11
Association of Research Libraries
Research Library Issues 290 2017
their research topic, choose keywords, evaluate articles and
websites, and learn other key research skills. We gathered Internet
IDs for students who met with a peer research consultant.
Outcomes measures: Students’ learning outcomes. The
dependent variables of interest in this study included first-year
students’ self-reported development of three learning outcomes:
critical thinking and analytical skills, written communication
skills, and reading comprehension skills. In the SERU survey,
students were asked to indicate their skill levels when they started
at the university and their current skill levels on a scale from 1
(very poor) to 6 (excellent). We subtracted students’ skills when
they started at the university from their current skills to develop
variables measuring students’ growth or regression in those areas.
Data Analyses
We utilized propensity score matching techniques in SPSS 23.0
using the procedures outlined by Thoemmes.30 We began by using
binary logistic regression to compute propensity scores for individual
students. We used the binary variable of using the library (yes or
no) as a dependent variable and the independent covariates listed
above in the regressions to calculate the probabilities of using a
library resource at least once. Next, we used 1:1 nearest neighbor
matching, meaning that each student in the treatment condition
is matched to a student in the untreated condition who has the
most similar estimated propensity score. We matched without
replacement and discarded all units that fell outside of the area
of common support to avoid extrapolation to units that were so
dissimilar that no comparisons could be made to other units.31
Next, it was important to check whether the matching procedures
balanced the distribution of variables in both the treatment and
control groups. We examined the standardized mean differences
(the mean differences between the two groups divided by the
Previous Page Next Page