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

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

Comparisons to a variable standard<br />

In <strong>the</strong> study originally reporting <strong>the</strong> compatibility effect Nuerk et al. (2001) observed a<br />

compatibility effect with response latencies following incompatible <strong>number</strong> pairs being<br />

reliably longer than latencies for compatible pairs (759 ms vs. 728 ms). Additionally, <strong>the</strong><br />

compatibility effect was reliably moderated by unit distance as it was more pronounced for<br />

large unit distances as compared to small unit distances (52 ms vs. 12 ms; see Figure 2, Panel<br />

A).<br />

Holistic model<br />

ANOVA<br />

On <strong>the</strong> one hand predicting empirical RTs on <strong>the</strong> basis <strong>of</strong> <strong>the</strong> holistic model data<br />

yielded a R 2 <strong>of</strong> .42, which was reliable [r = .65; F(1, 238) = 171.71, p < .001]. However,<br />

closer inspection revealed that important empirical effects cannot be accounted for by <strong>the</strong><br />

holistic model. In <strong>the</strong> item ANOVA for <strong>the</strong> data produced by <strong>the</strong> holistic model <strong>the</strong>re was<br />

only a reliable main effect <strong>of</strong> decade distance [F(1, 232) = 211.04, p < .001]. This indicated<br />

that according to <strong>the</strong> model data <strong>number</strong> pairs with large decade distance were responded to<br />

significantly faster than <strong>number</strong> pairs with a small decade distance (710 ms vs. 778 ms). Yet,<br />

nei<strong>the</strong>r <strong>the</strong> compatibility effect nor its interaction with unit distance were statistically reliable<br />

[main effect: F(1, 232) < 1; interaction: F(1, 232) < 1, see Figure 2, Panel B]. Moreover, no<br />

o<strong>the</strong>r main effect or interaction was found to be significant [all F < 1.59, all p > .21]. Taken<br />

toge<strong>the</strong>r, this indicated that a reliable correlation <strong>of</strong> modelled and empirical RT data <strong>of</strong> its<br />

own is not a sufficient index <strong>of</strong> model validity.<br />

Regression<br />

<strong>The</strong> final model <strong>of</strong> <strong>the</strong> stepwise regression analysis on <strong>the</strong> model data produced by <strong>the</strong><br />

holistic model accounted for a considerable amount <strong>of</strong> variance [adj. R 2 = .84, R = .92, F(4,<br />

252

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

Saved successfully!

Ooh no, something went wrong!