######################## ## REPLICATION OF LEWANDOWSKY ET AL 2012 ############################### ##GET DATA source("http://www.climateaudit.info/scripts/psychology/lewandowsky_utilities.txt") lew=get.data(dset="lew") ################### ##FACTOR ANALYSIS ################# ##"FREE-MARKET" #For free-market items, a single factor #comprising 5 items (all but FMNotEnvQual) accounted for 56.5% of the variance; the #remaining item loaded on a second factor (17.7% variance) by itself and was therefore #eliminated. pc=factanal(bak[,grep("FM",name0)],factors=2) pc$loadings #%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% #Loadings: # Factor1 Factor2 #FMUnresBest 0.771 FMNotEnvQual 0.995 # FMLimitSocial 0.555 0.293 # FMMoreImp 0.753 # FMThreatEnv 0.863 # FMUnsustain 0.839 Factor1 Factor2 SS loadings 2.924 1.095 Proportion Var 0.487 0.183 Cumulative Var 0.487 0.670 #%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% ##CO2 #The 5 climate change items (including CauseCO2) loaded on a common factor #that explained 86% of the variance; all were retained. pc=factanal(lew[,name0[ grep("CO2",name0)[1:5] ]],factors=2) pc$loadings #%%%%%%%%%%%%%%%%%%%%%%%%% Loadings: Factor1 Factor2 CO2TempUp 0.801 0.469 CO2AtmosUp 0.755 0.584 CO2WillNegChange 0.544 0.807 CO2HasNegChange 0.492 0.739 CauseCO2 0.652 0.605 Factor1 Factor2 SS loadings 2.175 2.126 Proportion Var 0.435 0.425 Cumulative Var 0.435 0.860 #%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% ##CONSPIRACY #For conspiracist ideation, two #factors were identi ed that accounted for 42.0 and 9.6% of the variance, respectively, with #the items involving space aliens (CYArea51 and CYRoswell ) loading on the second factor #and the remaining 10 on the f rst one (CYAIDS and CYClimChange #Items loading on each factor were summed to form two #composite manifest variables. The two composites thus estimate a conspiracist construct #without any conceptual relation to the scienti c issues under investigation. pc=factanal(lew[,c(13:15,17:24,26)],factors=3) pc$loadings #%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% Loadings: Factor1 Factor2 Factor3 CYNewWorldOrder 0.253 0.460 0.217 CYSARS 0.331 0.467 0.249 CYPearlHarbor 0.537 0.307 0.131 CYMLK 0.704 0.293 0.101 CYMoon 0.128 0.456 0.212 CYArea51 0.221 0.352 0.730 CYJFK 0.694 0.190 0.251 CY911 0.501 0.470 0.191 CYRoswell 0.220 0.299 0.872 CYDiana 0.354 0.565 0.253 CYOkla 0.399 0.127 0.108 CYCoke 0.345 0.372 0.159 Factor1 Factor2 Factor3 SS loadings 2.206 1.762 1.675 Proportion Var 0.184 0.147 0.140 Cumulative Var 0.184 0.331 0.470 #%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% ## OTHER SCIENCE All SEM's were performed with M-PLUS using ordinal coding of the manifest variables, with the consensus responses binned into 9 categories with approximately equal numbers. The acceptance-consensus item pairs (e.g., CauseHIV { ConsensHIV ) were entered into an SEM with two correlated latent variables, one capturing the common variance of all \belief" questions, and the other representing all \consensus" questions ( ordinal coding of the manifest variables, with the consensus responses binned into 9 categories with approximately equal numbers. Items loading on each factor were summed to form two composite manifest variables. The fact that acceptance of climate science (CauseCO2) and perceived consensus among climate scientists (ConsensCO2) loaded onto their respective latent variables together with other very di erent scienti c propositions suggests that respondents did not gauge consensus among climate scientists, and evaluate climate science, independently of their views of other, unrelated domains of scienti c inquiry. The acceptance-consensus item pairs (e.g., CauseHIV { ConsensHIV ) were entered into an SEM with two correlated latent variables, one capturing the common variance of all \belief" questions, and the other representing all \consensus" questions (see Figure 1). Pairwise correlations between the residuals for each belief-consensus pair represented content-speci c covariance. All SEM's were performed with M-PLUS using ordinal coding of the manifest variables, with the consensus responses binned into 9 categories with approximately equal numbers. The model t reasonably well, 2(5) = 53:71, p < :0001, CFI= :989, RMSEA= :092 (90% CI: .071 { .115). People's content-general inclination to accept science was associated with content-general perceived scienti c consensus; r = :43, Z = 12:76, p < :0001, over and above the content-speci c links represented by the pairwise correlations. The fact that acceptance of climate science (CauseCO2) and perceived consensus among climate scientists (ConsensCO2) loaded onto their respective latent variables together with other very di erent scienti c propositions suggests that respondents did not gauge consensus among climate scientists, and evaluate climate science, independently of their views of other, unrelated domains of scienti c inquiry. Rather, their perception of consensus and their endorsement of scienti c ndings about the climate reected in part a content-independent disposition to perceive scienti c consensus, and a correlated Figure 1. Latent variable model for the relationship between perceived consensus among scientists and acceptance of scienti c propositions. All correlations and loadings are signi cant and standardized. Manifest variable labels are explained in Table 2. See text for further explanation. #try to group into bins: 9 even bins not possible g= function(y, target=c(0,25,90,99,100) ) { M=length(target) v=rep(M,K) for(j in (M-1):1) { v[y