Appendix B — Spending a Training Budget
Professional development funds (both from DSTT and possibly from your employer) typically have a deadline attached and sometimes little guidance. The common outcome is that a team identifies what it wants in the final weeks, discovers a procurement constraint it did not know about, and either buys whatever can be approved quickly or returns the money.
Plan training purchases around the team’s needs and the procurement deadline. Section 22.3 covers staff development, and Appendix A lists free material to consider alongside paid options.
These guidelines are written for a general public health practitioner with a training budget looking to continue upskilling and professional development, and not necessarily guidelines specific for the DSTT training budget. Check with DSTT program staff for any DSTT-specific guidance.
B.1 Principles
Start with the skills the team needs for its current work.
Buy skills that transfer. SQL, R fundamentals, and version control will remain useful in five years, and will remain useful if the agency replaces its BI platform, its warehouse, or its surveillance system. Training in a particular interface may need updating when the product changes.
The same reasoning applies to AI training. A general course covering how these models work, what they do well, and where they fail will age better than instruction in the prompt syntax of whichever assistant the agency licensed this year. Chapter 26 explains why understanding the method matters more than knowing the interface.
Choose training that includes practice. For SQL, look for exercises in which learners write queries and inspect the results.
B.2 Skills Worth Buying
SQL comes first. It is chronically underrated and becomes unavoidable as agencies move surveillance data into warehouses (Chapter 17). An analyst who can only receive extracts depends on whoever writes them, which adds days to every ad hoc question. Include hands-on practice.
R fundamentals follow, specifically functions. Most public health analysts learn R as a sequence of scripts executing top to bottom. The transition from scripts to functions is what makes testing (Chapter 4), packaging (Chapter 14), and substantive code review (Chapter 27) possible, and it helps analysts reuse and test their work.
Version control comes next. It is inexpensive to learn relative to how much it changes a team’s working practice (Chapter 1).
Data visualization and communication follow, including selecting a visual encoding and structuring a message (Chapter 28). Start with the free reproducible reporting documentation before paying for additional instruction. General AI literacy completes the list.
B.3 Courses and Subscriptions
Public-health-specific providers. Applied Epi offers Intro to R for Applied Epidemiology and related courses taught by epidemiologists, with examples such as line lists and case counts. CSTE offers training tied to its own programs. These catalogs are narrow. Where an agency restricts purchases to public-health-approved content, the GIS or Shiny course a team actually wants may be unavailable, so check eligibility before selecting a course.
General platforms. DataCamp is strong for hands-on SQL and R practice in a browser sandbox requiring no local installation, which can help when staff cannot install software locally. Check current prices and whether the advertised rate requires annual billing. Coursera hosts the Johns Hopkins Data Science Specialization and the Executive Data Science sequence, both described in Appendix A. Udemy and Codecademy are inexpensive and uneven; evaluate the specific course you intend to use.
R-focused training companies. Jumping Rivers and comparable firms sell instructor-led private training. The per-seat cost is high, and a single-day workshop delivered to an entire team can represent better value than five individual subscriptions, with content built around the team’s own data.
Free and community options. The Carpentries publish workshop curricula on R, Git, and the shell that a team can work through independently. Workshops for Ukraine sells individual sessions on specific topics for a small donation and maintains a deep back catalog. The Epidemiologist R Handbook is free and is the most directly applicable R reference for this audience.
Conferences. posit::conf offers R and Python sessions for practitioners, CSTE’s annual conference offers opportunities to meet other public health practitioners, and vendor summits are useful primarily to organizations already committed to that vendor. Include registration, travel, and staff time when comparing conference costs. Where the choice is one conference trip for one person against a year of hands-on training for the team, the training is the better purchase.
Before purchasing a subscription, determine whether the state health department, a university partner, or the library system already holds an enterprise license. Several agencies pay for these platforms at the department level without informing program staff.
B.4 Books
Appendix A covers the technical R books worth owning, most of which are free online. Books remain worth purchasing for a team shelf, and they are the easiest item to buy quickly against a closing deadline.
Beyond the R canon, consider books on communication: Storytelling with Data and its companion exercise book, Talking Health, and Edward Tufte’s visualization books. For R specifically, R for Data Science (Wickham et al. 2023), the functions and control flow chapters of Advanced R, and Hands-On Programming with R cover the fundamentals identified above. Where anyone on the team works in Python, Python for Data Analysis is the standard reference.
B.5 Consulting
Some of what a team needs is not sold as a course. Where the gap is one specific workflow, a pipeline that needs to be made reproducible, a dashboard nobody has time to finish, or training built around your own data, a short consulting engagement can be a better use of the money than another subscription.
I am available for that kind of work, including after a team’s DSTT program year ends. I have worked with DSTT teams for years as a project coach, and I consult on data science and bioinformatics across a range of fields. I am a professor in the University of Virginia School of Data Science, and I was previously a professor of public health sciences in the UVA School of Medicine. The fastest way to reach me is stephenturner.us/email.
B.6 Procurement Constraints
Confirm the purchasing rules before committing to a course or event.
Establish what “spent” means to your finance office. Committed and spent are different states. Some offices count a purchase order issued before the deadline; others require the invoice paid and reconciled. Ask in writing and ask early, because the answer changes what can be purchased.
Establish whether a purchasing blackout applies. Many local governments freeze non-essential purchasing near the end of a fiscal year, which frequently coincides with training deadlines. Do not assume a personal purchase will be reimbursed. Obtain written approval before anyone incurs an expense.
Map travel restrictions before planning travel. Out-of-state travel bans are common and sometimes absolute, including for conferences, which can require an entire stipend to be spent in state. Virtual conference registration and subscription training are the usual substitutes when travel is unavailable.
Determine whether registering now for a later event counts. Registering and paying for a conference before a spending deadline for an event occurring months afterward is often possible, and whether a finance office accepts it varies. Ask, and hold a backup plan, since travel approval and scheduling can still prevent attendance after the money is committed.
Plan reconciliation where multiple payers are involved. A program sponsor, a county finance office, a grant, and a personal reimbursement within one purchase produce four sets of documentation, and reconciliation across several payers can consume more staff time than the training itself. Retain every receipt and record which fund each purchase draws on as it occurs.
Begin earlier than appears necessary. Approval lead times exceed expectations consistently. Sole-source justifications, vendor onboarding, and tax documentation for a small training company can require weeks. Six weeks before a deadline is late.
A delayed approval can leave a team choosing from an approved catalog in the final week, even when none of the courses meets its needs. Verify the approval path before reviewing the catalog.
B.7 A Worked Plan
Consider $3,000, six weeks, and a three-person team.
In week one, confirm the rules: what constitutes spending, whether a blackout applies, whether travel is permitted, and what catalog restrictions exist. In parallel, ask each staff member which of the skills above most limits their current work.
In week two, order books if the required approvals are in place. Allocate roughly $300 in this example.
In weeks two and three, submit the subscription request. Reserve roughly $500 for subscriptions in this example, subject to a current quote and comparison with free practice materials. Submit early, because vendor onboarding is the slow component.
In weeks three through five, allocate the remaining $2,200 to the team’s highest-priority training need: a private instructor-led R or SQL workshop for the full team, a course seat for each analyst, or virtual conference registration for each. Obtain quotes and start approval for instructor-led training in week one; its lead time may exceed six weeks.
In week six, reconcile. Do not plan to purchase anything in week six.
This plan excludes individual conference travel deliberately. Travel could consume much of this example budget. Purchase travel when the objective is the professional relationships, and state that as the objective.
B.8 Requesting Next Year’s Funding
Tie the request to a specific improvement. For example: “Three SQL course seats will help analysts query the warehouse directly. We currently wait four days for each extract request.” Use measured delays from your own team.
Record what was purchased and what changed as a result. Where last year’s version control training is the reason a weekly report survived an analyst’s departure, document that outcome in this year’s request. The continuity argument in Chapter 23 and Section 22.6 is the strongest case available for training funds, because turnover is a risk leadership is already concerned about.