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.

Appendix C<br />

On <strong>the</strong> interrelation <strong>of</strong> problem size and distance in <strong>the</strong> stimulus set used<br />

As can be seen from Figure F <strong>the</strong> interrelation between <strong>the</strong> two reliable predictors <strong>of</strong><br />

item RT produced by <strong>the</strong> holistic model (i.e., difference between <strong>the</strong> logarithms <strong>of</strong> <strong>the</strong> two<br />

<strong>number</strong>s and problem size) is curvilinear ra<strong>the</strong>r than linear. However, <strong>the</strong> multiple linear<br />

regression analysis is not able to account for this. Closer inspection <strong>of</strong> <strong>the</strong> figure shows that<br />

<strong>the</strong> curvilinear relation means that up to a problem size <strong>of</strong> about 65 <strong>the</strong> difference between <strong>the</strong><br />

logarithms is positively correlated with problem size. Only for items with a problem size<br />

larger than this <strong>the</strong> correlation becomes negative.<br />

Relation difference <strong>of</strong> logarithms and problem size<br />

0.7<br />

0.6<br />

Curvilinear: y = -0.0004x 2 + 0.04x - 0.70<br />

R 2 = 0.30<br />

Difference <strong>of</strong> logarithms<br />

0.5<br />

Linear: y = -0.004x + 0.52<br />

R 2 = 0.08<br />

0.4<br />

0.3<br />

0.2<br />

0.1<br />

0<br />

20<br />

-0.1<br />

30 40 50 60 70 80 90 100<br />

-0.2<br />

Problem size<br />

Figure F: Simple linear and curvilinear fitting <strong>of</strong> <strong>the</strong> interrelation <strong>of</strong> <strong>the</strong> two reliable<br />

predictors <strong>of</strong> item RT difference between <strong>the</strong> logarithms and problem size<br />

At <strong>the</strong> same time, inspection <strong>of</strong> Figure G reveals that item RT decreases with<br />

increasing problem size up to exactly <strong>the</strong> same point <strong>of</strong> a problem size <strong>of</strong> about 65; <strong>the</strong>reby,<br />

indicating a reversed problem size effect. Only for items with a problem size above 70 a<br />

regular problem size effect <strong>of</strong> item RT increasing toge<strong>the</strong>r with problem size can be observed.<br />

284

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

Saved successfully!

Ooh no, something went wrong!