Lab.2_Moderation and Mediation__Analysis and Interpretation_SR_02032022

docx

School

University of Alberta *

*We aren’t endorsed by this school

Course

505

Subject

Economics

Date

Apr 3, 2024

Type

docx

Pages

Uploaded by ChiefRhinoceros2142

1 EDPY 505 Please submit your preliminary responses by 11:20a.m. on Feb. 3, 2022 here: https://forms.gle/iYxRFYWmceNTMiHk9 You will receive a copy of your responses via email. Paste your responses into Word, revise them, if need be, and submit the Word document on e-class by Tuesday, Feb. 8. LAB 2 Moderation and Mediation Definitions A moderation model proposes that the relationship between two variables depends on a third variable. A mediation model proposes that the independent variable influences the mediator variable, which in turn influences the dependent variable. Why is it important to study moderation and mediation? From a theoretical point of view, they both provide a deeper understanding of the relationship between independent and dependent variables. From a data mining point of view, when you expect an IV to be significant, but it is not, and when the homoscedasticity assumption is violated, you may want to check for a moderation effect. Last updated on Feb. 2, 2022

2 EDPY 505 There is no overall effect, but the drug does reduce pain for male participants. When your treatment does not work, you don’t simply say it does not work. You ask for which subgroups it works, and for which subgroups it does not work. When heteroscedasticity occurs like above, you can often fit several regression lines into the scatter plot. It may indicate that there is a moderator. When you have many predictors in a regression, and the result shows that some theoretically important variables are not significant, consider checking for mediation. For example, in a study of low-income families’ physical health, after putting both housing condition and mental health Last updated on Feb. 2, 2022

3 EDPY 505 in a regression, housing condition is not a significant predictor. Can we conclude that housing condition is not an important variable for physical health? Can we tell policy makers to ignore the housing condition of low-income families? No. In this case, a mediation analysis reveals that housing condition significantly predicts mental health, which significantly predicts physical health. The indirect effect is significant as well. This is why when it comes to decision making, we cannot simply conduct a regression analysis and rule out any non-significant variables. How to test moderation Construct a new interaction variable, which equals to IV*moderator, and do the following regression: DV = b 0 + b 1 IV + b 2 moderator + b 3 IV ∗ moderator + error IV and moderator are centered to have means of 0. For moderation analyses using the process macro, use model number 1. How to test mediation Classic approach:  Check if IV significantly predicts DV  Check if IV significantly predicts mediator  Include both IV and mediator in the regression. Check 1) if the mediator is significant and 2) if the regression coefficient of IV is smaller in absolute value than before. Indirect effect approach:  Sobel test assumes normality of a*b in Figure 10.4 (page 1 of this document) and is NOT very robust to violation of normality. Since a*b is only normally distributed given large sample sizes, Sobel’s test is inaccurate most of the time. (Please refer to the lecture reading about moderation and mediation for more information.)  Compute a bias-corrected bootstrap confidence interval, the preferred method. For mediation analyses using the process macro, use model number 4. Last updated on Feb. 2, 2022

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

4 EDPY 505 In this lab, we are going to explore the following topics:  How to conduct a moderation analysis using SPSS and interpret the corresponding results  How to conduct a mediation analysis using SPSS and interpret the corresponding results Installing custom dialog boxes in SPSS Download and install file: Download the file process.spd (version 4.0) from eClass. If a new version is available, download it here: www.processmacro.org Save this file onto your computer. Open SPSS, select “Extensions,” “Utilities,” “Install Custom dialog” and open the “Process.spd”. Once the installation is complete, you’ll find that the PROCESS menu has been added to the existing regression menu (Analyze->Regression). Moderation Data The Video Games.sav data file contains data from 442 youths. It measures their aggressive behaviour (Aggression), callous unemotional traits (CaUnTs), and the number of hours per week they play video games (hours). Scenario Past research shows video games are linked to increased aggression in youths (Anderson & Bushman, 2001). Callous unemotional traits such as lack of guilt, lack of empathy, and callous use of others for personal gain are also linked to aggression (Rowe, Costello, Angold, Copeland, & Maughan, 2010). Now, researchers want to know if the strength or direction of the relationship between game-playing and aggression is affected by callous unemotional traits.  A. Which model would you recommend? Moderation o A.1 What is the outcome variable (Y)? Aggression o A.2 What is the independent variable (X)? Video-game playing  B. Which variable is the moderator? Callous unemotional traits  C. Please write the equation. (See p. 3 of this document.) Aggression = b0+(b1+b3Callous) Gaming + b2Callous + error Last updated on Feb. 2, 2022

5 EDPY 505 1. Running the analysis Open the Video Games.sav data file in SPSS. Click Analyze->Regression->Process, by Andrew F.Hayes Select the outcome variable and drag it to the corresponding box. Similarly, select and drag the predictor variable and the moderator variable ( W ). Select Model Number 1 which means moderation model. Click Options. Last updated on Feb. 2, 2022

6 EDPY 505 Check (1) Generate code for visualizing interactions which is helpful for interpreting and visualizing the simple slopes analysis. Select (2) Heteroscedasticity-consistent inference: HC3 (Davidson-MacKinnon). (3) Mean center for construction of products: Only continuous variable that define products. Doing so centers the predictor and moderator for you. (4) Moderation and conditioning if p < .05. (5) Conditioning values: -1SD, Mean, +1SD. (6) Select Johnson-Neyman output. Click Continue. Click OK. Last updated on Feb. 2, 2022

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

7 EDPY 505 2. Interpreting output  R-square increase due to interaction(s):  D. Is the model significant? Yes, p<0.001  E. Is there a significant interaction effect? Yes, it is Int_1 p=0.002 Last updated on Feb. 2, 2022

8 EDPY 505 The table above shows us the results of three different regressions: the regression for time spent gaming as a predictor of aggression when the callous trait value is 1) -9.6177; (2) at the mean value of callous traits and (3) 9.6177. We interpret the first models as: When callous traits are low, there is a non-significant negative relationship between time spent gaming and aggression, b = −0.091, 95% CI [−0.299, 0.117], t = −0.86, p = .392.  F. Please write the interpretation for the second model. When callous traits are at the mean, there is a non-significant slightly positive relationship between time spent gaming and aggression, b=0.1696, 95% CI [ .020, .319], t=2.23, p=0.026.  G. Please write the interpretation for the third model. When callous traits are high, there is a significant positive relationship between time spent gaming and aggression, suggesting a moderation effect for the amount of callous traits b=.430, 95% CI [.231, .628], t=4.256, p<.001. [PS: Should you wish to learn more about “Conditional effects of the focal predictor…” this was formerly labeled “Conditional effect of X on Y at values of the moderator(s).”] ************************************************************************** Last updated on Feb. 2, 2022

9 EDPY 505  H. Look at the output of the Johnson–Neyman method. What are the boundaries of the zone of significance? < -17.1002 & > -.7232  I. Look at the Effect column. How does the strength of relationship between time spent gaming and aggression change as callous-unemotional traits increase? It shows the negative slope at low levels of callous-unemotional traits, and positive slope at high levels of callous-unemotional traits Last updated on Feb. 2, 2022

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

10 EDPY 505 Mediation Data This data file, Massar et al. (2011).sav, records data from 83 women aged from 20 to 50 (Age) completed questionnaire measures of their tendency to gossip (Gossip) and their sexual desirability (Mate_Value). Scenario One school of thought is that gossip is used as a way to derogate sexual competitors – especially by questioning their appearance and sexual behaviour. For example, if you’ve got your eyes on a guy, but he has his eyes on Jane, then a good strategy is to spread gossip that Jane has a massive pus-oozing boil on her stomach and that she kissed a smelly vagrant called Aqualung. Apparently, men rate gossiped-about women as less attractive, and they were more influenced by the gossip if it came from a woman with a high mate value (i.e., attractive and sexually desirable). Karlijn Massar and her colleagues hypothesized that if this theory is true then (1) younger women will gossip more because there is more mate competition at younger ages; and (2) this relationship will be mediated by the mate value of the person (because for those with high mate value gossiping for the purpose of sexual competition will be more effective).  J. Which model would you use? Mediation  K. Which variable is the mediator? Mate value of the person (high mate value gossiping for the purpose of sexual competition will be more effective) Last updated on Feb. 2, 2022

11 EDPY 505 1. Running the analysis Click Analyze->Regression->Process, by Andrew F.Hayes (select the most updated version) Select the outcome variable and move it to the corresponding box. Similarly, select and move the predictor and mediator variables. Activate the Model number dropdown list and select 4 which means mediation model. Click Options. Last updated on Feb. 2, 2022

12 EDPY 505 Last updated on Feb. 2, 2022

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

13 EDPY 505 Check 1) Show total effect model which produces the direct effect of the predictor on the outcome and 2) Pairwise contrasts of indirect effects. Select the remaining process options as shown in the screenshot above. Continue. OK. ************************************************************************** Outcome: MaVal Model Summary R R-sq MSE F df1 df2 p .3815 .1455 .6312 13.4522 1.0000 79.0000 .0004 Model coeff se t p constant 3.7981 .2366 16.0558 .0000 Age -.0266 .0073 -3.6677 .0004  L. Based on the above output, does age significantly predict sexual desirability? yes  M. How much variance of sexual desirability is explained by age? 14.5% Last updated on Feb. 2, 2022

14 EDPY 505 ************************************************************************** Outcome: Gossip Model Summary R R-sq MSE F df1 df2 p .4614 .2129 .7988 10.5468 2.0000 78.0000 .0001 Model coeff se t p constant 1.1963 .5495 2.1771 .0325 MaVal .4546 .1266 3.5921 .0006 Age -.0113 .0088 -1.2753 .2060 The above output shows the results of the regression of gossip predicted from both age and sexual desirability.  O. Can age significantly predict gossip with sexual desirability in the model? yes  P. Can sexual desirability significantly predict gossip? yes Last updated on Feb. 2, 2022

15 EDPY 505 ************************** TOTAL EFFECT MODEL **************************** Outcome: Gossip Model Summary R R-sq MSE F df1 df2 p .2875 .0827 .9191 7.1180 1.0000 79.0000 .0093 Model coeff se t p constant 2.9230 .2855 10.2397 .0000 Age -.0234 .0088 -2.6680 .0093 The above output shows the total effect of age on gossip (outcome). The total effect is the effect of the predictor on the outcome when the mediator is not present in the model.  Q. When sexual desirability is not in the model, can age significantly predict gossip? yes  R. How much variance of gossip can be explained by age? 8.27%  S. What does the negative coefficient indicate? Negative b value shows that when mate value is not included in the model, age has a significant relationship with gossip Last updated on Feb. 2, 2022

Your preview ends here

Eager to read complete document? Join bartleby learn and gain access to the full version

Access to all documents
Unlimited textbook solutions
24/7 expert homework help

16 EDPY 505 ***************** TOTAL, DIRECT, AND INDIRECT EFFECTS ******************** Total effect of X on Y Effect SE t p -.0234 .0088 -2.6680 .0093 Direct effect of X on Y Effect SE t p -.0113 .0088 -1.2753 .2060 Indirect effect of X on Y Effect Boot SE BootLLCI BootULCI MaVal -.0121 .0057 -.0258 -.0034 Based on the above output  T. What is the effect of age on gossip in isolation? b=-.02  U. What is the effect of age on gossip when sexual desirability is included as a predictor as well? b=-0.01  V. What is the indirect effect of age on gossip? b=-0.01  W. What is the bootstrapped confidence interval for the indirect effect? -0.0258 & -.0034  X. Do the confidence intervals contain 0? no  Y. What does it indicate? there is an indirect effect of mat value as a mediator on the relationship between age and gossiping  Z. What is the size of the indirect effect? b=-0.012  AA. Is there a significant mediation effect? yes Last updated on Feb. 2, 2022

Lab.2_Moderation and Mediation__Analysis and Interpretation_SR_02032022

Related Documents