27.12.2013 Views

The influence of the place-value structure of the Arabic number ...

The influence of the place-value structure of the Arabic number ...

The influence of the place-value structure of the Arabic number ...

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

epresented separately also in <strong>the</strong> internal representation <strong>of</strong> two-digit <strong>number</strong>s as <strong>the</strong> authors<br />

observed a reliable compatibility effect for comparisons <strong>of</strong> a given <strong>number</strong> to an internally<br />

memorized standard (RT for compatible comparisons: 492 ms vs. incompatible comparisons:<br />

504 ms, see Figure 3, Panel A). Fur<strong>the</strong>rmore, this compatibility effect was found to be<br />

particularly driven by <strong>the</strong> compatibility effect for comparisons with a large unit distance (+ 16<br />

ms) while it could not be obtained for comparisons with a small unit distance (+ 8 ms).<br />

Holistic Model<br />

ANCOVA<br />

A regression analysis revealed that modelled RT data accounted for only 10% <strong>of</strong> <strong>the</strong><br />

variance in <strong>the</strong> empirical RT data [r = .32; F(2, 86) = 9.73, p < .01]. Fur<strong>the</strong>rmore, analyzing<br />

<strong>the</strong> RT estimates produced by <strong>the</strong> holistic model showed that nei<strong>the</strong>r <strong>the</strong> main effect <strong>of</strong><br />

compatibility nor its interaction with unit distance was statistically reliable [all F < 1].<br />

Additionally, separate analyses for comparisons for large and small unit distances revealed<br />

that <strong>the</strong> compatibility effect was not significant in both analyses [both F < 1] indicating no<br />

modulation <strong>of</strong> <strong>the</strong> compatibility effect by unit distance (see Figure 3, Panel B). Never<strong>the</strong>less,<br />

<strong>the</strong> <strong>influence</strong> <strong>of</strong> <strong>the</strong> covariate problem size was significant in all analyses [two-way ANOVA:<br />

F(1, 39) = 14.53, p < .001; large unit distance: F(1, 9) = 6.22, p < .05; small unit distance:<br />

F(1, 29) = 9.16, p < .01]. Thus, as has been <strong>the</strong> case for comparisons to a variable standard<br />

(see above) <strong>the</strong> holistic model could not account for <strong>the</strong> unit-decade compatibility effect.<br />

Regression<br />

<strong>The</strong> final regression model for <strong>the</strong> data produced by <strong>the</strong> holistic model accounted for a<br />

significant amount <strong>of</strong> variance [adj. R 2 = .31, R = .59, F(2, 37) = 9.67, p < .001] and included<br />

<strong>the</strong> predictors absolute problem size and difference between <strong>the</strong> logarithmic magnitudes <strong>of</strong> <strong>the</strong><br />

two to-be-compared <strong>number</strong>s (see Table 4). Evaluation <strong>of</strong> <strong>the</strong> beta weights revealed that<br />

260

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!