2.3. Results
2.3.3. Executive function performance
In order to explore whether verbal input from others influenced children’s executive skills, two different levels of certainty were examined (i.e., certainty vs. standard). Thus, performance on executive measures is shown as a function of age and certainty in Table 2.4. As can be seen in Table 2.4, no significant effects of linguistic manipulation were found in either measure. Age did not significantly contribute to differences in performance on either the Day/Night or DCCS task.
Table 2.4
Descriptive statistics for the EF tasks as a Function of Age
Younger 3 Older 3
Measures Standard Certainty Standard Certainty Range Day/Night
No. of correct trials 12.6 (4.28) 10.6 (5.10) 11.8 (4.63) 12.2 (3.93) 2-16 DCCS
No. of correct trials 1.14 (2.19) 1.20 (2.68) 3.25 (2.93) 3.50 (2.85) 0-6
Note. EF: Executive function. Day/Night: n = 5 and n = 9 in 3.0-year-olds, and n = 17 and n = 15 in 3.5-year-olds, for standard and certainty, respectively. DCCS: Dimensional Change Card Sort, n = 7 and n = 5 in 3.0-year-olds, and n = 16 and n = 16 in 3.5-year-olds, for standard and certainty, respectively. Standard deviations are in parentheses.
The Day/Night task
Among 57 children, 11 children were excluded from analysis due to being absent from the session (2 children) or not completing the task (9 children: see the method section). Therefore, 46 children were included in the analyses (22 = standard and 24 = certainty). The distribution of the correct trials was slightly negatively skewed with skewness of -2.67 (see Figure A.2 in the Appendix). Although the data were negatively skewed, the Explore function in SPSS did not show any outliers. Thus, two methods were used to meet normality. First, a square root data transformation was carried out; however, the transformation repeatedly produced outliers and did not meet normality. Hence, the second means was tried. Although the initial data did not show outliers, the lower scores were recoded until the data showed normality (see Tabachnick & Fidell, 1996). The scores under six were all changed to
approximately six (5.65 to 5.95) and then the data were normally distributed. However, a preliminary analysis on the recoded data showed no difference in the results from an analysis with the initial data; thus the original data without any transformation were reported.
The main objective was to explore the effects of certainty on executive skills. In order to explore the effects, a two-way analysis of covariance (ANCOVA) with age and certainty as main effects and an interaction was conducted taking account of verbal skills. As would be expected, the results showed that the covariate, verbal skills, was significantly related to the levels of performance, F(1, 41) = 7.19, p = .011, ηp2 = .15. However, there were
no significant main effects of age, F(1, 41) = .02, p = .88, ηp2 = .001, or certainty, F(1, 41)
= .000, p = 1.00, ηp2 = .002, or the interaction, F(1, 41) = 1.00, p = .32, ηp2 = .02. That is, the
children’s inhibitory control by the Day/Night task was not significantly influenced by age or linguistic expressions.
Across the 16 trials, the correct responses ranged between 2 and 16, and 11 children, who accounted for 23.9%, performed correctly on all the test trials (2 children were from the younger group). Thirty out of 46 children correctly responded in more than 12 out of 16 trials (65.2% pass rate) in total. Compared to Korean 3-year-olds (3.0-years: 14.7 and 3.5-years: 14.9) in Oh and Lewis (2008) (Experiment 2), the participants in this study were less likely to perform the task (3.0-years: 11.3 and 3.5-years: 12.0). However, it seems that Korean
children in this study performed better than English children (3.0-years: 5.63 and 3.5-years: 8.61) in Oh and Lewis. Similarly, it is apparent that the Korean 3-years’ performance showed better levels than those of North American children (approximately 48% pass rate of 3-years) in Carlson (2005). Taken together, although the 3-year-olds in this study performed at a lower level than other Korean children, they tended to perform better than those in Western cultures.
The Dimensional Change Card Sort task
Forty-four children performed on the DCCS task (23 = standard and 21 = certainty), and three children were excluded due to being absent. The distribution of the correct number in the post-switch trials was platykurtic which was driven by binary performance either sorting cards successfully or making errors. The distribution of cards sorted correctly is shown in Figure A.3 in the Appendix. Any methods for data transformation did not meet normality; however, following the procedure of Frye et al. (1995), a univariate analysis was employed. Again, a two-way ANCOVA was conducted to test influences of linguistic
expressions along with age as main effects including an interaction, and vocabulary loaded as a continuous covariate. The results showed that age had a trend towards significant effects,
F(1, 39) = 3.52, p = .068, ηp2 = .08. However, there was no significant main effect of the
certainty expression, F(1, 39) = .004, p = .95, ηp2 = .00. In contrast to the findings in
Day/Night, verbal skills did not explain a variance of DCCS, F(1, 39) = .92, p = .34, ηp2 = .02.
Taken together, as found in the Day/Night task, the certainty expression did not contribute to enhance cognitive flexibility in performing the DCCS task.
According to the criteria of Kirkham et al. (2003) which regarded five out of six trials in the post-switch trials as having passed, 19 out of 44 children passed the task (two 3- year-olds and 17 3.5-year-olds). That is, 17% of 3.0-year-olds and 53% of 3.5-year-olds passed the task (43% pass in total). The performance was comparable to English children (3.0-years: 14% and 3.5-years: 58%) in Oh and Lewis (2008) and North American 3-year- olds (42% pass rate) in Kirkham et al. Taken together, it is apparent that Korean 3-year-olds’ ability to suppress inappropriate responses tended to be better than those in other cultures, whereas their switching ability was comparable to their counterparts.