Social · Behavioral · Health Sciences

Rigorous methods.
Collaborative
research.

I work with researchers in the social, behavioral, and health sciences on the quantitative side of their studies: designing the analysis, fitting the models, and making the results clear enough to publish, present, and fund.

Some projects are a single conversation about model fit. Others run from the first draft of an aim through the published paper. Both are welcome, and both start the same way - with a short email about what you are working on.

Across the life of a study

Join at any point; earlier is usually better

Design

Framing aims as models, choosing waves and measures, and sizing the sample before anything is collected.

Measurement

Establishing that a scale holds up - factor structure, reliability, and invariance across groups and time.

Analysis

Fitting the model the question calls for, with the missing data, nesting, and attrition handled honestly.

Writing

Methods and results sections, tables and figures, and replies to the reviewer who questions the approach.

Funding

Analysis plans, power, and budget language for the next proposal - often the same study, one stage on.

Services

Most engagements are some combination of these four

Analysis and interpretation

Running the models, or making sense of output you already have - what the fit statistics support, what they do not, and how to say it in the language of your research question.

Manuscripts

Writing and structuring the quantitative sections, preparing tables and figures to journal specification, and drafting responses to methodological reviews.

Mentoring and teaching

Working through a method with a student, a lab, or a department - including annotated syntax you keep and can run again on the next study.

Talks and posters

Slides, posters, and diagrams for conferences and defenses, and a rehearsal of the methods questions that tend to follow them.

Grant support

Proposals are rarely turned down for the idea. They are turned down for an approach section that does not convince a reviewer the design can answer the question.

There is more room to help a month or two out than in the final week, though I have done both.

Statistical analysis plans

Each aim tied to a model and a contingency, written so a methodologist on the panel can follow it.

Power and sample size

Justification built on the model you will actually fit, with attrition and clustering accounted for.

Writing and editing

Approach and measurement sections drafted or tightened so the design’s strengths are explicit.

Budget and review

Statistical effort costed across the project years, plus a candid critique before you submit.

Methods

Grouped by the question they answer - you do not need to know which one you need

How does something change over time?

Latent growth curve models, multilevel modeling, latent transition analysis, and repeated-measures designs.

Are there distinct subgroups in the sample?

Latent class analysis, latent class growth analysis, growth mixture modeling, and profile comparison.

Is the measure capturing the construct?

Confirmatory factor analysis, item response theory, measurement invariance, and scale development.

What connects one variable to another?

Structural equation modeling, path analysis, mediation and moderation, and regression.

When does an event occur?

Discrete-time and continuous-time survival analysis, with time-varying covariates.

Analyses are run in Mplus, R, SAS, Stata, SPSS, or LISREL - whichever your team already works in - and you get the annotated syntax back with the results.

Tell me what you are working on.

A few sentences is enough to start. If it is not something I can help with, I will say so and point you somewhere better.

Get in touch