Purpose: This study aims to examine the impact of the informational value of feedback choices (confirmatory versus critical feedback) on students' performance, their choice to revise and the time they spend designing posters and reading feedback in a computer-based assessment game, Posterlet. Design/methodology/approach: An empirical correlational study was conducted to collect the choices to seek confirmatory or critical feedback and to revise posters in a poster design task from 106 grade 8 students from a middle school in California via Posterlet. Findings: The results of the study show that critical uninformative feedback is associated with students' performance, and critical informative feedback is associated with their learning strategies (i.e. feedback dwell time and willingness to revise), while confirmatory informative feedback is negatively associated with both performance and learning strategies. Research limitations/implications: The study controlled the choice students were given regarding the valence of their feedback but not regarding the informational value of their feedback. Additionally, the study was conducted with middle-school students, and more research is needed to ascertain whether the results generalize to other populations. Practical implications: The findings can be used to balance the design of the informational content of feedback messages to support student performance in an open-ended, creative design task. This study may also inform the design and implementation of agents (e.g. virtual characters) able to provide user-adaptive feedback for online interactive learning environments. Originality/value: This study constitutes the first research to examine the informational value of feedback that is chosen rather than received, the latter being the prevalent model of delivering feedback in education.