In collaborative learning, the intuition "the more device, the merrier" is somehow widely acknowledged, but little research has investigated the relationship between device-student ratio and the learning outcome. This study aims to investigate not only the main effect of different device-student ratio, also to identify the moderators in the learning context including task complexity, external script availability and students' familiarity to the collaboration settings. A three-round quasi-experiment was conducted in a primary school in mainland China, 130 fifth-grade students from four classes participated. Group worksheet including conceptual understanding and problem-solving tasks were used to collect participants' inquiry performance. Repeated measures ANOVA was employed in data analysis. Findings indicate that 1:m device-student ratio could be beneficial, and external scripts, and prior collaboration experience could moderate such effect. The different effect of 1:m device-student ratio to 1:1 is only significant in the situation when students are faced with relatively simple task, and the effect size is larger when external script is present. When the task is more complicated, such effect of device-student ratio would only emerge after a period of collaboration. This finding challenged the intuition that one-to-one device-student ratio could be better. Related discussions and recommendations to teaching were made.