library(psy) # cronbach
cat("\014") # cleans screen
rm(list=ls(all=TRUE)) # remove variables in working memory
setwd("C:/Users/evazquez/Downloads") # sets working directory
## PRETEST 1 VIVIDNESS
MainStudy<-read.csv("pretest 1 vividness.csv", skip=0, header=T)
boxplot(MainStudy$exposure.time.FB,MainStudy$exposure.time.TW,MainStudy$exposure.time.YT)
t.test(MainStudy$exposure.time.YT,MainStudy$exposure.time.TW,paired = T)
##
## Paired t-test
##
## data: MainStudy$exposure.time.YT and MainStudy$exposure.time.TW
## t = 0.9817, df = 44, p-value = 0.3316
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -2.681169 7.773925
## sample estimates:
## mean of the differences
## 2.546378
t.test(MainStudy$exposure.time.YT,MainStudy$exposure.time.FB,paired = T)
##
## Paired t-test
##
## data: MainStudy$exposure.time.YT and MainStudy$exposure.time.FB
## t = 0.018366, df = 44, p-value = 0.9854
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -6.565129 6.685884
## sample estimates:
## mean of the differences
## 0.06037778
t.test(MainStudy$exposure.time.FB,MainStudy$exposure.time.TW,paired = T)
##
## Paired t-test
##
## data: MainStudy$exposure.time.FB and MainStudy$exposure.time.TW
## t = 0.7958, df = 44, p-value = 0.4304
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -3.809773 8.781773
## sample estimates:
## mean of the differences
## 2.486
## Exposure time is equivalent for all SMP
boxplot(MainStudy$vividness.FB,MainStudy$vividness.TW,MainStudy$vividness.YT)
t.test(MainStudy$vividness.YT,MainStudy$vividness.TW,paired = T)
##
## Paired t-test
##
## data: MainStudy$vividness.YT and MainStudy$vividness.TW
## t = -2.6931, df = 44, p-value = 0.009978
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -0.68806873 -0.09904238
## sample estimates:
## mean of the differences
## -0.3935556
t.test(MainStudy$vividness.YT,MainStudy$vividness.FB,paired = T)
##
## Paired t-test
##
## data: MainStudy$vividness.YT and MainStudy$vividness.FB
## t = 1.1488, df = 44, p-value = 0.2568
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -0.1339213 0.4890324
## sample estimates:
## mean of the differences
## 0.1775556
t.test(MainStudy$vividness.FB,MainStudy$vividness.TW,paired = T)
##
## Paired t-test
##
## data: MainStudy$vividness.FB and MainStudy$vividness.TW
## t = -3.097, df = 44, p-value = 0.003398
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -0.9427625 -0.1994597
## sample estimates:
## mean of the differences
## -0.5711111
## Only two levels of vividness are found
## MAIN STUDY
MainStudy<-read.csv("DatasetJRIM.csv", skip=0, header=T)
summary(MainStudy$EXPOSURE.TIME)
## Min. 1st Qu. Median Mean 3rd Qu. Max.
## 30.84 48.31 62.93 61.60 72.83 98.44
MainStudy$Time2<-ifelse(MainStudy$EXPOSURE.TIME<=62.93,1,2)
MainStudy$Time3<-ifelse(MainStudy$Time==1,"1.Low",ifelse(MainStudy$Time==2,"2.Med","3.High"))
MainStudy$Vividness<-ifelse(MainStudy$X205=="Twitter","Vivid","NonVivid")
cronbach(cbind(MainStudy$FREQ.BUY,MainStudy$IMPORTANCE,MainStudy$FREQ.BUY.ONLINE))
## $sample.size
## [1] 900
##
## $number.of.items
## [1] 3
##
## $alpha
## [1] 0.8612343
## >.86 INVOLVEMENT
summary(lm(WOM~INVOLVEMENT+Vividness+Time+GENDER.1.IS.F+INCOME+EDU+RACE+YROFBIRTH,MainStudy))
##
## Call:
## lm(formula = WOM ~ INVOLVEMENT + Vividness + Time + GENDER.1.IS.F +
## INCOME + EDU + RACE + YROFBIRTH, data = MainStudy)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.5966 -1.8916 0.2395 1.6036 5.6609
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 24.197479 14.234907 1.700 0.0895 .
## INVOLVEMENT 0.265714 0.035169 7.555 1.03e-13 ***
## VividnessVivid -0.338262 0.146469 -2.309 0.0211 *
## Time 0.110448 0.098727 1.119 0.2636
## GENDER.1.IS.F 0.029924 0.139967 0.214 0.8308
## INCOME -0.030113 0.035679 -0.844 0.3989
## EDU -0.042312 0.036125 -1.171 0.2418
## RACE -0.050695 0.084754 -0.598 0.5499
## YROFBIRTH -0.010418 0.007157 -1.456 0.1459
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.066 on 891 degrees of freedom
## Multiple R-squared: 0.07189, Adjusted R-squared: 0.06355
## F-statistic: 8.627 on 8 and 891 DF, p-value: 2.23e-11
## Positive effect of enduring involvement - H1: APPROVED
results=lm(WOM~INVOLVEMENT+Vividness*Time3+GENDER.1.IS.F+INCOME+EDU+RACE+YROFBIRTH,MainStudy)
anova(results)
## Analysis of Variance Table
##
## Response: WOM
## Df Sum Sq Mean Sq F value Pr(>F)
## INVOLVEMENT 1 244.5 244.541 57.7150 7.681e-14 ***
## Vividness 1 23.7 23.650 5.5818 0.018363 *
## Time3 2 7.0 3.521 0.8310 0.435938
## GENDER.1.IS.F 1 0.0 0.026 0.0063 0.937000
## INCOME 1 3.2 3.238 0.7643 0.382213
## EDU 1 5.7 5.725 1.3512 0.245388
## RACE 1 1.4 1.407 0.3320 0.564614
## YROFBIRTH 1 9.0 9.025 2.1300 0.144799
## Vividness:Time3 2 41.6 20.816 4.9128 0.007553 **
## Residuals 888 3762.5 4.237
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## ANCOVA Vividness and time have interaction effects - H2: APPROVED
interaction.plot(MainStudy$Time3,MainStudy$Vividness,MainStudy$WOM)
qqnorm(results$res)
plot(results$fitted,results$res,xlab="Fitted",ylab="Residuals")
hist(results$res)
MainStudy3High<-subset(MainStudy,MainStudy$Time3=="3.High"&MainStudy$Vividness=="Vivid")
MainStudy3Low<-subset(MainStudy,MainStudy$Time3=="3.High"&MainStudy$Vividness=="NonVivid")
t.test(MainStudy3High$WOM,MainStudy3Low$WOM)
##
## Welch Two Sample t-test
##
## data: MainStudy3High$WOM and MainStudy3Low$WOM
## t = -3.358, df = 160.79, p-value = 0.0009801
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -1.6041322 -0.4160637
## sample estimates:
## mean of x mean of y
## 3.865079 4.875177
## Nonvivid content produces greater effect than vivid content when cognition is high - H3: APPROVED
MainStudy2High<-subset(MainStudy,MainStudy$Time3=="2.Med"&MainStudy$Vividness=="Vivid")
MainStudy2Low<-subset(MainStudy,MainStudy$Time3=="2.Med"&MainStudy$Vividness=="NonVivid")
t.test(MainStudy2High$WOM,MainStudy2Low$WOM)
##
## Welch Two Sample t-test
##
## data: MainStudy2High$WOM and MainStudy2Low$WOM
## t = 1.1793, df = 237.69, p-value = 0.2395
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -0.1756087 0.6994325
## sample estimates:
## mean of x mean of y
## 4.563910 4.301998
## Vivid content produces greater effect than nonvivid content when cognition is medium - H4: REJECTED
MainStudy1High<-subset(MainStudy,MainStudy$Time3=="1.Low"&MainStudy$Vividness=="Vivid")
MainStudy1Low<-subset(MainStudy,MainStudy$Time3=="1.Low"&MainStudy$Vividness=="NonVivid")
t.test(MainStudy1High$WOM,MainStudy1Low$WOM)
##
## Welch Two Sample t-test
##
## data: MainStudy1High$WOM and MainStudy1Low$WOM
## t = -2.3359, df = 178.5, p-value = 0.02061
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -1.2484857 -0.1050298
## sample estimates:
## mean of x mean of y
## 3.815261 4.492019
## Nonvivid content produces greater effect than vivid content when cognition is low - H5: APPROVED
MainStudy1.2.3.High<-rbind(MainStudy1High,MainStudy2High,MainStudy3High)
summary(aov(WOM~Time3,MainStudy1.2.3.High))
## Df Sum Sq Mean Sq F value Pr(>F)
## Time3 2 38.9 19.434 4.127 0.0171 *
## Residuals 297 1398.6 4.709
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
aov.out<-aov(WOM~Time3,MainStudy1.2.3.High)
TukeyHSD(aov.out)
## Tukey multiple comparisons of means
## 95% family-wise confidence level
##
## Fit: aov(formula = WOM ~ Time3, data = MainStudy1.2.3.High)
##
## $Time3
## diff lwr upr p adj
## 2.Med-1.Low 0.74864873 0.03363705 1.46366041 0.0377029
## 3.High-1.Low 0.04981832 -0.74128039 0.84091703 0.9879432
## 3.High-2.Med -0.69883041 -1.41121667 0.01355585 0.0558571
## Two levels of inverted U-shape
t.test(MainStudy1High$WOM,MainStudy2High$WOM)
##
## Welch Two Sample t-test
##
## data: MainStudy1High$WOM and MainStudy2High$WOM
## t = -2.5417, df = 181.56, p-value = 0.01187
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -1.3298286 -0.1674689
## sample estimates:
## mean of x mean of y
## 3.815261 4.563910
t.test(MainStudy2High$WOM,MainStudy3High$WOM)
##
## Welch Two Sample t-test
##
## data: MainStudy2High$WOM and MainStudy3High$WOM
## t = 2.2488, df = 171.72, p-value = 0.02579
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## 0.08544365 1.31221717
## sample estimates:
## mean of x mean of y
## 4.563910 3.865079
t.test(MainStudy1High$WOM,MainStudy3High$WOM)
##
## Welch Two Sample t-test
##
## data: MainStudy1High$WOM and MainStudy3High$WOM
## t = -0.14889, df = 163.91, p-value = 0.8818
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -0.7104985 0.6108618
## sample estimates:
## mean of x mean of y
## 3.815261 3.865079
## Two levels of inverted U-shape confirmed
results=lm(WOM~INVOLVEMENT+Time3+GENDER.1.IS.F+INCOME+EDU+RACE+YROFBIRTH,MainStudy1.2.3.High)
anova(results) ## ANCOVA
## Analysis of Variance Table
##
## Response: WOM
## Df Sum Sq Mean Sq F value Pr(>F)
## INVOLVEMENT 1 108.09 108.085 24.4080 1.317e-06 ***
## Time3 2 28.51 14.253 3.2186 0.04144 *
## GENDER.1.IS.F 1 5.27 5.272 1.1905 0.27614
## INCOME 1 0.06 0.063 0.0142 0.90512
## EDU 1 0.03 0.032 0.0073 0.93197
## RACE 1 0.11 0.109 0.0247 0.87527
## YROFBIRTH 1 6.74 6.743 1.5227 0.21821
## Residuals 291 1288.63 4.428
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## Vivid content will produce an inverted U shape patter when sizing their effect on digital engagement. H6: APPROVED
qqnorm(results$res)
plot(results$fitted,results$res,xlab="Fitted",ylab="Residuals")
hist(results$res)
MainStudy1.2.3.Low<-rbind(MainStudy1Low,MainStudy2Low,MainStudy3Low)
summary(lm(WOM~INVOLVEMENT+Time3+GENDER.1.IS.F+INCOME+EDU+RACE+YROFBIRTH,MainStudy1.2.3.Low))
##
## Call:
## lm(formula = WOM ~ INVOLVEMENT + Time3 + GENDER.1.IS.F + INCOME +
## EDU + RACE + YROFBIRTH, data = MainStudy1.2.3.Low)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.8083 -1.7981 0.2374 1.5077 5.3994
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 17.648130 17.069394 1.034 0.302
## INVOLVEMENT 0.241217 0.043497 5.546 4.42e-08 ***
## Time32.Med -0.160178 0.207772 -0.771 0.441
## Time33.High 0.276061 0.246597 1.119 0.263
## GENDER.1.IS.F -0.091436 0.169651 -0.539 0.590
## INCOME -0.036465 0.043341 -0.841 0.400
## EDU -0.067941 0.043801 -1.551 0.121
## RACE -0.080978 0.104950 -0.772 0.441
## YROFBIRTH -0.006641 0.008587 -0.773 0.440
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.039 on 591 degrees of freedom
## Multiple R-squared: 0.06928, Adjusted R-squared: 0.05668
## F-statistic: 5.499 on 8 and 591 DF, p-value: 1.024e-06
## Vivid content will produce an inverted U shape patter when sizing their effect on digital engagement. H7: REJECTED
results=lm(WOM~INVOLVEMENT+Time3+GENDER.1.IS.F+INCOME+EDU+RACE+YROFBIRTH,MainStudy1.2.3.Low)
qqnorm(results$res)
plot(results$fitted,results$res,xlab="Fitted",ylab="Residuals")
hist(results$res)
