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

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

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

Human Resources Department for hotels in China, where I implemented an employee<br />

satisfaction survey for all staff in 1996, using a 5-point Likert-scale technique for data<br />

collection. This survey was repeated in 1997 using a forced-choice “agree-disagree”<br />

technique with similar outcomes as to preferences. Also in 1997, the hotel chain<br />

implemented an employee satisfaction survey produced by a pr<strong>of</strong>essional surveyresearch<br />

firm, using Likert-type responses. Described in previous studies (Littrell,<br />

2002a, 2002b, <strong>and</strong> 2003) LBDQXII questionnaires were again administered in 2002 to<br />

the staff <strong>of</strong> these hotels in China, with subjects demonstrating underst<strong>and</strong>ing <strong>of</strong> the<br />

process <strong>and</strong> the intent <strong>of</strong> the surveys using the Likert technique. Using the LBDQXII<br />

the results indicated reliable <strong>and</strong> valid responses from the subjects (Littrell, 2003).<br />

Error or Data?<br />

Baumgartner <strong>and</strong> Steenkamp (2001) provide evidence <strong>of</strong> response bias as a source <strong>of</strong><br />

contamination in questionnaire ratings, noting that they threaten the validity <strong>of</strong><br />

conclusions drawn from marketing research data. They investigated five forms <strong>of</strong><br />

stylistic responding: acquiescence <strong>and</strong> dis-acquiescence response styles, extreme<br />

response style/response range, midpoint responding, <strong>and</strong> non-contingent responding.<br />

Using data from large, representative samples <strong>of</strong> consumers from eleven countries <strong>of</strong><br />

the European Union, they found systematic effects <strong>of</strong> response styles on scale scores as<br />

a function <strong>of</strong> two scale characteristics, the proportion <strong>of</strong> reverse-scored items <strong>and</strong> the<br />

extent <strong>of</strong> deviation <strong>of</strong> the scale mean from the midpoint <strong>of</strong> the response scale. The<br />

correlations between scales can be biased upward or downward depending on the<br />

correlation between the response style components.<br />

Systematic errors in response bias in research are errors introduced into a measurement<br />

by some factor that has persistent directional effects on the characteristic being<br />

measured, or the process <strong>of</strong> measurement (Patel, Harrison <strong>and</strong> McKinnon, 2002). Smith<br />

(2004a) points out that since initial studies <strong>of</strong> response sets or response bias, many<br />

cross-cultural researchers have approached culture-specific response style as a source <strong>of</strong><br />

error in the making <strong>of</strong> cross-cultural comparisons <strong>and</strong> that ways must thus be found to<br />

discount it (e.g., Leung <strong>and</strong> Bond, 1989; Harzing, 2006). H<strong>of</strong>stede (1980) used withinsubject<br />

st<strong>and</strong>ardization averaging the agreements that each respondent records within<br />

the full range <strong>of</strong> items. Deviation scores such as these are sometimes st<strong>and</strong>ardised by<br />

dividing by the st<strong>and</strong>ard deviation (subject-based z-scores). Smith (2004a) points out<br />

196

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

Saved successfully!

Ooh no, something went wrong!