Thursday 12 December 2013

ISIR Worst Performance Rule


It is hardly in the popular “How to boost your IQ” genre to title a symposium the “Worst Performance Rule” but that gives you an indication of the media savviness of us intelligence researchers. It would have been better to entitle it: “How to boost your already high performance by avoiding a few silly errors”. Any, too late to talk the fearsome Dodonovas into a proper PR campaign.;

This is what they have to say about their symposium:

To obtain a reliable measure of the speed of information processing, elementary
cognitive tasks are designed to include multiply repeated trials, for which
response times are recorded. This produces a set of individual-level RTs, which
do not necessarily equally well predict cognitive ability. A frequently observed effect is
that when individual-level RTs are ordered from fastest to slowest, the slowest RTs
produce higher correlations with cognitive ability than the fastest RTs. This effect is
called the worst performance rule (WPR). An explanation for WPR suggested by
previous studies has implied that individuals differ in efficiency of memory and
attention, or in some basic characteristics of neuron functioning, and this is what matters when both ECTs and intelligence-like tests are performed. However, previous studies on WPR have been too rare to accumulate consistent evidence of its causes,
confounding the variables and the implications for an analysis of the association
between intelligence and the speed of processing of various ECTs.
Taken together, the three papers of this symposium reconsider the issue of different
RT-IQ associations reported for best- and worst-performance trials and discuss their
possible causes and implications for studies examining the association between
cognitive ability and the speed of information processing.

In the first paper, Yulia Dodonova analyzes a set of elementary cognitive tasks and questions whether WPR is indeed present in these tasks and whether it can be explained via the effect of confounding variables and statistical artifacts.

In the second paper, Natalie Borter discusses the implications of WPR when
analyzing individual performance in tasks of varying complexity. Such tasks imply that individual differences in the speed of basic constant processes and task-specific
experimentally-induced processes can be analyzed at the latent level, and their
associations with cognitive ability can be evaluated. However, as shown in this study,
averaging across best and worst trials within each complexity level can mask another
source of variance, which can also be meaningful and can provide additional insights
into the associations between the speed of task processing and intelligence.

Finally, a paper by Yury Dodonov analyzes WPR-like effects in the context of
speed-accuracy relations that are always present when a participant performs speeded
tasks. This study suggests that accuracy rate is another factor that must be considered in any WPR-type analysis of associations between individual-level response times and intelligence.

And then lunch.

No comments:

Post a Comment