· web viewif the results were caused by object-response correspondence, we should expect only...

41
ACTION SIMULATION 1 SUPPLEMENTARY MATERIALS Experiment 1 We presented product pictures with a left or right orientation and asked participants to indicate their intention to purchase the product. We manipulated the orientation of the response scale by presenting a horizontal scale to one group of participants and a vertical scale to another group of participants. If the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically. Moreover, the effect should not differ between right-handers and left-handers. If, on the other hand, participants’ purchase intentions are affected by the activation of potential actions, we should expect that purchase intentions are higher for products oriented towards the dominant hand than towards the non-dominant hand, and this should not be affected by scale orientation. Method

Upload: dangmien

Post on 11-Mar-2018

215 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 1

SUPPLEMENTARY MATERIALS

Experiment 1

We presented product pictures with a left or right orientation and asked participants to

indicate their intention to purchase the product. We manipulated the orientation of the response

scale by presenting a horizontal scale to one group of participants and a vertical scale to another

group of participants. If the results were caused by object-response correspondence, we should

expect only an effect when the scale is oriented horizontally but not when it is oriented

vertically. Moreover, the effect should not differ between right-handers and left-handers. If, on

the other hand, participants’ purchase intentions are affected by the activation of potential

actions, we should expect that purchase intentions are higher for products oriented towards the

dominant hand than towards the non-dominant hand, and this should not be affected by scale

orientation.

Method

Participants. A group of 103 participants were recruited through Amazon’s Mechanical

Turk. Two participants failed to answer the handedness questions, and were excluded from the

analysis, leaving a total of 101 participants (34 females, Mage = 31.0, age range 18-62, 9 left-

handed).

Materials. Photographs of 12 different products were used (sprayer, stapler, blow dryer,

drill, frying pan, screwdriver, kettle, tennis racket, thermos cup, pocket knife, electric shaver,

iron). Each photograph was mirrored to create two orientations for each product. Participants

saw only one version of each product. Six products were shown left-oriented and six were shown

Page 2: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 2

right-oriented. Across participants and scale conditions each product was presented equally often

in both orientations.

Procedure. Participants were instructed to indicate how likely it was that they would

purchase the product. Above each product the question “How likely would you be to buy this

[product name]?” was displayed. They rated the likelihood on a scale from 1 (very unlikely) to 9

(very likely). The response scale was presented below the photograph as a row of nine buttons,

ordered from 1 to 9. In the horizontal condition the buttons were oriented horizontally, with 1 on

the left and 9 on the right. In the vertical condition, the response buttons were oriented vertically,

with 1 at the top and 9 at the bottom. Participants indicated their rating by clicking the button of

their choice. Table 1 presents an overview of the critical aspects of the procedure for all

experiments. Trials were presented in random order, with the restriction that the picture of the

water kettle was always presented on the first trial. Elder and Krishna (2012) presented only one

product to each participant. Presenting the same object first allowed us to perform an additional

analysis on first responses only.

Results

For each participant the mean likelihood ratings per condition were calculated. The

means per condition are shown in Figure 1. The data were analyzed with a repeated measures

ANOVA. Because the ANOVA p-values cannot be used to evaluate the evidence in favor of the

null hypothesis, we further analyzed the effects using the Bayesian Information Criterion (BIC,

see Masson, 2011; Wagenmakers, 2007). The BIC is an estimate of the posterior probabilities for

the null and the alternative hypothesis, given the data. These probabilities add up to 1, so for

example a pBIC(H0|D) = .80 for the null entails a pBIC(H1|D) = .20 for the alternative, which means

that the null is four times more likely than the alternative, given the data.

Page 3: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 3

The analyses showed no higher ratings for products oriented towards the dominant hand

than for products oriented towards the non-dominant hand, that is, no interaction between

product orientation and handedness, F(1, 97) < 0.01, p = .993, ηp2 = .00, pBIC(H0|D) = .91. We

then asked if a higher preference for right-oriented than left-oriented products occurred for

horizontal scales but not for vertical scales. No such effect was found; there was no interaction

between product orientation and scale orientation, F(1, 97) = 2.58, p = .112, ηp2 = .03, pBIC(H0|D)

= .73, nor an interaction between handedness, product orientation, and scale orientation, F(1, 97)

= 1.19, p = .279, ηp2 = .01, pBIC(H0|D) = .84. There was also no main effect of product orientation,

F(1, 97) = 0.14, p = .708, ηp2 = .00, pBIC(H0|D) = .90. Analysis of responses to only the first trial

also showed no interaction between product orientation and handedness, F(1, 93) = 0.01, p

= .944, ηp2 = .00, pBIC(H0|D) = .91, nor an interaction between handedness, product orientation,

and scale orientation, F(1, 93) = 2.26, p = .136, ηp2 = .02, pBIC(H0|D) = .75. Thus, we did not

replicate the match effect found by Elder and Krishna (2012). In Experiment 3 of their study,

they found that the match effect was present for positively valenced products, but reversed for

negatively valenced products. To investigate the relation between general product liking and

match effect, we calculated the correlation between average rating and the difference between

right and left oriented version (a positive difference means a preference for the right-oriented

product) for each product. For left-handers, a negative correlation indicates greater match effects

for positive products, and for right-handers, a positive correlation indicates greater match effects

for positive products. This analysis did not show a significant relation between match effect and

average rating, r(10) = -.41 for right-handers and r(10) = -.05 for left-handers. Although these

correlations were not significant, presumably because the number of objects was low, the

Page 4: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 4

negative correlation suggests that for right-handers match effects were larger for products that

had lower average ratings.

Thus, we found no evidence that participants’ purchase intentions were increased for

products that were oriented towards their dominant hand, nor did we find evidence that purchase

intention ratings were affected by the spatial correspondence of product orientation and response

location. To investigate why we did not replicate Elder and Krishna’s findings, we compared our

procedure to that used in their experiments. One difference between our experiment and theirs

that might be responsible for the differences in findings is that in our experiments, the product

remained visible while participants indicated their purchase intention, whereas in Elder and

Krishna’s study, the product disappeared before participants could indicate their purchase

intention. Previous research has shown that object affordances are activated during object

perception, and it is assumed that affordances are activated by visual information from objects.

Moreover, affordances affect behavior mostly within about 500 milliseconds after stimulus onset

(Makris, Hadar, & Yarrow, 2011). Therefore, we expected that any effect that is the result of

affordances should be strongest while an object is still visible. In the next experiment, however,

to make the task as similar as possible to Elder and Krishna’s studies, the product was no longer

visible when participants indicated their purchase intention.

In addition, to further investigate a potential effect of object-response orientation

correspondence, we also manipulated the direction of the response scales between participants.

We added a horizontal and a vertical condition in which the end points of the Likert scale were

reversed so that 1 meant high intention to buy and 9 meant low intention to buy. If activation of

affordances influenced intention to purchase, scale direction should not influence the results. If

Page 5: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 5

object-response correspondence influenced intention to purchase, however, the effect should

reverse for the reversed horizontal scale direction.

Experiment 2

Method

Participants. A group of 200 participants were recruited through Amazon’s Mechanical

Turk (75 females, Mage = 30.7, age range 18-67, 24 left-handed). Scale orientation and scale

direction were manipulated orthogonally, resulting in four conditions. Fifty participants were

assigned randomly to each condition.

Materials and Procedure. The materials and procedure were the same as in Experiment

1, with two exceptions. The product photograph was presented without the question and response

scale. Participants pressed the space bar when they were done viewing the object. Then the

photograph disappeared and the question and response scale were presented and participants

indicated their intention to purchase. The second change from Experiment 1 was that in two

additional conditions the scale direction was reversed so that 1 meant very likely to buy and 9

meant not likely to buy.

Results

Ratings on the reversed scales were recoded to the same scale as the regular scales, so

that higher numbers meant higher intention to purchase. For each participant the mean ratings

per condition were calculated. The means per condition are shown in Figure 2. The data were

analyzed with a repeated measures ANOVA. The analyses showed no match effect; there was no

interaction between product orientation and handedness, F(1, 192) = 0.22, p = .643, ηp2 = .00,

Page 6: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 6

pBIC(H0|D) = .93, nor an main effect of product orientation, F(1, 192) = 0.21, p = .651, ηp2 = .00,

pBIC(H0|D) = .93. We then investigated if a reversed scale resulted in higher likelihood ratings for

left-oriented products and lower likelihood ratings for right-oriented products than the regular

scale. In addition, this effect should occur for horizontal scales but not for vertical scales.

Contrary to these predictions, there was no interaction between product orientation, scale

orientation, and scale direction, F(1, 192) = 1.12, p = .292, ηp2 = .01, pBIC(H0|D) = .89. The means

suggest that there was a difference between ratings for left- and right-oriented products in the

horizontal, reversed scale condition, but this was not significant and the pattern was opposite to

what was expected. The correlation between average rating and the difference between right and

left oriented version for each product was r(10) =.18 for left-handers and r(10) = -.06 for right-

handers, again showing no higher match effect for more positive products. In sum, the results of

Experiment 2 show no match effects when products were oriented towards the dominant hand,

nor an object-response correspondence effect.

In the next experiments we further investigated whether spatial correspondence between

product orientation and response has an effect on the ratings. Bub and Masson (2010) found that

correspondence between object orientation and response had a larger effect when participants

had to make a reach action. In the previous experiments, and in several of the experiments

reported by Elder and Krishna, participants responded by moving and clicking a mouse, which

presumably did not involve a reach action. Moreover, holding an object in the hand might

actually interfere with activation of affordances for different objects. In one experiment, Elder

and Krishna did not find orientation match effects in intention to purchase ratings when

participants were holding a clamp in their dominant hand (see also Shen & Sengupta, 2012, for

effects of manually held objects on product evaluation), although in their other experiments

Page 7: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 7

holding a computer mouse did not seem to abolish the match effects. However, to investigate if a

match effect would show up if participants did not hold anything in their hands, in the next

experiments participants had to use the numbers 1 to 9 on their keyboard to respond, which

involved a reach action. In Experiment 3 the regular scale was used, where higher numbers

represented higher intention to purchase. In Experiment 4 the scale was reversed, so that higher

numbers represented lower intention to purchase. If purchase intention is influenced by hand

preference, we should find higher intentions for objects that are oriented toward the dominant

hand than toward the non-dominant hand in both experiments. If, on the other hand, purchase

intention is influenced by object-response correspondence, we should find higher purchase

intentions for right-oriented than left-oriented products in Experiment 3 but the reverse, higher

purchase intentions for left-oriented than right-oriented products, in Experiment 4.

Experiment 3

Method

Participants. A group of 103 participants were recruited through Amazon’s Mechanical

Turk (50 females, Mage = 33.5, age range 18-69, 14 left-handed).

Materials and Procedure. The materials and procedure were the same as in Experiment

2, except that participants typed a number between 1 and 9 to indicate their intention to purchase.

As a result, there was no manipulation of scale orientation.

Results

For each participant the mean likelihood ratings per condition were calculated. The

means per condition are shown in Figure 3. The data were analyzed with a repeated measures

Page 8: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 8

ANOVA. Likelihood ratings were higher for right-oriented products than for left-oriented

products, F(1, 101) = 10.78, p = .001, ηp2 = .10, pBIC(H0|D) = .05. The effect of orientation was

not different for right- and left-handed participants, F(1, 101) = 1.37, p = .244, ηp2 = .01, pBIC(H0|

D) = .84. Further analyses showed that both groups rated right-oriented products higher than left-

oriented products, t(13) = 2.46, p = .027 and t(88) = 2.85, p = .005 for left- and right-handers

respectively. The correlation between average rating and the difference between right and left

oriented version for each product was r(10) =.15 for left-handers and r(10) = -.45 for right-

handers, again showing no higher match effect for more positive products. When we analyzed

only responses to the first item, we obtained no significant effect of product orientation, F(1, 99)

= 0.723, p = .397, ηp2 = .01, pBIC(H0|D) = .87 nor an interaction between handedness and product

orientation, F(1, 99) = 0.01, p = .922, ηp2 = .00, pBIC(H0|D) = .91.

In sum, the results show higher intentions to purchase right-oriented than left-oriented

products, and this effect did not differ between right-handers and left-handers. This result

indicates that the effect of product orientation is more likely due to the correspondence between

object orientation and scale orientation than to hand preference, because in the latter case the

effect should have reversed for left-handers. When the object was oriented towards the left,

participants may have had a tendency to move their hand toward the left, resulting in a lower

rating, and if the product was oriented towards the right, participants may have had a tendency to

move their hand toward the right, resulting in a higher rating, simply because the scale had lower

numbers on the left and higher numbers on the right. It is also possible, however, that

participants preferred right-oriented products for some other reason. To investigate the role of

scale orientation we used the same procedure in Experiment 4, except that we reversed the

response scale. If the effect in Experiment 3 was due to object-response orientation

Page 9: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 9

correspondence, we should find higher intentions to purchase for left-oriented than right-oriented

products in Experiment 4.

Experiment 4

Method

Participants. A group of 100 participants were recruited through Amazon’s Mechanical

Turk. One participant failed to answer the handedness questions, and was excluded from the

analysis, leaving a total of 99 participants (32 females, Mage = 31.8, age range 19-63, 15 left-

handed).

Materials and Procedure. The materials and procedure were the same as in Experiment

3, except that the rating scale was reversed, so that 1 represented a high intention to purchase and

9 a low intention to purchase. In addition, all products were presented in random order.

Results

For ease of comparison, ratings were recoded to the regular scale, so that higher numbers

meant higher intention to purchase. For each participant the mean likelihood ratings per

condition were calculated. The means per condition are shown in Figure 4. The data were

analyzed with a repeated measures ANOVA. Likelihood ratings for right-oriented products and

left-oriented products did not differ, F(1, 97) = 0.62, p = .434, ηp2 = .01, pBIC(H0|D) = .88, nor did

object orientation interact with handedness, F(1, 97) = 0.05, p = .832, ηp2 = .00, pBIC(H0|D) = .91.

Thus, we did not find a reversed effect compared to Experiment 3, but we also did not replicate

the higher ratings for right-oriented products. To investigate if the effect of orientation was

affected by the direction of the rating scale, we analyzed the combined data from Experiments 3

Page 10: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 10

and 4. This analysis showed a significant interaction between orientation and experiment, F(1,

198) = 7.87, p = .006, ηp2 = .04, pBIC(H0|D) = .22.

These results suggest that the intention to purchase ratings were higher for right oriented

products than for left-oriented products as the result of correspondence between object

orientation and response orientation. If an object was oriented to the right, participants may have

had a tendency to reach to the right, resulting in a response that was more to the right. Because

responses to the right resulted in higher numbers than responses to the left, it seemed that

participants had a higher intention to purchase right-oriented than left-oriented products.

However, when the scale was reversed such that lower numbers meant a higher intention to

purchase, the effect disappeared. The results also showed that there was no difference between

right-handers and left-handers. Thus, we again found no evidence that participants have higher

intentions to purchase products that are oriented toward their dominant hand than toward their

non-dominant hand. A more consistent explanation is that participants still had a tendency to

give a spatially corresponding response.

It is a bit puzzling, however, that the effect did not reverse when the response scale was

reversed. Although we obtained an interaction between object orientation and experiment, the

results also showed that the effect was absent in Experiment 4. Therefore, before concluding that

the effect of object orientation in Experiment 3 was due to the correspondence between object

and response orientation we wanted to replicate this finding. In Experiment 5 we combined the

scale directions of Experiments 3 and 4 in a between subjects design to investigate the

interaction between object orientation and scale direction.

We also wanted to check for another factor that may have affected the results. In

Experiments 3 and 4 we had assumed that participants used the row of number keys at the top of

Page 11: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 11

their keyboard to type their responses. It is possible, however, that some participants used the

number key pad on their keyboard instead. In that case, the correspondence between object

orientation and response orientation would be more complicated. Therefore, we asked

participants which number keys they had used for their rating after the experiment was finished

so that we could use this information in the analyses.

Experiment 5

Method

Participants. A group of 190 participants were recruited through Amazon’s Mechanical

Turk (101 females, Mage = 34.2, age range 18-63, 28 left-handed). Scale direction was

manipulated between participants.

Materials and Procedure. The materials and procedure were the same as in Experiments

3 and 4. At the end of the experiment in addition to the demographic questions we asked whether

participants had used the number keys at the top of the keyboard or the number pad.

Results

Ratings on the reverse scale were recoded to the regular scale, so that higher numbers

meant higher intention to purchase. Three participants did not exclusively use the row of number

keys for their answers. We present the analyses of only the data from the other 187 participants.

Analyses including those three participants showed the same pattern of results. For each

participant the mean likelihood ratings per condition were calculated. The means per condition

are shown in Figure 5. The data were analyzed with a repeated measures ANOVA. Likelihood

ratings for right-oriented products and left-oriented products did not differ, F(1, 183) = 0.33, p

Page 12: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 12

= .565, ηp2 = .00, pBIC(H0|D) = .92, nor did object orientation interact with handedness, F(1, 183)

= 0.13, p = .722, ηp2 = .00, pBIC(H0|D) = .93, or scale direction, F(1, 183) = 0.22, p = .639, ηp

2

= .00, pBIC(H0|D) = .92. Thus, we again did not replicate the match effect, nor did we replicate the

object-response correspondence effect. The correlation between average rating and the difference

between right and left oriented version for each product was r(10) = -.11 for left-handers and

r(10) = -.28 for right-handers, again showing no higher match effect for more positive products.

In Experiment 6 we again tried to replicate the results from Experiment 3 with a slightly

changed procedure by leaving the product photograph in view while participants gave their

response. Comparison of the results from Experiments 1 and 2 showed no effect of removing the

product from view before the rating was given. As we discussed in the introduction to

Experiment 2, it seemed likely that the effect of object orientation is strongest in early stages of

processing when the object is still in view. Thus, in Experiment 6 we aimed to replicate the

results of Experiment 3 that participants gave higher ratings to right-oriented than left-oriented

products while these objects remained in view.

Experiment 6

Method

Participants. A group of 100 participants were recruited through Amazon’s Mechanical

Turk. Three participants failed to answer all questions, and were excluded from the analysis,

leaving a total of 97 participants (40 females, Mage = 32.4, age range 18-58, 14 left-handed).

Page 13: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 13

Materials and Procedure. The materials and procedure were the same as in Experiment

3, except that the product remained in view while participants typed a number between 1 and 9 to

indicate their intention to purchase.

Results

For each participant the mean likelihood ratings per condition were calculated. The

means per condition are shown in Figure 6. The data were analyzed with a repeated measures

ANOVA. Likelihood ratings for right-oriented products and left-oriented products did not differ,

F(1, 95) = 0.16, p = .692, ηp2 = .00, pBIC(H0|D) = .90, nor did object orientation interact with

handedness, F(1, 95) = 1.15, p = .286, ηp2 = .01, pBIC(H0|D) = .85. Thus, we did not replicate the

results from Experiment 3 that higher ratings were given to right-oriented than left-oriented

products. The correlation between average rating and the difference between right and left

oriented version for each product was r(10) =.30 for left-handers and r(10) = -.35 for right-

handers, again showing no higher match effect for more positive products.

To sum up the results so far, in six experiments we have not found any evidence that

participants indicated higher intentions to purchase products that are oriented towards their

dominant hand than towards their non-dominant hand. Thus, we have not replicated Elder and

Krishna’s results, even though our experiments had very high power to find such effects.

Moreover, the Bayesian analyses showed that we had positive to strong evidence for the null

hypotheses that there were no interactions between handedness and product orientation.

We also explored the alternative hypothesis that higher ratings might be due to spatial

correspondences between the product orientation and the response scale. Only one result in

Experiment 3 was consistent with this hypothesis, however, we did not replicate this effect in

Experiments 5 and 6, which suggests that the result in Experiment 3 may have been a false

Page 14: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 14

positive. Thus, we also did not find reliable evidence that spatial correspondences influenced the

intention to purchase ratings.

There is one more difference between Elder and Krishna’s study that might be relevant.

In all but one of their experiments they presented food products, whereas so far we have

presented only inedible artifacts such as tools and kitchen utensils. Perhaps food products invoke

stronger intentions to act than other objects, and thereby cause orientation match effects. Elder

and Krishna found similar effects for a picture of a thermos cup, but a cup may still be close

enough to food to invoke such effects. In the next experiment we replaced the pictures of

inedible products from Experiments 1 to 6 with pictures of food and drink.

Experiment 7

Method

Participants. A group of 100 participants were recruited through Amazon’s Mechanical

Turk (25 females, Mage = 30.4, age range 18-66, 12 left-handed).

Materials. Photographs of 12 different food products were used (cream of spinach soup,

chocolate cake, hot chocolate, ice cream, mashed potatoes, pizza slice, popsicle, rice, sandwich,

spaghetti, tea, vanilla yogurt). Each photograph showed an affordance on one side (e.g., a

utensil, a cup handle, or the product itself that was oriented toward one side). Each photograph

was mirrored to create two orientations for each product.

Procedure. The procedure was the same as that of the horizontal regular scale in

Experiment 2. Participants saw the product photograph and pressed space to continue. Then the

photograph disappeared and the question and response scale were presented. Participants

Page 15: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 15

indicated their rating by clicking the button of their choice. Trials were presented in random

order, with the restriction that the picture of the spinach soup was always presented on the first

trial.

Results

For each participant the mean ratings per condition were calculated. The means per

condition are shown in Figure 7. Likelihood ratings for left-oriented and right-oriented products

did not differ, F(1, 98) = 1.21, p = .275, ηp2 = .01, pBIC(H0|D) = .84. Object orientation did not

interact with handedness, F(1, 98) = 2.23, p = .138, ηp2 = .02, pBIC(H0|D) = .76. The ratings for

only the first object (spinach soup) showed no effect of object orientation, F(1, 96) = .22, p

= .638, ηp2 < .01, pBIC(H0|D) = .90, nor an interaction with handedness, F(1, 96) = .83, p = .363,

ηp2 = .01, pBIC(H0|D) = .87. The correlation between average rating and the difference between

right and left oriented version for each product was r(10) =.18 for left-handers and r(10) = .07

for right-handers, again showing no higher match effect for more positive products. In sum, the

results of Experiment 7 show no match effects when products were oriented towards the

dominant hand, nor an object-response correspondence effect. They do show a general higher

average rating to left-oriented than right-oriented products, although this was a weak effect, and

the Bayesian analysis showed that the data did not support either the presence or absence of a

left-oriented advantage.

In the next experiment, rather than test participants through Mechanical Turk, we tested

participants in the lab. This allowed us to better control the environment in which participants

performed the task. In all experiments so far we have presented the left and right oriented objects

in random order. Thus, orientation of the handle switch several times during the experiment,

Page 16: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 16

which may have attenuated the effect of orientation. In Experiment 8 we blocked the products by

handle orientation.

Experiment 8

Method

Participants. A group of 95 students at the Erasmus University participated for course

credit (67 females, Mage = 20.6, age range 17-34, 10 left-handed).

Materials. The materials and procedure were the same as in Experiments 3 and 4, with

one exception: The items were presented in two blocks of trials. One block consisted of six right

oriented items and the other block consisted of six left oriented items. The order of blocks and

assignment of products to blocks were counterbalanced across participants.

Results

Ratings on the reverse scale were recoded to the regular scale, so that higher numbers

meant higher intention to purchase. For each participant the mean ratings per condition were

calculated. The means per condition are shown in Figure 8. Likelihood ratings for left-oriented

and right-oriented products did not differ, F(1, 91) = 2.80, p = .098, ηp2 = .03, pBIC(H0|D) = .70.

Object orientation did interact with handedness, but in the opposite direction from what was

expected. Left-handers rated right-oriented products higher than left-oriented products, and right-

handers seemed to rate left-oriented products slightly higher than right-oriented products, F(1,

91) = 4.81, p = .031, ηp2 = .05, although the Bayesian analysis indicated that the null and

alternative hypothesis were equally likely, pBIC(H0|D) = .46. Scale direction did not interact with

object orientation, F(1, 91) = 1.62, p = .207, ηp2 = .02, pBIC(H0|D) = .81, or object orientation and

handedness, F(1, 91) = 1.21, p = .275, ηp2 = .01, pBIC(H0|D) = .84. A separate analysis on the data

Page 17: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 17

in the first block only also did not show an effect of match, t(93) = .15, p = .883. In sum, the

results of Experiment 8 show no match advantage when products were oriented towards the

dominant hand, nor an object-response correspondence effect. Rather, they do indicate a reversed

match effect, although the Bayesian analysis showed that the data did not support either the

presence or absence of a match effect.

Additional Analyses

To investigate if our failure to obtain a match effect might be explained by the specific products

that we used, we performed exploratory analyses in which we tested the match effect for the six

‘best’ items. We used the data from Experiments 1 to 3 to identify the six products that showed

the largest average positive difference between the match and mismatch condition. These items

were iron, stapler, sprayer, blowdryer, tennisracket, and knife. Next, we again tested the match

effect in Experiments 4 to 6 and 8 in which the same stimuli were presented, using the ratings for

only these six items. Note that this resulted in unequal observations per condition at the

participant level, because in our counterbalancing method five of the six items happened to be in

one of the two subsets. The results showed no positive match effect, in fact, in all experiments

the ratings were numerically higher in the mismatch than in the match condition, although this

difference was significant only in the data from Experiment 6, t(96) = 2.48, p = .015.

Page 18: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 18

References

Bub, D. N., & Masson, M. E. J. (2010). Grasping beer mugs: On the dynamics of alignment

effects induced by handled objects. Journal of Experimental Psychology: Human

Perception and Performance, 36, 341-358. doi:10.1037/a0017606

Elder, R. S., & Krishna, A. (2012). The "visual depiction effect" in advertising: Facilitating

embodied mental simulation through product orientation. Journal of Consumer Research,

38, 988-1003. doi:10.1086/661531

Makris, S., Hadar, A. A., & Yarrow, K. (2011). Viewing objects and planning actions: On the

potentiation of grasping behaviours by visual objects. Brain and Cognition, 77, 257–264.

Masson, M. E. J. (2011). A tutorial on a practical bayesian alternative to null-hypothesis

significance testing. Behavior Research Methods, 43, 679-690. doi:10.3758/s13428-010-

0049-5

Shen, H., & Sengupta, J. (2012). If you can't grab it, it won't grab you: The effect of restricting

the dominant hand on target evaluations. Journal of Experimental Social Psychology, 48,

525-529. doi:10.1016/j.jesp.2011.11.003

Wagenmakers, E. M. (2007). A practical solution to the pervasive problems of p values.

Psychonomic Bulletin and Review, 14, 779-804. doi:10.3758/BF03194105

Page 19: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 19

Table 1

Overview of Experiments

Exp Picture Response

mode

Scale

orientation

Scale direction

1 Visible during

response

Mouse click Horizontal or

Vertical

1 – 9: low - high intention

2 Disappears before

response

Mouse click Horizontal or

vertical

1 – 9: low - high intention or

1 – 9: high - low intention

3 Disappears before

response

Numeric key

press

Horizontal 1 – 9: low – high intention

4 Disappears before

response

Numeric key

press

Horizontal 1 – 9: high – low intention

5 Disappears before

response

Numeric key

press

Horizontal 1 – 9: low – high intention or

1 – 9: high - low intention

6 Visible during

response

Numeric key

press

Horizontal 1 – 9: low - high intention

7 Disappears before

response

Mouse click Horizontal 1 – 9: low - high intention

8 Disappears before

response

Numeric key

press

Horizontal 1 – 9: low – high intention or

Page 20: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 20

Figure 1. Mean likelihood ratings in Experiment 1 for left and right oriented products. Error bars

show SEM.

Page 21: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 21

Figure 2. Mean likelihood ratings in Experiment 2. Error bars show SEM. Ratings on the

Reversed scale were recoded to a regular scale such that higher numbers represent higher

likelihood.

Page 22: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 22

Figure 3. Mean likelihood ratings in Experiment 3. Error bars show SEM.

Page 23: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 23

Figure 4. Mean likelihood ratings in Experiment 4. Error bars show SEM.

Page 24: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 24

Figure 5. Mean likelihood ratings in Experiment 5 for the regular and reversed direction scales.

Error bars show SEM.

Page 25: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 25

Figure 6. Mean likelihood ratings in Experiment 6. Error bars show SEM.

Page 26: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 26

Figure 7. Mean likelihood ratings in Experiment 7 for food products. Error bars show SEM.

Page 27: · Web viewIf the results were caused by object-response correspondence, we should expect only an effect when the scale is oriented horizontally but not when it is oriented vertically

ACTION SIMULATION 27

Figure 8. Mean likelihood ratings in Experiment 8 for the regular and reversed direction scales.

Error bars show SEM.