoring Groups 1 Common causes 0 hen there re 397 a all IS not spuriouS Correlation y Q IIQ VL it X i a 39 E O 5 o E l O l 1 O O z or 39 XI o 39 0 Anxiety an make it appear like there is a 0 Not Depressed Depressed W A What Mimi ni CAI D Regression on i j 3 12815 vw yd a a ll 3 mi rwmi as i g Variable ma 7 A 08 si t Regression on I Variable b p A D r ang inert My li39 awnmil VM tM xLL l w wic UK 7 R Riki W at germs Sm w rmnsvw c volswi Common causes can make it appear like there is a correlation even when there really is not spurious correlation insomnia O O Anxiety Q Not Depressed Depressed 7 A D 14 Regression on i Variable blpl AM 08 sigj Regression on Variable b p 7 A D Common causes can make it appear like there is a correlation even when there really is not spurious correlation TN Depressed Depressed 7 A D l CU kmu iq a 7 quotE mar Regressron on l E i quotif e 7 y 8 r Variable b p A 08 sig Regression on 7 Variable b p A p Anxrety Common causes can make it appear like there is a correlation even when there really is not spurious correlation Not Depressed g Depressed 939 7 A 397 l l D l FEE Regression onl l g Variable b p A 08 sig 5 MW clots J Whllcdsla A y y NH MWW 7 Regressron onl r twea v 39 bl b Wielflmlll Pm 6 V p ertwsl A 7 0 ns Anxiety 39 D DDS Sig WWW M mm dCl V lel imwlm aim x wwwka WU WC etc rmx v W WW W ltf3 m M rw and s W vilwp r as F 9W a egiiim WCV0630 Common causes oa Insomnia n make a re 39 39 stronger than it is greSSIon appear even 9 NotDepressed Depressed N i 1 A Anxiety Regression on 1 Variable b pj TA i i J Regression on IA Variabid b pJ W draw m W51 E39Vic we Mirmoi w ih ro fadmm minrod IreUMWMP MN i8 Shuwi 39b Nominme oz WW1 r 01mm we WEWMWP 1 gm Common causes can make a regression appear even stronger than it is insomnia Anxiety 0 Not Depressed 39139 Depressed Regression on I i Variable b pj i i A roars Regression onlJ Variable b pi i A O4sii in iposisigj 12815 l L 9amp5th wwwf OMJ QCV EMWWM r0 Wt 0W dl39liYCC m Citkl ik Wmlu mtli Wl I M lw wartime lnsomnia quot639 Common causes can sometimes make an overall reg ression appear to be in the opposite direction f S39im QSQ m lS Banaidplx 0 Not Depress Depressed M C I C C C C Q A n Xi ety Slimew s WWW o eukmrlo a ipped Norm 39 39 oJr Whmw o n twv W3W aw loud Willilbw law gamma lm Wrtm wrong mto mwt M WWMMENP oelurwn M diam 39w it 39 7quot M While CWuses can sometimes make an overall regression appear to be in the opposite direction Insomnia 7 0 Not Depressed m p Depressed A 3910 I rRegression onl J Vairiablel b l d V A 08 sigJ Regression on l 4 Anxiety H 12815 Alternative Causes Anxiety 7 Key 0 Not Depressed Depression 9139 Depressed quot i 39 Regression Within a Group Overall 9 Regression t I Ignoring Groups Insomnia l depme it uwvmte wr to damw mania Accounting for Alternagiluegagses can make a causai relation easier to detect and make the regression more S39gmflcant39 3 ii 0 Not Depressed a 3 Depressed o r i i A i m T a E D i Q r J I i 13 O Regression on I E i i i 39 b i x g o a Vanabie p C r i quot Q 0 i A i i o f O i g 0 Regressron on i Va abiei b p 39 f 0 w A i i f D i Anxiety lam ll JAM M t JKLMMIA OC Wy r ttg m tkmt H w z QWPL y 5 Wit WWW LL End39le lovei WW rwj C J 9 dmmc Accounting for Alternative Causes can make a causal relation easier to detect and make the regression more Slgnmcant39 39 Not Depressed Depressed E C E o I E 0 o liRegression on i 0 Variable b p A D Anxnety J Accounting for Alternative Causes can make a causal relation easier to detect and make the regression more significant quot 0 Not Depressed i quot Depressed on w A O g 0 o 939 Regression on I J E o x g 0 v Variablel b p g a 39 039 A 04 ns v d x H 0 F Regression on I 0 Variable b p O 39 A 04 sug Anxiety D DDS Sig U 73 I 1Lquotf d Insomnia Moderators Anxiety Key 0 Not Depressed Depression Depressed Regression Within a Group Overall Regression Ignoring Groups 4r ale wsuum MatMd lava Wll lwmmw lain Immeth 5L NHBWW W H W H between the moderator and the main cause Moderators change the strength of the relation between the main Cause and Effect This is often called an interaction 0 Not Depressecj Depressed T A D l V 0 g Q 0 Regression on I g O 0 Variable b pj E o 139 o 9 39 A l l J 0 7 Regression onl J Variable b pJ O 77 A r g A H Xi ety Interaction A39D 12815 Moderators change the strength of the r main Cause and Effect This is often call between the moderator and the main cause elation between the ed an interaction 0 Not Depressed Depressed 39 1 quot math 949939 JKW farther in C remnantup 5E5 umme iW UW WXU I 39 39 E Lei0W w Wkev39 CE 1M m i 53 3m 0w S W It E 4 Wm H QWMNAO 7 Regression 0L Q63 khwv i Variable b p A z 70 Anxiety Wmac on A39D Moderators change the strength of the relation between the main Cause and Effect This is often called an interaction 1 lug N WAW between the moderator and the main cause 0 Not Depressed 939 Depressed w w la quotits W fihf m 02 7 r up H E i egressmn on x 5 Wariable b p quan F A 08 Sig in a 39 a MYESW l5 Regression on i WWW v 39 bl b 7 39 ts lltulal a39la 9 P webth A r 06 sig dngw a V D pos sig Anx iety interaction ND 04 39 V f if egg dwm does 1qu g r M 6mm Lmtukmwel diagrng q elw M1 islets tv new we 39 r i39 ve memu Wu migmd xm 14 l ns wdcwtw
Are you sure you want to buy this material for
You're already Subscribed!
Looks like you've already subscribed to StudySoup, you won't need to purchase another subscription to get this material. To access this material simply click 'View Full Document'