26.11.2012 Views

comparative value priorities of chinese and new zealand

comparative value priorities of chinese and new zealand

comparative value priorities of chinese and new zealand

SHOW MORE
SHOW LESS

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

� Linear regression will be employed to test the significance <strong>of</strong> SVS <strong>value</strong><br />

dimensions as predictors <strong>of</strong> LBDQXII preferred leader behaviour dimensions.<br />

SELECTION OF ANALYSIS TOOLS<br />

Scales originally designed by Likert (1932), hence called “Likert scales”, are widely<br />

used in social science research, especially in measuring attitudes, opinions, <strong>and</strong> beliefs<br />

(Jacoby, 1971). Likert proposed an interval summated scale. Some argue that the scale<br />

is ordinal in nature (Allen <strong>and</strong> Seaman, 2007; Hodge <strong>and</strong> Gillespie, 2003). Allen <strong>and</strong><br />

Seaman (2007) state that analysis <strong>of</strong> Likert scales should employ non-parametric<br />

procedures; there is a multitude <strong>of</strong> articles disagreeing. Blaikie (2003) observes that it<br />

has become common practice to use Likert-type categories as constituting interval-level<br />

measurement. Rarely is it made clear by authors whether they are aware that some<br />

would regard this as illegitimate; generally no statement is made about an assumption <strong>of</strong><br />

interval status for Likert data; usually no argument made in support.<br />

Fier-Walsh <strong>and</strong> Toothaker (1974) demonstrate that when normality <strong>and</strong>/or homogeneity<br />

<strong>of</strong> variance are doubtful, as in this study, the parametric ANOVA F-test is the<br />

recommended procedure for testing hypotheses about means. Boos, <strong>and</strong> Brownie (1995)<br />

demonstrate that the ANOVA F-test is also the recommended procedure when<br />

assumptions <strong>of</strong> normality <strong>and</strong> homogeneity <strong>of</strong> variance are violated <strong>and</strong> the number <strong>of</strong><br />

treatments or variables is large, such as the ten <strong>and</strong> twelve dimensions in this study.<br />

Review <strong>of</strong> research literature using Likert-type scales indicated that the consensus <strong>of</strong><br />

researchers has been to use parametric statistical analyses. While a psychometric theory<br />

purist can question whether this approach violates assumptions <strong>of</strong> procedures,<br />

parametric tests will be employed to conform to past publication <strong>of</strong> results.<br />

Analysis <strong>of</strong> Variance <strong>and</strong> Covariance<br />

Analysis <strong>of</strong> variance <strong>and</strong> covariance test for differences between the means samples to<br />

examine the probability that they are drawn from the same population. The hypotheses<br />

for the comparison <strong>of</strong> my two independent groups are:<br />

Ho: The samples come from populations with identical means<br />

Ha: They samples come from different populations with different means<br />

206

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

Saved successfully!

Ooh no, something went wrong!