Choose Your Own Conjecture: An Analysis of Social Stories™ Text

Background:  Social Stories™ strategies are used in schools, often to address the behaviors of children with a clinical autism diagnosis. Each intervention package comprises materials that are written as a short story for a specific person and procedures that are chosen for convenient implementation in a specific place. These features have widespread appeal but intervention studies have varied results; this discrepancy may reflect the confounding influence of Social Stories™ text variables.

Objectives:  To characterize the full text of Social Stories™ research interventions using conceptual guidelines for storybook development and conventional measures of text difficulty.

Methods:  Intervention studies published between 1993 and 2009 were identified by comprehensive pearl growing. Social Stories™ passages with full text were transcribed and summarized in terms used by intervention developers (e.g., perspective sentences, first-person voice) and those used generally to measure readability (e.g., Flesch-Kincaid grade level), decodability (e.g., percentage of high-frequency words), and other aspects of text difficulty (e.g., mean sentence length).

Results:  Social Stories™ were used in 38 intervention studies including 26 studies that provided the full text of 48/48 passages, 7 studies that provided the full text of 8/38 passages, and 5 studies that provided the full text of 0/15 passages. Written for more children than adolescents, this corpus has questionable readability (mean Flesch-Kincaid grade level: 6.6; median: 6.0; range: 1.2-9.4). Only 25/56 passages contained the low proportion of directive sentences specified as a guideline; however, the high frequency of flexible language (e.g., try, sometimes) matches another guideline for Social Stories™ construction.

Conclusions:  The characteristics of some Social Stories™ are varied. Research reports should provide the full text of intervention materials and research reviews should consider the relation of efficacy, text attributes, and other Social Stories™ variables.