Overview
Experiment 1 tested on the effect of context range non-overlap on occurrence of assimilation or contrast. In essence, this two-stage main experiment was to test for any movement of the contexts and target’s representative values, as well as changes in their respective boundary values pre- and post-context effect. While Chien et al.’s (2010) Dimensional Range Overlap Model suggest that an overlap/non-overlap between the context/target ranges will cause the occurrence of an assimilation or contrast effect, we also infer from Hsiao’s (2002) Reciprocity Hypothesis that both contexts will have a mutual effect on each other when presented at once, thereby resulting in concurrent shifts towards or away from each other. Therefore using the same fixed context range width with the same relative distance between the presentative values of both contexts, we manipulated Prime 1 and 2 to have a 0.5 non-overlap in between them in Stage 1 of Experiment 1, and a comparison of whether this degree of non-overlap will change will be done with results collected in Stage 2.
Hypothesis Recap
This experiment shall focus on testing for a context effect, which we expect would be a contrast, arising on a lack of dimensional range overlap between two contexts.
Hypothesis 1: When prime 1 and 2 have different valences and there is no overlap between the two primes initially, then prime 1 and prime 2 will contrast away from each other.
Figure 5.1 Hypothesis 1
Participants and Design
A total of 31 young adults aged between 20 and 30 years old were recruited to participate in the study; most of these participants were undergraduates and postgraduate students from National Taiwan Normal University, while the rest were recruited via online postings. Participants signed up voluntarily for one of the 32 sessions, each lasting about ten to fifteen minutes, held over four days of the study, in exchange for a lucky draw chance to win NTD500 worth of Uni-President vouchers.
The three-section Experiment 1 was carried out by way of a within-participants design, with dimensions retained from the pretest, as well as new dimensions included. Similar to the design of our pretest, the first section of the questionnaire served as a practice for participants to get used to giving representative value and range ratings, as well as providing items they can think of within a given range rating. An initial non-overlap of 0.5 was manipulated Prime 1 and 2, within our manipulated range 2.5-4.5 and 5-7 respectively, by way of the computer programme. Part 3 required participants to give ratings again after both contexts have been mutually influenced. The target apparel brand advertisement will not be shown in this set of questionnaire at all.
Procedure
The study questionnaire of Experiment 1 was saved on one of the four computers we placed in our study room at National Taiwan Normal University. As there was a lack of study rooms and computers to accommodate all 31 students at once, we held the experiments in 8 different sessions over four days, with about three to four students in each session. Each participant was randomly assigned to a seat, with one of four computers was pre-set with Experiment 1 questionnaire.
In the beginning of the questionnaire, participants were told they would be taking part in a study to test their reading and cognitive abilities of various advertisements and articles shown in magazines currently existing in the market. As with the pretest, participants were told in the first section, the practice exercise, that they would first be tested on their
observation and sensitivity abilities, and asked to practise giving one-point representative values, interpretational dimensional ranges with upper and lower bounds, as well as to think of an item within a given range and dimension. The first two sample questions in this practice consisted of two dimensions, “hardness of objects” and “sweetness of food”, followed by two practice questions on dimensions of “hardness of objects” and “swiftness of animals’
movement”.
In the second part of the questionnaire, participants were told they would be tested on their imagination abilities, and were subsequently asked to provide apparel brands, which would be used as our context 1 and 2 in the next section. In order to reduce the likelihood of participants guessing the purpose of our questionnaire, we first asked participants to provide an apparel brand for context 1 as fitting within our given range of 2.5-4.5 prestige level (low) as possible, before asking for irrelevant car and handphone brands within the range of 5-6 (moderate) and 8-10 (high) on dimensions “safety level” and “price level” respectively. A psychological test meant to disrupt their concentration from previous questions was then inserted, followed by asking for another apparel brand for context 2, as best fitted within the prestige range of 5-7 (moderate). The section ended by requiring participants to provide another irrelevant fast-food chain and handphone brand within the health and fashion range of 1-3 (low) and 7-9 (high) respectively.
Before the commencement of the third section, a Sudoku game was placed to reduce participants’ accessibility of previously given ratings.
In the beginning of section 3, participants were told that the computer programme has chosen the clothing industry for them to answer subsequent questions on. Context 1 and 2, which they had provided in Section 2, were brought up then and participants were asked to give representative values and dimensional range ratings for both contexts again.
Results
Experiment 1
Manipulation check. After taking all the data into consideration for an analysis, results from our within-participants t-test (Table 5.1) showed that the lower and upper bounds of the two contexts; the lower bound of Prime 1 (M = 2.48) did not differ significantly from its initially manipulated lower bound of M = 2.50; (t(30) = -0.08, p = 0.94), as neither was its upper bound of M = 4.63, compared to the initial M = 4.50; t(30) = 0.58, p = 0.57. Similarly, the lower (M = 5.27) and upper bounds of Prime 2 (M = 7.08) were also not statistically different from the initial value of 5 and 7 respectively; t(30) = 1.28, p = 0.21 and t(30) = 0.38, p = 0.71 correspondingly.
A negligible shift in the non-overlap range between Prime 1 and 2 after their mutual effects were also reflected; though the non-overlap range slightly widened as we had expected (M = 0.65; t(30) = 0.76, p = 0.45), it was not significantly different from its initial non-overlap range 0.5, suggesting that contrast effect did not occur as we had expected.
Prime 1 and 2 evaluations. Paired sample t-tests were also conducted to compare the representative values (results presented in Table 5.1 and Figure 5.2); the positive context, Prime 2, generated slightly higher prestige ratings (M = 6.13) than its initial rating (M = 6.00);
t(30) = -0.78, p = 0.44, while the negative context, Prime 1 also produced higher post-context prestige ratings (M = 3.68) than its context-free ratings (M = 3.50); t(30) = -0.79, p = 0.44.
Further comparisons were made on the representative value differences of the Stage 1
context-free situation between both primes (M = 2.50) with that of Stage 2 (M = 2.45), and found that both primes did not in fact move away from each other simultaneously; t(30) = 0.19, p = 0.85, given the initial lack of overlap in between them. Our Hypothesis 1 was henceforth not confirmed.
Table 5.1
Means for the two primes and the target of Experiment 1
Prime 1 Representative Value Lower Bound Upper Bound
Stage 1 Mean 3.50 2.50 4.50
Stage 2 Mean 3.68 2.48 4.63
Prime 2 Representative Value Lower Bound Upper Bound
Stage 1 Mean 6.00 5.00 7.00
Stage 2 Mean 6.13 5.27 7.08
Prime 1 and 2 Non-overlap Range Distance between Representative Values
Stage 1 Mean 0.50 2.50
Stage 2 Mean 0.65 2.45
Figure 5.2 Diagram of movements of Prime 1 and 2 in Experiment 1, indicating respective representative values, lower and upper bound ratings.
Further Analyses
Manipulation check. We selected those sets of data, 13 in all, with which had a post-context effect overlap range that fulfilled our expectations (> 0.5) for further analyses. From the results generated and depicted in Table 5.2, we can see that the slight shift in the upper bound of Prime 1 and significant shift in the lower bound of Prime 2, both in opposite directions as expected, as well as the shortened context ranges of both Prime 1 and Prime 2 from their initially manipulated range of 2, consequentially created a significantly larger non-overlap range with an M = 1.69; t(12) = 5.72, p<0.001.
Prime 1 and target evaluation. Results once again did not confirm any statistically
substantial movements in the lower (M = 2.54) and upper (M = 4.15) bound of the negative context, Prime 1. However, the lower bound of Prime 2 showed significant movement higher up the scale (M = 5.85 in relation to the initial M = 5.00; t(12) = 2.42, p < 0.05, though its slightly higher upper bound of M = 7.31was not proven statistically compared to the initial M
= 7.00; t(12) = 0.74, p = 0.48.
There was also no change at all in the representative value of Prime 1, from Stage 1 to Stage 2 in our paired sample t-test, with changes almost as negligible in Prime 2 too. The distance between representative values of Prime 1 and 2 also vaguely differ from Stage 1 (M
= 2.50) to 2 (M = 2.65; t(12) = -0.36. p = 0.73), though it did widen as we expected it to in the occurrence of a contrast effect. (Please refer to Table 5.2 and Figure 5.3)
Table 5.2
Means for the two primes of Experiment 1 on selected data sets
Prime 1 Representative Value Lower Bound Upper Bound
Stage 1 Mean 3.50 2.50 4.50
Stage 2 Mean 3.50 2.54 4.15
Prime 2 Representative Value Lower Bound Upper Bound
Stage 1 Mean 6.00 5.00 7.00
Stage 2 Mean 6.15 5.85 7.31
Prime 1 and 2 Non-overlap Range Distance between Representative Values
Stage 1 Mean 0.50 2.50
Stage 2 Mean 1.69 2.65
Figure 5.3 Diagram of movements of Prime 1 and 2 on selected data sets in Experiment 1, indicating respective representative values, lower and upper bound ratings.