|
Throughout my career I have had an opportunity to be involved
in a variety of research and applied projects, including participation in development of new applications, grant-funded research
projects, as well as my own research.
Values and Politics: Personal Value Priorities as Predictors of
Psychological and Instrumental Political Engagement (Doctoral
Thesis).
The
values voters and political polarization in the United States, drastic
changes in core societal values in Europe, the rise of religious
extremism, and other phenomena witnessed by most post-industrial
societies during past decades confirms that values continue to play an important
role in political and social life. However, our knowledge of how and to
what extent values influence political life of modern societies is
very limited. In my dissertation I examined the impact of personal
values on individual's general interest in politics and participation
in political activities. I used modified version of the Schwartz's
Values Inventory to measure values, and Roper's Survey of Political
Trends to measure political participation; relationships between the
variables of interest were examined using structural equation
modeling.
Thesis (
PDF)
| Appendices (
PDF) .
Application of Weighted Deviations Heuristic
to a Selection Model with Multiple Diversity Constraints
(Pre-Doctoral Project, equivalent of Master Thesis, under supervision of Charles Lewis, Educational Testing Service/Fordham University).
Most of decision support systems used by college admissions
offices in the US are based on simple models that operate with
quantitative information, since quantification of all aspects of the
selection process is the best guarantee of its objectivity. Such models, however, become inefficient when in addition to
selection criteria, multiple diversity constraints are imposed on
non-academic factors, such as gender, ethnicity or work experience
of the candidate. . The problem of incorporating diversity constraints into
selection model can be resolved by introducing methods that allows for
measuring and representing diversity quantitatively. In this
paper, we consider weighted deviations heuristic model that allows
facilitation of diversity in admissions decisions by integrating selection
criteria and diversity constraints. More efficient alternative to
conventional admission models, weighted deviations heuristic allows
quickly assemble from the large pool of applicants the class with
desired properties. This paper investigated how differences in prior
distributions of scores on selection variables affect the diversity of
the final class, and how adding diversity constraints affects the
"quality" of the class.
Topological Structures and Games with Mixed
Strategies: Application to Stock Market Games (Bachelor Diploma
Thesis). This paper examined mathematical models of two common
stock-market situation: assembly of optimal investment portfolio and
corporate takeover with minimal stock purchase.
Differential Performance of Test Items
by Geographical Regions (paper presentation, submitted for
National Councils for Measurement in Education (NCME) conference,
2008, in collaboration with Deanna Morgan, College Board).
Differential item functioning (DIF) occurs when examinees of the same ability level from
different groups have a different probability of answering given item
correctly. DIF analyses have been traditionally used for examining the
fairness of tests with regard to specific populations, including but
not limited to gender, age, ethnic, cultural and language groups. The
objective of this research was to compare the performance of items of
a nationally administered test across geographical regions and
individual states within the continental United States. The premise of
this study was that certain geographically-based dissimilarities among
examinees may potentially result in differential performance of
certain items across various parts of the US. DIF analysis was performed on individual items
across geographical regions (College Board classification) and
individual states, as indicated on the examinees’ records. Items were
flagged for DIF based on the results of both MH and SIBTEST criteria.
Further analyses, such as examination of item impact, and analyses of
response patterns, have been performed to confirm DIF and identify the
possible causes of DIF in some items.
Download
(
PDF) .
Comparison of latent class structure of SAT Reasoning Test Math and
Critical Reading Item Responses for Students with and without
Disabilities and across Disability Groups (paper, presented at
the annual meeting of the American Educational Research Association (AERA),
2007, in collaboration with Deanna Morgan, College Board). To
date, various approaches have been employed for assessing test
performance across certain groups. Traditional methods involve
comparison of individual item scores and total test scores across the
groups of interests, using ANOVA, MANOVA, t-test, hierarchical linear
modeling, and other univariate and multivariate statistical
techniques. However, such methods often may help to detect only
quantitative differences. An alternative approach would be to examine
the latent structures of the test responses, looking for structural
differences across groups of interest at the parameter level. This
more comprehensive analysis could help to detect important qualitative
differences, which may be a sign of compromised construct validity of
the test. This study examined both the quantitative and qualitative differences, by comparing the discrete latent structures of the test across examinees with and without disabilities, and examining the possible association between the latent ability class membership and disability status and type of disability.
Measuring Job Satisfaction among Graduate Assistants (in
collaboration with Anthony Benners, Fordham University). The
purpose of this project was to develop and validate a new instrument
for measuring various aspects of job satisfaction among the graduate
and research assistants. Project was partially funded by grants from
the Applied
Psychological Measurement/Assessment Systems, Inc. and Sodexo
Campus Services.
In
addition to theoretical research, I was involved in development of
numerous products and solutions at Mediamark Research & Intelligence, including a number of
attitudinal and behavioral segmentations and consumer typologies.
|
