Corrigendum to “Walk the number line – An embodied training of numerical concepts” [Trends Neurosci. Educ. 2/2 (2013) 74–84]

The authors regret: In the published version of this article, we did not centralize the control variables (e.g., general cognitive ability, working memory capacity, etc.) before evaluating the differential training effects in an ANCOVA incorporating these variables as covariates where necessary. This led to partially misleading results. A subsequent reanalysis of the data by ANCOVAs using centralized values of the covariates indeed led to changes in the results. These will be described in the following with direct reference to the respective section of the article. 3.1 Specific training effect on children's spatial numerical representation Differential training effects – raw estimation improvements: An ANCOVA incorporating general cognitive ability and verbal working memory capacity as covariates did not yield a significant main effect of training condition, F(1, 30)1⁄40.41, p1⁄4 .53. Differential training effects – standardized estimation improvements: An ANCOVA considering verbal working memory capacity as covariate did not indicate a significant main effect of training condition, F(1, 31)1⁄40.05, p1⁄4 .82. 3.2 Transfer effects to addition performance 3.3.2 Carry problems Differential training effects – efficiency scores: An ANCOVA considering general cognitive ability as a covariate revealed a marginally significant main effect of training condition after centralizing the covariates, F(1, 12)1⁄43.25, p1⁄4 .097, d1⁄40.74. This result indicates a tendency towards a more pronounced training gain after the embodied compared to the control training. Differential training effects – error rates: An ANCOVA incorporating writing speed as a covariate did not indicate a significant main effect of training condition, F(1, 12)o1, p1⁄4 .93. In the following we also report statistical details which changed for the other evaluation tasks. Please note that in these cases reanalyses did not lead to changes of the interpretation of the results in the original article. 3.3 Training effects for the other evaluation tasks 3.3.1 Symbolic number comparison Distance effect – z-transformed RT: F(1, 31)1⁄43.26, p1⁄4 .08, d1⁄40.50 3.3.3 Place-value understanding Tedi math subtest 8 – PR: F(1, 31)1⁄42.00, p1⁄4 .17 Furthermore, we erroneously reported arc-sin transformed values in Table 2 depicting error rates for carry addition problems. Correct values are embodied: Mean1⁄4 1%, sd1⁄423; control: Mean1⁄4 5%, sd1⁄425. Considering the results of the reanalysis we have to amend our interpretations. The results of the reanalysis did not substantiate a reliable differential training effect regarding children's raw and standardized improvements in number line estimation. This indicates that both trainings – embodied and control – had reliable but comparable beneficial effects on children's number line estimation performance. Nevertheless, as reported following the original analyses, the interaction with the covariate general cognitive ability suggests that the embodied training may be specifically suited for children with lower general cognitive abilities. Moreover, regarding the transfer effects to performance in mental addition, there are hardly any amendments to our interpretation: a tendency towards a stronger improvement of a subgroup of children regarding carry addition problems after the embodied compared to the control training was still present after reanalyzing the data. As there were no further changes, especially with regard to simple addition tasks, we are confident that our original interpretation is still valid. The authors would like to apologize for any inconvenience caused.