Convergent vs. Discriminant Validity: Which Is More Important?

If I had to prioritize one form of validity in structural equation modeling, I would choose convergent validity, because without it, the factors themselves are not meaningful enough to evaluate discriminant validity in the first place. Convergent validity ensures that the items meant to measure the same construct are actually related and share a substantial portion of variance. For example, in my thesis, I measured “academic motivation” using items related to effort, persistence, and interest. If these items do not converge well—meaning low factor loadings or an AVE below .50—then the construct itself becomes unstable. In that case, it would not matter how distinct it is from other constructs such as academic stress or self-efficacy, because the underlying factor would be poorly defined.