When estimating the effect of side-quest engagement on in-game purchases, begin with the treatment, outcome, and other variables central to the causal question. A set is causally sufficient when it contains every common cause of each pair of variables in that set. This supports a DAG whose stated causal relationships do not omit shared ancestors relevant to the included variables. section 3.3.1 section 11.2
Build a causally sufficient set by expanding the variables needed for the causal question:
- Start with the treatment, outcome, and any other variables needed to formulate the intervention or explanation. section 3.3.1
- Trace causal lineages backward and add shared ancestors that cause pairs of included variables. section 3.3.1
- After including a common cause, omit earlier common causes on the same paths when the included variable already intercepts those paths. section 3.3.1
- Add instruments, effect modifiers, mediators, or communication variables only when the causal analysis or explanation benefits from them; they are not required for causal sufficiency. section 3.3.1 section 11.2
Online gaming causal DAG
For the effect of Side-Quest Engagement on In-Game Purchases, include Guild Membership, Prior Experience, Time Spent Playing, and Player Skill Level because they are common causes of variables in the analysis. Add Side-Quest Group Assignment and Customization Level as potentially useful instruments. Treat Total Inventory cautiously: it is a collider between In-Game Purchases and Won Items and can introduce bias if used as a predictor for causal effect estimation. section 11.2
