11.03.2014 Views

thesis_Daniela Noethen_print final - Jacobs University

thesis_Daniela Noethen_print final - Jacobs University

thesis_Daniela Noethen_print final - Jacobs University

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Multilevel Investigation of Antecedents of Knowledge Sharing and Seeking in Teams<br />

2.6.7. Limitations<br />

The present study has some limitations that should be noted. First of all, we only examine<br />

knowledge transfer within teams. Consequently, our results cannot be generalized to transfer<br />

across team borders. More generally, the generalizability is limited since data were collected<br />

from two German companies in a cross-sectional design. Accordingly, our sample might not<br />

be representative for all organizations, and inferences cannot be made about the direction of<br />

effects.<br />

Moreover, we did not differentiate between different kinds or characteristics of<br />

knowledge. The influences might have differed for sharing tacit versus explicit knowledge<br />

(Nonaka, 1994; Nonaka & Takeuchi, 1995; Nonaka & von Krogh, 2009), more versus less<br />

ambiguous knowledge (Szulanski, 1996, 2000; Szulanski, Cappetta, & Jensen, 2004), or more<br />

versus less codifiable knowledge (Reagans & McEvily, 2003). However, these distinctions<br />

have already been well researched and are widely accepted, and introducing a third distinction<br />

in the present study would not only have complicated matters in a non-negligible manner, but<br />

would also have rendered the parsimonious approach to measurement impossible.<br />

Furthermore, limitations with respect to the operationalization of constructs have to be<br />

mentioned; independent as well as dependent variables were operationalized via self-reports,<br />

and the scales we used were very parsimonious, measuring constructs with only two to four<br />

items.<br />

First, using only one measurement source can entail common method bias that can inflate<br />

or deflate the estimations of relationships between constructs (Podsakoff, MacKenzie, Lee, &<br />

Podsakoff, 2003). To control common method bias, we took several steps recommended by<br />

Podsakoff and colleagues (2003) when designing the study’s procedure: we protected<br />

respondents’ anonymity and reduced evaluation apprehension as much as possible.<br />

Furthermore, we improved scale items by reducing ambiguity through the clarification of key<br />

73

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!