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.

<strong>the</strong> middle <strong>the</strong> beneficial effect <strong>of</strong> a central <strong>number</strong> smaller than <strong>the</strong> arithmetical middle was<br />

more pronounced than for close distances (930 ms vs. 203 ms). Finally, <strong>the</strong> three participant<br />

groups did not differ reliably in terms <strong>of</strong> RT [F(2, 15) < 1].<br />

Errors: In <strong>the</strong> error analysis a main effect <strong>of</strong> group was observed [F(2,15) = 6.83, p <<br />

.01]. Bonferroni-Holm corrected t-tests showed that neglect patients committed reliably more<br />

errors than control patients [+5.0 %, t(10) = 3.41, p < .01, one-sided] as well as healthy<br />

controls [+5.0 %, t(10) = 3.12, p < .01, one-sided]. Additionally, <strong>the</strong> interaction <strong>of</strong> distance to<br />

<strong>the</strong> middle and size relative to <strong>the</strong> middle was significant [F(1, 15) = 5.23, p < 05.]. This<br />

indicates that for triplets with a central <strong>number</strong> numerically close to <strong>the</strong> arithmetical middle<br />

rejection was more error prone when than when <strong>the</strong> central <strong>number</strong> was smaller than <strong>the</strong><br />

middle compared to when it was larger than <strong>the</strong> middle (6.4 % vs. 3.9 % errors, respectively).<br />

However, this pattern was reversed for triplets with a central <strong>number</strong> far from <strong>the</strong> middle<br />

(smaller: 1.9 % vs. larger: 6.1 %). <strong>The</strong>re were no fur<strong>the</strong>r significant main effects <strong>of</strong> <strong>the</strong><br />

manipulated item characteristics [all F < 1.16, all p > .30 or interactions <strong>of</strong> group with any <strong>of</strong><br />

<strong>the</strong>m [all F < 1.11, all p > 13].<br />

Bisected triplets (RT and error data):<br />

Decade crossing: In line with <strong>the</strong> hypo<strong>the</strong>sis, <strong>the</strong> interaction <strong>of</strong> decade crossing and<br />

participant group was reliable for errors [F(2, 15) = 6.53, p < .01], but not for response<br />

latencies [F(2, 15) < 1]. Bonferroni-Holm corrected t-tests were conducted to directly test <strong>the</strong><br />

hypo<strong>the</strong>sis <strong>of</strong> neglect patients’ performance being impaired most for triplets crossing into <strong>the</strong><br />

next decade. <strong>The</strong>refore, <strong>the</strong> absolute effects <strong>of</strong> decade crossing were contrasted for neglect<br />

patients vs. patient controls [t(10) = 0.90, p = .20, one-sided] as well as for neglect patients vs.<br />

healthy controls [t(10) = 2.82, p < .05, one-sided b ]. This indicates that <strong>the</strong> increase <strong>of</strong> errors<br />

b Note: Again, similar results were obtained when directly contrasting <strong>the</strong> effects <strong>of</strong> decade crossing in an oneway<br />

ANOVA with subsequent post-hoc Dunnett t-tests: neglect vs. patient control p = .37, neglect vs. healthy<br />

control p < .01.<br />

227

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

Saved successfully!

Ooh no, something went wrong!