• Non ci sono risultati.

Cluster analysis: identification of intuitive and analytical participants

HOW INTUITIVE AND ANALYTICAL STYLES ARE RELATED: COGNITIVE AND DECISION PROFILES

2.2 STUDY 2

2.2.3 Results

2.2.3.5 Cluster analysis: identification of intuitive and analytical participants

Subscales Subsamples Mean SD F(1,288), p η2 Students 0.54 0.53

1. Reflective

Workers 0.88 0.58 15.784 <.001 .053 Students 1.52 1.11

2. Intuitive

Workers 1.02 1.11 8.460 <.005 .029 Tab. 2.29 – CRT: One-way ANOVA comparing scores obtained by occupation subsamples

Statistical mean differences emerged on both the reflective and intuitive indexes.

In particular, students scored lower than workers on the reflective index, and, on the contrary, obtained higher scores than workers on the intuitive index.

Furthermore, a one-way ANOVA was computed in order to compare scores obtained on each index by the different age groups. Results are reported in Table 2.30.

Subscales Age Mean SD F4,284) p η2

18-19 0.52 .50

20-29 0.55 .57

30-39 0.84 .58

40-49 0.78 .63 1. Reflective

50-59 0.87 .23

3.453 <.01 .047

18-19 1.55 1.10

20-29 1.47 1.16 30-39 1.02 1.02 40-49 1.31 1.30 2. Intuitive

50-59 1.25 .46

1.651 .162 .023

Tab. 2.30 – CRT: One-way ANOVA comparing scores obtained by age subsamples

Only on the reflective index statistical differences emerged. Specifically, 18-19 and 20-29 years groups scored lower than the older groups.

2.2.3.5 Cluster analysis: identification of intuitive and analytical

identified two different pattern of correlations, the number of clusters was fixed at two. The variables included in the cluster analysis were the ones which turned out to be openly part of one of the two pattern as emerged in the previous correlational analyses (cfr. 3.3.2). All the components of the Maximization Scale were excluded because of the unclear dimensionality of the instrument. From a theoretical point of view, the maximizing construct should be unidimensional, but, as emerged in both Study 1 and 2, its different components, which are supposed to be just simple examples or facets of the unidimensional construct, showed a trend that is anything but univocal. Moreover, it emerged that not all the components have the same influence in determining the global maximizer score. As a consequence, the whole maximizing construct does not seem to clearly be in an univocal way on the side of either the intuitive or the analytical style and for that reason it was excluded by the cluster analysis which was aimed at identifying only people with an extreme intuitive and analytical profile.

Even if for a different reason, also CRT was not included in this analysis. In that case it was excluded because the instrument was not consistently related with the other scales. Table 2.31 reports the final cluster centres which shows the mean abundance of each variable in each of the clusters. Basing on the variables in each cluster it was possible to give to each cluster a descriptive name, which, specifically, corresponded to “analytical style” as for cluster 1, and “intuitive style” as for cluster 2.

Clusters

1 2

1. GDMS Rational 20.18 16.67

2.GDMS Intuitive 14.61 18.04

3. GDMS Dependent 17.49 16.91

4. GDMS Avoidant 11.84 14.19

5. GDMS Spontaneous 8.03 11.26

6. PID Deliberative 36.93 30.84

7. PID Intuitive 29.86 33.93

8. SOLAT Left 11.05 6.76

9. SOLAT Right 9.55 15.82

Tab. 2.31 – Final clusters centers

Table 2.32 reported the number of samples in each cluster. Almost about half participants turned out to belong to the analytical style cluster, and the other half to the intuitive one.

1 – analytical style 143 Clusters

2 – intuitive style 146 289 Tab. 2.32 – Number of cases in each cluster

In order to test whether the variables included in cluster analysis are suitable to maximize the differences among cases in the different clusters F tests for each variable are computed. Results, which are reported in Table 2.33, supported this hypothesis.

Cluster Error Mean Square Df Mean Square Df

F p

1. GDMS Rational 868,994 1 6,689 281 129,905 <.001 2.GDMS Intuitive 830,092 1 8,812 281 94,203 <.001 3. GDMS Dependent 256,925 1 16,526 281 15,448 <.001 4. GDMS Avoidant 386,776 1 16,369 281 23,628 <.001 5. GDMS Spontaneous 737,589 1 7,437 281 99,177 <.001 6. PID Deliberative 2623,088 1 16,134 281 162,578 <.001 7. PID Intuitive 1172,498 1 17,781 281 65,942 <.001 8. SOLAT Left 1300,398 1 13,664 281 95,169 <.001 9. SOLAT Right 2773,897 1 16,094 281 172,360 <.001 Tab. 2.33 –F-tests for each variable

Through cluster analysis each participant of the whole experimental sample was assigned to one of the two clusters, the analytical or intuitive style cluster.

However, we were interested in identifying, within the total sample, the subsample of participants showing either an extreme intuitive or analytical profile.

We considered the results obtained through cluster analysis the first step of this identification procedure. In fact, after the splitting up of the total sample into two groups, the analytical and the intuitive ones, two distinct procedures were activated.

Firstly, the difference between the intuitive and rational GDMS scales, the intuitive and deliberative PID scales, and the right and left SOLAT scales were calculated. By sticking to the reference literature (Betsch, 2004; Schunk &

Betsch, 2006), only those participants who obtained an absolute difference value higher than 8 on every instruments were identified as analytical or intuitive (e.g.

PID-difference = -10, GDMS-difference = -9, SOLAT-difference = -12 : the participant was classified as analytical; but PID-difference = -3, GDMS-difference = -9, SOLAT-GDMS-difference = -12: the participant was classified neither as analytical, nor as intuitive).

Secondly, the median values for each instrument were calculated (PID-D = 34;

PID-I = 32; GDMS-Rational = 19; GDMS-Intuitive = 16; SOLAT-Left = 9;

SOLAT-Right = 12). Only those participants who scored simultaneously above the analytical median value and under the intuitive median value on every instrument were classified as analytical. Conversely, those participants who scored simultaneously above the intuitive median value and under the analytical median value on every instrument were classified as intuitive.

Lastly, only those participants whose scores met simultaneously both the above-mentioned criteria were selected. Namely, the “extreme” analytical participants were those who, within the analytical group as identified through cluster analysis, were classified as analytical according both the difference and the median computation methods. Conversely, the “extreme” intuitive participants were those who, within the intuitive group as identified through cluster analysis, were classified as intuitive according both the difference and the median computation methods. Following this procedure 13 participants were classified as intuitive people (2 males and 11 females; ranged from 19 to 32 years, mean age = 22.9 yrs), whereas 16 participants as analytical people (2 males and 14 females; ranged from 18 to 48 years, mean age = 27.1 yrs). These people constituted the experimental sample of the Study 5.