nadine without video
noah castelo
March 5, 2017
setwd("~/Documents/Dropbox/Research/Bernd")
p<-read.csv ("nadine_without_videos.csv", header=T, sep=",")
p<-subset(p, attn=="attention" |attn== "Attention" |attn== "ATTENTION")
p$e[p$E0M0==1|p$E0M1]<-0
p$e[p$E1M0==1|p$E1M1]<-1
p$m[p$E0M0==1|p$E1M0]<-0
p$m[p$E0M1==1|p$E1M1]<-1
p$e[p$e0m0AI==1|p$e0m1AI]<-0
p$e[p$e1m0AI==1|p$e1m1AI]<-1
p$m[p$e0m0AI==1|p$e1m0AI]<-0
p$m[p$e0m1AI==1|p$e1m1AI]<-1
p$a[p$e0m0AI==1|p$e0m1AI==1|p$e1m0AI==1|p$e1m1AI==1]<-0
p$a[p$E0M0==1|p$E0M1==1|p$E1M0==1|p$E1M1==1]<-1Manipulation checks - emotion:
summary(aov(emo_1~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 45.5 45.45 9.685 0.00201 **
## e 1 256.0 256.02 54.555 1.08e-12 ***
## a 1 8.1 8.09 1.725 0.18994
## m:e 1 1.8 1.78 0.380 0.53796
## m:a 1 1.2 1.21 0.257 0.61219
## e:a 1 3.1 3.12 0.664 0.41558
## m:e:a 1 29.6 29.55 6.298 0.01253 *
## Residuals 356 1670.7 4.69
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1There’s a 3 way interaction above… the “a” factor is algorithm vs. AI descripton. It seems that when it’s called AI, emotion and memory interact (memory affects perceived emotion when emotion module is off, but not when it’s on.) When it’s called algorithm, there’s no such interacton. This is just for the manipulation check.
AI<-subset(p, a==0)
summary(aov(emo_1~m*e, data=AI))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 26.1 26.05 5.038 0.0260 *
## e 1 156.4 156.39 30.246 1.26e-07 ***
## m:e 1 22.2 22.16 4.286 0.0398 *
## Residuals 184 951.4 5.17
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1lowe<-subset(AI, e==0)
summary(aov(emo_1~m, data=lowe))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 31.9 31.90 6.33 0.014 *
## Residuals 77 388.1 5.04
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1hie<-subset(AI, e==1)
summary(aov(emo_1~m, data=hie))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 0.4 0.381 0.072 0.788
## Residuals 107 563.3 5.265alg<-subset(p, a==1)
summary(aov(emo_1~e*m, data=alg))## Df Sum Sq Mean Sq F value Pr(>F)
## e 1 92.4 92.36 22.084 5.32e-06 ***
## m 1 23.8 23.75 5.679 0.0183 *
## e:m 1 8.9 8.94 2.137 0.1456
## Residuals 172 719.3 4.18
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Memory:
summary(aov(memory_1~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 5393 5393 1404.466 <2e-16 ***
## e 1 1 1 0.346 0.557
## a 1 3 3 0.740 0.390
## m:e 1 0 0 0.091 0.763
## m:a 1 1 1 0.381 0.537
## e:a 1 0 0 0.058 0.811
## m:e:a 1 1 1 0.140 0.708
## Residuals 356 1367 4
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1How human?
summary(aov(human_1~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 239.5 239.51 44.191 1.12e-10 ***
## e 1 2.6 2.59 0.477 0.490
## a 1 2.1 2.10 0.388 0.534
## m:e 1 3.1 3.06 0.565 0.453
## m:a 1 0.2 0.20 0.037 0.847
## e:a 1 5.3 5.28 0.974 0.324
## m:e:a 1 4.4 4.41 0.813 0.368
## Residuals 356 1929.5 5.42
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1Liking Nadine? 2 way interaction between AI/algorithm and emotion. The emotion module increases liking when it’s called AI, but not when it’s called algorithm.
summary(aov(like_1~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 330.3 330.3 54.481 1.12e-12 ***
## e 1 5.8 5.8 0.954 0.3293
## a 1 6.6 6.6 1.094 0.2962
## m:e 1 0.0 0.0 0.000 0.9869
## m:a 1 8.1 8.1 1.329 0.2497
## e:a 1 20.8 20.8 3.428 0.0649 .
## m:e:a 1 6.4 6.4 1.056 0.3049
## Residuals 356 2158.2 6.1
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1summary(lm(like_1 ~ e, data=AI))##
## Call:
## lm(formula = like_1 ~ e, data = AI)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.9266 -1.9494 0.0734 2.0506 6.0506
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 3.9494 0.3017 13.092 <2e-16 ***
## e 0.9772 0.3962 2.467 0.0145 *
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.681 on 186 degrees of freedom
## Multiple R-squared: 0.03168, Adjusted R-squared: 0.02647
## F-statistic: 6.084 on 1 and 186 DF, p-value: 0.01454summary(lm(like_1 ~ e, data=alg))##
## Call:
## lm(formula = like_1 ~ e, data = alg)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.377 -2.033 -0.033 1.967 5.967
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 4.3765 0.2777 15.758 <2e-16 ***
## e -0.3435 0.3862 -0.889 0.375
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.561 on 174 degrees of freedom
## Multiple R-squared: 0.004525, Adjusted R-squared: -0.001196
## F-statistic: 0.7909 on 1 and 174 DF, p-value: 0.3751Comfort using Nadine as a social companion:
summary(aov(tasks_1~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 43.3 43.25 6.647 0.0103 *
## e 1 9.1 9.11 1.400 0.2376
## a 1 7.6 7.61 1.170 0.2801
## m:e 1 15.2 15.16 2.329 0.1278
## m:a 1 0.4 0.42 0.064 0.8000
## e:a 1 1.2 1.24 0.191 0.6626
## m:e:a 1 7.4 7.38 1.134 0.2877
## Residuals 356 2316.5 6.51
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1As a receptionist:
summary(aov(tasks_2~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 474.1 474.1 66.255 6.72e-15 ***
## e 1 1.4 1.4 0.193 0.661
## a 1 6.9 6.9 0.966 0.326
## m:e 1 3.9 3.9 0.543 0.462
## m:a 1 4.9 4.9 0.685 0.408
## e:a 1 10.3 10.3 1.434 0.232
## m:e:a 1 2.2 2.2 0.307 0.580
## Residuals 356 2547.5 7.2
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1retail salesperson:
summary(aov(tasks_3~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 165.7 165.69 26.147 5.18e-07 ***
## e 1 7.0 7.05 1.112 0.292
## a 1 0.3 0.26 0.041 0.840
## m:e 1 0.0 0.00 0.001 0.979
## m:a 1 0.9 0.88 0.139 0.709
## e:a 1 3.5 3.53 0.558 0.456
## m:e:a 1 1.5 1.46 0.230 0.632
## Residuals 356 2255.8 6.34
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1job interviewer:
summary(aov(tasks_4~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 229.9 229.90 36.887 3.21e-09 ***
## e 1 3.3 3.33 0.534 0.465
## a 1 0.2 0.21 0.033 0.856
## m:e 1 0.0 0.00 0.001 0.979
## m:a 1 14.8 14.83 2.379 0.124
## e:a 1 0.8 0.83 0.134 0.715
## m:e:a 1 2.3 2.27 0.364 0.547
## Residuals 356 2218.8 6.23
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1purchase nadine for personal life
summary(aov(purchase_1~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 103.0 103.02 15.094 0.000122 ***
## e 1 0.4 0.42 0.061 0.804780
## a 1 3.4 3.44 0.505 0.477903
## m:e 1 0.0 0.03 0.005 0.944322
## m:a 1 1.0 1.03 0.151 0.697421
## e:a 1 1.9 1.87 0.274 0.600956
## m:e:a 1 1.9 1.86 0.272 0.602309
## Residuals 356 2429.7 6.83
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1for business life. main effect of AI/algorithm. People are MORE interested in purchasing for business like when it’s called algorithm!
summary(aov(purchase_2~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 251.9 251.93 33.676 1.44e-08 ***
## e 1 5.6 5.60 0.749 0.3875
## a 1 32.3 32.30 4.318 0.0384 *
## m:e 1 3.1 3.06 0.409 0.5227
## m:a 1 18.1 18.11 2.421 0.1206
## e:a 1 17.7 17.70 2.366 0.1249
## m:e:a 1 2.1 2.10 0.281 0.5963
## Residuals 356 2663.2 7.48
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1t.test(AI$purchase_2, alg$purchase_2)##
## Welch Two Sample t-test
##
## data: AI$purchase_2 and alg$purchase_2
## t = -1.9339, df = 350.4, p-value = 0.05393
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## -1.1750273 0.0098919
## sample estimates:
## mean of x mean of y
## 2.627660 3.210227willingness to pay
summary(aov(wtp~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 24.0 23.998 22.471 3.09e-06 ***
## e 1 0.3 0.338 0.316 0.574
## a 1 0.1 0.135 0.126 0.723
## m:e 1 0.8 0.816 0.764 0.383
## m:a 1 0.3 0.300 0.281 0.597
## e:a 1 1.8 1.824 1.708 0.192
## m:e:a 1 0.0 0.003 0.003 0.954
## Residuals 356 380.2 1.068
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1threat to jobs
summary(aov(threat_1~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 10.2 10.171 1.370 0.243
## e 1 4.1 4.074 0.549 0.459
## a 1 6.3 6.282 0.846 0.358
## m:e 1 0.8 0.779 0.105 0.746
## m:a 1 12.8 12.778 1.721 0.190
## e:a 1 4.5 4.480 0.603 0.438
## m:e:a 1 12.9 12.867 1.733 0.189
## Residuals 356 2643.4 7.425threat to lives
summary(aov(threat_2~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 0.7 0.701 0.098 0.755
## e 1 3.1 3.139 0.437 0.509
## a 1 2.1 2.129 0.296 0.587
## m:e 1 2.6 2.579 0.359 0.550
## m:a 1 17.7 17.670 2.457 0.118
## e:a 1 9.2 9.156 1.273 0.260
## m:e:a 1 0.8 0.771 0.107 0.744
## Residuals 356 2559.8 7.190is unemployment caused mostly by robots or by outsourcing?
summary(aov(unemployment_1~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 0.8 0.762 0.179 0.673
## e 1 1.7 1.656 0.388 0.534
## a 1 3.7 3.725 0.873 0.351
## m:e 1 4.4 4.363 1.023 0.313
## m:a 1 5.6 5.624 1.318 0.252
## e:a 1 2.9 2.910 0.682 0.409
## m:e:a 1 8.2 8.217 1.926 0.166
## Residuals 356 1518.8 4.266uncanny valley
p$uncanny<-(p$emoreact_6 + p$emoreact_5 + p$emoreact_4)/3
summary(lm(uncanny~m*e*a, data=p))##
## Call:
## lm(formula = uncanny ~ m * e * a, data = p)
##
## Residuals:
## Min 1Q Median 3Q Max
## -4.1304 -2.1304 -0.3172 1.8083 7.4524
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 4.1304 0.3818 10.819 <2e-16 ***
## m -0.9183 0.5907 -1.555 0.1209
## e -1.0808 0.5370 -2.012 0.0449 *
## a -0.9109 0.5561 -1.638 0.1023
## m:e 1.1859 0.7744 1.531 0.1266
## m:a 0.6155 0.8154 0.755 0.4509
## e:a 0.8137 0.7673 1.060 0.2897
## m:e:a -1.2878 1.1010 -1.170 0.2429
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 2.589 on 356 degrees of freedom
## Multiple R-squared: 0.02691, Adjusted R-squared: 0.007773
## F-statistic: 1.406 on 7 and 356 DF, p-value: 0.2015interaction.plot(p$m, p$e, p$uncanny)
emotional reaction to nadine:
p$emoreact<-(p$emoreact_1+p$emoreact_2+p$emoreact_3)/3
summary(aov(emoreact~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 170.4 170.37 27.511 2.69e-07 ***
## e 1 13.4 13.42 2.167 0.142
## a 1 3.4 3.38 0.546 0.461
## m:e 1 1.6 1.61 0.260 0.611
## m:a 1 0.0 0.00 0.000 0.999
## e:a 1 0.7 0.75 0.121 0.729
## m:e:a 1 3.9 3.89 0.628 0.429
## Residuals 356 2204.6 6.19
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1“cognitive” reaction to nadine (how useful, appropriate)
p$cog<-(p$describenadine_1+p$describenadine_2+p$describenadine_3)/3
summary(aov(cog~m*e*a, data=p))## Df Sum Sq Mean Sq F value Pr(>F)
## m 1 624.8 624.8 127.133 <2e-16 ***
## e 1 0.0 0.0 0.000 0.987
## a 1 0.3 0.3 0.064 0.800
## m:e 1 6.4 6.4 1.296 0.256
## m:a 1 0.7 0.7 0.142 0.706
## e:a 1 11.8 11.8 2.394 0.123
## m:e:a 1 2.4 2.4 0.488 0.485
## Residuals 356 1749.7 4.9
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1interaction.plot(p$m, p$e, p$cog)