In a formal model, a variable whose value is determined by a specified manipulation or combination of independent variables and/or other intervening variables, without any hypotheses about the existence of unobserved entities or processes, and that plays a part in explaining the value of a dependent variable, although it has no factual content apart from the empirical relationships that it summarizes. A typical example is the concept of broad heritability (h2), which is defined as the ratio of genetic variance VG to total phenotypic variance VP, so that, formally, h2 (broad) = VG/VP. Another example is that of the therapeutic ratio of a drug, defined as the ratio of the median lethal dosage (LD-50) to the median effective dosage (ED-50). Hullian learning theory has scores of intervening variables. The term was introduced in 1938 by the US psychologist Edward C(hace) Tolman (1886–1959), and the distinction between hypothetical constructs and intervening variables was first made explicit in 1948 by the US psychologists Kenneth MacCorquodale (1919–85) and Paul Everett Meehl (1920–2003)—see hypothetical construct.