First, a normality analysis was performed using the Kolmogorov-Smirnov test. The results showed that there was normality of the data distribution.
The results on the effectiveness of biofeedback training show that, from the HRV measurement at the beginning of the training to the final assessment, all participants in the experimental group learned to breathe in a prolonged and slow manner, F(1.97) = 176.26, p = 000; \({\eta }_{p}^{2}\)= 0.372. This learning occurred in all educational cycles F(2, 297)=11.10, p=.000,\({\eta }_{p}^{2}=\).070 but as the interaction shows F(2,297)=21.05; p=.000; \({\eta }_{p}^{2}=0.124\) it was the students in the second cycle (Cycle 2=HRV session 1: M=23.78;HRV session 5: M= 94.99) who showed a greater improvement from the first HRV measurement to the last measurement after the end of the programme (Cycle 1=HRV session 1: M = 26.61;HRV session 5: M= 49.77; Cycle 3=HRV session 1: M=13.16;HRV session 5: M= 54.84).
Performance in the attention test was assessed by scoring one point for each mark made, whether correct or incorrect. In addition, omissions or unmarked stimuli were also scored. Thus, to examine the impact of HRV biofeedback training on the attentional performance of girls and boys in the d2 test, we analysed the hit rate, omissions, task concentration and total test effectiveness (see Table 1). Results were analysed with ANOVAS for mixed designs, 2 (Group: control, experimental), x 2 (Assessment: pre- and post-training), x 3 (Educational cycle: first, second and third) with the variables Group and Educational cycle as independent measures and Assessment as repeated measures. Post-hoc comparisons were performed with the Bonferroni test and pairwise comparisons with Student's t-test.
Table 1
Means and standard deviations in brackets of the experimental and control groups in the pretest and posttest for the 3 cycles of primary education.
| | | Experimental | Control |
|---|
| | | Cycle 1 | Cycle 2 | Cycle 3 | Cycle 1 | Cycle 2 | Cycle 3 |
| | | M(DT) | M(DT) | M(DT) | M(DT) | M(DT) | M(DT) |
| TN-E | pre | 57.13 (26.13) | 96.76(41.13) | 136.79(44.96) | 73(27.00) | 103.14(32.16) | 153.85(33.73) |
| post | 85.32(27.06) | 120.02(38.39) | 165.68(42.41) | 68.87(30.29) | 118.92(41.23) | 154.35(54.39) |
| CON | pre | 43.23(28.50) | 89.56(48.20) | 132.75(47.36) | 42.78(37.28) | 95.92(35.04) | 148.36(45.41) |
| post | 76.57(30.28) | 116.04(39.87) | 163.51(44.75) | 51.72(50.75) | 121.09(37.07) | 163.67(37.75) |
| TO | pre | 243.11(27.89) | 201.90(41.49) | 162.21(44.95) | 225.22(26.66) | 195.86(31.94) | 145.09(33.82) |
| post | 213.22(27.46) | 167.40(56.43) | 127.86(49.23) | 229.97(30.31) | 152.42(69.10) | 116(55.81) |
| TOTR | pre | 57.13(26.13) | 96.76(41.13) | 136.79(44.96) | 73(27.00) | 103.14(32.16) | 153.85(33.73) |
| post | 85.32(27.06) | 120.55(37.64) | 165.68(42.41) | 68.87(30.29) | 123.59(34.42) | 165.78(34.89) |
* Total correctly processed (TN-E), CON concentration (total hits-errors), TO total omissions and TOTR total responses - (omissions + errors).
Total correctly processed
The Total correctly processed (TN-E) measure refers to the number of relevant characters marked correctly. The number of Total Answers was higher in the post evaluation than in the pre evaluation (119.17 vs. 98.27), F(1, 490) = 89.08, p = 0.001, \({\eta }_{p}^{2}\)=0.154. Although the Group factor was not significant, it interacted with the Evaluation factor, F(1, 490)=45.53, p=0.991, \({\eta }_{p}^{2}\)= 0.083. Thus, although in the initial evaluation, the control group obtained more correct scores in the attention test than the experimental group (105.48 vs 94.01), t(552)=-2.97, p = 0.004, d=0.268) after training, in the final evaluation there were no significant differences between the control and experimental groups (113.61 vs 118.62), t(523)=1.86, p=0.236, d=0.18. However, each of the groups improved in the total number of correct scores from the initial assessment to the final assessment, showing a greater impact of training on the experimental group compared to the control group (control t(164)=-3.58, p = 0.000, d=-0.220; experimental t(330)=-15.81, p = 0.000, d=-0.580). In addition, the educational cycle factor was significant, F(2,490)=120, p = 0.000, \({\eta }_{p}^{2}\)=Post hoc comparisons with the Bonferroni test showed that as the age of the students increases, the performance in the number of correct answers increases in the attention test. Thus, students in cycle 1 performed worse (M=71.31) than students in cycle 2 (M=111.40) and cycle 3 (M=152.81). Moreover, students in cycle 3 also outperformed students in cycle 2 (p=0.000). Finally, the interaction Group x Educational Cycle x Assessment was significant, F(2,490)=10.33, p=0.000, \({\eta }_{p}^{2}\)=An important aspect to highlight is the special positive impact shown by the intervention on Cycle 1 in relation to the improvement in the total number of correct scores obtained. On the one hand, and recalling that the experimental group obtained worse scores in total hits than the control group in the pretest, it should be noted that cycle 1 is the only cycle of the experimental group that shows an improvement in the scores in relation to the control group (85.32 vs. 68.87), t(113)=2.86, p = 0.006, d=0.574). Moreover, in cycle 1 the experimental group obtains a statistically significant improvement from the first assessment to the last assessment with a large effect size (t(66)=-8.135, p = 0.000, d=-1.204) higher than the rest of the educational cycles which obtain moderate effect sizes (cycle 2: t(216)=-11.82, p = 0.000, d=-0.610; cycle 3: t(44)=-6.95, p = 0.000, d=-0.689).
Omissions
The Omission measure refers to the total number of relevant items not checked. The number of Omissions was lower in the post evaluation than in the pre evaluation (165.949 vs. 201.002), F(1,521) = 99.89, p = 0.001, \({\eta }_{p}^{2}\)=0.161. The Group x Evaluation interaction was also significant F(1,521)=5.43, p=0.002, \({\eta }_{p}^{2}\)=0.020. Thus, in the initial evaluation the control group committed a lower number of omissions than the experimental group (193.94 vs. 205.02), t(552)=2.96, p = 0.003, d=0.272, while in the final evaluation no statistically significant differences between the control group and the experimental group (162.51 vs. 171.61), t(553)=1.64, p = 0.101, d = 0.143. Although both the control and experimental groups improve performance from the initial assessment to the final assessment, the improvement is more relevant in the experimental group (control: t(179)=6.98, p = 0.000, d = 0.606; experimental: t(346) = 13.28, p = 0.000, d = 0.775). The factor educational cycle was significant, F(2,521)=123.91, p = 0.000, \({\eta }_{p}^{2}\)=Post hoc comparisons with the Bonferroni test showed that as the age of the students increased, performance in the omissions dimension improved and the number of omissions decreased. Thus, students in cycle 1 performed worse (M=227.586) than students in cycle 2 (M=180.1664) and cycle 3 (M=139.203). Moreover, students in cycle 3 also outperformed students in cycle 2 (p=0.000). We observed significant interactions between Assessment time x Educational cycle F(2,521)=9.82, p = 0.000, \({\eta }_{p}^{2}\)=.036 and between Assessment time x Group x Educational cycle F(2,521)=7.94, p = 0.000, \({\eta }_{p}^{2}\)=0.030. Recalling that the experimental group obtained worse scores than the experimental group in the pretest, in the posttest the experimental group in cycle 1 obtained better scores (213.22 vs. 229.97), t(113)=-2.89, p = 0.005, d= -0.580). Moreover, the experimental group of cycle 1 obtains better scores in the posttest than in the pretest with a large effect size (t(66)=7.97, p = 0.000, d=1,230) higher than in the rest of the educational cycles (cycle 2: t(230)=9,916, p = 0.000, d= 0.746; cycle 3: t(46)=6.49, p = 0.000; d=0.753).
Concentration
The Concentration measure is a measure of accuracy calculated by subtracting errors from the total number of hits. Analyses showed that there was a higher Concentration in the final evaluation than in the initial evaluation (115.646 vs. 89.886), F(1,485) = 162.090, p = 0.000, \({\eta }_{p}^{2}\)=0.250. The factor Group was not significant but interacted with the factor Evaluation, F(1,485)=10.04, p=0.000, \({\eta }_{p}^{2}\)=0.028. Thus, it is observed that there were significant differences (p<.05) between the control and experimental groups in the initial evaluation (85.86 vs 93.52) but not in the final evaluation (113.84 vs 111.88). The variable Educational cycle was also significant, F(2,485)=149.89, p = 0.000, \({\eta }_{p}^{2}\)=Post hoc comparisons with the Bonferroni test showed that as the age of the students increases, performance in the concentration dimension increases. Thus, as in the rest of the dimensions analysed, there was a significant improvement (p=0.000) as the participants progressed through the educational cycle (first cycle M=53.63; second cycle M= 08.25; third cycle M=152.10). The interaction Group x Educational level was also significant, F(2,485)=3.56, p= 0.029, \({\eta }_{p}^{2}\)=0.1 and the interaction Evaluation x Group x Educational Level F(2, 485)=5.46, p=0.05,\({\eta }_{p}^{2}\)=0.22. Thus, as with the rest of the dimensions analysed, it can be seen that in the first cycle biofeedback training is more effective than in the rest of the educational cycles. It is also observed that taking into account that the experimental group obtained worse scores than the control group in the pre-test, in the post-test cycle 1 obtains better scores in Concentration than the control group (76.56 vs. 51.71). Also in cycle 1 there is a statistically significant improvement in the Concentration measure in the posttest with respect to the pretest and with a large effect size (t(66)=-9.44, p=0.000, d=-1.213) higher than in the rest of the cycles (cycle 2: t(216)=-11.09, p=0.000, d=-0.621; cycle 3: t(44)=-8.07, p = 0.000, d=-0.741).
Total Test Rate Effectiveness (TOTR)
This measure measures the overall effectiveness of the test by subtracting the sum of omissions and errors from the total number of responses. The TOTR measure was higher in the post than in the pre assessment (120.63 vs. 98.03), F(1, 484) = 150.286, p = 0.000, \({\eta }_{p}^{2}\)=0.237. Although the Group factor was not significant, it interacted with the Evaluation factor, F(1,484)=35.51, p=0.000, \({\eta }_{p}^{2}\)=Thus, although in the initial evaluation the control group scored better in the TOTR variable than the experimental group (105.48 vs 94.01), t(552)=-2.927, p = 0.004, d=0.268) after training, in the final evaluation there were no significant differences between the control and experimental groups (117.67 vs 118.96), t(516)=0.314, p=0.753, d=0.029. However, each of the groups improved their performance in this dimension from the initial assessment to the final assessment, showing a greater impact of training on the experimental group compared to the control group (control t(159)=-6.69, p = 0.000, d=-0.324; experimental t(329)=-16.49, p=.000, d=-0.588). The factor Educational cycle was also significant, F(2,484)=130.83, p = .000, \({\eta }_{p}^{2}\)=0.351. Post hoc comparisons with the Bonferroni test showed that as the age of the participants increases, the performance in the total effectiveness of the test increases. Thus, students in cycle 1 performed worse (M=71.55) than students in cycle 2 (M=111.48) and cycle 3 (M=154.45). Moreover, students in cycle 3 also outperformed students in cycle 2 (p=.000). Finally, the interaction Group x Educational Cycle x Assessment was significant, F(2,484)=10.07, p=0.000, \({\eta }_{p}^{2}\)=0.051. Thus, as with the rest of the variables analysed, it can be seen that biofeedback training is more effective in cycle 1. Remembering that the experimental group also obtained worse scores in the total effectiveness of the test, cycle 1 in the post-test obtained better results than the control group (85.32 vs 68.87), t(113)=2.86, p=0.006, d=0.574). And in relation to the improvement of the total test effectiveness scores of the experimental group in the posttest compared to the pretest, cycle 1 also obtains a very large effect size (t(66)=-8.13, p=0.000, d=-1.204) higher than the rest of the cycles (cycle 2: t(215)=-12.63, p = 0.000, d=-0.631; cycle 3: t(44)=-6.95, p = 0.000, d=-0.689).