Skip to main content
SpringerLink
Log in

Findings

  • Chapter
  • First Online:
Design of Incentive Systems

Part of the book series: Contributions to Management Science ((MANAGEMENT SC.))

  • 2048 Accesses

Abstract

The findings are presented according to the three objectives formulated in the introduction. First incentive and sorting effects are compared (Cf. Sect. 6.1), then, separately from sorting effects, incentive effects are considered in order to specifically observe influences of contingency factors on productivity levels (Cf. Sect. 6.2), and finally, single determinants of the decisions for the available contracts leading to sorting effects are analyzed (Cf. Sect. 6.3).

Our senses don’t deceive us, our judgment does.

Johann Wolfgang von Goethe (1749 Frankfurt am Main – 1832 Weimar)

Cited and translated according to von Goethe (1986), p. 408: “Die Sinne trügen nicht, das Urteil trügt.”

This is a preview of subscription content, log in via an institution to check access.

Access this chapter

Log in via an institution
Chapter
USD 29.95
Price excludes VAT (USA)
  • Available as PDF
  • Read on any device
  • Instant download
  • Own it forever
eBook
USD 84.99
Price excludes VAT (USA)
  • Available as EPUB and PDF
  • Read on any device
  • Instant download
  • Own it forever
Softcover Book
USD 109.99
Price excludes VAT (USA)
  • Compact, lightweight edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info
Hardcover Book
USD 109.99
Price excludes VAT (USA)
  • Durable hardcover edition
  • Dispatched in 3 to 5 business days
  • Free shipping worldwide - see info

Tax calculation will be finalised at checkout

Purchases are for personal use only

Institutional subscriptions

Notes

  1. 1.

    In Sects. 6.1 and 6.2 for presentation purposes the figures stated in the text (besides R2) contain two decimal places and may be rounded off as compared to the figures stated in the tables. In Sect. 6.3 for computation purposes concerning the logistic models mostly three decimal places are given.

  2. 2.

    Cf. Anderson (1986), p. 191; Jaccard et al. (1997), p. 21.

  3. 3.

    The significance values of the F tests which are given in the tables’ notes are calculated by SPSS. Manually calculated values might differ slightly, because of round-off errors. In this case, for R1 = 0.406, R2 = 0.504, k1 = 3, k2 = 5 and N = 165, F equals 9.48.

  4. 4.

    In the subsequent analysis these indicators will only be mentioned if abnormalities are detected.

  5. 5.

    Cf. Rogan and Keselman (1977), p. 493.

  6. 6.

    Cf. Glass and Sanders (1972), p. 245; Refer to Wilcox (1995) for a review on statistical power and effect sizes of ANOVA.

  7. 7.

    It shall be noted that one-way ANOVA analysis does not indicate differences between all the six treatment groups. In order to be in a position to perform the one-way ANOVA on the productivity data the assumption of normal distribution needs to be met, additionally. Analyzing the productivity data the Kolmogorov-Smirnov goodness-of-fit test indicates that normal distribution can be assumed for each of the six treatment groups at the 0.05 confidence level.

  8. 8.

    Lazear (2004), p. 21; See also Chiappori and Salanié (2003) or Eriksson and Villeval (2008).

  9. 9.

    This also shows the success of the stratified random assignment procedure described in Sect. 3.2.2. The small differences in the mean of these groups are due to no-shows in Round 2 of the experiment. Four participants assigned to AFix, two participants assigned to APie and four participants assigned to ABud did not show up.

  10. 10.

    On average, as intended by the random assignment procedure, the participants in the assignment treatment (N = 83) show equally high levels of skill as the participants in the self-selection group (N = 82).

  11. 11.

    However, the skill level is also slightly lower.

  12. 12.

    The influence of skill on the decision for a compensation scheme is tested directly in Sect. 6.3.

  13. 13.

    Cf. Derfuss (2009).

  14. 14.

    Cf. Bonner (2008), p. 223.

  15. 15.

    Still, the self-selection groups together have the same variance as the assignment groups together, as the self-selection groups vary around means that are relatively wide apart from each other.

  16. 16.

    Cf. Eriksson et al. (2009).

  17. 17.

    Cf. Eriksson et al. (2009), p. 544.

  18. 18.

    An alternative rationale considering percentages might be formulated as follows: On average productivity in comparison to skill increases by 24 % under the budget-based scheme. Through sorting participants in SBud are 40 % more skilled than participants in ABud. In this case, performance difference induced by sorting is approximately 1.7 times higher than performance difference induced by the incentive scheme (0.4/0.24).

  19. 19.

    If one only observes the assignment condition, the result concerning the interaction terms is qualitatively equal.

  20. 20.

    Accommodating magnitude of incentive and interaction of magnitude of incentive and need for achievement in the analysis yields statistically equal results. Performing the analysis only with participants working under the piece rate scheme, for which the model is supposed to be particularly valid, does not lead to significances, either.

  21. 21.

    Cf. Beckmann and Heckhausen (2006), p. 171.

  22. 22.

    In this two variable context the Pearson correlation is the same as β in linear regression.

  23. 23.

    When considering a need for achievement level (x-axis) by relative expectation (y-axis) scatter plot, a higher variance in goals set for lower need for achievement individuals in contrast to higher need for achievement individuals cannot be supported, which is why Atkinson and Litwin (1960)’s finding that lower need for achievement individuals tend to set unrealistic (high) goals is not supported in this context.

  24. 24.

    As the correlation is still below 0.9, it is believed that the assumption of the absence of multicollinearity is not violated to an extent which results in unreliable statistical procedures (Cf. 5.2.3).

  25. 25.

    MAGOI, magnitude of incentive, MAGOI = 1, fixed pay, MAGOI = 2, piece rate pay, or MAGOI = 3, budget-based pay.

  26. 26.

    Cf. Cacioppo and Petty (1982).

  27. 27.

    Accommodating magnitude of incentive as well as interaction of magnitude of incentive and need for cognition in the analysis yields statistically equal results. Performing the analysis only with participants working under the piece rate scheme, for which the model is supposed to be particularly valid, does not lead to significances, either.

  28. 28.

    One needs to be aware that R.LOC.P and R.LOC.C are not merely inverse scales of R.LOC.I, but are based on different context-related items. This may be a reason why R.LOC.P and R.LOC.C are not merely inversely related with the same variables as R.LOC.I is related with.

  29. 29.

    Cf. Spector (1982), p. 486; Ammon (2006), p. 106.

  30. 30.

    Regarding Round 1 (skill session) in which the anagram task is relatively new to the participants, a positive relationship between R.LOC.I and performance cannot be found either (Pearson’s R = −0.04, p = 0.31). A negative relationship as in the productivity session cannot be evidenced either.

  31. 31.

    Unexpectedly, R.LOC.P turns significant at the 0.1 level. This significance is assumed to be a statistical artifact and is not further investigated.

  32. 32.

    Perception of task difficulty (G.7) is not included, because of the cannibalization with skill. Interest is not included, because of its interactive nature.

  33. 33.

    The general trend that the higher the skill, the higher is the incentive in the contract selected (moving from fixed to budget-based pay), is also supported by an additionally performed ordinal regression (BSKILL = 0.155, Wald = 33.75, p = 0.00).

  34. 34.

    The estimate of the 2 versus 3 comparison is derived from the two other comparisons (1 vs. 2, 1 vs. 3) such that BSKILL is 0.105 (= 0.239 – 0.134). The value suggests that skill influences the decision between piece rate and budget-based pay such that the higher the skill, the more likely an individual selects the budget-based pay in contrast to the piece rate pay. The predictor is significant at the 0.01 level (p = 0.00, Wald = 8.48). As suggested by Backhaus et al. (2006), p. 476, this test is performed by changing the reference category of the multinomial testing procedure in SPSS.

  35. 35.

    The odds ratio is defined by the quotient of two probabilities, whose total is one such that the higher probability is divided by the smaller probability. The odds ratio of 4 indicates probabilities of 0.8 to 0.2. If the probabilities are equal (0.5), the odds ratio is 1. Refer to Backhaus et al. (2006), pp. 434 ff. for further explanation.

  36. 36.

    The odds ratio is automatically calculated by SPSS, which is why manually calculated figures can involve round-off errors.

  37. 37.

    In analogy, a participant is 10 % more likely to select the budget-based instead of the piece rate contract if the participant’s skill level increases by h = log(1.5)/log(1.111) = 3.85 anagrams from the respective threshold of 56 anagrams.

  38. 38.

    Note that around P(MAGOI = 2) = 0.5 a linear relationship is assumed, which is not appropriate at probability levels towards P(MAGOI = 2) = 1. However, such simplification facilitate the interpretation of effect sizes.

  39. 39.

    The 2 vs. 3 comparison is not significant.

  40. 40.

    Cf. Brunstein and Heckhausen (2006), pp. 156 ff.; Bonner (2008), p. 204.

  41. 41.

    Cf. Vecchio (1982).

  42. 42.

    As the set goals were stated by the participants after they had known their incentive scheme, the set goals cannot be used to predict the choice for a scheme.

  43. 43.

    Cf. Schroeder et al. (2005), p. 67.

  44. 44.

    R.IABSOLUTE is statistically equivalent to R.LOC.I. It holds that R.LOC.I = R.IABSOULTE/8. The R.IABSOULTE scale is taken for illustration purposes in Fig. 6.12. The R.IABSOULTE scale is taken for analysis, because the inventories R.LOC.C and R.LOC.P do not lead to statistically significant predictions, which is why they are excluded from tabular as well as textual consideration.

  45. 45.

    This categorization follows MacCrimmon and Wehrung (1985b), p. 10.

  46. 46.

    For a more comparative data analysis and discussion in this respect refer to Fehrenbacher and Pedell (2012).

  47. 47.

    Cf. Dohmen and Falk (2011); Waller and Chow (1985); Ackerberg and Botticini (2002).

  48. 48.

    In general, the risk perception levels between the assignment and self-selection conditions are not found to be significantly different.

  49. 49.

    A.RISK1 with M6 and J2.RISK3 with M.RISKPERCEPTION do not have significant Pearson correlations.

  50. 50.

    The Pearson correlation between J2.RISK3 and F.PFI is 0.07 (p = 0.19).

  51. 51.

    Levene’s test does not reject the homogeneity of variance assumption of the gender groups (p = 0.44).

  52. 52.

    An independent samples t test evaluating differences in mean M.RISKPERCEPTION levels between men and women results in (t(162) = 1.64, p = 0.05 1-tailed). An independent samples t test evaluating differences in mean M.6 levels between men and women results in (t(163) = 1.49, p = 0.08 1-tailed).

  53. 53.

    The risk perception levels were measured after the participants had selected or had been assigned to their compensation scheme.

  54. 54.

    Cf. Baron and Kenny (1986), p. 1177.

  55. 55.

    Cf. Baron and Kenny (1986), p. 1177.

  56. 56.

    German Socio Economic Panel 2004 cited in Dohmen and Falk (2011); For a current discussion of the gender wage gap based on the German Socio Economic Panel 1999–2006 refer to Al-Farhan (April, 2010). The differences in compensation between the public and private sectors is in line with theory, which predicts that workers, who are paid on a variable basis, earn more than fixed rate workers, because they demand a risk premium (Cf. Brown 1992, p. 366).

  57. 57.

    As the B values and the marginal effects are not standardized, they need to be regarded in relation to the variation and range of the measures in order to judge relative importance. The SDs are as follows: SKILL = 16.44, E.NFA = 1.99, Q.NFC.14 = 10.88, R.IABSOULTE = 4.14, J2.RISK3 = 1.48.

References

  • Ackerberg DA, Botticini M (2002) Endogenous matching and the empirical determinants of contract form. J Polit Econ 110(3):564–591

    Article  Google Scholar 

  • Ammon S (2006) Commitment, Leistungsmotivation, Kontrollüberzeugung und erlebter Tätigkeitsspielraum von Beschäftigten. in Unternehmen und Behörden im Vergleich, Berlin

    Google Scholar 

  • Anderson CH (1986) Hierarchical moderated regression analysis: a useful tool for retail management decisions. J Retailing 62(2):186–203

    Google Scholar 

  • Atkinson JW, Litwin GH (1960) Achievement motive and test anxiety conceived as motive to approach success and motive to avoid failure. J Abnorm Soc Psychol 60:53–62

    Google Scholar 

  • Backhaus K, Erichson B, Plinke W, Weiber R (2006) Multivariate analysemethoden: eine anwendungsorientierte einführung, 11th edn. Heidelberg, Berlin

    Google Scholar 

  • Baron RM, Kenny DA (1986) The moderator-mediator variable distinction in social psychological research: conceptual, strategic, and statistical considerations. J Pers Soc Psychol 51(6):1173–1182

    Article  Google Scholar 

  • Beckmann J, Heckhausen H (2006) Motivation durch erwartung und anreiz. In: Heckhausen J, Heckhausen H (eds) Motivation und handeln. Springer, Heidelberg, pp 105–142

    Chapter  Google Scholar 

  • Bonner SE (2008) Judgment and decision making in accounting. Pearson/Prentice Hall, Upper Saddle River

    Google Scholar 

  • Brown C (1992) Wage levels and method of pay. Rand J Econ 23(3):366–375

    Article  Google Scholar 

  • Brunstein J, Heckhausen H (2006) Leistungsmotivation. In: Heckhausen J, Heckhausen H (eds) Motivation und Handeln, 3rd edn. Springer, Heidelberg, pp 144–192

    Google Scholar 

  • Cacioppo JT, Petty RE (1982) The need for cognition. J Pers Soc Psychol 42:116–131

    Article  Google Scholar 

  • Chiappori J-C, Salanié B (2003) Advances in economics and econometrics. testing contract theory: a survey of some recent work. In: Dewatripont M, Hansen LP, Turnovsky SJ (eds) Advances in economics and econometrics. theory and applications; eighth world congress. Cambridge University Press (Econometric Society monographs, No. 35), Cambridge, pp 115–149

    Chapter  Google Scholar 

  • Derfuss K (2009) The relationship of budgetary participation and reliance on accounting performance measures with individual-level consequent variables: a meta-analysis. Eur Account Rev 18(2):203–239

    Article  Google Scholar 

  • Dohmen T, Falk A (2011) Performance pay and multi-dimensional sorting: productivity, preferences and gender. Am Econ Rev 101(2):556–590

    Article  Google Scholar 

  • Eriksson T, Villeval MC (2008) Performance-pay, sorting and social motivation. J Econ Behav Organ 68(2):412–421

    Article  Google Scholar 

  • Eriksson TO, Teyssier S, Villeval M (2009) Self-selection and the efficieny of tournaments. Econ Inq 47(3):530–548

    Article  Google Scholar 

  • Fehrenbacher DD, Pedell B (2012) Disentangling incentive effects from sorting effects: an experimental real-effort investigation. Working paper # 2012–08, Risk management and decision processes center, The Wharton school, University of Pennsylvania

    Google Scholar 

  • Glass GV, Sanders JR (1972) Consequences of failure to meet assumptions underlying the fixed effects analyses of variance and covariance. Rev Educ Res 42:237–288

    Article  Google Scholar 

  • Jaccard J, Turrisi R, Wan CK (1997) Interaction effects in multiple regression, 9th edn. Sage, Newbury Park

    Google Scholar 

  • Lazear EP (2004) Output-based pay: incentives, retention or sorting. In: Polachek SW (ed) Accounting for worker well-being. Elsevier JAI (Research in labor economics, 23), Amsterdam, pp 1–25

    Chapter  Google Scholar 

  • MacCrimmon KR, Wehrung DA (1985) A portfolio of risk measures. Theor Decis 19:1–29

    Article  Google Scholar 

  • Rogan JC, Keselman HJ (1977) Is the ANOVA F-test robust to variance heterogeneity when sample sizes are equal?: an investigation via a coefficient of variation. Am Educ Res J 14(4):493–498

    Article  Google Scholar 

  • Schroeder LD, Sjoquist DL, Stephan PE (2005) Understanding regression analysis: an introductory guide. Sage, Newbury Park, Calif

    Google Scholar 

  • Spector PE (1982) Behavior in organizations as a function of employee’s locus of control. Psychol Bull 81(3):482–497

    Article  Google Scholar 

  • Vecchio RP (1982) The contingent-noncontingent compensation controversy: an attempt at a resolution. Hum Relat 35:449–462

    Article  Google Scholar 

  • Waller WS, Chow CW (1985) The self-selection and effort effects of standard-based employment contracts: a framework and some empirical evidence. Account Rev 60(3):458–476

    Google Scholar 

  • Wilcox RR (1995) ANOVA: a paradigm for low power and misleading measures of effect size? Rev Educ Res 65:51–77

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

  1. Institute of Business Administration, Universität Stuttgart, Stuttgart, Germany

    Dennis D. Fehrenbacher

Authors
  1. Dennis D. Fehrenbacher
    View author publications

    You can also search for this author in PubMed Google Scholar

Rights and permissions

Reprints and permissions

Copyright information

© 2013 Springer-Verlag Berlin Heidelberg

About this chapter

Cite this chapter

Fehrenbacher, D.D. (2013). Findings. In: Design of Incentive Systems. Contributions to Management Science. Physica, Berlin, Heidelberg. https://doi.org/10.1007/978-3-642-33599-0_6

Download citation

Publish with us

Policies and ethics

Search

Navigation