The use of intrinsic–extrinsic dichotomy exists ubiquitously in the classic literature of motivation research, both incentives and human needs. In addition to Maslow (1954 Maslow, A. 1954. Motivation and Personality, New York: Harper & Row.  [Google Scholar]), Herzberg (1966 Herzberg, F. 1966. Work and the Nature of Man, Cleveland, OH: World Publishing.  [Google Scholar]), and McGregor (1960 McGregor, D. 1960. The Human Side of Enterprise, New York: McGraw-Hill.  [Google Scholar]), Lawler (1971 Lawler, E. E. 1971. Pay and Organizational Effectiveness: A Psychological View, New York: McGraw-Hill.  [Google Scholar]) developed his extrinsic–intrinsic rewards dichotomy, indicating that rewards extrinsic to individuals reflect lower-order physiological, safety, and social needs whereas intrinsic rewards can satisfy higher-order self-esteem and self-actualization needs. Wilson's (1973 Wilson, J. Q. 1973. Political Organizations, New York: Basic Books.  [Google Scholar]) material/tangible incentives and solidary/intangible incentives were designed to satisfy extrinsic needs and intrinsic needs respectively.

In the original SDT by Dyan and Deci (2000a, 2000b), integrated regulation exists between intrinsic motivation and identified motivation. Integrated motivation represents the most autonomous form of extrinsic motivation. Integration occurs when individuals bring new regulations into congruence with their values and needs, and integrated forms of motivation share some similarities with intrinsic motivation such as an internal locus of control and high autonomy. However, integrated motivation is hard to be differentiated from identified motivation. In providing examples, Vallerand and Ratelle (2004) used some cases of educational psychology. The example of integrated regulation is: “A ballet dancer might choose not to go to a party with friends in order to be in shape for dance class early on the next morning.” The example of identified regulation is: “A boy in high school who decides to get up an hour earlier to review his chemistry notes because he feels it is personally important to do so.” In both cases, the value of activities can be both identified by the actors and integrated into their value systems. As a result, many recent SDT studies do not discuss integrated motivation (Gagné et al., 2010 Gagné, M., Forest, J., Gilbert, M.-H. H., Aubé, C., Morin, E. and Malorni, A. 2010. The Motivation at Work Scale: Validation Evidence in Two Languages. Educational and psychological measurement, 70: 628–46. [Crossref], [Web of Science ®] , [Google Scholar]; Sheldon et al., 2003 Sheldon, K. M., Turban, D. B., Brown, K. G., Barrick, M. R. and Judge, T. A. 2003. “‘Applying Self-Determination Theory to Organizational Research”. In Research in Personnel and Human Resources Management, Edited by: Martocchio, J. J. and Ferris, G. R. Vol. 22, 357–93. Oxford: Elsevier Science Ltd. [Crossref] , [Google Scholar]). The present study follows this line.

Vallerand and Ratelle (2004) provided examples for introjected regulation and external regulation as well. The example of introjected regulation is: “A woman may vote municipal elections because she feels she has to, because it is her duty as a citizen.” The examples of external regulation are “getting money” and “avoiding a parent's reprimands.”

The awarded values (+3, +1.5, −1. −2, −3) are not random, but instead, grounded in theories and previous studies. In two earlier studies using SDI (Green-Demers et al., 1997 Green-Demers, I., Pelletier, L. G. and Ménard, S. 1997. The Impact of Behavioural Difficulty on the Saliency of the Association between Self-Determined Motivation and Environmental Behaviours. Canadian Journal of Behavioural Science, 29: 157–66. [Crossref], [Web of Science ®] , [Google Scholar]; Tremblay et al., 2009 Tremblay, M. A., Blanchard, C. M., Taylor, S., Pelletier, L. G. and Villeneuve, M. 2009. Work Extrinsic and Intrinsic Motivation Scale: Its Value for Organizational Psychology Research. Canadian Journal of Behavioural Science, 41: 213–26. [Crossref], [Web of Science ®] , [Google Scholar]), values assigned to intrinsic motivation, integrated motivation, identified motivation, introjected motivation, external motivation, and amotivation were +3, +2, +1, −1, −2, and −3 respectively. However, integrated motivation in the present study is deemed a part of identified motivation because of their conceptual similarity (please refer to Note 2). As a result, the value awarded to identified motivation in this study is (2+1)/2 = 1.5.

RFI = .87 (suggested cut-off >.90); NFI = .96 (suggested cut-off >.90); FGI = .98 (suggested cut-off >.90); CFI = .93 (suggested cut-off >.90); RMSR = .01 (suggested cut-off  < .08); RMSEA = .09 (suggested cut-off  < .08).

In fact, the value of Cronbach's alpha in most of the previous research of work motivation, such as Wright's (2004, 2007) empirical studies, was below .70. Cortina (1993 Cortina, J. M. 1993. What is Coefficient Alpha? An Examination of Theory and Applications. Journal of Applied Psychology, 78: 98–104. [Crossref], [Web of Science ®] , [Google Scholar]) indicates that the level of alpha is a function of the size of the average correlations among items and can be large despite a wide range of item intercorrelations. For instance, for a 3-item scale with α = .80, the average correlation is .57; for a 10-item scale with α = .80, the average correlation is only .28. In other words, internal consistency estimates are relatively invariant when many items are pooled.

The alpha values for identified motivation, introjected motivation, external motivation, and amotivation are .64, .70, .79, and .61 respectively. There is no alpha value for service motivation as this category has only one item.

Some statisticians treat the sum-up index as an ordinal variable and argue that Spearman's Rho is more suitable than Pearson's correlation. The Spearman's Rho values are .14, .08, −.09, −.11, and −.24 respectively, a pattern in line with the Pearson's correlation.

Some statisticians treat the sum-up index as an ordinal variable and argue that Kruskal–Wallis equality-of-populations rank test is more suitable than t-tests. We conducted Kruskal–Wallis tests, generating results similar to t-tests.