Poster #28 - Evaluating Stability of Third Grader Responses to a Recently-Developed Semantic Organization Task

• In Event: 1-005 - Poster Session 01
  In Poster Session: PS 01 Section - Cognitive Processes

Integrative Statement

The Spatial Arrangement Method (SpAM) is a recently developed measure of semantic organization that can be used with young children. The task entails arranging (either two or many) images on a 10x10 grid such that things that “go together” or are “the same kind of thing” are placed closer together, while things that “don’t go together” or aren’t “the same kind of thing” are place farther apart. The distances between images placed by subjects are interpreted as a proxy for the individuals’ semantic distance between these two concepts. SpAM is particularly useful for collecting a large number of semantic comparisons very quickly (when there are many items in a single trial), and for gaining insight into group differences in semantic organization (e.g., Unger et al., 2016). However, as for any newly-developed cognitive measure, this task must be evaluated for reliability and validity across subjects and age ranges to understand how data generated in this way may be appropriately used and interpreted. In the current study, we assessed the consistency 50 8- to 10-year-old children’s responses on the SpAM task across two time points, with two different cross-classifiable sets of images (19 of foods and 20 of living things). First, we evaluated group consistency by comparing average distances across subjects for each unique pair of items at the two time points. Both image sets’ pairwise distances were moderately correlated (rlivingthings = 0.479, rfood = 0.522). We also compared the group’s average distances for different types of image pairs (thematically and taxonomically related, taxonomically related, thematically related or unrelated) and found them to be very consistent between the two time points (see Figure 1). Second, stability in individual responses was evaluated. Individual subjects’ pairwise distances between the time points were uncorrelated (r ̅livingthings = 0.08, r ̅food= 0.09), and the distribution of similarities between distance matrices generated from subjects’ matched responses at times 1 and 2 were comparable to those between distance matrices generated from two random subjects’ responses (pliving things = 0.217, pfood = 0.223). Children’s taxonomic scores at time 1 were weakly correlated with their taxonomic scores at time 2 (rtax,livingthings = 0.312, rtax,food = 0.3488; see Figure 2). These findings suggest that the Spatial Arrangement Method, when many items are arranged on a single trial, yields consistent results at a group level but less so at the individual level.

Authors

• Patience Stevens, Carnegie Mellon University
  Presenting Author

• Catarina Vales, Carnegie Mellon University
  Non-Presenting Author

• Anna Fisher, Carnegie Mellon University
  Non-Presenting Author