Desktop version

Home arrow Business & Finance

  • Increase font
  • Decrease font

<<   CONTENTS   >>

Data Analysis

Chapter overview

Chapter 5 is concerned with the analysis of the gathered data. Reliability measures and inter-item correlation indices are outlined for each scale. A confirmatory factor analysis is pursued for the new self-leadership scale, the Self-Leadership Skills Inventory (Furtner & Rauthmann, in prep.). Descriptive statistics provide insight in structure of followers’ response behavior of perceptions of leadership behavior and leader-member exchange. Finally, data is examined for heteroskedasticity, multi- collinearity, and common method variance.

Descriptive Statistics and Reliability

In the following section, descriptive characteristics of the follower sample regarding questions on perceived leadership behavior are outlined. For analysis, the response scheme ranged from 1 to 5 (1= not at all, 2 = once in a while, 3 = sometimes, 4 = fairly often, 5 = frequently, if not always). For all responses by followers (n = 372) means, standard deviations, skewness, and kurtosis were computed. Furthermore, scales were tested for reliability indicated by Cronbach alpha scores and mean inter-item correlations. Literature suggests this procedure if multiple- indicator measures are used (Bryman & Bell, 2011, p. 160).

Cronbach’s alpha is commonly used as a statistical method to test an instrument for reliability. Alpha coefficients are interpreted as functions of interrelatedness of items, so-called internal consistency (Cortina, 1993). Reliability is said to be high if a scale (or a set of items) produces similar results under consistent conditions (Field, 2013, p. 708). Interpretation of Cronbach alpha scores has yet often been subject to discussion. Whereas Kline (1999) argues that values of .80 are acceptable for cognitive measures, psychological scales can result in values even lower than .70 and may still be regarded as satisfactory. Nunnally (1978) even states that alpha values of .50 may be regarded sufficient. Cortina (1993) points out that internal consistency is often confused with homogeneity. Homogeneity yet explains the degree of unidimensionality (Green, Lissitz & Mulaik, 1977). In order to provide a clear distinction Cortina (1993) defines alpha as following:

It is a function of the extent to which items in a test have high communali- ties and thus low uniqueness. It is also a function of interrelatedness, although one must remember that this does not imply unidimensionality or homogeneity. (Cortina, 1993, p. 100)

© Springer Fachmedien Wiesbaden 2017

N. Poser, Distance Leadership in International Corporations,

Advances in Information Systems and Business Engineering,

DOI 10.1007/978-3-658-15223-9_5

Literature on mean inter-item-correlation is divided on interpretation of correlation values. Clark and Watson (1995, p. 114) recommend inter-item correlations to range between .15 and .50, whereas an earlier study proposes values between .20 and .40 (Briggs & Cheek, 1986). The researchers explain that narrower constructs might be subject to higher inter-item correlation than others. High item correlations usually claim that a construct possesses a high degree of internal consistency. On the other hand, extendedly high correlations among measures may indicate that items are describing the construct too narrowly by causing redundancy of the remaining items (Briggs & Cheek, 1986, p. 114).

<<   CONTENTS   >>

Related topics