Two selection hypothesis testing tasks and nine evaluation hypothesis testing tasks which varied systematically with respect to causality, response alternatives, and context continuity were administered to two samples of adults to determine the effects of those variables and to determine the degree to which subjects (Ss) reasoned with material conditional, material biconditional, partial, or partial plus biconditional connectives. On the causal selection task the majority of Ss responded “yes” to all four instances, while on the evaluation tasks all three variables were found to affect performance. Ss responded consistently with the material biconditional connective especially when the tasks called for them to state their expectations or were written with context continuity. Results were interpreted to support the hypothesis that hypothesis testing reasoning in causal contexts begins with the material biconditional connective and involves schemata of controlling variables, probability, and correlations. A strictly “logical” falsification strategy is not viewed as operative. Implications for teaching students how to effectively test hypotheses in science classes are discussed.