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.

adjustment <strong>of</strong> <strong>the</strong> degrees <strong>of</strong> freedom. To fur<strong>the</strong>r investigate in what way a carry operation<br />

<strong>influence</strong>d <strong>the</strong> <strong>number</strong> <strong>of</strong> regressions (i.e., leftward saccades to a previous interest area)<br />

initiated from ei<strong>the</strong>r tens or units <strong>of</strong> both <strong>the</strong> second summand as well as <strong>the</strong> result a 2 x 2<br />

ANOVA was conducted discerning <strong>the</strong> factors digit identity (decade vs. unit digit) and carry<br />

(required vs. not required). Finally, despite <strong>the</strong> origin <strong>of</strong> <strong>the</strong> regressions also <strong>the</strong> target <strong>of</strong> <strong>the</strong><br />

regressions was appraised. For regressions from <strong>the</strong> second summand to <strong>the</strong> first summand a 2<br />

x 2 ANOVA with <strong>the</strong> factors digit identity (decade vs. unit digit) and carry (required vs. not<br />

required) was conducted while regressions originating from <strong>the</strong> result were analyzed by a 2 x<br />

2 x 2 ANOVA involving <strong>the</strong> factors problem element (first vs. second summand), digit<br />

identity (decade vs. unit digit), and carry (required vs. not required).<br />

RESULTS<br />

Participants who performed at or below chance level in at least one <strong>of</strong> <strong>the</strong><br />

experimental conditions where excluded from fur<strong>the</strong>r analyses. This affected one participant.<br />

For <strong>the</strong> o<strong>the</strong>r participants error rates ranged from 1.0 % to 13.3 % with <strong>the</strong> mean at 7.0 % and<br />

a standard deviation <strong>of</strong> 3.4 %. Only response latencies followed by a correct classification<br />

were considered for <strong>the</strong> RT analysis. Additionally, a trimming procedure eliminated all<br />

latencies shorter than 200 ms and longer than 5000 ms in a first step. Subsequently, all RTs<br />

falling below or above three standard deviations <strong>of</strong> an individual participant’s mean were<br />

excluded from <strong>the</strong> analyses in a second step. For <strong>the</strong> analyses <strong>of</strong> <strong>the</strong> eye fixation data only<br />

data from trials already included in <strong>the</strong> RT analysis were considered.<br />

Classification performance<br />

ANOVA: <strong>The</strong> ANOVA showed that both response latencies as well as error rates were<br />

reliably <strong>influence</strong>d by problem size [RT: F(1, 18) = 86.26, p < .001; errors: F(1, 18) = 16.09,<br />

p < .001, see Figure 1, Panel A]: addition problems with a relatively larger problem size<br />

70

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

Saved successfully!

Ooh no, something went wrong!