News on the Applied Statistics Lab (ASL)

 A “lean” version of the ASL is up and running!  There are three faculty directors of the ASL: Arne Bathke (Director, email: arne@uky.edu), Connie Wood (Assoc. Dir. with focus on Statistics in Agricultural Research, email: cwood@email.uky.edu), and Heather Bush (Assoc. Director with focus on Statistics in Medical Research, email: heather.bush@uky.edu).  You can contact them directly (e.g., via email) to describe the statistical question or problem.  For standard questions, the problem will be referred by one of the directors to one of the (currently) four ASL graduate student research assistants under the supervision of the respective faculty director. For complex analyses the faculty directors will take a more active role directly in the analysis and, as appropriate, enlist other faculty members in the Department of Statistics and Biostatistics. In this latter approach, the hope is that the question is sufficiently novel to develop into a collaborative grant application and a sustained research collaboration. One of the consultants, Xiaofei Wang, a PhD student in Economics, will cover consulting on Stata issues. 
 

Workshop Updates!

Dec. 2nd & 3rd: Research Talk, James Gimpel, Political Scientist, University of Maryland.

○ Dec. 2, Thursday: GIS workshop from 3-4:30 (1645 POT): "New Directions in the Study of Political Geography" Demonstration of GIS software (ArcGIS, GeoDa) for a general audience.

○ Dec. 3, Friday: Research Talk from 2-4 (18th Floor POT, West End Room): "Voter Migration and the Geographic Sorting of the American Electorate"

February 18, Grant Workshop, UK Panel of Grant Experts, NSF, NIH, Other. 2-4 pm, Young Auditorium

○ Reception afterward at the Faculty Club, 4:30.

○ Panel: Tom Janoski (Sociology), Genia Toma (Martin School), Jim Ziliak (Economics), Thomas Kelly (Behavioral Science), Lawrence Gottlob (Psychology, at NSF), Janice Almasi (Education Curriculum and Instruction) Anna Secor (Geography).  Panel members have served recently on reviewer panels for various funding agencies (e.g., NSF, NIH), and will give faculty and grad students advice on what elements of grant proposals add to, or subtract from a successful review of an external grant proposal.  

November 30th (Tuesday), 4-5 pm, CB 110, Andrew Gelman (Statistics, Columbia U), “Information Visualization vs. Statistical Graphics,” Skype broadcast.

Andrew Gelman, known worldwide for his work on statistical methods with social science applications will be presenting this seminar via Skype from his office at Columbia. Andrew’s lecture should be of interest to social scientists, computer scientists, and many others. Please forward this message to anyone you think might be interested. For those of you who follow Gelman’s Blog, this should be extremely useful!  Thanks to the Statistics Dept. for arranging this! 

QIPSR Workshops, 2010-2011

Dec. 2nd & 3rd, Research Talk, James Gimpel, Political Scientist, University of Maryland.

○ Dec. 2, Thursday: GIS workshop from 3-4:30 (1645 POT): "New Directions in the Study of Political Geography" Demonstration of GIS software (ArcGIS, GeoDa) for a general audience.

○ Dec. 3, Friday: Research Talk from 2-4 (18th Floor POT, West End Room): "Voter Migration and the Geographic Sorting of the American Electorate"

February 18, Friday, Grant Workshop, "An Insider's Guide to Social Science Grants"
  2:00-4:00 pm, Whitehall Classroom Building 102
     ○ Panel: Janice Almasi (Education Curriculum and Instruction), Tom Janoski (Sociology), Genia Toma (Martin School), Jim Ziliak (Economics), Matthew Webster (Behavioral Science), Lawrence Gottlob (Psychology, NSF), Anna Secor (Geography)
   4:30 pm-, Faculty Club Reception, Conference & First Ladies Room

March 9, Wednesday, Methods Workshop, Daniel Hopkins (Political Scientist, Georgetown)
  3:00-5:00 pm, President's Room, Singletary Center
  Workshop on automated content analysis of digitized text. (Co-sponsored with Communications College)

March 10-11, Thursday-Friday, Conference on “Immigration Policy in an Anti-Immigrant Era”
  10:00 am-4:00 pm, Thursday, President's Room, Singletary Center
  10:00 am-1:00 pm, Friday, 230 New Student Center
1. Douglas Massey (Demographer, Sociologist), Princeton. Co-author of “Brokered Boundaries: Creating Immigrant Identity in Anti-Immigrant Times, 2010.”
2. Paul Sniderman (Political Scientist, Stanford). Research on attitudes toward immigrants and immigration in Italy, Netherlands, and (currently) Denmark. 
3. Michael Jones Correa (Political Scientist, Cornell), author of “Latino Lives in America: Making It Home.”
4. Daniel Hopkins (Political Scientist, Georgetown), author of several articles on how national trends influence local reactions to immigration policy.
(Co-sponsored with UK Center for Poverty Research.)

Summer Statistical Workshops
 May 17th to 19th (Tuesday to Thursday), Gatton
1. Paul Voss (Sociologist, UNC, Chapel Hill, Odum Institute) will present a 3-day workshop on spatial regression analysis.  Paul introduces R, some Bayesian modeling and a simple software package, GeoDa, as well.  Morning sessions are devoted to conceptual topics, with afternoon sessions in the Gatton computer lab. (Co-sponsored with the Tracy Farmer Institute and Statistics.)

May 23rd to 25th (Monday to Wednesday), Gatton
2. Alan Acock (Oregon State) will present a 3-day workshop on multilevel modeling using Stata that extends to longitudinal analysis.  Alan is author of "A Gentle Introduction to Stata" (the 3rd ed. includes a section on multiple imputation).   Morning sessions are devoted to conceptual topics, with afternoon sessions in the Gatton computer lab. (Co-sponsored with UK Center for Poverty Research.)

Week of June
3. J.S. Butler (Martin School) will present a 2-day workshop on programming in Stata. (Co-sponsored with the Martin School)

SEM course offered next Spring

Hongwei "Patrick" Yang (Educational Policy Studies and Evaluation) will be offering EPE 773 Section 4: Applied Structural Equation Modeling with an application in latent growth curve modeling (LGM) next spring. Free SPSS AMOS software will be used. Some basic statistics is assumed (sample statistics, sample moments, hypothesis testing, regression models) and such topics will be reviewed at the beginning of the course. Matrix algebra is not required because the course is applied.

Spring 2011: EPE 773 Section 4
Applied Structural Equation Modeling (SEM):
With an application in growth curve modeling
SEM is a highly popular technique for analyzing survey/instrument data and understanding relationships between latent and observed variables. SEM is also an important tool for analyzing longitudinal data. The course aims to help students understand and become competent users of SEM. SPSS AMOS will be used in the course.
Topics include the following:
1. Path analysis of observed variables
2. Measurement model under confirmatory factor analysis
3. General structural equation models containing both measurement & structural models
4. Growth curve modeling from the SEM perspective for analyzing longitudinal data

Finalizing Statistical Workshops, 2010-11

Based on your input and many discussions, I’ve narrowed down the list of potential topics for statistical workshops to those that would inspire the greatest interest and attendance. 3 of the first 4 listed below would be my preference. I also met with the Director of Michigan’s ICPSR summer methods program who recommended several good presenters.  I’d also like your advice on scheduling the workshops.  My own feeling is that the best opportunity for scheduling presenters and maximizing attendance would be to schedule most workshops during the early summer, well before the ICPSR in July.  The last two summers, the Martin School and Economics scheduled workshops the week of June 1 in Event Count Models (J.S. Butler), Spatial Econometrics (Chris Bollinger & Jihai Yu) and Frontier Functions (J.S. Butler).  I propose we shoot for that week again this summer, though I’m open to other recommendations.  As before, please add more ideas or comment on the suggestions below by sending me an email or leaving a comment.

• Intro to R programming and Statistical Graphs (Bill Jacoby, Political Science, MSU). Morning and afternoon sessions, 1 day. The idea would not be to get people to convert to R, but to learn the basics to be able to run the huge store of statistical programs from an R shell. Most beneficial would be graphing and multiple imputation programs, which are vastly superior to those available in, say, STATA. In my field there is emerging a huge emphasis on presenting statistical (e.g., regression) results in graphs instead of tables, but Stata and Excel are often not up to the task.
• Handling Missing Data: Multiple Imputation. (Tenko Raykov, College of Education, MSU). I think this workshop would have widespread appeal across departments and could be handled in one session. My preference would be to have the presenter show people how to use Gary King’s Amelia software, which is applicable to time-series and comparative settings. Then the presenter would show people how to combine the multiply imputed datasets in doing analysis in STATA (and SPSS?).
• Multilevel Analysis, a popular request. (Tenko Raykov, College of Education, MSU). Also, some people would like to have an extension to longitudinal, panel and growth curve models.  Question:  what software should the presenter use? HLM is popular but clunky, at best. Stata has its relative advantages over HLM in post-estimation analysis and graphing, but can be much slower than HLM for larger datasets.
• Spatial Econometrics. Paul Voss, Odum Institute (emeritus in rural sociology at the University of Wisconsin-Madison, expert in spatial analysis and spatial regression). At ICPSR, he does a week-long course, "Introduction to Spatial Regression Analysis," that he co-teaches with Katherine Curtis, that uses free software--GeoDa and R. This would build on the workshop presented by Chris Bollinger and Jihai Yu (Economics) using Stata last summer. This is likely to be more time-intensive and expensive, and we would need support from other units across campus. (Note: there is a 3-day workshop on using ArcGIS software for which there is a $700 registration fee. Our interest, however, is in spatial econometric analysis. Consult the Geospatial Science and Technology (GST) working group website for more info on ArcGIS software, which can be downloaded for free at UKIT.)
   
• Matching. This is increasingly popular in political science, but I didn’t hear a lot of requests for it outside my own department.
• Structural Equation Modeling. Requested by a few. Primarily used in Sociology and Psychology these days.  One of the consultants (Adam) in the SSTARS Center/Applied Statistics Lab is very helpful with SEM.  I think the best software is AMOS in SPSS, for which UK has a university site license, either via the discount at SSTARS Center or for free for A&S faculty and grad students.
• Computer Content Analysis. Increasingly essential in social science as a method of coding digitized text for analysis.  NSF sponsored a summer workshop and a useful website of vimeo and paper postings at the University of Washington.  This might be a good topic for next summer.  

Ideas for Annual Conferences

Add more ideas or comment on the suggestions below by sending me an email or leaving a comment.

Policy-oriented conference ideas: planning around policy themes with multi-disciplinary interest and a quantitative focus:

• Citizenship, Immigration, Migration
• Justice
• Inequality
• Health
• Education

Broad methods conference ideas (OR simply a speaker series in causal inference):

• Causal Inference in different disciplines; see example of a workshop on Research Design for Causal Inference at Northwestern, 8-2010
• Survey Research, the future of

A&S Software Licensing

  I didn’t know this, so I’m passing it along to fellow A&S faculty.  All A&S faculty, staff and graduate students can get the following software installed on our personal computers, for FREE:  SPSS, SAS, SmartDraw, JMP, MatLab, and Qualtrics Survey Software.  Among  the many products that are part of the SPSS package, AMOS, is an excellent structural equation modeling (SEM) program. For a brief description of these packages see the SSTARS Center webpage here. Otherwise, check the web.  Simply contact the A&S Help Desk (support.as.uky.edu or ASHelp@uky.edu) to get the installation package and installation key.

Applied Statistics Laboratory (ASL)

     The Statistics Department (with the cooperation of the Biostatistics Dept.) will launch the Applied Statistics Laboratory (ASL), supported by Arts & Sciences and the Office of the Vice President for Research, with Arne Bathke (arne@email.uky.edu) as its director, to provide statistical consulting for the university. The SSTARS Center (i.e., Social Science Teaching and Research Statistics) will be integrated into the ASL.  One of the consultants, Xiafei Wang, a PhD student in Economics, will cover consulting on STATA issues. I have every confidence that Xiafei knows her stats and her STATA.
     The Applied Statistics Laboratory will not provide consulting for data management, which will be handled by UKIT. Many social scientists have a need for data management services to help them archive data they have collected (e.g., with ICPSR) and post it to their website so other scholars can use it.  I’m meeting with Vince Kellen in IT to find out how social science faculty can work with IT to get help with data management.  I’ll keep you posted. 

Ideas for Statistical Workshops

1-2 days during the academic year, or up to 4 days during the summer. Add more ideas or comment on the suggestions below by sending me an email or leaving a comment.

• Handling Missing Data: Multiple Imputation
• R programming in social science research, see example
• Beyond Regression Tables: How to Effectively Present Quantitative Research A Workshop at Kansas University’s IPSR.
• Item Response Theory
• Web Survey Implementation & Introduction to Red Cap, UK’s Web Survey Software
• Multilevel Analysis
• Structural Equation Modeling
• Matching
• Spatial Analysis: Chris Bollinger and Jihai Yu (Economics) gave a 4-day Stata Workshop last summer (files are linked on Chris’s website). Is there interest in bringing someone in to give a presentation using a different software platform more specifically geared to spatial analysis, like ArcGIS or other Environmental Systems Research Institute products?
• Bayesian Statistics: An Introduction for Social Scientists
• Longitudinal Data Analysis Workshop, see example
• Dyad analysis: used in Psychology and International Relations

Ideas for Grant Workshops

Add more ideas or comment on the suggestions below by sending me an email or leaving a comment.

• Proposal development: Please come prepared with an idea you would like to develop (Project Development Office)
• Finding the right funding agency (PDO rep, A&S officer, state govt. agency, individual faculty from different social science units and fields)
• Invite NSF officers in social science divisions to UK (e.g., Sociology, Political Science, Law & Society)
• Invite representatives from the social science divisions of private foundations to UK (e.g., Pew Foundation, Ford Foundation, Robert Wood Johnson)
• NSF graduate student dissertation enhancement grants