Network Analysis of the Vanderbilt Fatigue Scale for Adults

The mathematical tools of network science were used to examine the English version of the Vanderbilt Fatigue Scale for Adults with Hearing Loss (VFS-A) to illustrate one way in which this approach could be used in the speech, language, and hearing sciences.

Nodes represented items in the scale, and connections between nodes represented the correlation in the responses to those items, forming a fully connected, web-like network. Various measures of network structure at the microlevel (i.e., node strength), and at the mesolevel (i.e., communities and subsets of nodes based on node strength) revealed information about individual nodes/items and groups of items, respectively, that could not be observed using more conventional analysis techniques.

Specifically, average node strength distinguished between the pattern of VFS-A responses among listeners with no/slight hearing loss and listeners with mild hearing loss, even though listening-related fatigue, as measured using conventional summed scores, was not statistically different between these groups. This highlights the importance of the relationship between items and not just the item score itself. Although the communities that were detected did not replicate the four subscales in the VFS-A, our analysis of subsets of nodes based on node strength illustrates how the network approach might be used to reduce the number of items in a large scale to a more focused set of nodes/items that might be useful for evaluating treatment effects.

We discuss other ways the network approach could be useful to researchers and clinicians, and can advance the speech, language, and hearing sciences.