An open access article that can be dowloaded and read. 👍 https://www.sciencedirect.com/science/article/pii/S2405844026007565
Click on images to enlarge for easy reading
Abstract
In the present study, we aimed to provide a first estimate of the general relationships among math skills through network analysis. Using clinically validated instruments, we examined the performance of a group of 166 typically developing Italian children attending 4th and 5th grade in several math abilities (mental and written calculation, arithmetic facts retrieval, magnitude processing, number transcoding, knowledge of computation procedures, and computation strategies), and domain-general factors (working memory, processing speed, verbal fluency, and visuospatial reasoning). A first network, based on math tasks, indicated that mental calculation is more associated with automatization in retrieving arithmetic facts, while written calculation is more associated with magnitude processing. Both mental and written calculation are strongly related to computation strategies, a central node in the network. A second network indicated that domain-general factors appear peripheral in the network (except for visuospatial reasoning), without direct associations with calculation abilities.
Comments from IQs Corner’s blogmaster:
I love when newer psychometric network analysis methods are applied to cognitive+achievement variables. No one methodology answers all questions, but PNA offers unique advantages that has the potential to improve cognitive-ach intervention research. As I’ve stated elsewhere (McGrew, 2023)
- In the current context, the primary value of these descriptive models is their ability to function as a bridge to theory formation and the ability to hypothesize, and empirically test or statistically simulate, potential causal mechanisms in the network (Borsboom et al. 2021; Haslbeck et al. 2021) (McGrew et al. 2023, p. 5). PNA models can be used to generate causal hypotheses between abilities measured by individual node measures, offering insights regarding the most likely influential targets (or target systems) for intervention (Haslbeck et al. 2021; McGrew et al. 2023).
- Traditional statistical prediction models of achievement, such as multiple regression, provide few clues regarding potential complex causal relations between and among variables. The PNA cognitive-achievement interpretations offered here, although speculative, when informed by the extant substantive research and theoretical literature, have greater potential to elucidate the complex relations between and among CHC cognitive and achievement constructs. The descriptive PNA models (Figures 1 and 2) can be explored with various tools from network science (e.g., exploratory and confirmatory PNA; exploratory stepwise search algorithms to guide the removal or addition of nodes to improve the model; in silico mathematical simulations where changes in network nodes are statistically modified [or constrained] to see how the effect propagates through the entire network and potentially reveals causal mechanisms in the network; etc.) (Epskamp et al. 2017; Haslbeck et al. 2021; Lunansky et al. 2022) (McGrew et al. 2023, p. 6). PNA could assume a pivotal role in improving CHC cognitive-achievement relations SEM modeling research as it acts as a natural interface between correlation and causality . . . [as] the typical attempt to determine directed SEMs from correlation structures in fact appears somewhat haphazard in psychology, a historical accident in a field that has been prematurely directed to hypothesis testing at the expense of systematic exploration (Epskamp et al. 2017, pp. 924). PNA methods could facilitate CHC SEM modeling via the systematic identification of relations between multiple variables unfettered by concerns for direct causal relations, reciprocal causation, latent common causes, semantic overlap between items [variables], or homeostatic coupling of parameters (Epskamp et al. 2017, p. 925).
