Meeting 21Nov13

Present: Thornton-Wells, Wang, Urbano,Tapp, Dietrich, Li

Agenda: Discuss focus and organization of new 'Biostatistics and Bioinformatics Core' in U54 proposal
  1. Tricia / Warren to try to schedule spring seminar speakers
  2. Tricia to get Core use survey from Jon Tapp / Tim Stafford - create REDCap survey with additional questions
  3. Warren - working on multiple imputation for Paul Yoder - paper "imputation with integrity" about when/how to use imputed values in analyses
  4. Warren - helping Teddi Walden's students/postdocs with data management for grants
  5. Rick - has Synthetic Derivative (SD) data on individuals with Down syndrome diagnosis
    • working with ICD9s and CPT codes with the help of a 2-month internship student based on APA guidelines for what should be monitored, what comorbidities are common
    • -->no API so must be queried manually
    • --> ICD9s can come with hierarchical/nested codes but CPT codes do not; medications cannot be queried across subjects
    • --> IRB app is in for a study on the Research Derivative (RD) which would permit use of identified data, which would could be linked to state databases, linking to parents, etc.
  6. Rick - NY Behavioral Research Institute - national survey on parents with DS - has your child had a precipitous decline in function

Meeting 17Oct13

Present: Harrell, Lambert, Thornton-Wells, Tapp, Wang, Dietrich, Li
Agenda: Discuss changes in core leadership and focus of new 'Biostatistics and Bioinformatics Core' in U54
  • New core director Tricia Thornton-Wells
    • Much experience in VKC research, genetics, informatics, imaging, neuro, neuro-psychiatric, biological basis for disorders, Alzheimers
    • More focus on leveraging synthetic derivative, BioVU
  • Need to transfer workshop coordination to VKC admin until possible Julie LoundsTaylor may rejoin the core
  • Research data management remains a challenge
  • Need to ascertain how the core will interact with the internal R grant in the U grant
  • May be beneficial to reach out to 2-3 investigators who might profit from BioVU to and help them through the process of getting information about available genotypes/data
  • See how daily clinics can help
  • Consider a survey to gauge potential use of the core for the U grant

Meeting 15Aug13

Present: Dietrich, Harrell, Wang, Ben Saville, Lambert
  • Discussed possible change in core leadership
  • Mary said we should consider childhood obesity as a theme for the U54
  • Discussed dimensions of the U54 that need to be explored, e.g. BioVU, genomics, any need for staff statistical support, role of bioinformatics
  • Warren discussed some of the EEG work he's done in the core
  • Lily discussed her projects with Blythe Corbett - use of theater in autism therapy; also sleep problems with autistic children

Meeting 18Jul13

Present: Gilbert, Tapp, Harrell
  • Jenny starting new position with Metro Schools Aug 5
  • Scott Burns presentation Monday 12:30 on REDCap APIs
  • Need to determine how observational methods are tied into new U grant core structure

Meeting, 20Jun13

Present: Gilbert, Harrell, Tapp, Lambert, Li, Wang
  • Discussed U54 planning
    • For the within-program R01 can special methodologist % effort be charged in a separate budget
  • Scott Burns will talk about REDCap APIs e.g. Python in July; Mario Davidson will talk about questionnaire design in September
    • Other ideas: SEM, ordinal response models, standard nonpar. tests and how to jump to repeated measures data
  • Discussed move of Biostat offices

Meeting 16May13

Present: Gilbert, Tapp, Lambert, Wang
  • P30 Progress report was submitted successfully
  • This summer there will be a new Kennedy Center grant application submitted
  • Workshop survey response was improved after implementing a Redcap survey to poll attendees
  • Discussed speakers for next year's quantitative workshops/seminars

Meeting 21Feb13

Present: Gilbert, Tapp, Lambert, Harrell
  • E. Dykens has prepared a dataset for the upcoming "live" workshop
  • Discussed possibly core restructure for resubmission of P grant which will become a U grant
    • Research data management
    • Pre-studios
  • Had 52 participants to last workshop - multi-level sem (Preacher)
  • Discussed ways to increase workshop survey response
  • Discussed teaching material paradigms and Github

Meeting 20Dec12

Present: Gilbert, Wang, Harrell, Li, Lambert
  • Update on workshops - Bingshan Li will present in May on latest trends in human genetics; Lambert-Key postponed until Fall 2013
  • General discussion on what methodologies we should be teaching
    • How to think
    • What methods are out there and what to look out for (e.g., repeated measures - don't use ordinary analysis; know that other tools exist)
    • Not how to solve a particular problem; but how to interpret statistical results
    • Oversimplified toy examples may not serve the audience well
    • Avoid the technical side (see Freedman, Pisani, Purves)
  • General outline
    • Measurement
      • What they mean, what you should strive for (precision, information content, noise, technical replication, inter-observer variability, bias)
      • Preserving information - categorization cost
      • Artefacts
      • Problems with measuring change, % change, relative vs. absolute; when to take ratios, logs
      • Bland-Altman plots
      • What makes a good Y?
      • Summary measures incl. odds ratio
    • Data collection and recording
    • Experimental design
    • Study design
      • What designs to avoid, e.g., pre-post
    • Analysis "why" - width over depth
      • Are there commonly used analyses that students should be able to do? Perhaps reducing serial data to summary measures and doing simple Wilcoxon-type analyses
    • Interpretation of descriptive and analysis results stats
      • Many real examples
    • Reproducible research
    • What to expect when working with a biostatistician
  • See if interactive tools would meaningfully augment this material if web-based

Meeting 15Nov12

Present: Gilbert, Wang, Tapp, Lambert, Harrell, Dietrich
  • Reported on Lily's promotion
  • "Live" data analysis session to kick off in March - analysis of Dr Dykens' cortisol data by Frank
  • Partner workshop in May 2013 by Lambert & Key
  • Should we strive for APA continuing education credits?
  • Discussed REDCap progress
  • Mary may have a module pre-packaged for skewed, small datasets as analyzed in Special Education
    • She will demo some of her materials at the next meeting
  • Discussed various short course training modalities

Meeting 20Sept12

Present: Gilbert, Lambert, Wang, Tapp

Topics: Talked about talk invitation for Dan Byrne.

From Warren: SPSS version 21 is now full featured with new no cost options.

Discussed user interface issues with different statistical packages.

Meeting 16Aug12

Present: Gilbert, Lambert, Wang, Tapp, Li * Topics: Ergonomic glasses for regular computer users; comparing multiple means * Updates: Jon is starting a paper on sequential analysis with Paul Yoder * Training: Video from Harrell and Chen Mini-Workshop now on the web; Paul Harris will present 9/4 on What's New & What's Next for REDCap

Meeting 19Jul12

Present: Gilbert, Wang, Tapp, Lambert, Dietrich
  • Issue (Tapp, Dietrich): Users of RedCap have serious problems with panel data, with no good solution in sight. Many use Excel for data entry!
  • Computer in meeting room crashed during meeting

Meeting 21Jun12

Present: Gilbert, Wang, Harrell, Tapp, Lambert
  • Issue (Tapp): stuttering research- longitudinal, observation of fidgeting - dichotomized
    • Record event start, stop, so duration is captured
    • Two observers, kappa on smallest time unit (one second); problem with non-independence of multiple spells for one subject
    • Could summarize as the total time in agreement
    • Could summarize as count of # fidgeting spells per hour
    • Or total time fidgeting - compute mean absolute difference between two observers
    • Let d vary from 1 to 10 seconds; for each d summarize the proportion of seconds in which fidgeting occurred; compute mean absolute difference between two observers' d-second proportion of time fidgeting; plot vs. d
    • If consider a single summary measure, we have an R function to do over all possible pairs of observers
  • Statistics reference library - collection of good articles we've given clients - is it worth the trouble? May be hard to keep up to date. Latest article at present is 2004.
  • Workshops booked out until Jan 2013
    • What about a "Let's look at your data" exploratory session

Meeting 17May12

Present: Lambert, Wang, Harrell, Gilbert
  • Discussed ways to convince investigators not to dichotomize variables
  • Paul Harris will present a workshop on REDCap in September
  • Partners Lecture in November with Paul Yoder and John Tapp
  • Discussed current consultations

Meeting 19Apr12

Present: Gilbert, Lambert, Tapp, Harrell, Dietrich, Li
  • Frank will be the external reviewer for the UNC stats core
  • Warren has received input from all but one person for the progress report
  • Frank and Cindy Chen are speaking on July 6 on missing data and Frank is presenting May 3 on diagnostic evaluation
  • Ideas for next academic year: Chris Preacher - SEM, mediation, moderation; Paul Harris - REDCap; Partners lecture e.g. Warren and client, Sasha and Natalie Miater; Paul Yoder and Jon - observational measurements; Jonathan Schildcrout - repeated measures with categorical outcomes
  • Upcoming Stata course - Jeffrey Blume
  • Shifts in % efforts - Lily down, Warren up

Meeting 15Mar12

Present: Gilbert, Wang, Harrell, Tapp, Lambert
  • Lily is giving a training workshop Mar 26
  • Cindy Chen and Frank are giving a 1/2 day workshop on missing data on July 6
  • VKC will be co-sponsoring a Bayesian workshop
  • Another workshop is happening - on Stata
  • De-briefing from external consultant for Core C Peg Burchinal
    • Many positives
    • Main area for improvement is documenting the impact/productivity of Core C by scientific area/ typesof services rather than by investigator/manuscript
      • Would be a good idea to get a sample from another institution that was well done
    • We are not clear on the comment about statistician time conflicts when investigators are in greatest need
      • Perhaps the comment related more to who in the core can accept funding from derivative grants; there is frequently no grant to drop at the moment of need
      • Need to see if there is flexibility in temporarily lowering the percent effort in the core

Meeting 16Feb12

Present: Gilbert, Wang, Harrell, Tapp, Urbano, Lambert, special guest Sonya Sterba
  • Went over preparation for expert consultant visit
  • Sonya is in Psych & Human Development with interest in SEM, latent variable models, longitudinal
  • Sonya gave a workshop on power in SEM models
  • Starting in the fall is a new quantitative forum
  • Discussed opportunities for interaction
  • Discussed clinics and other resources and new github/wiki resources
  • Jennifer raised the possibility of having workshops in the summer; need suggestions for speakers
  • International R conference at Vanderbilt

Meeting 19Jan12

Present Wang, Lambert, Urbano, Tapp

* Discussion of consultant visit

Brainstorming ideas of activity during visit...
  • Each consultant meets individually with consultant and explains what they do in this role and respond to questions from the consultant
  • Have consultant review last VU P30 grant proposal section
  • Frank and director (Dr. Dykens) discuss overall strategy of the core and operations
  • Consultant to provide P30 application text describing her operations and procedures as an example
  • Discuss consultant's views of the continuing future of P30 MRDD Center grants at NIH

Most Recent Consults

Jennifer Gilbert, VKC Core C Educational Director: As the educational director, I coordinate seminars with the purpose of educating researchers and their staff on various statistical or methodological techniques. The most recent seminar was on Jan. 31. Dr. Sun-Joo Cho from the Quantitative Methods & Evaluation Program at Peabody and Dr. Jennifer Gilbert of the Department of Special Education presented an hour-long introductory talk on "What item response models can do". There were 52 individuals in attendance representing 19 departments. In addition to seminar coordination, I help KC investigators and their students manage data and apply appropriate models to answer their research questions; this sometimes includes walking individuals through building models and/or interpreting results. Most recently, a research associate and postdoctoral fellow needed help building and interpreting growth models that were associated with a parent-training efficacy study.

Warren Lambert, Core C Behavioral Coordinator, logged 3 consults on January 19. KC members & investigators are underlined:

1.Hands-on analysis and presentation of results to team for Len Bickman’s project on military PTSD (3 hrs).Starting July 1, 2011 Bickman began paying 20% of Lambert’s salary after a number of free consults from Core C.Team needs statistician to do analysis for planned articles.

2. Analytic plan and power analysis on Nathalie Maitre’s grant application using ERP to detect cognitive development of infants in neonatal intensive care (3.5 hrs).In January 2012 an ERP paper by Key, Lambert, Aschner & Maitre was accepted by Psychophysiology. Lambert is a clinical psychologist learning about ERP, not an expert on its analysis.

3. Lambert chaired the Core C meeting on Jan 19 (1 hr), filling in for Frank Harrell who had to travel.

Lily Wang

My latest collaboration is with Dr Blythe Corbett. The title of our manuscript is "Peers, play and performance for enhancing social interactions in youth with autism spectrum disorders". In this paper, we evaluated social interaction skills and stress responsivity in children with autism spectrum disorders (ASD) before and after a novel two-week intervention using behavioral and theatrical approaches.

Jon Tapp Tapp most-recent consult was with Erik Carter, a new Kennedy Center investigator, to specify equipment and create a coding scheme for observation and analysis using the MOOSES system for Erik’s observation project related to educational and transition services for adolescents with intellectual disabilities. Addition Jon helped with interpretation of inter-rater agreement, logistical issues, and file management methods for the project.

Most Recent Consult (Dietrich)

* Resulted in a manuscript currently under review entitled:

Measuring Progress toward Achieving Comprehensive Family Centered Care for Children with Special Health Care Needs: A Comparison of Three Samples

* Abstract:

Over 20% of households in the US care for a child with special health care needs. Much effort has been directed to developing care systems that meet core objectives of the US Maternal Child Health Bureau (MCHB) for CSHCN. Better understanding of the challenges these families face is needed, particularly those already receiving services from state and non-profit agencies focused on these children’s needs. This study assessed experiences of families (N=816) with a CSHCN receiving services. Their perceptions were compared to data from the national sample (N=40,840) and TN subsample (N=794) of the NS-CSHCN 2005-2007 to determine the degree of achievement of the core MCHB objectives. Family challenges reported centered on time, finances, and family interactions. Comparisons indicate similar perceptions of family satisfaction with services but differences related to coordination of care. Findings highlight successes and places for improvement in providing optimal health care services for this special population.

* My roles:
  • Helped them map the questions in their study (families receiving services) to the survey questions included in the NS-CSHCN survey.
  • Facilitated downloading and working with the NS-CSHCN so that we generated actual statistical summaries from the original data (rather than using online interactive survey aggregate tools).
  • Consulted with them on some methods that could be used to test for the statistical significance of differences observed among the summary stats in each of the respective sample.
  • Collaborated with the PI and other co-authors on the manuscript. *

Meeting 17Nov11

Present: Wang, Gilbert, Harrell, Li, Urbano, Tapp, Lambert
  • Rick Urbano reported
    • Dev disabilities training program renewed
    • Multiple research databases moved to central server and encrypted
  • External review now scheduled for 21Feb12
  • Jennifer reported that today's workshop was well attended (Sterba: Power Analysis for Latent Variable Models)
    • Sonya would be interested in coming to one of our future meetings
  • Discussed briefly the tracking issue
    • Ask Core A to give us a listing of grant proposals we've been involved in
    • Would also be good to get a list of "big" VKC investigators
  • Need more ideas for sessions for June's useR conference at VU
  • Discussed respect for methodologists
    • Think about a roundtable discussion on respect for each other's disciplines

Meeting 20Oct11

Present: Lambert, Tapp, Li, Harrell, Gilbert
  • Outside reviewer Margaret Burchinel still being scheduled; will be after the holidays
  • IRT talk in January needs feedback on basics vs. more complex; may start with more basics
  • Nov 17 talk is set
  • Jon will start using a help desk ticket system for observational data consults
  • Need to keep track consults
  • How can we contact VKC investigators to find out about current research about which we can be more involved?
    • Many opportunities for improving data quality/management
    • What about better advertising REDCap clinics?
    • Talk to ED about whether certain data management aspects should be mandated
  • Jon raised an interesting question about testing his observational software

Meeting 21Jul11

Present: Dietrich, Lambert, Harrell, Li, Gilbert, Wang, Tapp
  • Discussed Biostatistics plan for requiring pre-specified analysis plan
  • Jennifer reported that Sasha Keys will be presenting on ERP on Sep 8; Sonya Sterba Nov 17 on latent class analysis; Shun-Joo Cho and Jennifer will be presenting on item response analysis Jan 31
  • Frank talked about March of Dimes proposal
  • Warren is working on a paper with Sasha Keys
  • Need to get the word out more about daily clinics

Meeting 16Jun11

Present: Taylor, Gilbert, Harrell, Wang, Lambert, Tapp
  • Jennifer Gilbert has taken over as organizer of the workshop series
  • Jon Tapp will be doing more web development
    • Discussed R graphics for the web, e.g. using ggplot2; can also use base graphics and lattice, RStudio
    • New model proposed for "simple as wiki" R interface; goal is for power calculators etc.
    • Also see Rook (Jeffrey Horner's new approach)
    • REDCap backend
  • Julie's K award is a reality
  • Julie has given Jennifer a list of workshop suggestions - have presenters for about 5 workshops; need one for May
    • Possible new topics: nonlinear effects (e.g., splines); design issues such as intent-to-treat, could tie in to compliance
    • Lily could give one on GWAS - gene pathway analysis; Lily had given one on pathway analysis for microarrays already
    • Interrupted time series; diagnostic research and measuring utility of diagnostic tests
  • More demand for quant psych PhDs
  • Biostat MS and PhD program starts in the fall - opportunities for those getting PhDs in other fields to get MS in biostat while doing so
  • Margaret Birchinal (quant psych; UC Davis) will be the new core external adviser

Meeting 19May11

Present: Taylor, Harrell, Lambert, Dietrich, Wang, guest Jim Steiger (Quant Psych)
  • Jim talked about R and simulations in teaching
  • Discussed value of teaching ANOVA not using regression
  • Jim is experimenting with Demo Builder software package for creating flash movies (vs. Camtasia)
  • Gave examples of what can go wrong in unprincipled statistical analysis

Meeting 21April11

Present: Li, Harrell, Lambert, Wang
  • Warren turned in noncompetitive Core C P30 chapter, so we're done. Thanks to all for providing information.
  • Julie Taylor has very likely won a K-Award. Congratulations Julie!
  • Warren will chat again with Nina Martin
  • Still need external stat consultant
  • Tony Maupin is an A/V specialist at VKC who may assist in video recording of workshops; we need to develop modern course development skills
  • Took a first look at RStudio - the manipulate package that comes with it has some interesting possibilities for teaching through the use of interactive graphics

Meeting 17Mar11

Present: Dietrich, Harrell, Lambert, Taylor, Wang
  • Discussed P30 competing renewal
    • Each core member needs to update activities, from 2010-present
    • Papers, training, consultations, software
  • Training
    • May want to contact Nina Martin who was a doctoral student with Singer & Willett; has material on longitudinal data analysis

Meeting 17Feb11

Present: Dietrich, Lambert, Harrell, Wang, Li
  • P30 yearly report to be submitted in May.
    • Warren will send data each of us needs to update.
    • Need to get a new external consultant; one possibility is the one biostatistician who heads a dev. core outside of VU.
    • The competitive renewal of the P30 should be drafted by May 2012.
  • We may consider pushing for professional data managers to increase overall data quality.
  • Pre-Biostatistics course: take a look at the book Designing Clinical Research (byStephen B Hulley).
  • Mary discussed her online courseware. It's time to make statistical teaching material available on WWW.
  • See also the BBC production Joy of Stats.
  • Kennedy Center Core A has an AV expert we can consult: tony.maupin@vanderbilt.edu
  • UCLA has "screen movies" with voice: http://www.ats.ucla.edu/stat/stata/notes_old/movies/IntroStata1.html.
  • UCLA has a Camtesia instructional movie on how to make movies: http://www.ats.ucla.edu/stat/seminars/MakingMovies/default.htm

Meeting 19Nov10

Present: Dietrich, Taylor, Harrell, Lambert, Urbano, Tomarken
  • Frank updated on the biostat grad program
  • Sonya Sterba will attend next month
  • There have been discussion about impact of VKC, and publication indexes
  • Eva Pantaleoni is doing a stats core presentation on GIS/pollution/traffic and death and Down's syndrome
    • Would be useful to link up with other air pollution researchers dealing with complexities of TN monitoring station data
  • Discussed the Bayesian statistics workshop
  • Discussed pressures to torture data with a multiplicity of analyses
    • Department of BIostatistics is finalizing a policy on pre-specified statistical analysis plans
  • Being aware of severe design flaws before analysis commences
  • Challenges of analyzing symptomatic improvement when there is a floor effect
  • Possibility of launching a short course/workshop on "Pre-Statistics" - study design and measurement; descriptives

Meeting 15Oct10

Present: Tomarken, Taylor, Wang, Li, Harrell, Lambert, Tapp, Dietrich, Jennie Gilbert
  • Jenny Gilbert joined us - a recent grad under Don Compton specializing in late emerging reading disabilities
  • Research with Don Compton, Doug & Lynn Fuchs
  • Designing & evaluating reading research, now extending to math problems in addition to reading
  • Issue of forming 3 overlapping groups (reading, math, both)
  • Are kids with rdmd difference with those with rd or md - working memory, underlying cognitive processes
  • General approach: Is there an interaction between degree of md and degree of rd in predicting e.g. working memory or attention?
  • Easiest to do separate analysis of the various cognitive features
  • Could cluster left hand side variables - simplest result would be that all but one LHS variable is redundant given the remaining one
  • Full multivariate analysis with canonical variates would not impose meaningful structure on the LHS
    • seemingly unrelated regression may have something to offer
  • Could compare univariate results across response variables
  • Would also be reasonable to flip LHS and RHS; would allow interactions in cognitive functions
  • How do we better bring theory into the design of the analysis?
  • If there is interest in a single cognitive function measure on the LHS, can fit a flexible nonadditive model:
    • attention = f(reading) + g(math) + h(reading, math)
    • f = p basis functions (e.g., regression spline or linear and quadratic)
    • similar for g; tensor spline for h with all cross-products of basis functions representing f and g
  • Warren discussed core director's meeting - need for publication metric
    • also discussed meta-analysis of research publications to tease out impact of VKC type centers

Meeting 24Sep10

Present: Taylor, Wang, Urbano, Lambert
  • Warren has an investigator who wants to submit for CTSA funds to pay him to do her analyses. She is affiliated with the KC and we discussed whether the CTSA would fund this biostats support, given that she is a KC investigator. We decided this is something that Frank will be able to answer.
  • Julie talked about the upcoming 2.5 day Bayes workshop. Everything is in place and people are enthusiastic.

Meeting 17Aug10

Present: Taylor, Harrell, Tomarken, Wang, Dietrich, Tapp, Urbano
  • Update on Core Directors' Meeting
    • Course in biology for statisticians
    • Workshop on meta-analysis methods
  • Item response theory, contrast with standard test theory, factor analysis - Sun-Joo Cho ; also could include a bit about questionnaire design
  • Warren and Kathleen Lane are doing a general survey design presentation in March 2011
  • More about reproducible research/research integrity
  • Propensity analysis for lay people
  • Andy will be giving a psych lunch seminar on analysis of count data
  • Geographical information systems and spatial analysis methods
  • Handling clustering; what about correlations across animals?
  • Lily gave a talk on GWAS and pathway analysis which may be good for VKC
  • Julie has 2 workshops left to schedule this calendar year: possibilities: count data, meta-analysis (Mark Lipsey - policy, program evaluation, psych); see Scott Maxwell in Psych Meth
  • ERP: Psych in A&S: Geoff Woodman, Sean Polyn; possible collaborators
  • Carissa Cascio has submitted a K award grant related to ERP training
  • Sar Peters; both may be working with Sasha Keys
  • Interest in bringing in a national speaker; need groups to share expenses

Meeting 25Jun10

Present: Taylor, Harrell, Lambert, Xu, Tomarken
  • Tracking was discussed at last core directors' meeting
  • Discussed the last methods workshop, about data collection/quality
  • Discussed REDCap for animal and other data
  • Discussed ideas for training - Julie has sent out an e-mail asking for input
    • Bayesian statistics - might merge with Chris Fonnesbeck
    • Regression discontinuity analysis - contact Dave Cordray and Patrick Arbogast
    • Partner to bring in outside speakers; one idea is Singer (longitudinal)
    • Need for alternatives to repeated measures categorical data (possibly Andy and Jonathan Schildcrout)
  • Be thinking strategically about whether more of our effort should be diverted to training
    • Psychiatry is getting more interest in invigorating training in research methodology

Meeting 28May10

Present: Harrell, Lambert, Tapp
  • Discussion of quant core workshop on data quality
  • Warren brought up Don Guthrie's comments about activity logging plus items from the Deutsch paper
  • Need to finalize thinking on the purpose of gathering the data, e.g., easier to write progress reports, improving quality of progress reports, staffing planning, core director understanding of activities

Meeting 23Apr10

Present: Harrell, Li, Taylor, Wang, Lambert, Tapp, Dietrich
  • Rant about Microsoft
  • Lei Xu sent the following update: Sasha, Warren and I still keep the monthly meeting. Sasha gave me the raw data and We plan to run the ICA analysis on the raw data using EEGlab. Right now I have difficulty transforming the raw data into some format that I can read it into R and EEGlab. If I couldn't figure it out I probably will try the massive univariate approach first using the average data. At the same time I am running simulations to test the false positives of the ICA approach.
  • Discussion of tracking consults
    • Tracking ideas being discussed in CTSA
    • Need for conistency of tracking across consultants
    • Difficulty of determining which investigators are covered under the P30
    • Consider having a standard form to be filled out at the start of a consult
    • Will try a paper system for now, with information on the P30-mentioned grant this request pertains to
  • Julie asked about a challenge grant and need to be proactive in data collection design and data management at the beginning

Meeting 26Mar10

Present: Tapp, Lambert, Urbano, Lounds-Taylor, Tomarken
  • Warren Lambert announcement that P30 progress report is due soon. Consultants will need to provide data on services provided and any generative items such and published article references to be included.
  • Tapp reports that investigator satisfaction survey will be sent to investigators soon for core service ratings reports. Listing of service provided have been sent out for editing.
  • Brainstormed about speakers for the next year quantitative lecture series.

Meeting 26Feb10

Present: Li, Tapp, Lambert, Xu, Wang, Harrell, Tomarken, special guest Sasha Key
  • Lei Xu and Sasha Key presented their collaboration on independent component analysis (ICA) in ERP EEG analysis
  • Lei showed the performance of the method and its software when the data consist of nothing but noise, raising concerns about the validity of the method

Items Sent by Julie Taylor

Seminars

  • The March seminar is Ana Regina Anarde and Katie Smith presenting their work on data management. I think this is going to be a really nice topic for Kennedy Center researchers and I am going to try to push it as much as possible.
  • My May seminar just canceled, so I am looking for a speaker. There is some interest at the KC in learning about bioVU so I might see if Lou Muglia is available then (or can send somebody). Although its a bit of a stretch, I think learning about bioVU fits into the "methodology" part of the lecture series.
  • I am also going to advertise the upcoming KC ethics seminar about statistics and publication to the quant listserve.
  • I am planning on approaching Theresa Scott to present an advanced R workshop over the summer. She has expressed some interest in doing this in the past, and there is certainly interest from the KC end.
  • I would also like to think about bringing in an outside statistician for next year (May 2011?) who can give a series of talks like Dr. Kraemer. Maybe we can talk about that at our March meeting.

Consulting

  • My consulting in the past few months has involved fairly basic issues. Nothing too complicated with a lot of people stopping in for one-time consultations on relatively minor questions.
  • I manned the KC Science Day poster - there was some interest and I was able to talk about our core services and pass out a few cards.

Meeting 22Jan10

Present: Li, Harrell,Xu
  • Work has started with Lei, Lily, and Matt (a recently graduated graduate student)
  • Decided to postpone Lei's presentation until Sasha and Warren could attend

Meeting 18Dec09

Present: Tomarken, Wang, Harrell, Li, Lambert, Taylor
  • Discussed upcoming panel discussion on research ethics in January involving Frank, Elisabeth Dykens, and Karoly Mirnics
    • difficulty with the amount of information that review committees need to process
    • publication bias towards 'significant' results
    • how do we get across the issue that it saves time to do things right the first time
      • .... if you get caught
    • reviewers appreciate authors' honesty about limitations
      • overselling a point may cause a long revision cycle
    • need examples of egregious and gray zone examples
      • single most frequent problem is filtering out 'negative' results; multiplicity issues
      • many researchers have difficulty tolerating ambiguity
      • examples from "The Tipping Point" by Gladwell; need to simplify, e.g., human traits
    • what is wrong with studying 6 mice?
    • many power calculations emphasize one-at-a-time feature analysis
    • when is it better to not use $ rather than use it when the study is grossly underpowered
    • need more thought to two-phase strategies (screening -> validation)
    • how to deal with journal space limitations
    • where do methodologists draw the line? How good is good enough?
    • selection of methods on general principles
  • Briefly discussed upcoming Key-Xu-Lambert presentation this forum
  • Chun pointed out a new article by Keith Baggerly - see Cancer Letter - where a major flaw was found related to a data shift

Meeting 20Nov09

Present: Lambert, Tapp, Harrell, Dietrich, Li, Xu, Taylor
  • Discussed software such as R in SPSS, R for Excel
    • R for Excel is giving new users an easier introduction to R
    • SPSS, Stata, and SAS all have R interfaces
  • Jon is re-writing the observation system software from scratch
  • Potential consult from a neurosci grad student
    • Need to see if qualifies for support from the core
    • Could have him come back to a future Friday clinic when a couple of us can attend
  • Discussed Andy Tomarken's idea to present current consults during our meetings, ideally with the investigator present
    • Lei Xu and Sasha Key will likely make the first presentation, for 30-45 min. in our January meeting
    • A future possibility is Jon demonstrating event data collection introduced by a research problem
  • Announced Chun Li's promotion

Meeting 23Oct9

Present: Wang, XU, Tapp, Lambert, Dietrich, Taylor, Tomarkin
  • Discussed using redcap for logging consults and logging in general
  • Wished Frank as speedy recovery

Meeting 18Sep9

Present: Wang, Xu, Tapp, Lambert,Tomarken, Taylor, Urbano
  • Logged in a figured out Wiki editor
  • Discussed two books 'R through Excel' and 'ggplot2' Warren brought
  • Discussed using RedCap for logging. Tapp will explore initial setup so he can sleep better.
  • Seminars are set for semester
  • Harrell seminar was well reviewed, has generated some consults
  • Tapp wrote a macro for matching spectral output and Lily is helping with the comparison stats on the resulting output
  • Urban reported on P30 supplement success to support efforts or Cole and Jeremy on 'linkage' project

Meeting 28Aug9

Present: Taylor, Harrell, Lambert
  • Discussed ideas for the upcoming nonparametric statistics seminar

Meeting 24Jul09

Present: Wang, Dietrich, Harrell, Urbano, Tapp, Lambert, Tomarken
  • Frank raised the issue of whether the core should support an investigator who is not a KC member (but whose mentor is)
    • Investigator should apply to be a KC member to get services
  • Core members reported on current activities
    • Warren: various projects with students and Pediatrics; wrote first responses to P30 critique
    • Jon: a methodology book is coming out packaged with his software; has been thinking about logging consultant activities
      • Tim had an issue with using the google e-mail keyword tagging approach
      • Jon will try another approach
      • DIscussed data encryption issue
    • Rick: submitted CDC development epidemiology grant related to congenital heart defects and Down syndrome; supplement to P30 to continue to develop database linkage software
    • Mary: working mainly with 3 SON faculty
    • Lily: brain gene expression and physical activity; Neurology supplemental app - metabolites rel. to melatonin in autism; grant submitted to DOD on genetics of insomnia for children with autism; methods paper related to genetics association and pathway assessment; discussed multinomial SNP problem
    • Andy: several projects underway, e.g. a student of Paul Yoder's on effect size in a matched pairs design (particular problem with confidence intervals); consulting problems are leading to methods papers such as one related to context insensitivity - group x stimulus interaction replaced with a more focused hypothesis; cfa, different models for count data

Meeting 26Jun09

Present: Dietrich, Harrell, Tomarken
  • Discussed repeated measurement analysis roadmap
  • Discussed a multi-level factor analysis problem

Meeting 22May09

Present: Tapp, Harrell, Wang, Lambert, Tomarken
  • Priority of score 129
  • Discussion of pink sheets from review of P30 renewal application
    • Isn't Lei funded on the grant?
      • Should we move some of Lily's funding?
    • Methods publications
      • Need to pay special attention to compiling lists of methods research that is most relevant to P30
    • % effort for Frank
  • Task tracking/documenting services
    • Only Warren is using the Work Logger (worklog2)
    • It would be good for Frank to know more about the current areas of research being emphasized, and to have information related to load balancing
    • At what level do we want to collect information (down to "aha" moments?)
    • RedCap vs. wiki vs. event recorder

Meeting 22Apr09

Present: Tomarken, Dietrich, Urbano, Harrell, Wang, Xu, Tapp, Lambert
  • Meeting planning going well for Dr Kraemer
  • CTSA supplement for study monitoring
  • Discussed developmental epi data sources
  • Better defining VKC investigators qualifying for core services
  • P30 application is highly likely to be funded
  • Andy mentioned some new ERP research opportunities that may be of interest to Lei
  • Mary has been working on constructing an online course in stat (2nd semester PhD in Nursing)
  • Need to improve RedCap handling of longitudinal data

Administrative Support Link

Things to do

  • All: Read Deutsch article suggested by Frank
  • JB: get approval for imaging conference
  • JB: plan graphics 2-day series
  • JB: plan versioning/reproducible results series for Fall
  • WL: Get P30 application to Frank by EOD TH AP 12
  • WL: Work with Claudia C on announcements + brochure on clinics

Meeting 29Jan07

Present: Warren Lambert, Lily Wang, Chun Li, Jenni Blackford, Jon Tapp, Rick Urbano, Frank Harrell

Training

  • Almost every month, internal consultants, invitees from other departments, grad students
    • Attendance excellent; around 240 unique persons
    • Have emphasized genetics, microarray
    • Upcoming one on psychometrics by Warren, graphics by Frank
    • Generated interests persisting beyond the session
    • No survey of attendees at present
  • Interest in short course on imaging
  • Conference on complex data analysis last summer
  • Other areas for wish list: Epi stats, GIS, gen epi
  • May be good to emphasize training that changes practice or gives investigators ideas for new directions
  • Keep a lookout for opportunities for consultants to jointly give talks with investigators regarding cutting edge work, e.g., Chun Li and Jim Sutcliffe
  • Master lectures: need to start this; could have biostat lecturers give another lecture to VKC
  • Biostat weekly seminars on lookout for good outside speakers to invite
  • Any VKC member who thinks of biostat topic of interest can send the request to Frank to send to all biostat faculty asking for volunteers
  • Need basics of data entry (e.g. Dan Byrne)

Recommended Reading and other mass mailings

  • Warren has e-mailed a few interesting readings
  • Occasionally it may be a good idea to remind investigators about the consultants; putting out a brochure may be a good idea
  • Web site has been updated
  • Regarding biostat clinics, add message to Monday morning message (if doesn't get lost) and a link from the web page; can pass out flyer at seminars
    • Need to determine barriers to clinic participation so that investigators find coming to clinic less unnerving
    • Can be confusing when consultants disagree
  • Article finder: Yanna and Jessica worked on this; could link to references from biostat author checklist; send suggestions for new articles for biostat site to Frank
  • Jenni can send a sample query to Frank; look at http://kc.vanderbilt.edu/articledb

Quality Control

  • VKC investigators asked to rate stats core personnel once per year, by type of consultation
  • Think about quick post-consultation surveys, possibly with a reward, e.g., coffee shop gift certificate
  • Tracking of publications may be important; need to link with core

Dealing with Unique Requests

  • E.g., factor analysis of ERPs which may require some simulation; Jon may be willing to help with some of it
    • Need to find out how to get volunteers who may be interested in collaborating on longer-term projects

Worklog

  • Jon and Warren are redoing, esp. with regard to fee for service
  • Jon is thinking about adding a ticketing capability; gmail gtd (getting things done) may be worth examining as a model
  • Need to capture more info about purpose and fulfillment of consult
  • Need to track usage by area (P30 is targeted to developmental research) but don't want to be exclusionary
  • Staffing seems to be about right at present

Future Meetings

  • Monthly
  • Brief reports of current activities and challenges including methodologic and educational issues
  • Discuss how to deal with senior investigators who want to do methodologically unsound things

Meeting 26Feb07

Present: Warren Lambert, Lily Wang, Chun Li, Jenni Blackford, Jon Tapp, Rick Urbano, Andy Tomarkin, Frank Harrell

Documenting Work | List

  • Document generative work
    • Plus how did we advance the scientific or methodologic field; innovation
  • E-mail users to gather data
  • Get info from clients on 3/1, 10/1
  • Get paragraphs, citations
  • Need more categories for work log
    • design
    • analysis planning meeting
    • data analysis and programming
    • methods research * May be best to wait about a week and ask the client to go to an anonymous web survey site
    • survey could also ask whether questions were not answered and prompt a repeat visit
  • People are not being required to use work logger; concern was expressed about doing this in midyear
  • Recommend continue to use work logger through end of June 2007; by end of March a dry run of the progress report should be done to determine if there are large gaps in data
  • Each consultant should keep track of generative consultations. Consultant should ask questions (to ourselves and to clients) related to
    • how did the consultant change the nature or direction of the project or how things are measured
    • unique contributions
    • new methodologic advances
    • co-authorship

Improving Quality of Data Analysis Post-Consultation

Biostat Clinic

Global P30 Issues

  • Should not be collaborators (co-authors) on 100% of projects

Planning for Next Cores Meeting

  • Present short paragraphs with generative meeting examples
    • can pull from last P30 application
  • Enhance general core interactions
    • John Allison - sample size justifications routine
    • Ray, Denise - check that investigators have contacted core C
    • Develop checklist for basic scientists

Meeting 26Mar07

Present: Warren Lambert, Lily Wang, Jenni Blackford, Jon Tapp, Rick Urbano, Frank Harrell

Items from 26Mar07 Core Directors Meeting

  • Warren and Frank will meet with service coordinators to discuss what a consult means and when to get it
  • Weekly clinic announcements can go in the Monday Morning Message
    • Need to rotate message (clinic days) and perhaps have them once/month
    • Frank needs to go to dept. faculty meetings
  • For the non-competing renewal we need to wait for the pink sheets in order to respond
    • Update training list, pubs, usage stats, grants, generative examples, including top 10 peds article
  • Follow-up of data analysis quality and correct interpretation of our advice
    • Reproducible analysis
      • Need to educate researchers about this, e.g., when teaching SPSS
      • Version control is also important
      • Need to develop a seminar for VKC: 3-part series. Start with moving people away from Excel into a scriptable system such as SPSS, finish with R and LaTeX and R and OpenOffice and introduction to version control; get across the point about whether a local machine is the logical place to hold data and code vs. a server
    • Think about a randomized trial in which an MS data analyst is assigned to 1/5 of the consults and outcomes are compared
    • Investigators should keep records of analysis decisions, options, etc., in a lab notebook
  • Worklogger is still being used in other cores but not the time component
  • Each consultant should send out an e-mail every 4 months to their clients asking for updates regarding any publications arising from the consultation

Education During Consultation

Deutsch R, Hurwitz S, Janosky J, Oster R: The role of education in biostatistical consulting. Stat in Med 26 :709-720; 2007.
  • biostatistical education;consulting;statistical consultant;teaching;clinics;"both the need and the opportunity exist for specialized biostatistical instruction during one-on-one sessions between a consulting biostatistician and physicians, medical students, and research staff. Academic researchers are ideally positioned to absorb this kind of training when they initiate a request for assistance with their own research project."

Meeting 30Apr07

Present: Warren Lambert, Lily Wang, Jon Tapp, Frank Harrell, Andy Tomarkin
  • Jon presented a consult request for an image analysis related to auditory neural circuitry
  • Discussion of Deutsch paper
    • The paper pushed several things we're already doing but it's nice to have this as a reference
    • Time requirement is our enemy, especially for complex problems
    • Extreme consulting has a role - clients see how we work and don't think of us as button-pushers
    • Need follow-up survey of how our suggestions are followed
    • Warren has asked for their survey forms to see if should be incorporated into worklogger
    • Outcomes should be measureable but can the client generalize what she's learned to the next (similar) problem * We might consider a dataset deposit web page
  • Feasibility of a summer conference on complex data analysis, perhaps to coordinate with the "biomarker summit" suggested by research support services
    • How far-ranging can the topics be?
    • What about an informal analysis challenge to a panel of quantitative experts?
    • Review Rubin's work on making observational data analysis more scientific
    • The funding and promotion system is part of the problem
  • Jon, Warren, and Frank should meet with Tim to unify thinking on revisions to work logger

Meeting 25Jun07

Present: Warren Lambert, Chun Li, Jon Tapp, Frank Harrell, Andy Tomarkin
  • Need feedback on graphics course
  • Talked about planning a reproducible analysis workshop
  • Progress being made on work logger
  • Discussed the common situation of having to analyze longitudinal, correlated data and difference between random effects and generalized least squares approach
    • Andy talked about the distinction between repeated measures and serial data and the need to know the correlation pattern induced in Y
    • Random slopes and intercepts and induce a strange heteroscedastic model
    • Random effects may not be normally distributed
  • Would it be worth writing guidelines for which serial data methods to use in which situations?
    • How to take into account the feasibility of the client doing the analysis?
  • What book(s) should we recommend? Singer and Willet is a good one, so is Charles Davis, Verbeke and Molenberghs, and Diggle, Liang, Zeger
  • Discusssed analytic plans for grant proposals

Meeting 30Jul07

Present: Warren Lambert, Chun Li, Frank Harrell, Jenni Blackford
  • Graphics course evaluations and in and will be summarized
  • Does the core have funds we can access? Warren will request information.
  • Think about a reproducible analysis workshop in the fall
    • Might have an intro lecture on repro research in general
    • Present what is wrong with the spreadsheet model
    • Present how to comment commands
    • Then start with the basics of command-saving in SPSS/Stata
    • Finish up with Sweave
  • Chun will present on interactions at the Stats and Methodology seminar series

Meeting 24Sep07

Present: Warren Lambert, Chun Li, Frank Harrell, Jenni Blackford, Andy Tomarkin, Jon Tapp, Lily Wang, Richard Urbano
  • Jenni reported on training efforts
    • training this past summer got good evaluations; several persons have been able to use Terri's guides
    • 3-day Imaging workshop evaluations being analyzed; anecdotal comments very positive
    • DVDs are offered, as well as slides offline
    • Chun Li kicked off fall training on interaction
    • KC Communications Core met; going to every other month schedule from monthly
    • Jenni will be doing an SPSS introduction to scripting and reproducible results in January; to be followed by Terri in May - reproducible research using R
    • Warren will be doing a primer on repeated measures analysis in March
    • We would be interested in Terri Scott giving more in-depth workshops on R/Sweave
    • Andy is interested in doing a workshop next fall
    • We'll try to get Tony Brown to give a talk on missing data
  • Warren reported on the core budgets and available funds for core personnel
    • KC will be offering full-time members $1000/yr (prorated for part-time) for professional meetings etc.; need to find out if it pertains to non-college personnel
    • P30 budget may have some funds other than for salaries effective 1Jul07 (core is now funded as long as other cores)
  • Jon is working on a a new work logger application; some metrics still need to be decided
  • VKC investigators vary in their ability to collect and enter data well; for those involved with the GCRC there is a system that is easy to use; investigators define their own data dictionary. We should also consider having a presentation on Rhubarb.
  • There is an opening for a database administrator at VKC
  • Frank reported on the R Data Deposit resource being developed in Biostatistics
  • Various database/web resources/study volunteer databases are being developed at VKC
  • At upcoming VKC planning retreat in October we might talk about collaborations with other centers and resources for travel/communication related to this; could also discuss enhancements to between-core collaborations
    • Jon will try to document which other centers are using his software
  • Need to re-invigorate the effort to get VKC investigators to come to Biostat clinics

Meeting 29Oct07

Present: Warren Lambert, Chun Li, Frank Harrell, Jenni Blackford, Jon Tapp, Lily Wang, Richard Urbano
  • From KC Strategic Planning Meeting 8-9 Oct07
    • Much discussion of electronic communication options
  • From KC Core Directors Meeting 11Oct07: Planning for P30 renewal due Dec08
    • Pat Levitt requested a staff training session on WIKI to see if over-reliance on long emails can be reduced by shared WIKI documents at VKC.
      • Need to have some hands-on experience
      • Could start by having quarterly reports put on wiki rather than mailed in
      • Possible uses: reporting, meeting minutes, strategic planning, educational materials, Center memory, P30 draft, procedures, policies, personnel
    • In May 2008 Core C will turn in a progress report on its services, including quantitative descriptions. In August 2008 Core C will turn in its section on the competitive renewal for the P30 grant. Emphasize service, quality control, cost-effectiveness as well as innovation.
    • One issue is cost effectiveness for biostat FTEs. Can justify an increase by the fact that central M.S. biostatisticians can be trained in a variety of analytic approaches; each investigator cannot get needed expertise on their own
    • Need to estimate current need for M.S. and fraction of investigators currently needing such service
    • Collaboration across cores is important
    • Think of ways to collaborate with other DDRC statisticians
  • How do we convince people to quit using Word?

Meeting 26Nov07

Present: Warren Lambert, Chun Li, Frank Harrell, Jenni Blackford, Jon Tapp, Richard Urbano, Andy Tomarkin
  • Discussed recent consulting requests
    • Depression: SNPs vs clinical phenotypes
    • Test construction/tuning, psychometrics
    • Training
    • Manuscript resubmissions, helping grad students
    • Help with data capture
    • Much help to beginners, e.g., grad students
    • Have benefitted from http://www.ats.ucla.edu/stat/examples/alda site esp. for HLMs and longitudinal data analysis
    • Questions specific to imaging data; some persons are using imaging software for analysis, difficult to figure out what the software is doing
      • Documentation for Brain Voyager is terrible, SPM somewhat better
    • Some complex needs where the validity of the analytic procedure is in question and simulations are needed
    • Development of coding schemes to code observational data streams
    • Launching web surveys using Survey Gold
    • Teaching grad students how to move from GUI to generating commands
  • Discussion of having students learn software before learning stats basics
  • Need to have more intro to imaging analysis; many potential investigators
  • Need to have more consistent biostat introductory training to graduate students; also principles of data collection/management

Meeting 28Jan08

Present: Warren Lambert, Chun Li, Frank Harrell, Jenni Blackford, Lily Wang, Richard Urbano, Andy Tomarkin
  • Jenni's K award is likely and she has received a Hobbs grant; training activities done until end of March
    • New appointee in VKC/Pediatrics is a likely good fit
  • Report from VKC Core Directors' Meeting 19Dec07
    • Analyzing (via simulation and other methods) the validity of analytic procedures (such as those used in software provided with devices; see point from 26Nov07 meeting) should be thought of as generative. We should expand and document such efforts. Imaging is a key area. So is stat software that comes with devices.
    • We should seriously consider offering a service of helping researchers use the RedCap research data management system developed by Paul Harris in Biomedical Informatics
  • Discussed good research database practice and reproducible analysis; relates to better training of new generation of researchers
  • Would be good for Kiley to put out a sheet listing consultants and their expertise
  • Lily reported on R01 award on sleep and autism; Beth Malow PI
  • Old work logger is still running
  • Still need VKC investigators to come to biostat clinics
  • NEW Large EEG data sets * I met with Sasha Key,PhD, today. She runs KC's eeg lab in MRL. Sasha and Pat Levitt want more statistical input to analyze existing datasets and also to develop better methods, e.g. data reduction, canned software evaluation etc. I've done a lot of behavioral multivariate analysis (PCA, EFA etc) but have little knowledge of eeg data reduction. Pat is hoping that new solutions to problems can be developed and published ("generative"). Sasha is going to a conference, but when she returns she'd like to set up a meeting with me & Jenni (she doesn't know). Shasha would like to be able to meet with her researchers and a stats consultant knowledgable about eeg (ERPs). --WL Jan 31, 2008

Meeting 25Feb08

Present: Warren Lambert, Chun Li, Frank Harrell, Jenni Blackford, Lily Wang, Richard Urbano, Andy Tomarkin, Jon Tapp
  • Jenni announced the funding of her K01 award
  • Frank announced that the Department of Biostatistics has decided to make secondary appointments to Warren and Andy
  • Chun's R01 grant is likely to be funded
  • Rick reported that a HRSA grant has been awarded related to Down's syndrome
  • Warren, Rick, and Wendy Stone got an extension on a grant
  • Update on EEG/ERP analysis (see above); need to know how many researchers are using each of these two methods; encourage researchers to come to Tuesday or Friday clinics. Warren will meet with Sasha Key to make a list of investigators.
  • Think about whether Lei should get some P30 funding, especially if she could have access to more real neuroimaging data; see also if she could work on EEG/ERP analysis
  • Jon will talk to Paul Yoder and think about simulations; Warren will talk further with Sasha about who needs help.
  • Lily is working on microarrays with Christine Konradi
  • Look for opportunities for collaboration in pediatrics/genetics (Marshall Sumner, etc.)
  • Weekly seminar for Psychiatry/PhD Psych postdocs would be a good idea; Jenni is doing a journal club.
  • Progress report first draft due April 7; competitive renewal will be submitted to DC in December but our due date will be about a month earlier. Warren will distribute the old progress report soon. Tim Stafford will have a dump of the work logger data made in a few days.
  • Core C statisticians should submit their revisions by the March Core meeting to give Warren a week to incorporate changes and Frank time to review. The progress report is merely an update, but the competitive renewal in the fall needs to be extremely well done.
  • For next meeting look at a new extract of work logger; Andy will make his own table summarizing his KC activities
  • Julie Lounds Taylor (dev psych, quant background, large datasets) could be an asset; would need to have others cultivate contacts for workshops. Pat Levitt and Tim Stafford are open to the suggestion that Julie replace Jenni.
  • Progress on VKC Communications Committee; need to add sections to http://devepi.vanderbilt.edu/wiki

Meeting 31Mar08

Present: Warren Lambert, Chun Li, Jenni Blackford, Lily Wang, Richard Urbano, Andy Tomarkin, Jon Tapp.
  • Frank unexpectedly called out of town.
  • Training: Mary Dietrich and Julie Lounds Taylor have joined the Core. Mary has a standing meeting. Frank, we re-schedule this meeting? Mary and Julie will replace Jenni running the training program of Core C. Mary has long experience at VU, and Julie is a new PhD from Wisconsin. Jenni will try to set up a partners lecture soon.
  • Warren met with Sasha Key and Lei Xu to discuss cross-core collaboration on EEG/ERP analysis. Tentative plan: Sasha would provide datasets, Lei&Warren do draft analyses, then get review (eg from Core C statisticians) on whether methods could be improved. The first problem will be a PCA on EPR, a popular approach in the field.
  • Rick and Bob Hodapp are involved with a machine learning problem.
  • Jon's new worklogger - Discussed logging process. The new Work Looger provides very similar information as the old, and not everything we need to evaluate Core services. For the competing renewal, next month Warren will submit consult categories that would enable to Core to evaluate its services.
  • Rick working on collaboration with Waisman Center on Downs syndrome adults and death.
  • Core C Progress Report. Got updates from Jenni, Lily, Jon, Andy. Chun will submit a pargraph about his activities. Warren will edit, update, and give draft to Frank this Thurs.

Meeting 19May08

Present: Lambert, Tapp, Harrell, Wang, Taylor
  • Introduced new member Julie Taylor; Julie is especially interested in longitudinal data analysis
  • We discussed Terri Scott's workshop (last Friday) on reproducible research (45 attendees)
  • We discussed the problem of too short deadlines for analyses and statistical reports; reproducible reports can help
  • A future workshop might be a 2-day R course (by Terri); need to gather data on need
  • Also need to keep assessing need for an MS statistician to assist investigators with data analysis (for next P30)

Meeting 23Jun08

Present: Lambert, Harrell, Wang, Taylor, Urbano, Tomarkin, Dietrich, Tapp

Report of 12Jun08 Core Directors Meeting

  • Things to emphasize in P30 renewal
    • Core provides open access to multiple expertises
    • It provides training to access methodologies
    • Methodologic workshops, sharing best practices
    • Regular core staff meetings
  • Format of renewal should follow Karoly Mirnics' model
  • Will ask Don Guthrie once again to be a reviewer
  • Draft 1 is due Aug 1
  • Mention that Rick serves in an advisory capacity and serves to foster communications between cores
  • Section on observational consultation should be separate from stat
  • Consultants need to follow through on publications of clients; pay attention to joint publications
    • Be sure to include methodologic articles
  • How can we enhance QC? (communication accuracy, satisfaction (ceiling effect))
  • See if we can increase Lei's % effort
  • What about following up on workshops for impact? A follow-up survey listing all the workshops and assessing attendance, usefulness, did it change the way the participant does or thinks about things? Add a needs assessment.
  • Impact on grant applications could also be considered
  • Clinics
  • Mary and Julie are following up on an invitation for an outside speaker

Meeting 25Jul08

Present: Harrell, Wang, Taylor, Tomarkin, Dietrich, Tapp, Xu, Li
  • We can see if Lei Xu can work with the Neuroscience Services core (Mirnics)
  • Julie reported on the survey of seminar attendees; the response rate was 42%
  • Group picture
  • Frank is set to work on P30 application; late additions/changes should be e-mailed to him

Meeting 29Aug08

Present: Tapp, Lambert, Dietrich, Harrell, Li, Wang, Urbano, Taylor, Xu, Tomarkin
  • Discussion upcoming visit by HK May 2009; will include Grand Rounds in Psychiatry
  • Warren updated on the P30 which is now in good shape; Jan will do more editing
  • Talked about difficult consultations
  • Training update

Meeting 21Sep08

Present: Lambert, Dietrich, Harrell, Li, Xu, Wang, Taylor, Tomarkin
  • Beefed about the economy
  • Warren showed slides for the cores workshop
  • Warren is working on the new round of P30 edits
  • Some members of the group are fed up with power analysis and its associated game playing; let's emphasize computing of detectable effect sizes and expected margins of error instead of estimating sample sizes
  • Look at Kraemer's 2006 article on misuse of effect sizes from pilot studies
  • Mary raised a time series issue; referred to Patrick Arbogast and Rafe Donahue
  • Frank reported on progress on the new biostat graduate program being developed
  • Chun received the NOGA for his R01
  • Need to finalize topics for HK visit

Meeting 24Oct08

Present: Lambert, Dietrich, Harrell, Li, Xu, Taylor, Urbano, Tapp
  • Discussed leadership change for VKC, and potential move of Biostatistics Department; P30 application is in good shape
  • Warren discussed cost effectiveness of methodology support; we reviewed a proposed insert into the P30
  • Regarding additions to P30 budget, Warren will check with Tim
  • Julie described a graduate student's analytical problems
  • Frank urged consultants to encourage clients to attend biostat clinics, and for consultants to attend clinics to get methodologic backup
  • Kraemer's visit is well planned

Meeting 14Nov08

Present: Wang, Dietrich, Harrell, Xu, Lambert, Tapp, Taylor
  • Spoke about VKC leadership change and what we could do to help
    • We will invite Dr Dykens to one of our core meetings
    • Provide her with a log of current activities (last few months)
  • Julie will invite Paul Harris to give a workshop on RedCap and research data management
  • Jon mentioned that a new book by Paul Yoder on observation methods will include a demo version of his software

Meeting 19Dec08

Present: Wang, Dietrich, Lambert, Tapp, Taylor, Xu, Urbano, Tomarken
  • P30 is about done. WIll be submitted to NICHD in Dec
  • Discussed first draft of VKC Science Day poster about Core C. We will include paragraphs on behavioral, biostats, observational, and population dataset consulting
  • We don't have an agreed-upon logging method for 2009. Jon will show us the work logger and Rick will bring an example of google GTD to the next meeting
  • New book to recommend: RA Muenchen, R FOR SAS AND SPSS USERS. Springer Nov 2008.

Meeting 23Jan09

Present: Tomarkin, Tapp, Harrell, Taylor, Urbano, Wang, Dietrich, Li, Dykens
  • Discussed the economy
  • Science Fair is next Thursday; try to attend our poster and see other facilities
  • VKC Director Dykens joined the group and discussed the following
    • Successful P30 application; more emphasis on expanding beyond basic science than the past P30
    • The Competition
    • Interim Associate Director Lou Muglia (Peds, neurosci, Wash U)
    • NICHD branch rename Intellectual and Developmental branch
    • P30 funds cores but has a mandate bigger than that, including services to the community
    • Diversity of programs already in place at VKC is a plus
    • So is new facilities such as the new Vivarium

Meeting 26Mar09

Present: Tapp, Harrell, Taylor, Wang, Lambert, Tomarkin
  • Ethical question: A hypothetical PI says that in a journal article you tell the truth, the whole truth, and nothing but the truth (like in court). But, she says, a grant application is a marketing document and applicants don't have to tell the whole truth. It's perfectly OK to withhold evidence (e.g. unsuccessful fishing). What do people think?
    • See http://www.medscape.com/viewarticle/588146
    • Don Rubin's publications in 2007-8 relate to making observational studies more scientific
    • Technical helps: parallel analyses on random data; bootstrap entire analysis history; alternative explanation analysis; more detailed analysis plans in grant proposals; simulations of performance of a new method
    • Point out the roots of bias
*Programer: Warren Lambert;
*Date: Jan 18, 2012 ;


* ***********************************************************;
* ***********************************************************;
OPTIONS NOFMTERR formchar = '|----|+|---+=|-/\<>*'
nocenter
date
pagesize = max
linesize = max
number
probsig = 2
NOgwindow
nofmterr
noMprint
Mprint
compress = yes
compress = no
obs = max
errors = 3
threads
cpucount = 2
msgLevel = i
mautosource
sasautos = (sasautos, " R:\Projects\Army\Data\DataSAS\Macros ")
mAutoLocDisplay ;

* ***********************************************************;
* ******************************************************************************************************** ;
%Let WSpace = "============================================================================>" ;
%Let WShort = "=========================>" ;
title4 ;
title4 &WSpace " " ;
* ******************************************************************************************************** ;
* ******** Physical health symptoms from question 8a that we will use in the physical distress scale ;
%Let PhysHealth =
Smfever Smcough Smbreathing Smheadache Smweakness Smmuscle Smjoints Smback_pain
Smnumbness Smhearing Smtearing Smvision Smchest_pain Smdiarrhea Smrash
Smother_cond
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;

%Let Behavioral = Smnightmares Smavoid_situations Smon_guard Smdetached Smconflicts
Smlittle_interest Smfeeling_down ;

%Let M_Exhaustion = SMDECISIONS SMTIRED SMDISTRACTED SMIRRITABLE SMMEMORY ;


* ***********************************************************;
* *** The other variables we included in the global mental health distress scale;

*Let OthDistress = Smnightmares Smavoid_situations Smon_guard Smdetached Smconflicts
Smlittle_interest Smfeeling_down ;
*************************************************************** ;

* ***********************************************************;
* *** The overall mental health question #4 on the PDHRA is smdiff_emo ;

* ***********************************************************;
%Let Overall = smdiff_emo ;

* ***********************************************************;
%Let Phys_Q8a = Smdizzy Smtired Smdistracted Smmemory Smdecisions Smirritable ; smrisktaking ;

***********************************************************;
* ** Overall physical question on the PDHRA is question #3 variable ;
%Let PhysGlobal = smdiff_phy ;

* ***********************************************************;
* **************** Risk from an earlier meeting ;
%Let Risks = smsex
smtotaldep_oifoeff
smgrade
smmar_stat
iraq
afghan ; * * Not from Lauren this time ;

* ***********************************************************;
* **************** Referral from earlier meeting ;

%Let Referral = anyrefR medicalrefR nonmedicalrefR ;
* **** clref_primary clref_mental_primary clref_mental_special ;

%Let M_Exhaustion = SMDECISIONS SMTIRED SMDISTRACTED SMIRRITABLE SMMEMORY ;

%Let TheRest = SMDIFF_EMO SMDIFF_PHY SMDIZZY SMSEX SMTOTALDEP_OIFOEFF SMGRADE
SMMAR_STAT Iraq Afghan ;

%Let All3 = &PhysHealth &Behavioral &M_Exhaustion ;

* ***********************************************************;
* ***********************************************************;

%Let ArmyA = %Bquote(C:\Documents and Settings\lamberew\Desktop\PsychometricCEPI\ArmySorted.sas7bdat) ;

libname library "R:\Projects\Army\Data\DataSAS\Formats" ;

* ***********************************************************;
* ***********************************************************;

*options obs = 10000 ;

Data ArmyA ;
set "&ArmyA" (Keep = RandUniNumb &PhysHealth &Behavioral &M_Exhaustion &TheRest) ;

if RandUniNumb < 0.20 ;

if NMISS(of &PhysHealth ) = 0 ;
if NMISS(of &Behavioral ) = 0 ;
if NMISS(of &M_Exhaustion ) = 0 ;
*if NMISS(of &TheRest ) = 0 ;

title3 &WSpace "Examine scale with CTT" ;

label SMFEVER = "Fever "
SMCOUGH = "Cough "
SMBREATHING = "Breathing "
SMHEADACHE = "Headache "
SMWEAKNESS = "Weakness "
SMMUSCLE = "Muscle "
SMJOINTS = "Joints "
SMBACK_PAIN = "Back Pain "
SMNUMBNESS = "Numbness "
SMHEARING = "Hearing "
SMTEARING = "Tearing "
SMVISION = "Vision "
SMCHEST_PAIN = "Chest Pain"
SMDIARRHEA = "Diarrhea "
SMRASH = "Skin Rash "
SMOTHER_COND = "Other Phys"

SMNIGHTMARES = "Nightmares "
SMAVOID_SITUATIONS = "Thoughts "
SMON_GUARD = "On guard "
SMDETACHED = "Detached "
SMCONFLICTS = "Conflicts "
SMLITTLE_INTEREST = "Disinterest"
SMFEELING_DOWN = "Down "
SMDECISIONS = "Decisions "
SMTIRED = "Tired "
SMDISTRACTED = "Concentrate"
SMIRRITABLE = "Irritable "
SMMEMORY = "Memory " ;

format &PhysHealth &Behavioral &M_Exhaustion 1.0 ;

run ;

data Small (keep = &PhysHealth &Behavioral &M_Exhaustion ) ;
set ArmyA ;

run ;

Topic revision: r111 - 18 Dec 2023, IneSohn
 

This site is powered by FoswikiCopyright © 2013-2022 by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Vanderbilt Biostatistics Wiki? Send feedback