Beyond classification: how preschool children select attributes and construct variables in data modeling

Preschool children are known to notice multiple attributes in the same objects, but far less is known about how they select one of these as an organizing criterion and turn it into a category structure. This study examined which attributes preschool children noticed in an unstructured data set, which they selected as an organizing criterion, how they justified that selection, and how they formed, negotiated, and revised category structures as they applied the selected attribute to the data set.

In this qualitative multiple-case study, two focal groups (n = 4 each), drawn from the 11 children (mean age 5 years 8 months) in a four-week implementation in a Turkish public kindergarten, were each treated as a case. The analysis centered on one activity using museum-themed data cards with no predefined categories. Video recordings, transcripts, observation notes, and children's products were analyzed within and across cases.

The children noticed several attributes in the same data units, including object type, color, and material, but not every noticed attribute was used to organize the data. Within-group negotiation shaped attribute selection, and the groups structured the same data set using different criteria. They also differed in sustaining the chosen criterion: one applied it consistently across the data set, whereas the other revisited it. While building categories, they evaluated object-category fit, justified their decisions, and returned to the criterion when a case did not fit.

The findings suggest that constructing a variable may be less a single moment of selection than an iterative process of noticing attributes, negotiating a criterion, and testing the resulting structure, making pre-representational decision processes in early data modeling visible through the lens of attribute structuring.