A valid construct measure must balance breadth with unidimensionality

Evidence

Assuming we have a good conceptualization that can plausibly cover the field, it has to be objectively and validly measured. If the assessment is done through a psychological test, that test needs to be broad enough to cover the range of beliefs and attitudes pertinent to the underlying construct, yet cohesive enough to be essentially unidimensional. The farther we fall short on either score, the more grief we shall face further down the road. — Bob Altemeyer, The Authoritarian Specter, 634