Historically, effort has been viewed as aversive. Most supporting evidence comes from studies demonstrating increased force/effort requirements reduce operant responding. Changes in force/effort requirements, however, are often accompanied by changes in response definition when mechanical devices are used to define the response. As a consequence, responses measured at one point in a study may go unmeasured at other points. In an alternative approach, we used a continuous measurement strategy that provided a means to fix the threshold force defining the response class and simultaneously allowed independent manipulation of the force criteria required to produce reinforcement. Rats pressed a force transducer according to a fixed‐ratio 5 schedule of food delivery. The criterion force was systematically increased and decreased; the threshold for response detection was constant. When response rates included only criterion responses, overall rate decreased when force requirements increased. By contrast, when all responses, both those meeting force criteria and those that did not (above the threshold but below the criteria for reinforcement) were included in the rate calculation, increases in force increased response rate. Increases in force criteria also increased the maximum force (g) and time‐integral of force (g‐s) of operant behavior. Control conditions showed increases in responding could be explained by the emergence of subcriterion responses, irrespective of force. We conclude that prior results showing effort decreases response rates are due to an artifact arising from inadvertent changes in response definitions. Increases in effort may better be understood as changes in the response:reinforcer payoff owing to the emergence of a subcriterion response class.