By the same authors

From the same journal

From the same journal

Breadth versus depth: cumulative risk model and continuous measure prediction of poor language and reading outcomes at 12

Research output: Contribution to journalArticlepeer-review

Author(s)

Department/unit(s)

Publication details

JournalDevelopmental Science
DateAccepted/In press - 2 May 2020
DateE-pub ahead of print (current) - 22 Jun 2020
Number of pages13
Early online date22/06/20
Original languageEnglish

Abstract

This study examines whether, and how, multiple risks in early childhood are associated with an increased likelihood of a poor language or literacy outcome in early adolescence. Using data from 210 participants in the longitudinal Twins Early Developmental Study, we focus on the following risk factors at age four: family risk, and poor language, speech, emergent literacy and nonverbal skills. The outcomes of interest at age 12 are language, reading fluency, and reading comprehension. We contrast a ‘cumulative risk’ model, counting the presence or absence of each risk factor (breadth), with a model that also considers the severity of the early deficits (depth). A ‘cumulative risk index’ correlated modestly but significantly with outcome (r = .32-.40). Odds ratios confirmed that having many risk factors (3-6) confers a higher probability of a poor outcome (OR 7.86-17.71) than having one or two (OR 3.65-7.28). Logistic regression models showed that predictive validity is not improved by including information about the severity of each deficit. Even with rich information on children’s risk status at age 4, we can make only a moderately accurate prediction of the likelihood of a language or literacy disorder eight years later (Area Under the Curve = .74-.84; Positive Predictive Value = .33-.55, Negative Predictive Value = .86-.91). Taken together, and consistent with the idea of ‘cumulative risk’, these results suggest that breadth of risk is a core predictor of outcome, and furthermore that severity of early deficits does not add significantly to this prediction.

Bibliographical note

© 2020 The Authors

Discover related content

Find related publications, people, projects, datasets and more using interactive charts.

View graph of relations