includes is the right default for primitives: it handles NaN correctly and reads as a yes/no question. indexOf !== -1 is the older idiom and fails the NaN case because it uses strict equality. For a hot loop or membership check inside a render path, build a Set once with new Set(values) and call .has(x) per lookup; that turns repeated linear scans into average O(1). The biggest gotcha is OBJECT membership: [{id:1}].includes({id:1}) is false because the two object literals are distinct references. For "find by field value", reach for .some(predicate) or .find(predicate) instead.

For a flat object, Object.values(obj).includes(x) is the one-liner answer. For nested data (object inside object inside array), the recursive walker above checks the current node, recurses into every value (which works for arrays since Object.values(['a', 'b']) returns the items), and short-circuits on the first match. The walker uses strict equality at the leaf, so it finds primitives and the same-reference object but not a structurally equivalent copy; for that use js-object-deep-equal instead. Add a seen Set of visited nodes if your input might have cycles, otherwise the recursion will overflow.