Microsoft Word 2007 ichrie conference Proceedings Final-Final 06-06-07. doc

Data Collection
The revised instrument adapted from Brownell (1994) and Ng and Pine (2003) was formatted for electronic 
delivery via the Internet. Data were collected online over a 5 week period in Spring, 2006. The population for the 
study was comprised of hospitality undergraduate, graduate students, and educators (full time faculty and 
instructors) from four-year hospitality programs in the United States, and industry recruiters associated with 
hospitality programs. Three convenience sampling frames were obtained for this research: 1) hospitality students 
from six universities, 2) hospitality educators from thirteen universities, and 3) hospitality recruiters from three 
universities. To improve the response rate, a contact person was identified at each of the participating universities. 
The contact person introduced the study by distributing a letter that explained the purpose of the study and provided 
the URL link to access the survey. A reminder email message was sent to both university contacts and industry 
recruiters two weeks after the initial request. 
Data Analysis
Usable responses were obtained from a convenience sample of 232 participants, which included 110 
students, 62 educators and 60 recruiters. Prior to analysis, the data were examined for accuracy, missing data, and 
the assumptions of multivariate analysis. Since most questions in the survey were set as “required” questions, no 
missing data were found. Normality, linearity, and multicollinearity were evaluated and the results were satisfactory. 
Mahalanobis distance was performed to identify multivariate outliers. Using Mahalanobis distance with p <.001, a 
total of 6 cases (about 2%) were identified as multivariate outliers. These outliers appeared to be random in the data 
set and were deleted. A total of 226 cases were left for further analysis: 107 cases in the student group, 60 cases in 
the educator group, and 59 cases in the recruiter group. 
Structural equation modeling (SEM) analysis was performed to test the extent to which a priori pattern of 
factor loadings in pilot study represented the actual data. Since some of the indices were less than the recommended 
standard, a principal axis factor analysis (PAF) was computed to determine the underlying dimensions of 15 
facilitators, 15 constraints, and seven gender issues related to women’s career advancement. Factor analysis was 
conducted independently for the above three sets of variables, using the same procedures as in the pilot study. KMO 
values ranging from .80 to .85 were greater than the recommended criterion of .60 (Tabachnick & Fidell, 2001), 
which indicated that factor analysis was appropriate.