16.07.2014 Views

Leadership styles, mentoring functions received, and job-related stress

Leadership styles, mentoring functions received, and job-related stress

Leadership styles, mentoring functions received, and job-related stress

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.

LEADERSHIP, MENTORING AND STRESS 377<br />

Mentoring <strong>functions</strong> <strong>received</strong><br />

To assess the degree of <strong>mentoring</strong> <strong>received</strong> by the prote ge , we used two 10-item scales from Noe<br />

(1988) as indicators of <strong>mentoring</strong> <strong>functions</strong>: (a) career development (a ˆ 0.86), <strong>and</strong> (b) psychosocial<br />

support (a ˆ 0.87). Prote ge s were asked to indicate their extent of agreement with each item<br />

using a ®ve-point scale ranging from disagree strongly (1) to agree strongly (5). Hypothesis 4,<br />

which pertained to the moderating e€ect of <strong>mentoring</strong> <strong>functions</strong> <strong>received</strong>, was tested by creating<br />

two sub-samples based on a median split: low <strong>mentoring</strong> <strong>functions</strong> <strong>received</strong> sub-sample (scale<br />

score 53.85; n ˆ 103) <strong>and</strong> high <strong>mentoring</strong> <strong>functions</strong> <strong>received</strong> sub-sample (scale score 43.85,<br />

n ˆ 101).<br />

Job-<strong>related</strong> <strong>stress</strong><br />

Job-<strong>related</strong> <strong>stress</strong> perceived by the prote ge was measured using six items from Parasuraman<br />

(1977, `Sources <strong>and</strong> outcomes of organizational <strong>stress</strong>: a multidimensional study of the<br />

antecedents, <strong>and</strong> attitudinal <strong>and</strong> behavioral indices of <strong>job</strong> <strong>stress</strong>'. Unpublished doctoral<br />

dissertation, State University of New York at Bu€alo). Prote ge s were asked to judge how often<br />

they experienced situations described in each item. Each situation was measured on a ®ve-point<br />

frequency scale ranging from never occurs (1) to constantly occurs (5).<br />

The indicators of laissez-faire, contingent reward, <strong>and</strong> transformational leadership were<br />

expected to covary (see Bass, 1998), as were the indicators of <strong>mentoring</strong> <strong>functions</strong> <strong>received</strong> <strong>and</strong><br />

<strong>job</strong>-<strong>related</strong> <strong>stress</strong>, thereby indicating that they arise from their respective constructs. Therefore,<br />

these indicators were modelled as re¯ective.<br />

Control variables<br />

Theoretical work on <strong>mentoring</strong> (e.g., Kram, 1985; Murray, 1991) suggests that prote geÂ<br />

experience (i.e., age, <strong>job</strong> level, <strong>job</strong> tenure, education level), industry, <strong>and</strong> mentor gender can a€ect<br />

mentor behavior, <strong>mentoring</strong> <strong>functions</strong> <strong>received</strong>, <strong>and</strong> prote ge <strong>job</strong>-<strong>related</strong> <strong>stress</strong>. To control for<br />

these potential e€ects, age, <strong>job</strong> level, <strong>job</strong> tenure, education level, industry, <strong>and</strong> mentor gender<br />

were entered into the PLS model as covariates. Each covariate was modelled as a single-item<br />

indicator. PLS is insensitive to how the indicators of single-indicator constructs are modelled.<br />

Results<br />

Table 2 presents the scale means, st<strong>and</strong>ard deviations, <strong>and</strong> product moment correlations among<br />

the measures. PLS generates statistics to test the reliability <strong>and</strong> validity of latent constructs with<br />

two or more re¯ective indicators. Reliability was assessed by ®rst examining the factor loadings of<br />

indicators: a common rule of thumb is that the factor loadings should exceed 0.7 since this implies<br />

that less than half of the indicator's variance is due to error (Fornell <strong>and</strong> Larcker, 1981). Next, we<br />

computed each construct's composite scale reliability (Fornell <strong>and</strong> Larcker, 1981), a measure of<br />

internal consistency similar to Cronbach's alpha. Fornell <strong>and</strong> Larcker (1981) recommended using<br />

a criterion cut-o€ of 0.7 or more. Also, the average variance extracted by the construct<br />

from indicators was examined. For this criterion, Fornell <strong>and</strong> Larcker (1981) recommended using<br />

a cut-o€ of 0.5 or more.<br />

Copyright # 2000 John Wiley & Sons, Ltd. J. Organiz. Behav. 21, 365±390 (2000)

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

Saved successfully!

Ooh no, something went wrong!