The effect of the relation between target object and affective stimuli

The effect of the relation between target object and affective stimuli on the evaluation of the
affective stimuli (pairfocus10) – study report.
Purpose
In previous studies (pairfocus5, pairfocus9b), we tested the effect of action's outcome on the evaluation
of the action's object.
In pairfocus5, we exposed participants to four types of actions: Actions with positive outcomes
performed on positive stimuli (e.g., giving a cute animal to the participant); Actions with negative
outcomes performed on positive stimuli (e.g., taking a cute animal from the participant); Actions with
positive outcomes performed on negative stimuli (e.g., taking a nasty animal from the participant); and
actions with negative outcomes performed on negative stimuli (e.g., giving a nasty animal to the
participant). We presented the actions one at a time. After each presentation, participants evaluated
the stimulus that was the object of the action. Pairfocus9b was similar but, during the task, one cute
animal was always given to the participants, one nasty animal was always given to the participants, one
cute animal was always taken away from the participants, and one nasty animal was always taken away
from the participants and the participants evaluated the animals only in the end of the task. In
pairfocus9b participants rated the animals after the whole task, whereas in pairfocus5 they rated the
animals right after each action.
In both studies, we found that people evaluated objects more positively if the action that was
performed on the objects had a positive outcome than if the action had a negative outcome.
However, in both studies, the different actions were confounded with specific men (e.g., the same man
always took away the nasty animals), so we cannot determine whether the effect we found is due to the
valence of the outcome of the action or due to the valence of the man who performed the action
(although even in that case, the man's valence is based on the action's outcome).
To test whether the effects do not depend on different men performing each type of action, the present
study replicated pairfocus5, but one man performed all the actions.
Procedure (based on pairfocus5)
Participants learned about one target man who performed four actions:
(1) gave them cute animals, (2) gave them nasty animals, (3) took cute animals away from them and (4)
took nasty animals away from them.
Participants were instructed to learn if the man is positive or negative based on his actions.
In each trial, the man (image + name, randomly selected from a list of 6 men) appeared next to an image
of cute animals or nasty animals for 400ms, at which point a word was added between them: "gives
you" or "takes away from you", displayed for another 1600ms, until all stimuli disappeared.
Then we presented again the animal image with a question:
At the end of each trial, we showed for 500ms a countdown of miles left until participants finish the
task.
The game had 32 trials (8 of each action type). The learning task was divided to three blocks (in the first
block, 2 trials for each action type, and in the second and third blocks 3 trials for each action type). In
each block, the trials were presented in a random order.
The first block introduced participants with the task.
Instruction before the first block:
After the first block participants saw these instructions:
The image of the man was randomly selected for each participant from a pool of six men:
The name of the man was randomly selected for each participant from a pool of six names:
James, Kevin, Michael, David, Brian, and Mark
Cute animals:
Nasty animals:
Deliberate evaluation:
Next, participants completed a deliberate evaluation measure. The purpose of using this measure was
mainly to justify the instruction of the game, in which we ask the participants to form impressions of
James. Participants were asked how much they like the man.
Scale: “Dislike extremely “,” Dislike moderately “,” Dislike slightly “,” Neither like or dislike, "Like
slightly", "Like moderately", "Like extremely"
Brief IAT
In the end of the experiment the participants completed a race BIAT. The purpose of using the BIAT was
to justify this study as a project implicit study (i.e., we need to measure the so-called “implicit”
attitudes). The BIAT had 4 blocks of 20 trials and will compare White peoples and Black peoples.
These were the categories and stimuli:
White people:
Black people:
Pleasant: Joy. Love, Happy, and Good
Unpleasant: Horrible, Nasty, Bad, and Evil
Sample size
In pairfocus5 we found an effect of ηp2 =0.21. So we tried to replicate it with 200 participants (that gave
us more than 99% power).
*These results were computed for participants who finished all the tasks in the study, including the
measures that are not relevant to the main research question. However, in the manuscript, we
included all the participants who completed the main task. The results for these participants are
reported in the end of this report (the results were the same in the two samples).
Results
200 participants (out of 355 who started it) completed the study (Mage = 33.98, SD = 14.01, 68% women).
Evaluation of the affective stimuli:
Internal consistencies
Role
Cronbach α
Positive stimuli – positive action
0.81
Positive stimuli – negative action
0.86
Negative stimuli – negative action
0.77
Negative stimuli – positive action
0.85
Interestingly, the internal consistency of the stimuli evaluation is higher when the stimuli were taken
away than when they were given.
Mean evaluation as a function of stimuli valence and action valence:
Variable
N
Mean Std Dev Std Error
Lower 95% Upper 95%
CL for Mean CL for Mean
giveGdEv
200
3.52
0.43
0.03
3.46
3.58
takeGdEv
200
2.96
0.77
0.05
2.86
3.07
takeBadEv
200
2.14
0.78
0.06
2.03
2.25
giveBadEv
200
1.66
0.48
0.03
1.59
1.73
We clearly replicated the effect. The evaluation of positive stimuli is more positive when the stimuli
were the object of positive action (were given to the participants) than when they were the object of
negative action (were taken away from the participants), d = 0.89. The evaluation of negative stimuli is
more positive when the stimuli were the object of positive action (were taken away from the
participants) than when they were the object of negative action (were given to the participants), d =
0.74. These effects are larger than the effects we found in pairfocus5 (d = 0.60 for positive stimuli and d
= 0.51 for negative stimuli).
The 2 (stimuli valence) x 2 (action valence) repeated measures ANOVA, found a very strong main effect
of stimuli valence (ηp2 = .71), reflecting more positive evaluation of the positive stimuli than of the
negative stimuli. Most important, replicating our previous finding, there was a strong effect of action
valence (ηp2 = .30), reflecting more positive evaluation of stimuli that were the object of positive action
than of stimuli that were the objects of negative actions. Here are the details:
Source
DF
stim_val
Error(stim_val)
Source
Type III SS Mean Square F Value
1 359.4103920
359.4103920
199 142.9174899
0.7181783
DF
action_val
Source
54.0193503
54.0193503
199 123.4272816
0.6202376
DF
stim_val*action_val
500.45 <.0001
Type III SS Mean Square F Value Pr > F
1
Error(action_val)
Pr > F
87.09 <.0001
Type III SS Mean Square F Value Pr > F
1
0.23862535
0.23862535
Error(stim_val*action_val) 199 16.35675660
0.08219476
2.90 0.0900
(ηp2
Main effect of stimuli valence
= .71):
Evaluations as a function of stimuli valence:
Variable
N
Mean Std Dev Std Error
posStimEv
200
3.24
0.46
0.03
negStimEv
200
1.90
0.50
0.04
More positive evaluation of the positive stimuli than of the negative stimuli.
Main effect of action valence (ηp2 = .30):
Evaluations as a function of stimuli valence:
Variable
N
Mean Std Dev Std Error
posActionEv
200
2.83
0.44
0.03
negActionEv
200
2.31
0.47
0.03
More positive evaluation of stimuli that were the object of positive action than of stimuli that were the
objects of negative actions.
Because each stimulus appeared once with the relation "give" and once with the relation "take", we can
test the effect separately for each stimulus:
Positive stimuli:
Negative
action
2.89 (1.12)
t-test
Pos1
Positive
action
3.47 (0.72)
Pos2
3.54 (0.72)
3.01 (1.08)
t(198) = 6.59, p< .001, d = 0.57.
t(198) = 6.87, p< .001, d = 0.61.
Pos3
3.52 (0.58)
2.97 (1.06)
t(198) = 6.88, p< .001, d = 0.64.
Pos4
3.67 (0.64)
3.04 (1.16)
t(199) = 7.06, p< .001, d = 0.67.
Pos5
3.43 (0.60)
2.89 (1.04)
t(199) = 6.79, p< .001, d = 0.63.
Pos6
3.52 (0.64)
3.02 (1.08)
t(198) = 5.92, p< .001, d = 0.56.
Pos7
3.57 (0.64)
3.01 (1.06)
t(198) = 6.68, p< .001, d = 0.63.
Pos8
3.44 (0.65)
2.89 (1.05)
t(198) = 6.59, p< .001, d = 0.62.
Negative
action
1.52 (0.76)
t-test
Neg1
Positive
action
2.01 (1.13)
Neg2
2.10 (1.16)
1.44 (0.70)
t(198) = 7.38, p< .001, d = 0.68.
Neg3
2.10 (1.11)
1.62 (0.75)
t(199) = 5.58, p< .001, d = 0.50.
Neg4
2.00 (1.13)
1.56 (0.77)
t(198) = 5.41, p< .001, d = 0.45.
Neg5
2.38 (0.96)
2.01 (0.85)
t(198) = 4.65, p< .001, d = 0.40.
Neg6
2.30 (1.13)
1.78 (0.85)
t(199) = 5.57, p< .001, d = 0.52.
Neg7
2.13 (1.18)
1.53 (0.77)
t(198) = 6.40, p< .001, d = 0.60.
Neg8
2.16 (1.03)
1.81 (0.79)
t(198) = 4.76, p< .001, d = 0.38.
Negative stimuli:
The effect is reliable for each of the stimuli
t(198) = 5.11, p< .001, d = 0.50.
Test the effect of outcome valence for the first evaluation of each stimuli (between participants)
Stim
Positive action
Negative action
T-test
Neg1
1.78 (1.02)
1.61 (0.83)
t(195.98) = 1.29, p = .198, d = 0.18
Neg2
2.16 (1.19)
1.47 (0.75)
t(179.23) = 4.95, p < .001, d = 0.73
Neg3
2.14 (1.08)
1.61 (0.75)
t(166.09) = 3.97, p < .001, d = 0.61
Neg4
1.96 (1.12)
1.63 (0.82)
t(187.14) = 2.42, p = .016, d = 0.35
Neg5
2.29 (0.91)
2.01 (0.85)
t(197) = 2.21, p = .027, d = 0.31
Neg6
2.26 (1.12)
1.75 (0.80)
t(163.83) = 3.65, p < .001, d = 0.57
Neg7
2.05 (1.17)
1.55 (0.78)
t(166.2) = 3.55, p < .001, d = 0.55
Neg8
1.90 (0.95)
1.91 (0.81)
t(197) = -0.09, p = .929, d = -0.01
The effect is in the expected direction for 7 of the stimuli and reliable for 6 of them.
Stim
Positive action
Negative action
T-test
Pos1
3.45 (0.69)
3.13 (1.04)
t(179.28) = 2.59, p = .010, d = 0.38
Pos2
3.56 (0.75)
3.19 (0.94)
t(179.7) = 3.04, p = .002, d = 0.45
Pos3
3.53 (0.52)
3.28 (0.88)
t(158.82) = 2.41, p = .017, d = 0.38
Pos4
3.73 (0.68)
3.21 (0.98)
t(180.82) = 4.45, p < .001, d = 0.66
Pos5
3.38 (0.63)
3.03 (0.91)
t(166.63) = 3.08, p = .002, d = 0.47
Pos6
3.55 (0.64)
3.13 (0.99)
t(189.88) = 3.60, p < .001, d = 0.52
Pos7
3.58 (0.62)
3.13 (0.95)
t(168.35) = 3.92, p < .001, d = 0.60
Pos8
3.46 (0.61)
2.91 (1.05)
t(152.13) = 4.52, p < .001, d = 0.73
The effect is in the expected direction and reliable for all of the stimuli.
The other measures:
Evaluation of the man:
Analysis Variable : eman
N Mean Std Dev Std Error
192
3.66
1.04
0.08
Is very neutral (the scale was 1-7).
BIAT:
Analysis Variable : IAT
N Mean Std Dev Std Error Pr > |t| t Value
200
0.11
0.48
0.03
0.0020
3.14
There is overall preference for White people over Black people (d =0.22)
Results for participants who completed the main task:
223 participants (out of 308 who started it) completed the study (Mage = 33.84, SD = 13.95, 68% women).
Mean evaluation as a function of stimuli valence and action valence:
Variable
N
Mean Std Dev Std Error
Lower 95% Upper 95%
CL for Mean CL for Mean
giveGdEv
223
3.53
0.43
0.03
3.47
3.58
takeGdEv
223
3.00
0.76
0.05
2.90
3.10
takeBadEv
223
2.11
0.78
0.05
2.00
2.21
giveBadEv
223
1.65
0.49
0.03
1.59
1.72
We clearly replicated the effect. The evaluation of positive stimuli is more positive when the stimuli
were the object of positive action (were given to the participants) than when they were the object of
negative action (were taken away from the participants), d = 0.85. The evaluation of negative stimuli is
more positive when the stimuli were the object of positive action (were taken away from the
participants) than when they were the object of negative action (were given to the participants), d =
0.70.
The 2 (stimuli valence) x 2 (action valence) repeated measures ANOVA, found a very strong main effect
of stimuli valence (ηp2 = .72), reflecting more positive evaluation of the positive stimuli than of the
negative stimuli. Most important, replicating our previous finding, there was a strong effect of action
valence (ηp2 = .29), reflecting more positive evaluation of stimuli that were the object of positive action
than of stimuli that were the objects of negative actions. Here are the details:
Source
DF
stim_val
Error(stim_val)
Source
Type III SS Mean Square F Value
1 427.0811812
427.0811812
222 165.6451383
0.7461493
DF
action_val
Error(action_val)
Source
53.2210168
53.2210168
222 129.5303027
0.5834698
DF
stim_val*action_val
572.38 <.0001
Type III SS Mean Square F Value
1
0.32653712
0.32653712
Error(stim_val*action_val) 222 17.27665732
0.07782278
N
Mean Std Dev Std Error
posStimEv
223
3.26
0.47
0.03
negStimEv
223
1.88
0.50
0.03
More positive evaluation of the positive stimuli than of the negative stimuli.
Main effect of action valence (ηp2 = .29):
Evaluations as a function of stimuli valence:
Pr > F
4.20 0.0417
Main effect of stimuli valence (ηp2 = .72):
Evaluations as a function of stimuli valence:
Variable
Pr > F
91.21 <.0001
Type III SS Mean Square F Value
1
Pr > F
Variable
N
Mean Std Dev Std Error
posActionEv
223
2.82
0.43
0.03
negActionEv
223
2.33
0.46
0.03
More positive evaluation of stimuli that were the object of positive action than of stimuli that were the
objects of negative actions.
Because each stimulus appeared once with the relation "give" and once with the relation "take", we can
test the effect separately for each stimulus:
Positive stimuli:
Negative
action
2.94 (1.11)
t-test
Pos1
Positive
action
3.48 (0.72)
Pos2
3.54 (0.71)
3.03 (1.07)
t(221) = 6.85, p< .001, d = 0.56.
Pos3
3.52 (0.58)
3.01 (1.04)
t(221) = 6.85, p< .001, d = 0.60.
Pos4
3.67 (0.65)
3.07 (1.15)
t(222) = 7.09, p< .001, d = 0.64.
Pos5
3.43 (0.60)
2.92 (1.02)
t(222) = 6.89, p< .001, d = 0.60.
Pos6
3.53 (0.63)
3.05 (1.07)
t(221) = 6.15, p< .001, d = 0.54.
Pos7
3.58 (0.63)
3.03 (1.07)
t(221) = 7.11, p< .001, d = 0.62.
Pos8
3.46 (0.64)
2.93 (1.05)
t(221) = 6.76, p< .001, d = 0.60.
Positive
action
1.97 (1.11)
Negative
action
1.53 (0.78)
t-test
t(221) = 6.83, p< .001, d = 0.57.
Negative stimuli:
Neg1
t(221) = 5.04, p< .001, d = 0.45.
Neg2
2.03 (1.15)
1.43 (0.69)
t(221) = 7.26, p< .001, d = 0.63.
Neg3
2.05 (1.10)
1.61 (0.77)
t(222) = 5.64, p< .001, d = 0.46.
Neg4
1.95 (1.11)
1.55 (0.77)
t(221) = 5.27, p< .001, d = 0.41.
Neg5
2.35 (0.96)
2.02 (0.86)
t(221) = 4.47, p< .001, d = 0.36.
Neg6
2.27 (1.11)
1.78 (0.83)
t(222) = 5.69, p< .001, d = 0.49.
Neg7
2.10 (1.17)
1.54 (0.78)
t(221) = 6.43, p< .001, d = 0.56.
Neg8
2.14 (1.03)
1.78 (0.79)
t(221) = 5.17, p< .001, d = 0.39.
The effect is reliable for each of the stimuli
Summary
Replicating the finding from pairfocus5, people evaluated objects more positively if the action that was
performed on the objects had a positive outcome than if the action had a negative outcome.
Importantly, we found the effect even when one man preformed all of the actions. This suggests that
the effect is due to the valence of the outcome of the action and not due to the valence of the man who
performed the action.