Colloquia & Workshop Series

CORE Colloquia

The six colloquia target the university faculty members and graduate students, as well as practitioners in the field who are interested in scholarship on issues of survey research. The colloquium series addresses topics such as sampling designs, the use of surveys to measure impact, the psychological mechanisms behind formatting and wording that influence responses, and contemporary statistical analysis tools and procedures.


Is a Probability Sample Required?

Presenter: Dr. Lynne Stokes, SMU
Date/Time
: Friday, February 14, 12:00pm. - 1:00pm.
Location
: Simmons Hall, Room 144
Probability samples assure external validity, or the ability to generalize from a sample to a population of interest. But are they necessary for internal validity? Dr. Stokes discusses the difference in these being investigated for the purpose of improving non-probability sample's external validity, either at the sample selection or analysis stage. Dr. Stokes' presentation discusses these ideas in the context of the National Children's Study, a nationally representative longitudinal study of U.S. children, sponsored by NIH. NCS plans both probability and non-probability components as part of its data collection. For more information click for PowerPoint presentation.


Dynamic Panel Data Models with Irregular Spacing

Presenter: Dr. Daniel Millimet, SMU
Date/Time
: Friday, February 28, 12:00pm. - 1:00pm.
Location:
Simmons Hall, Room 144
With the increased availability of longitudinal data, dynamic panel data models have become commonplace. Moreover, the properties of various estimators of such models are well known. However, we show that these estimators breakdown when the data are irregularly spaced along the time dimension. Unfortunately, this is an increasingly frequent occurrence as many longitudinal surveys are collected at non-uniform intervals and no solution is currently available when time-varying covariates are included in that model. Dr. Millimet proposes several new estimators for dynamic panel data models when data are irregularly spaced and compare their finite sample performance to the naive application of existing estimators. Dr. Millimet's presentation illustrates the practical importance of this issue by turning to two applications on early childhood development.For more information click on PowerPoint Presentation


Design of a Probability of Food Banks in North Texas

Presenter: Doctoral Students, Department of Statistical Science, SMU
Date/Time
: Friday, March 7, 12:00pm. - 1:00pm.
Location:
Simmons Hall, Room 144
The SMU Center for Economic Studies was funded to investigate the determinants of food insecurity in North Texas. A sample of clients was to be selected from the food banks of the North Texas region. Professor Stokes's Graduate Sampling Class in Fall 2013 was asked to determine a sample design that would optimally allocate the data collection resources available for the study. This presentation describes the approach used to determine an adequate design and the results obtained for the food insecurity study. Lessons learned from this project will be discussed. For more information click on PowerPoint Presentation


Survey Sampling Weights and Item Response Modeling

Presenter: Dr. Paul Yovanoff, SMU
Date/Time
: Friday, March 28, 12:00pm. - 1:00pm.
Location:
Simmons Hall, Room 144
Measurement modeling of survey questionnaire responses is completed for various purposes, e.g. to understand underlying responses processes, to estimate latent variables, or perhaps to improve questionnaire formats. Survey respondents are often sampled such that sampling weights are used analytically for statistical representation of sampled populations. The colloquium will focus on the impact of sampling weights when modeling item response data. The procedures will be illustrated using the World Values Survey data (http://www.worldvaluessurvey.org). For more information click on PowerPoint Presentation


Analyzing Survey Item Data

Presenter: Dr. Michael Rodriguez, University of Minnesota
Date/Time
: Friday, April 11,  12:00pm. - 1:00pm.
Location:
Simmons Hall, Room 144 
Survey item response data tend to be qualitative but are not often analyzed using quantitative methods. At the item level, most survey data consist of categorical responses to Yes/No questions or rating scale questions (approve to disapprove, or never to often) which are ordinal. This requires special attention to the methods used to analyze such data - ordinal data require ordinal methods. The presentation will include a review of appropriate methods for summarizing categorical data, the use of the Chi-Square test of association, an example of a factor analysis of several items developed to comprising a single measure, and an example employing the Rasch model as a way to create a quantitative continuous scale from categorical items.For more information click on PowerPoint Presentation


Satisficing When Answering Questions: A Theoretical Explanation for a Wide Range of Findings in the Questionnaire Design Literature

Presenter: Dr. Jon Krosnick, Stanford University
Date/Time
: Friday, May 9, 12:00pm. - 1:00pm.
Location:
Dallas Hall, Room 306
During the last 100 years, numerous experiments have documented how different ways of formatting and wording a question in a questionnaire can yield different answers from respondents; remarkably, these experiments have clarified the psychological mechanisms responsible for many of these effects of question format and wording. Dr. Krosnick's presentation examines the theory of survey satisficing, which seeks to explain the psychological mechanisms at work, and reviews some of the large body of evidence consistent with this parsimonious account. Dr. Krosnick addresses the concerns of researchers who are interested in social interaction, information processing and decision-making, or maximizing reliability and validity when designing their own questionnaires. For more information click here and here