Characteristics of Education Doctoral Dissertation References, p. 7
Similar to Gooden (2001), this study found, across all institutions, that research
collections overwhelmingly contained the sources cited by doctoral students.
Journal
and magazine titles were checked in the online library catalogs of the institutions.
Of
the 196 references cited by Institution 1 candidates, 19, or 9.3%, were not
locally held,
90.7% were owned. Likewise, of the 298 references cited by Institution
2 students, 21,
or 7%, were not owned by the institution,
93% were owned. Of the 362 references
cited by Institution 3 students, only 11, or 3%, were not locally owned,
97% were.
To arrive at some explanation of student reliance on local collections
dissertation citations were scored for scholarliness, currency, and appropriateness
of
format. The criterion of scholarliness was scored based on journal prestige
within the
discipline and the field, presence or absence of peer review, and consideration
of
empirical, research-based studies rather than program descriptions. Citations were
also
rated on currency, or their timeliness of publication. The date of publication was
considered in context of type of material and usage in the literature review, and
the
raters recognized when currency was not an issue. Appropriateness, or
fit of the
material type to the topic being developed, was considered in relation to maturity
of the
field. Scores on the three criteria were averaged to arrive at an overall quality
rating.
Across all coded citations, the mean statistic for scholarliness was 2.70 (SD =
.80), skewness was .164 (SE = .057), and kurtosis was -.752 (SE = .114). Statistics
for
the remaining criteria include: currency (M = 2.63, SD = .56, skewness = -1.243, and
kurtosis = .560), appropriateness (M = 2.68, SD = .56, skewness = -1.534, and kurtosis
= 1.383), and quality score (M = 2.67, SD = .45,
skewness = -.398, and kurtosis = -
.478). Descriptive statistics for each criterion and by institution are shown in Tables 7
through 10. Scores were also submitted to the Lilliefors Significance Correction of the
Kolmogorov-Smirnov test of normality. Normality statistics are reported in Table 11,
and boxplots, see Figures 1 through 4, offer a graphic representation of the
distributions.
A Kruskal-Wallis test was conducted comparing the scores on coded citations
across institutions. A statistically significant
result was found for scholarliness (H(2) =
107.11, p < .01), indicating that the institutions differed from each other. Institution 2
averaged a placement of 774.37, while Institution 1 averaged a placement of 978.70
and Institution 3 averaged 1038.20. Currency also differed significantly (H(2) = 43.11,
p < .01) across institutions. Institution 2 averaged a
rank of 847.61 while Institution 1
averaged 918.41 and Institution 3 999.74. A statistically significant result was found
for appropriateness scores (H(2) ,--- 57.70, p < .01) when compared across institutions.
Institution 2, with an average rank of 829.82, was lower than Institution 3, at 975.81
and Institution 1, at 986.95. Quality scores were likewise significantly different (H(2)
= 150.32. p < .01). Institution 2 averaged 739.72
while Institution 1 averaged 988.36
and Institution 3 1068.03.
A one-way ANOVA was also calculated comparing each of the criteria across
institutions. For scholarliness scores, a statistically significant difference was found
(F(2,1839) = 52.36, p < .01). Tukey's HSD was calculated to determine the nature of
9