Overview
Experiment 3 was conducted to examine a mixed prime condition, namely one prime each on the positive and negative of the target. It will focus on the presence of a dimensional range overlap each prime has with the target, as well as the non-overlap that exist between both primes at the same time; the subsequent assimilation or contrast effect produced by the mixed prime condition in which caused the consequential moves would be the interest of this experiment.
Hypothesis Recap
Hypothesis 3: For the condition in which the two primes have different valences and each overlap with the target initially, but the positive prime overlaps with the target at a larger amount than the negative prime, and both primes do not overlap with each other, the target will be more likely to shift toward the positive prime than toward the negative prime.
Figure 7.1 Hypothesis 3
Participants and Design
A total of 30 participants, aged between 20 and 30 years old, took part in this study.
As participants were recruited altogether for all three experiments and randomly assigned to one of the four experimental conditions later, recruitment was the same as Experiment 1 and 2, where most of these participants were undergraduates and postgraduate students
approached in the school campus of National Taiwan Normal University, while the rest were recruited via online postings. Participants signed up voluntarily for one of the 32 sessions held over four days of the study, in exchange for a lucky draw chance to win NTD500 worth of Uni-President vouchers.
Similar too, was the within-subject design of the experiment, consisting of four sections in the VBA-written questionnaire. The assigned Prime 1 and 2 range ratings were manipulated and preset on the questionnaire programme to have and initial overlap range of 0.5 and 1, negative and positive of the target range offered by each individual participant in the previous section respectively. As with Experiment 1 and 2, fillers, by way of
psychological tests and Sudoku games meant to compel the use of participants’ cognitive abilities, and possibly disrupt their concentration and accessibility of previously primed item, were inserted in between the sections, in order to collect as unaffected Stage 1 contexts and target stimuli for Stage 2 use as possible.
Procedure
The first section of Experiment 3 was identical to Experiment 2, including the psychological test that concluded the section.
Section 2 of Experiment 3 was also similar to Experiment 2, only that as the condition of this experiment would require comparisons of both primes with each other and each with the target, participants were asked to provide two apparel brands (Prime 1 and 2) instead of
one, after rating the target advertisement (representative value and item boundaries).
Participants were told to input one apparel brand (Prime 1) within a specified range, followed by brands of two other irrelevant items, also within different given ranges. A Sudoku game was placed after. Participants were subsequently asked to give another apparel brand within a different stated range (Prime 2), also followed by brands of two other irrelevant items best fitted in listed ranges. The section was wrapped up with a different Sudoku game.
Finally, participants were made to believe that the computer has chosen the clothing industry for them to answer the following questions on. After being prompted of both Prime 1 and 2 brands they specified in the section before, they were also instructed to rate the
representative value and interpretative range on the “prestige level” dimension of Belìssimo (the target) and both primes of the condition again.
Results
Manipulation Check
Prime 1 and Prime 2. In the condition where the mutual effect of Prime 1 and 2 was examined, the paired sample t-test revealed that the non-overlap range and the distance between their representative values were significantly different after their mutual effect than before. The Stage 1 non-overlap range was 0.5 initially, differing significantly from the Stage 2 range rating of exactly 0; t(29) = -2.35, p < 0.05. Similarly, the Stage 2 representative value distance (M = 2.10) was also significantly higher than in Stage 1 (M = 2.55); t(29) = 1.99, p <
0.1. Essentially, the changes in the non-overlap range and representative value distance suggested a statistically significant assimilation effect between Prime 1 and 2, where we had initially expected them to repel from each other at first.
Prime 1 and target. Paired sample t-tests were once again conducted to compare the overlap range and distance of the representative values between the negative context, Prime 1, and
target. As depicted in Figure 7.2, Prime 1 and target have strongly assimilated towards each other, producing statistically significant Stage 2 overlap range (M = 1.03) compared with its pre-context effect one (M = 0.50); t(29) = 2.90, p < 0.01, as well as shorter distances between their representative values, with the Stage 2 distance (M = -0.87) substantially shorter than its Stage 1 distance (M = 1.40); t(29) = 7.30, p < 0.001. While the initial overlap range of Prime 1 and target would be expected to cause an assimilation effect to occur, we did observe from the post-context effect changes in representative values, lower and upper bound ratings that target did not shift towards Prime 1 as Prime 1 shifted greatly towards the target. The significance in the overlap and distance between the representative values of Prime 1 and target was a result of the major movement in Prime 1. (Please also see Table 7.1)
Prime 2 and target. Results from the paired sample t-tests used to test the overlap range and distance of the representative values between the positive context, Prime 2, and target
revealed consistent results. Firstly, the initial overlap range (M = 1.00) became smaller after they were mutually affected by each other (M = 0.83); t(29) = -0.72, p = 0.48. Next, the distance between their representative values were also shown to have lengthened in Stage 2 (M = 1.23) as compared to Stage 1 (M = 1.15); t(29) = -0.37, p = 0.71. Both the smaller overlap range and longer distance between presentative values suggest that both Prime 2 and target range have repelled from each other, though both were not supported statistically. This trend of contrast is in conflict with our expectation of an assimilation effect, to be caused by the larger overlap range (M = 1.00) manipulated in the study. (Please refer to Table 7.1 and Figure 7.2)
Evaluation
Prime 1. From the data analysed, it was revealed that only Prime 1 had all its Stage 2 representative value ratings as well as the upper and lower bounds significantly different
from those of its Stage 1. Its post-context affected representative value shifted substantially (M = 3.73) positively, as opposed to its initial value (M = 2.98); t(29) = -3.28, p < 0.01. Both lower and upper bound ratings of Prime 1 also showed great movements towards the higher end of the scale; the post-movement lower bound (M = 2.88) was significantly higher than the pre-influenced one (M = 1.98); t(29) = -4.27, p < 0.001, similar to that of its upper bound where the Stage 2 rating (M = 4.77) was also significantly higher than in Stage 1 (M = 3.98);
t(29) = -3.55, p = 0.001).
Prime 2. On the other hand, Prime 2 also displayed vague movement positive of the scale, though its post-context affected representative value was significantly higher in Stage 2 (M = 5.83) than in Stage 1 (M = 5.53); t(29) = -1.71, p < 0.1. Tthe lower and upper bound point-ratings were also higher, though not statistically supported; the lower bound was weakly higher after context effect (M = 4.77) than before (M = 4.53); t(29) = -1.11, p = 0.28, as with the higher upper bound (M = 6.68) which barely shifted rightwards of the scale from its initial upper bound ratings (M = 6.53); t(29) = -0.95, p = 0.35.
Target. Likewise, there were also negligible shifts made positive of the scale by the target range. Both lower and upper bound ratings of the target moved feebly higher; the
post-movement lower bound (M = 3.73) was insignificantly higher than the pre-influenced one (M
= 3.48); t(29) = -1.54, p = 0.13, similar to that of its upper bound where the Stage 2 rating (M
= 5.60) was also significantly higher than in Stage 1 (M = 5.53); t(29) = -0.38, p = 0.71). The representative values of the target before context-effect (M = 4.38) was also lower than its after it was influenced (M = 4.60); t(29) = -1.04, p = 0.31.
Examining the mutual effects of both contexts with the target as well as with each other, namely Prime 1 with Prime 2, Prime 1 with target, and Prime 2 with target, we derived the following results.
Overall Results and Discussion
The mixed prime condition of Experiment 3 was not supported as we had
hypothesised in Hypothesis 3. From the post-context effect data analysed, Prime 1 and 2 did not repel from each other as expected, given our 0.5 lack of overlap range between them.
Though Prime 2 did move feebly in the direction we hypothesised, Prime 1 did not move in the negative direction nor subsequently shift leftwards of the scale to create a non-overlap with the target. (Please refer to Table 7.1 and Figure 7.2)
With a larger initial overlap between Prime 2 and the target, we had expected that though a contrast effect would take place between Prime 1 and 2, resulting in their
subsequent moves negative and positive of the target respectively, an overlap between Prime 2 and the target would still exist, thereby leading them to assimilate towards each other and the latter would be more likely to shift towards the former thereafter. This is shown in the post-context effect results, where it appeared that the target range had a greater tendency to move towards Prime 2, shifting rightwards of the scale towards Prime 2, though this tendency was not confirmed statistically. (Please refer to Table 7.1 and Figure 7.2 for more details)
Table 7.1
Means for target, as well as Primes 1 and 2 in Experiment 3
Prime 1 Representative Value Lower Bound Upper Bound
Stage 1 Mean 2.98 1.98 3.98
Stage 2 Mean 3.73 2.88 4.77
Prime 2 Representative Value Lower Bound Upper Bound
Stage 1 Mean 5.53 4.53 6.53
Stage 2 Mean 5.83 4.77 6.68
Target Representative Value Lower Bound Upper Bound
Stage 1 Mean 4.38 3.48 5.53
Stage 2 Mean 4.60 3.73 5.60
Prime 1 and 2 Non-Overlap Range Distance between Representative Values
Stage 1 Mean 0.50 2.55
Stage 2 Mean 0.00 2.10
Prime 1 and
Target Overlap Range Distance between Representative Values
Stage 1 Mean 0.50 1.40
Stage 2 Mean 1.03 -0.87
Prime 2 and
Target Overlap Range Distance between Representative Values
Stage 1 Mean 1.00 1.15
Stage 2 Mean 0.83 1.23
Figure 7.2 Diagram of movements of the target, as well as Primes 1 and 2 in Experiment 3, indicating respective representative values, lower and upper bound ratings.
CHAPTER 8 GENERAL DISCUSSIONS AND CONCLUSION
Conclusion Review
From the results presented, most of our observed tendencies have been consistent with the hypotheses in the current research, though a couple of them remain to be confirmed.
In the first hypothesis where two non-overlap primes were presented, the lower of Prime 1 as well as both the lower and upper bound of Prime 2 shifted in the direction of a contrast, as hypothesised. These tendencies were however not significant; Hypothesis 1 was thus not confirmed. After further analyses were ran on data sets that met our expectations, contrast effect was made more apparent; all except the lower bound of Prime 1 shifted in repelling directions, significantly widening the range of non-overlap and supporting our hypothesis.
In the next result, besides the lower bound of the target that did not move accordingly towards Prime 1 when there was an initial overlap in between them, both target and negative context, Prime 1, was proven to have assimilated, with the resulting overlap range and distance between their representative values statistically supporting Hypothesis 2A.
Likewise in the similar hypothesised situation of Hypothesis 2B, apart from the upper bound of Prime 2, on the positive valence of target, which did not move accordingly towards the target when there was an initial overlap in between them, both target and Prime 2 showed tendencies of assimilation, but changes in the overlap range and distance between their representative values could not be proven statistically. Hypothesis 2B was thereby not confirmed. Only when further analyses were ran on selected data sets did assimilation effect become more obvious; all lower and upper bounds of target and Prime 2 moved towards each other, and Hypothesis 2B was supported as the degree of overlap range and distance between their representative values became significant.
Finally, when context effects between multiple contexts and their combined prime effect on target judgment was tested, the notion of assimilation effect when there was an initial overlap was supported but could not be observed in the case when there was an initial lack of overlap. As such, Hypothesis 3 was not supported.
Limitation and Suggestions
Procedural Design
One major limitation we encountered was with the lack of participant pool. Due to the limited time and budget, we opted for our questionnaire to be written on Excel-run VBA programme.
Two issues arose consequently during the testing stage of the programme; the VBA programme could neither work on outdated versions of Excel nor non-Traditional Chinese system language computers, and the file would be a relatively huge one to be sent to individual participants. As such, not only did the search of working computers that can accommodate the number of participants proved to be a challenge, the recruitment of
participants was equally trying. Participants had to be present physically at one of the 32 time slots given over four days, in a specific study lab with four laptops tried and tested with our questionnaire programme, opposed to the conventional way of approaching students in their respective lecture rooms during their class breaks The inconvenience greatly reduced their willingness to take part.
Lack of Motivation
Much as the use of computers was helpful in manipulating overlap ranges between the multiple contexts and target, as well as carrying out the whole experiment procedure, the process of recruiting participants was a tedious one. During the recruitment exercise in the school campus, participants were told that every completion of the main experiment would give them a chance to draw for NTD500 Uni-President gift certificates. That said, few
participants were willing to make a special trip for the main experiment. Attempts to recruit online via postings on popular forums (i.e. PTT) were also to no avail, leading to our subsequent trips to other different school campuses to recruit for participants and managing barely enough participants for each experiment.
Questionnaire Design
The questionnaires that were done on computer programmes were one of a new method, keeping many participants excited and intrigued during the entire course of the study. One shortfall of the questionnaire design was, however, the repetition of the Sudoku game in the final experiment, which many participants completed half-heartedly. Though the Sudoku games served as an interruption of concentration between sections, it would require a higher level of thinking skills to complete, which was unexpected by the participants in a
questionnaire. Most participants were not pleased with the repetition of the Sudoku games, and tend to fill in the blanks with random numbers after expressing their displeasure.
Enthusiasm was noted to have reduced, though they were on to their final few questions by then.
Suggestions
The use of the computer programme proved to be a benefitting one, though the programme could be modified to suit a wider variety of computers; the files could be sent to individual participants and returned upon completion, increasing the convenience of participants.
Alternatively, higher incentives could be given to increase the efforts they are willing to make for the study. In example, instead of a lucky draw chance, a guaranteed gift for every participant who completes the study would be more fruitful.
Research Contributions and Practical Implications
In the present study, we attempted to apply and extend the Dimensional Range Overlap Model (Chien et al., 2010) to conditions of multiple primes. By providing a more complete scenario to illustrate the underlying factor of multiple contexts, we hope to offer some practical consumer insights to the marketing and management field.
The current research realistically takes into consideration the countless of similar products consumers encounter with, and make their evaluations and judgments from every day. Though derived from the DROM, both researches differ with a crucial element; the DROM was experimented based on a single prime and its corresponding effect on the target, whilst the current research recognised the roles of multiple primes. Nonetheless, the crux of the model, the existence of an overlap or non-overlap, would determine the movement of perceptual judgment. By having a deeper understanding of how consumers form their evaluations and apply to the different scenarios and environmental settings, a marketer can strategize and manipulate contextual information to influence consumers’ perception towards a given product.
In example, marketers selling cars can choose who and what to put in their advertisements, be it a well-known celebrity endorsing the product, or arranging for the advertisements to be placed after that of a prestigious brand (i.e. Mercedes Benz).
Alternatively, marketers can also influence consumer perceptions in creating a favourable external contextual environment setting, by placing its brand of cars amongst more prestigious ones, or hold joint events with other highly acclaimed product brands.
Future Research
Despite the lack of confirmation in some of the hypotheses of our current research, we do note the overall shift tendencies were supportive of our general assumptions, and thereby believe that they are worthy of further exploration and research. The greatest limitation in our main experiment was in its procedure. Where the length and design of the questionnaire, as well as execution and environmental settings of the study were expressly positive, the recruitment of participants posed an unexpected challenge. As we had to recruit a
considerable number of participants to come forth at a specified time and venue, participant reluctance was apparent, given the lack of incentive and motivation. Further setbacks were met when we had to recruit our study participants within a limited time, especially in the period where schools were starting to dismiss students for summer break. If this issued is rectified, the experiment may potentially be a successful one.
In our final dual-primes condition experiment, we note that both Prime 2 and target had general movements according to what was hypothesised, but Prime 1 had an entirely unexpected assimilation move, significantly contradicting our assumptions. One reason proposed could be that the attitude certainty on the lower bound of Prime 1, in which overlapped with target, was so strong at the particular bound that instead of a contrast, as predicted, with the other prime in which it had no overlap with, it moved together with the target in which it had a strong overlap with. Researchers can examine for the factor of attitude certainty on one or both bounds of each prime and target evaluated.
Another area worth investigating would be the degree of symmetry in simultaneous shifts between two context ranges (Reciprocity Hypothesis, Hsiao 2002). While the
Reciprocity Hypothesis proved simultaneous shifts between two context ranges, it did not address the issue on whether both corresponding shifts would be congruent. Therefore,
another reason why Prime 1 could have assimilated with target instead, was possibly because its post-context effect with Prime 2 had resulted in a lesser than expected repel from Prime 2, thereby retaining an overlap with target, albeit a smaller one, after it supposedly contrasted.
Subsequent shifts would then see all three overlapped target and contexts ranges assimilating closer to each other, though target would ultimately end up higher up the scale, given the larger initially overlap range with Prime 2 manipulated. This assumption of incongruent shifts could possibly occur singly, or in conjunction with attitude certainty. Future research may likely include and investigate the role and influences of these suggested factors.
Last but not least, valence effect is not verified in the present research; we cannot determine if a prime placed to overlap with the target on its negative valence would result in a different contextual effect than a similar prime manipulated to overlap on the positive valence of the target. Therefore unlike in the current research, future research may consider a symmetrical setup for a more accurate examination, with the degree of overlap kept constant
Last but not least, valence effect is not verified in the present research; we cannot determine if a prime placed to overlap with the target on its negative valence would result in a different contextual effect than a similar prime manipulated to overlap on the positive valence of the target. Therefore unlike in the current research, future research may consider a symmetrical setup for a more accurate examination, with the degree of overlap kept constant