Affect and behavior in preschoolers during a tablet-based assessment: insights from observational coding of BELLA use

<jats:p>&lt;p&gt;&lt;strong&gt;Background. &lt;/strong&gt;Despite widespread tablet use in early childhood education, children's affective and behavioral experiences with digital assessments remain understudied. This study addresses gaps in understanding how preschoolers emotionally and behaviorally respond to a tablet-based learning tool.&lt;strong&gt; Methods. &lt;/strong&gt;We employed a mixed-method approach analyzing observational data from 161 children (ages 3-6) across three Houston preschools using the Bilingual English Language Learner Assessment (BELLA). Behavioral coding adapted from SPAFF and DPICS frameworks examined affect, engagement, digital skill, and social dynamics across sessions. Statistical analyses included descriptive analysis, McNemar tests for session comparisons, and multiple linear regression.&lt;strong&gt; Results. &lt;/strong&gt;During first exposure, engaged affect was most prominent (40.6%), with 51.6% demonstrating good digital skills. Between sessions, engaged affect significantly increased (&amp;chi;&amp;sup2; = 10.42, p &amp;lt; .01) while detached affect decreased (&amp;chi;&amp;sup2; = 4.02, p &amp;lt;.05). Digital proficiency improved across all levels. Multiple linear regression (R&amp;sup2; = .497, p &amp;lt; .001) revealed positive affect and digital skill as significant performance predictors, with age as the strongest covariate (&amp;beta; = .357, p &amp;lt; .001).&lt;strong&gt; Discussion. &lt;/strong&gt;Digital assessments capture both cognitive competency and real-time indicators of self-regulation and engagement. Findings highlight digital literacy's critical importance in early childhood education and demonstrate that emotional engagement significantly contributes to learning outcomes in technology-mediated environments.&lt;strong&gt; Limitations/Future Directions. &lt;/strong&gt;The study was limited by observational data and suburban private school sampling. Future research should incorporate structured protocols, video coding, time-stamped data collection, and more diverse samples to enhance generalizability and capture dynamic learning processes.&lt;/p&gt;</jats:p>