My research in survey methodology emphasizes the need to use analytical models and methods that are
consistent with the nature of the data we are analyzing, while recognizing that much of what we study are "moving targets"-that human
behaviors, opinion, values, and beliefs are dynamic and changing in modern societies. Categorical data models and mode effects require
that we take seriously the strengths and weakness of the data we use in our research. Cross-national trend surveys and my focus on the
topics of politics and religion is an effort to understand the dynamics of human populations in the 21st Century.
- Models for Categorical Data Analysis
Over the past 2-3 decades, as the computational speed of computers has grown, researchers have
developed increasingly sophisticated models for the analysis of categorical data. Categorical data-that is, data for which
averages are not meaningful (e.g., national origin, partisan identification, whether one has voted or not) -are of great concern
to survey research, since most of the data collected in surveys and polls are categorical in nature. My research
focuses on new models for analyzing and presenting such data in the modern survey setting.
- Cross-National and Trend Survey Research
Globalization has led to an increased awareness that public opinion, values,
and beliefs must be understood globally. A quarter of a century ago, most survey studies focused on the population
of only a single nation, at a single point in time. In the 21st Century, an expanding number of studies are focusing
on populations of several nations, and are attempting to chart the nature of change in these populations over time.
My research focuses on studying the nature of cross-national changes in opinions, values, and beliefs in publics worldwide.
- Mode Effects Research
One of the consistent findings in survey methodology is that the mode of data collection-that
is whether the interviews are conducted as in-person face-to-face surveys, by telephone surveys, by mail-out questionnaires,
or by internet/web surveys-may influence the responses we obtain. This reminds us that surveys are social interactions-that
we may respond differently to questions if we are asked by some one we see (face-to-face), only hear (telephone), or do not
see at all (mail-out and web surveys). Part of my research examines how the phenomenon of "mode effects" influences the nature
and quality of the research data we collect.
- Politics and Religion
Most research-even though methodologically directed-focuses on substantive topics of interest.
Most of my research has focused on the changing role of political and religious behaviors, attitudes, values, and beliefs over
the past 2-3 decades in the Americas and Europe.
- Survey Nonresponse
Response rates to surveys have been
falling in Western countries for the last few decades. Why this is
happening and the conditions under which it affects estimates made from
the survey are still unknown. My research focuses on understanding
when the likelihood of survey participation is related to the items measured
in a survey, whether correlates of survey participation change over time,
and how participation in cross-sectional surveys differs from participation
in longitudinal surveys.
- Intersection of Nonresponse and Measurement
Errors
A long standing hypothesis in the survey
literature is that people who participate in surveys are likely to give
poor quality answers. My research examines this hypothesis, drawing
on a newly developed conceptual framework for understanding the circumstances
under which response propensity and measurement error might be related.
The relationship between nonresponse propensity and measurement error
on behavioral, attitudinal, and knowledge questions are being examined
in light of this framework.
- Interviewer Effects
Interviewers in surveys have long been
known to have an effect on an individual's decision to participate in a
survey and the quality of answers that are provided. How an interviewer
acquires these behaviors is less well-understood. My research examines
how an interviewer's perceptions and behaviors change with increased experience
and over interactions with a respondent.
The overall goal of my research in survey methodology is to find ways to design surveys, both the implementation features
and questionnaires, that are respondent friendly in the sense that design features, regardless of mode, are used strategically
to encourage respondents to complete the survey and to help them better process and answer questions. While this oftentimes
means exploring strategies for optimal question and questionnaire design, it also means identifying common practices that
inhibit participation and/or processing of questions. To further complicate the picture, sometimes the lessons learned in one
mode translate quite nicely to another, but at other times the fundamental differences between modes mean that different strategies
need to be used in the different modes. In my research, I attempt to unravel this complicated tangle of design and mode effects
and to find ways to encourage survey response.
- Visual Design
With the growing popularity of the Internet as a survey mode, increasing attention has
been given to the effects of visual design on survey measurement. Research has shown that the visual design of survey
questions and questionnaires in both mail and web modes can significantly impact respondents¡¯ understanding of and ability
to process survey questions, and thus, their responses. My research in this area focuses on how various visual elements
(numbers, symbols, graphics, shapes, etc) and their properties (i.e., location, color, contrast, motion, etc.) affect the
response process. This leads me to an exploration of both how poor visual design can produce error and how good visual
design can be used strategically to minimize error and to encourage quality responses.
- Survey Mode Effects
Surveyors now have more modes available to them than ever before, but it is becoming
increasingly challenging to get respondents to participate in our surveys. To try to maintain or increase response rates,
many now find it necessary to use more than one mode to contact respondents and/or collect data, and oftentimes modes that
rely on entirely different types of communication (aural vs. visual) are used jointly. This relatively new tendency raises
the question of whether or not data collected in one mode can be combined with data collected in another for analyses.
It also raises interesting questions about how best to construct survey questions for use across modes (i.e., should the
questions be written exactly the same for both modes or should variations be allowed to account for the strengths and
weaknesses of each mode?). My research in this area explores under what conditions different survey modes yield comparable
or uncomparable data and what can be done at the design stage to minimize differences due to mode.
- Survey Implementation and Nonresponse
I am also currently working on a project that examines the effectiveness of different
implementation features on response rates. This area of research is important inasmuch as response rates are declining across
all of our major survey modes. As respondents get harder and harder to contact and to convince to participate, finding new and
innovative ways of gaining their cooperation becomes paramount.
