In an earlier post I compared the Deci-Ryan model and the theory of 16 basic desires on the scientific criterion of construct validity. I observed that mind-body dualism in theology, and the Deci-Ryan model of intrinsic-extrinsic motivation, divide motives into two similar groups, so that hunger is a motive of the body (Hull drive, extrinsic motivation), while curiosity is a motive of the mind (an intrinsic motive. ) When Deci realized this was an invalid model of motivation, he backed off and set on a course of multiple revisions to define the differences between the two kinds of motivation. In my previous blog I suggested that the Deci-Ryan group have failed to provide a satisfactory definition of intrinsic-extrinsic motivation -- their work fails the scientific criterion of construct validity -- because humans inherit a wide range of genetically diverse motives that do not divide into just two kinds. The research reported by my colleagues and I, for example, have validated scientifically (confirmatory factor analysis, measurement reliability, criterion validity, and concurrent validity) 16 different kinds of intrinsic motives.
Now I take up another scientific criterion, measurement reliability. Social psychologists have put forth two kinds of measures, questionnaires (self-report) and behavioral (observations of free choice behavior). They have reported reliability data for the self-report measures of intrinsic motivation. They have not shown the reliability of free choice measures.
The issue is crucial because the so-called paradoxical effects of reward do not occur on self-report measures. If you give an adolescent a reward for working on a puzzle, after being rewarded the adolescent says he likes solving puzzles, much as he had said before the reward. Deci-Ryan enthusiasts say, however, that despite this self-perception of intrinsic liking, post-reward the adolescent spends less time working on puzzles than prior to reward. We know that the self-report measure of liking is reliable, but we don't know if the behavioral free choice measure is.
In the 1970s Mark Lepper and his colleagues executed a now-famous study in which it was claimed that a "Good Player award" undermined the intrinsic interest of nursery school children in drawing with "magic marker" crayons. The measure in this study was how much time nursery school kids spent drawing. If the nursery school was anything like the ones I have visited, there were scores of little children running in a seeminly aimless manner around a large room filled with all sorts of toys and adults going in and out. How reliable is the time they spend with any one toy from day to day?
Richard Ryan, who seems to me like an impressive and sincere person, once asked me how I could be so critical of intrinsic-extrinsic motivation when it is published it top quality journals. Good question. How did the Journal of Personality and Social Psychology publish this work without asking for the reliability of the measures? Or construct validity?
In contrast, the theory of 16 basic desires is based on a measure I developed called the Reiss Motivation Profile. As summarized in Chapter 2 of my book "Normal Personality," and in peer reviewed journals, the reliabilitiies are comparable or a bit better than those reported for widely used personality tests such as MMPI. Today, many thousands of people find the RMP a meaningful tool for self-discovering their motives, how their motives connect to their traits, values, and relationships. (See my post 3 Pillars of Motivational Psychology).
In conclusion, the Deci-Ryan model is based on extensive empirical research and scholarship. Yet when we ask how the evidence stacks up against key scientific criteria, we find it is far weaker than many realize. The Deci-Ryan model is construct invalid, which is why it has been changed so many times since its early days as applied Bem self-perception theory. We also find that the behavioral measures of intrinsic motivation used by social psychologists testing the Deci-Ryan model may be unreliable.
Next Up: Validity data.