Multivariate research techniques are powerful and well-suited tools to analyze strategic management or organizational behavior constructs such as entrepreneurial orientation. This chapter will firstly present multivariate techniques of the so-called “first generation”, and discuss their strengths and weaknesses. Secondly, it will introduce and discuss further advanced “second-generation” multivariate research techniques, in particular variance-based and covariance-based structural equation modeling. A comparison of the techniques across generations will lead to the assessment that second-generation multivariate research techniques are more suitable to contribute as a methodology to the problem of influence of entrepreneurial orientation on technology transfer performance.


Structural Equation Modeling Entrepreneurial Orientation Manifest Variable Formative Indicator Interval Scaling 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 349.
    See Shook, Ketchen, Hult and Kacmar (2004) as to the development of multivariate research tech-niques and structural equation modeling in strategic management literature. The authors present an analysis of 92 strategic management studies published from 1994 to 2002. Despite the relatively wide-spread usage of multivariate research techniques, the authors critically comment that this usage has often been less than ideal, though, and that researchers might draw erroneous conclusions about rela-tionships among variables.Google Scholar
  2. 350.
    See Hoyle (1995b) for a general introduction into the concept of structural equation modeling.Google Scholar
  3. 351.
    Anderson and Gerbing (1988), pp. 411–412, argue that “although it is convenient to distinguish be-tween exploratory and confirmatory research, in practice this distinction is not as clear-cut. … Rather then strict dichotomy, then, the distinction in practice between exploratory and confirmatory analysis can be thought of that as an ordered progression.”Google Scholar
  4. 352.
    Fornell (1982), pp. 1–2.Google Scholar
  5. 353.
    See Fornell (1982), p. 2. In addition, Sheth (1971), pp. 13–17, and Kinnear and Taylor (1971), pp. 56–59, provide classifications of multivariate methods.Google Scholar
  6. 354.
    See McDonald (1996), p. 239. McDonald states that “a random variable is observable if and only if its values can be obtained by means of a real-world sampling experiment.”Google Scholar
  7. 355.
    See Bagozzi, Yi and Phillips (1991), p. 459, and Fornell (1982), p. 3. Fornell argues that “very few, if any, measures in the social (or natural) sciences are free from error. Variables such as earnings or sales, population growth, and age are subject to error because of inaccurate statistics, faulty record keeping, or imperfect coding. In ignoring these errors, the analyst runs the risk of obtaining biased pa-rameter estimates.”Google Scholar
  8. 356.
    See Fornell (1982), pp. 2–3, and Anderson and Gerbing (1988), pp. 411–412.Google Scholar
  9. 357.
    See Autio, Keeley, Klofsten and Ulfstedt (1997), who examined the entrepreneurial intent construct using SEM.Google Scholar
  10. 358.
    Chin (1998), p. 296.Google Scholar
  11. 359.
    It is beyond the scope of this dissertation to provide in-depth detail about the alternative multivariate techniques. Much more, this introduction serves to present a very brief general framework. The inter-ested reader is referred to Hänlein and Kaplan (forthcoming), p. 2, and Fornell (1982), pp. 3–4, for a more comprehensive introduction to these methods. In the following, only the models of LISREL and PLS will be examined to a greater extend.Google Scholar
  12. 360.
    See Lumpkin and Dess (1996), p. 152.Google Scholar
  13. 361.
    See Fornell (1982), pp. 3–4.Google Scholar
  14. 362.
    See Hänlein and Kaplan (forthcoming), p. 5, and Fornell (1982), p. 10.Google Scholar
  15. 363.
    Fornell (1982), p. 5.Google Scholar
  16. 364.
    See Fornell (1982). p. 5.Google Scholar
  17. 365.
    See Fornell (1982), pp. 6–7.Google Scholar
  18. 366.
    See Fornell (1982), p. 7.Google Scholar
  19. 367.
    See Knapp (1978), pp. 410–416, for an introduction into canonical correlation analysis, van den Wollenberg (1977), pp. 207–219 for an introduction into redundancy analysis, and Fornell (1979), pp. 323–338, for an introduction into ECCSA.Google Scholar
  20. 368.
    See Fornell and Larcker (1981), p. 39, who refer to Jöreskog (1967) and Jöreskog (1970) for the foundations of structural equation modeling.Google Scholar
  21. 369.
    See Hänlein and Kaplan (forthcoming), p. 4.Google Scholar
  22. 370.
    Hoyle (1995a), p. 1.Google Scholar
  23. 371.
    Shook, Ketchen, Hult and Kacmar (2004), p. 397.Google Scholar
  24. 372.
    Examples for covariance-based methods are LISREL, EQS, AMOS, SEPATH, and RAMONA. See Chin (1998), p. 295.Google Scholar
  25. 373.
    See Jöreskog (1970), Keesling (1972), and Wiley (1973) for the foundations of this approach.Google Scholar
  26. 374.
    Chin (1998), p. 297.Google Scholar
  27. 375.
    See Chin and Newsted (1999), p. 307. See also Diamantopoulos (1994) for an introduction into LISREL.Google Scholar
  28. 376.
    See Meier (2006), p. 73.Google Scholar
  29. 377.
    See Chin (1998), p. 314.Google Scholar
  30. 378.
    See West, Finch and Curran (1995), pp. 56–75, for issues regarding nonnormal variables.Google Scholar
  31. 379.
    See Jöreskog (1967), p. 443.Google Scholar
  32. 380.
    See Hu and Bentler (1995), pp. 76–99, for an introduction into model fit.Google Scholar
  33. 381.
    See Jöreskog and Sorbom (1982), p. 408.Google Scholar
  34. 382.
    See Kaplan (1995), p. 100.Google Scholar
  35. 383.
    See Bielby and Hauser (1977), p. 153. Not rejecting a false hypothesis is a type II error.Google Scholar
  36. 384.
    See Bagozzi (1981a), p. 380, and Fornell and Larcker (1981), p. 39.Google Scholar
  37. 385.
    See Bollen (1987), p. 375.Google Scholar
  38. 386.
    See Boomsma (1985), p. 345.Google Scholar
  39. 387.
    See Marsh, Hau, Balla and Grayson (1998), p. 187.Google Scholar
  40. 388.
    See MacCallum and Browne (1993), p. 540.Google Scholar
  41. 389.
    McDonald (1996), p. 240, states that “Partial Least Squares appears to be, currently, the most fully developed general system for path analysis with composites.”Google Scholar
  42. 390.
    See Bollen and Long (1992), p. 128.Google Scholar
  43. 391.
    See Wold (1989), p. vii.Google Scholar
  44. 392.
    See Chin (1998), p. 311.Google Scholar
  45. 393.
    Chin (1998), p. 303.Google Scholar
  46. 394.
    See Anderson and Gerbing (1988), p. 412.Google Scholar
  47. 395.
    Fornell and Cha (1994), p. 66.Google Scholar
  48. 396.
    See Chin, Marcolin and Newsted (1996), p. 31.Google Scholar
  49. 397.
    Chin (1998), p. 332.Google Scholar
  50. 398.
    See Chin (1998), p. 304.Google Scholar
  51. 399.
    Fornell and Cha (1994), p. 74.Google Scholar

Copyright information

© Betriebswirtschaftlicher Verlag Dr. Th. Gabler | GWV Fachverlage GmbH, Wiesbaden 2008

Personalised recommendations