IM-1 — Pacing Guide

Companion documents. This guide governs the clock. For how each deliverable is scored, see IM-2 (Grading Rubrics); for the errors to watch as the clock runs, see IM-3 (Common Student Pitfalls); for the layered-mentorship rhythm and how all-hands sessions are run, see IM-4 (Mentor-Session Facilitation & Guest Lectures); for anchor work and answer keys, IM-5; for the access, compute, licensing, and waiver constraints that shape any schedule you build, IM-6 (Equity, Access & Compliance). All cross-references are by name so this manual survives reordering.

The ASSIP 2026 cohort runs full-time Monday–Friday, 9:00 am–5:00 pm, from Monday June 22, 2026 through Wednesday August 12, 2026 (the Symposium). Orientation is Thursday June 18, 2026. Two federal holidays fall in the window — Friday June 19, 2026 (Juneteenth observance), the day before orientation, and Friday July 3, 2026 (Independence Day observance) — and the cohort takes those days off. Prof. Gao’s 2026 ASSIP slot is REMOTE only: every block in this guide is run over Zoom, recorded, and re-watchable from the cohort wiki.

This gives eight active program weeks (Jun 22 – Aug 12) with 37 working weekdays minus 1 holiday = 36 active weekdays, plus the half-day Symposium on Wed Aug 12. The textbook was built for exactly an eight-week intensive at roughly this hour count, so under the ASSIP calendar the eight book weeks map one-to-one onto the eight program weeks — no compression, no merging. What this guide gives you is the daily shape of each week: which block runs the curriculum, which block runs the mentor research project the camp deliverables actually demand, and which block runs the cohort sync that keeps a remote 14-student cohort glued together.

1. The eight-week clock

The eight weeks are not eight parallel modules; they are a single load-bearing arc, and the order is not negotiable. Weeks 1–4 build the toolkit from the ground up: probability and the logic of inference (W1), the OLS engine and its standard-error pathologies (W2), causal inference from potential outcomes through matching and instruments (W3), and the modern quasi-experimental toolkit — DiD, RD, synthetic control, shift-share — together with the recent literature showing the obvious estimators are often wrong (W4). Weeks 5–6 turn from running estimators to reading them: one empirical paper per day under a fixed Reader’s-Guide anatomy (W5), then text-as-data and a full module on using LLMs as a research co-pilot without fooling yourself (W6). Weeks 7–8 are the capstone: question to pre-analysis plan (W7), then execution, robustness, writing, the poster, the abstract, and the eight-minute talk at the Symposium (W8).

Each week earns the next. You cannot read a weak-instrument disaster in Week 5 if you did not derive 2SLS in Week 3; you cannot stress-test your own DiD in Week 8 if you did not meet the staggered-adoption crisis in Week 4. The dependency order is fixed.

A useful way to hold the whole arc in mind: Weeks 1–4 answer “what is a defensible number?”, Weeks 5–6 answer “how do I judge someone else’s number, including a machine’s?”, and Weeks 7–8 answer “can I produce a defensible number of my own and survive a referee, a Symposium audience, and the JSSR editor?” The terminal assessment (W8) is the only thing the camp was ever building toward; every earlier week is rehearsal for it.

2. The ASSIP daily block structure

The ASSIP day is eight hours, 9:00 am – 5:00 pm, with a one-hour lunch (12:00–1:00). Within those eight hours the cohort runs three repeating blocks, every weekday, plus a fourth block that lives only on Monday mornings:

9:00–12:00 — Curriculum block. Three hours of book content: chapter read-along, notebook live work, and problem-set clinic. This is where the textbook’s daily rhythm lives — one chapter + notebook + problem set per day, five days a week, five chapters per book-week.
12:00–1:00 — Lunch. Hard break. Off-camera. Resists the temptation to schedule “working lunches” — a remote cohort burns out faster than a residential one, and the lunch break is the single biggest dropout-prevention lever you control.
1:00–4:00 — Mentor research block. Three hours of work on the student’s own capstone project — data acquisition, EDA, replication, identification design, PAP drafting, robustness, writing, poster layout. Pod sub-leaders (the four PhD/RA mentors) hold open Zoom rooms during this block; Prof. Gao drops in by rotation. The mentor research block is the engine that produces the Symposium poster, the JSSR abstract, and the optional full paper.
4:00–5:00 — Cohort sync. A one-hour shared close to the day: ten-minute pod standups led by each sub-leader (each pod reports what they built, what is blocked, what they will do tomorrow); forty-minute rotating element — Tue/Thu peer-review pairs walking each other through their notebooks, Wed a guest-lecture or skill clinic, Fri a fifteen-minute show-and-tell where one randomly drawn student demos the week’s deliverable to the cohort. Mondays the cohort sync is replaced by the all-hands mentor session that opens the morning (see below).
Monday 9:00–10:00 — All-hands mentor session with Prof. Gao (IM-4). This sits inside the Monday curriculum block, pushing the chapter/notebook/PS work to 10:00–12:00 on Mondays only. Eight Mondays, eight all-hands sessions, each anchored to one of Prof. Gao’s papers (see IM-4 for the topic sequence). This is the cohort’s intellectual anchor of the week.

The week therefore costs each student 40 contact hours (8 hours × 5 days), of which roughly 15 are curriculum, 15 are mentor research, and 5 are cohort sync; the remaining 5 land in Monday’s all-hands mentor session and the buffer that Friday’s show-and-tell creates. This is the magnitude the eight-week book was written to.

Why the day is split this way

Three pedagogical pressures shape the schedule. First, morning curriculum, afternoon research matches the cognitive science: declarative learning (reading a chapter, deriving an estimator) goes best when working memory is freshest, and applied work (debugging your own pipeline, refining a threats table) is more resilient to mid-day fatigue. Second, the mentor research block is the camp’s actual deliverable engine — the Symposium poster, the JSSR abstract, and the 8-minute talk are all built in this block, not in the curriculum block; giving it three protected hours a day prevents the typical research-program failure where students “ran out of time” because the curriculum ate the project. Third, the cohort sync is the remote-format anti-isolation mechanism — fourteen students working alone in fourteen different homes is the worst-case dropout scenario; the 4:00 pm end-of-day check-in is what makes Tuesday morning’s start-up reliable.

3. The eight-week schedule, week by week

What follows is the mapping of the eight book-weeks onto the eight ASSIP program-weeks, with the weekday calendar and the holidays marked. Each week’s curriculum block runs the book’s five chapters + five notebooks + five problem sets across five days (one per day) and closes Friday with the end-of-week assessment graded against IM-2.

Program week	ASSIP dates	Book content	Mon AM mentor (Prof. Gao)	Friday deliverable
W1	Mon Jun 22 – Fri Jun 26	W1 — Probability & Inference (ch1.1–1.5, nb1.1–1.5, ps1.1–1.5)	M1: “What is a finding?”	W1 mini-sim (size of a t-test) — Assessment W1
W2	Mon Jun 29 – Thu Jul 2 (Fri Jul 3 OFF)	W2 — OLS Engine (ch2.1–2.5, nb2.1–2.5, ps2.1–2.5)	M2: “Why your SE is the whole ballgame”	W2 SE-choice replication — Assessment W2 (compressed to Thu; Fri off)
W3	Mon Jul 6 – Fri Jul 10	W3 — Causal I (PO/matching/IV) (ch3.1–3.5, nb3.1–3.5, ps3.1–3.5)	M3: “Natural experiments” (Deng, Gao & Kim 2020)	W3 weak-IV / AR-coverage task — Assessment W3
W4	Mon Jul 13 – Fri Jul 17	W4 — Causal II (DiD/RD/SC/shift-share) (ch4.1–4.5, nb4.1–4.5, ps4.1–4.5)	M4: “Detecting discrimination with a clean design” (Sun & Gao 2019)	W4 staggered-DiD task + first capstone idea memo — Assessment W4
W5	Mon Jul 20 – Fri Jul 24	W5 — Reading the Frontier I (Reader’s-Guide pack across five classics)	M5: “Anatomy of a JF paper”	W5 Reader’s Guide on an unseen paper — Assessment W5
W6	Mon Jul 27 – Fri Jul 31	W6 — Reading the Frontier II + AI co-pilot (ch6.1–6.5, nb6.1–6.5, ps6.1–6.5)	M6: “Text as data, AI without fooling yourself”	W6 OOS-validated classifier + leakage audit — Assessment W6
W7	Mon Aug 3 – Fri Aug 7	W7 — Question to Pre-Analysis Plan (ch7.1–7.5, nb7.1–7.5, ps7.1–7.5; Lab 7 = stand up the capstone repo)	M7: “How I pick a project — and kill one”	PAP + identification memo, filed as `pap-filed` tagged commit — Assessment W7
W8	Mon Aug 10 – Tue Aug 11; Wed Aug 12 = SYMPOSIUM	W8 — Execution, Robustness, Poster, Talk (ch8.1–8.5 + poster + abstract + talk)	M8: “Defending a result”	Symposium deliverables: poster (in-person + virtual breakout) + JSSR abstract + 8-min talk + repo

Two structural points the table does not surface. First, Week 2 is shortened by the Jul 3 holiday — the cohort loses Friday — and the week is reshaped so that the W2 assessment lands on Thursday Jul 2 rather than Friday, and the W2 cohort sync that would have been Friday’s show-and-tell is folded into Thursday’s 4:00 pm block. Second, Week 8 is dominated by the Symposium: Monday and Tuesday are the final execution-and-rehearsal sprint (dry-run defense on Tuesday afternoon, run by the sub-leaders and Prof. Gao), and Wednesday August 12 is the Symposium itself — an in-person poster session at GMU plus virtual breakouts for remote interns (this cohort), with audience including Mason faculty, mentors, families, College of Science leadership, and Mason News press.

What happens after Wed Aug 12

The four-week paper-refinement phase that the book’s W8 chapter (and the JSSR submission standard) imagines is outside the formal ASSIP window but inside the camp’s deliverable arc. Students who elect to write the optional full paper for JSSR continue to work with their pod sub-leader and with Prof. Gao for four weeks after the Symposium, on a revise-and-resubmit cycle (referee report → R&R memo → revision → final reproducibility check). The Symposium poster and JSSR abstract are the required deliverables; the full paper is opt-in. Plan grading of the 200-point terminal capstone rubric (IM-2 §4) at the end of that four-week refinement window for students who pursue it, and at the Symposium itself for those who do not.

A worked Tuesday, end to end (W3 Tuesday as the example)

To make the daily shape concrete, here is Tuesday July 7 — the second day of the Causal-I week, when the cohort is meeting the IV estimator for the first time:

9:00–9:10 — Daily kickoff. Sub-leader rotates the lead; reads the day’s “what we’re building” in one sentence (today: the first-stage F and why a plausible second stage is not enough) and the rubric row this day’s PS feeds.
9:10–10:30 — Chapter 3.2 read-along + nb3.2 live work. Cohort reads ch3.2 (the IV estimator and the two assumptions) under a shared screen, then live-runs nb3.2 — the worked 2SLS pipeline. Pod sub-leaders mute and watch their pod’s chat; questions surface through chat-first to keep the shared room from fragmenting.
10:30–10:40 — Break.
10:40–12:00 — ps3.2 clinic. Students attempt the problem set’s six problems in pods; the pod sub-leader floats. The two highest-signal problems are the first-stage F diagnosis and the Anderson–Rubin coverage check — IM-5 flags which ps3.2 problems separate strong from weak work.
12:00–1:00 — Lunch.
1:00–4:00 — Mentor research block. Students work on their own capstone projects. Most W3 students are still in the data-acquisition / replication phase; pod sub-leaders run drop-in office hours in their pod’s Zoom rooms; Prof. Gao floats by rotation (today: Pod 2’s afternoon).
4:00–4:10 — Pod standups. Each pod’s sub-leader runs a ten-minute round: what built today, what is blocked, what tomorrow brings.
4:10–4:50 — Tuesday peer-review pairs. Students pair across pods; each pair walks each other through their morning nb3.2 work and one problem from ps3.2. This is the cohort-glue move — it forces students out of their pod silo and into a fresh perspective.
4:50–5:00 — Cohort close. Sub-leader on rotation summarizes; flags Wed’s chapter (3.3, matching and propensity scores) and any wiki updates due.

Every weekday follows this shape; only the chapter, the live demo, the PS clinic, and the cohort-sync element rotate.

4. The layered-mentorship model

ASSIP 2026’s cohort is 14 students — at the high end of typical ASSIP group sizes (2–15). One faculty mentor cannot give 14 remote students the same depth of weekly attention a 4-student in-person group would receive; the cohort is therefore organized as a layered mentorship:

Prof. Lei Gao is the faculty mentor of record. He runs the Monday all-hands mentor session (9:00–10:00, recorded), is in the room for the Tuesday dry-run defense in Week 8, sets the Symposium poster standards, and rotates through pod afternoon office hours across the week.
Four PhD / RA sub-leaders (one per pod) are the daily owners of their pod’s progress. Each sub-leader: runs the 4:00 pm pod standup, holds the open drop-in office hour during the 1:00–4:00 mentor research block, grades their pod’s daily problem sets against IM-2, escalates any pod concern to Prof. Gao the same day, and is the named co-mentor on each of their pod’s three or four capstone posters.
Four pods of three or four students each are the working unit. Students stay in their pod Monday–Friday for standups and problem-set clinics; pods are reshuffled exactly once, at the start of W4, to break early cliques and to balance project topics — the second-half pod composition is the one that runs through the Symposium.

The layered model is what makes a 14-student remote cohort work. It also makes IM-4 — the mentor-session facilitation chapter — load-bearing in a new way: Prof. Gao runs the Monday all-hands session and is the escalation backstop, but the sub-leaders are the ones running daily research mentorship, and IM-4 now documents both layers.

5. What gets cut, and what never does

The ASSIP full-time calendar is generous enough that nothing in the book gets cut for time. What changes from a pure-residential design is:

The W2 Friday holiday compression. The W2 assessment moves to Thu Jul 2 and the W2 chapter load is front-loaded into Mon–Thu. No content drops; the rhythm tightens for one week only.
The Symposium deliverable replaces the standalone capstone defense. The 8-minute talk + poster
- JSSR abstract at the Symposium on Wed Aug 12 is the capstone presentation; the rubric in IM-2 §5 grades it live.
The optional full paper moves outside the program window. Students who opt to write the full paper for JSSR refinement run a four-week R&R cycle with their sub-leader and Prof. Gao after the Symposium; this is graded at the end of that window, not during the camp.

You protect, without exception: the dependency order of Weeks 1→4; the Monday all-hands mentor session every week; the W7 PAP-and-tag discipline (the pre-registration freeze is non-negotiable — breaking it voids the meaning of every Week-8 p-value); and the W8 reproducibility standard (make clean && make all).

6. A pacing checklist for the instructor

Before orientation (Thu Jun 18):

Provision WRDS seats for every student via GMU’s College of Science institutional subscription — coordinate with Dr. Amanda Haymond Still / cosassip@gmu.edu (IM-6). Lead time is the binding constraint; request the batch the week the cohort is announced.
Provision Hopper accounts for every student (Appendix B.4), and the camp’s Azure OpenAI allocation for the Week-6 AI module (IM-6).
Confirm the fee-waiver process has reached every Pell-eligible / reduced-lunch admit before the $25 application fee or the $1,299 tuition is invoiced (IM-6).
Stand up the cohort wiki (Notion, Confluence, or GitHub-hosted MkDocs — whichever the cohort knows) with one page per student, one page per pod, the eight weekly chapter folders pre-seeded, and the daily-standup template pinned. This is the cohort’s shared brain (IM-3).
Run FM-4 (Prerequisite Self-Test) the orientation afternoon (Thu Jun 18). Assign Appendix A/B remediation as async work for the orientation weekend to anyone who misses the calculus/probability/Python anchors.
Pin the environment (python=3.11 + the CONVENTIONS §5 stack) and confirm every notebook runs on a fresh env on the Zoom shared display.
Stand up GitHub Classroom and the capstone repo template (Lab 7 / Appendix B.2) so the pap-filed tag workflow is ready by W7.

Each day:

Confirm the prior day’s problem set was graded by the pod sub-leader against its rubric (IM-2) before the 9:00 kickoff, so the day’s chapter lands on a cohort already aware of yesterday’s highest-signal pitfall (IM-3).
Run the 4:00 pm cohort sync ruthlessly — no day ends without standups, even on Friday with the show-and-tell.
Push every chapter recording, every Monday mentor recording, and every cohort-sync recording to the wiki the same evening. The recording is the equity guarantee for any student who lost bandwidth, had a family emergency, or had an accommodation that needed asynchronous review (IM-6).

Each Monday:

Open with the all-hands mentor session (9:00–10:00); record; post-session reflection prompt assigned by 10:00 for Friday review.
Sub-leaders deliver one-sentence pod status to Prof. Gao by 9:00 — who shipped Friday’s deliverable, who needs intervention this week.

Each Friday:

Grade the Friday assessment against IM-2 by Saturday so the following Monday’s all-hands can debrief from the cohort’s own work, not generic examples.
4:30–4:50 show-and-tell: one randomly drawn student demos the week’s deliverable to the cohort. This is graded only as participation, but the cohort knows their name might be drawn — the possibility is what holds attention through Friday afternoon.

Week 8 specifically:

Tue Aug 11: dry-run defense for every student (sub-leaders + Prof. Gao). IM-4 gives the question protocol.
Wed Aug 12: Symposium. Virtual breakouts open by 9:00 ET; in-person poster session per the College of Science schedule.
Final week of the four-week paper-refinement window (for students writing the full paper): run make clean && make all on every capstone repo. It is the single most objective row in the terminal rubric, and it is the JSSR submission standard.

The clock is unforgiving precisely because the deliverable is real: these posters and abstracts head to a real Symposium audience and a real journal. Pace the camp so that the freeze holds, the code rebuilds, and every student can say the sentence the whole program exists to let them say truthfully — here is what I found, here is why you should believe it, and here is the command that lets you check me.

# IM-1 — Pacing Guide > **Companion documents.** This guide governs the *clock*. For how each deliverable is scored, > see **IM-2 (Grading Rubrics)**; for the errors to watch as the clock runs, see **IM-3 (Common > Student Pitfalls)**; for the layered-mentorship rhythm and how all-hands sessions are run, see > **IM-4 (Mentor-Session Facilitation & Guest Lectures)**; for anchor work and answer keys, **IM-5**; > for the access, compute, licensing, and waiver constraints that shape any schedule you build, > **IM-6 (Equity, Access & Compliance)**. All cross-references are by name so this manual survives > reordering. The ASSIP 2026 cohort runs **full-time Monday–Friday, 9:00 am–5:00 pm**, from **Monday June 22, 2026 through Wednesday August 12, 2026** (the Symposium). Orientation is **Thursday June 18, 2026**. Two federal holidays fall in the window — **Friday June 19, 2026 (Juneteenth observance)**, the day before orientation, and **Friday July 3, 2026 (Independence Day observance)** — and the cohort takes those days off. Prof. Gao's 2026 ASSIP slot is **REMOTE only**: every block in this guide is run over Zoom, recorded, and re-watchable from the cohort wiki. This gives **eight active program weeks** (Jun 22 – Aug 12) with **37 working weekdays minus 1 holiday = 36 active weekdays**, plus the half-day Symposium on Wed Aug 12. The textbook was built for exactly an eight-week intensive at roughly this hour count, so under the ASSIP calendar the eight book weeks map **one-to-one onto the eight program weeks** — no compression, no merging. What this guide gives you is the *daily* shape of each week: which block runs the curriculum, which block runs the mentor research project the camp deliverables actually demand, and which block runs the cohort sync that keeps a remote 14-student cohort glued together. --- ## 1. The eight-week clock The eight weeks are not eight parallel modules; they are a single load-bearing arc, and the order is not negotiable. Weeks 1–4 build the toolkit from the ground up: probability and the logic of inference (W1), the OLS engine and its standard-error pathologies (W2), causal inference from potential outcomes through matching and instruments (W3), and the modern quasi-experimental toolkit — DiD, RD, synthetic control, shift-share — together with the recent literature showing the obvious estimators are often wrong (W4). Weeks 5–6 turn from *running* estimators to *reading* them: one empirical paper per day under a fixed Reader's-Guide anatomy (W5), then text-as-data and a full module on using LLMs as a research co-pilot without fooling yourself (W6). Weeks 7–8 are the capstone: question to pre-analysis plan (W7), then execution, robustness, writing, the poster, the abstract, and the eight-minute talk at the Symposium (W8). Each week earns the next. You cannot read a weak-instrument disaster in Week 5 if you did not derive 2SLS in Week 3; you cannot stress-test your own DiD in Week 8 if you did not meet the staggered-adoption crisis in Week 4. The dependency order is fixed. A useful way to hold the whole arc in mind: Weeks 1–4 answer "what is a defensible number?", Weeks 5–6 answer "how do I judge someone else's number, including a machine's?", and Weeks 7–8 answer "can I produce a defensible number of my own and survive a referee, a Symposium audience, and the JSSR editor?" The terminal assessment (W8) is the only thing the camp was ever building toward; every earlier week is rehearsal for it. --- ## 2. The ASSIP daily block structure The ASSIP day is **eight hours, 9:00 am – 5:00 pm**, with a one-hour lunch (12:00–1:00). Within those eight hours the cohort runs three repeating blocks, every weekday, plus a fourth block that lives only on Monday mornings: - **9:00–12:00 — Curriculum block.** Three hours of book content: chapter read-along, notebook live work, and problem-set clinic. This is where the textbook's daily rhythm lives — one chapter + notebook + problem set per day, five days a week, five chapters per book-week. - **12:00–1:00 — Lunch.** Hard break. Off-camera. Resists the temptation to schedule "working lunches" — a remote cohort burns out faster than a residential one, and the lunch break is the single biggest dropout-prevention lever you control. - **1:00–4:00 — Mentor research block.** Three hours of work on the student's own capstone project — data acquisition, EDA, replication, identification design, PAP drafting, robustness, writing, poster layout. Pod sub-leaders (the four PhD/RA mentors) hold open Zoom rooms during this block; Prof. Gao drops in by rotation. The mentor research block is the engine that produces the Symposium poster, the JSSR abstract, and the optional full paper. - **4:00–5:00 — Cohort sync.** A one-hour shared close to the day: ten-minute pod standups led by each sub-leader (each pod reports what they built, what is blocked, what they will do tomorrow); forty-minute rotating element — Tue/Thu peer-review pairs walking each other through their notebooks, Wed a guest-lecture or skill clinic, Fri a fifteen-minute show-and-tell where one randomly drawn student demos the week's deliverable to the cohort. Mondays the cohort sync is replaced by the **all-hands mentor session** that opens the morning (see below). - **Monday 9:00–10:00 — All-hands mentor session with Prof. Gao** (IM-4). This sits *inside* the Monday curriculum block, pushing the chapter/notebook/PS work to 10:00–12:00 on Mondays only. Eight Mondays, eight all-hands sessions, each anchored to one of Prof. Gao's papers (see IM-4 for the topic sequence). This is the cohort's intellectual anchor of the week. The week therefore costs each student **40 contact hours** (8 hours × 5 days), of which roughly 15 are curriculum, 15 are mentor research, and 5 are cohort sync; the remaining 5 land in Monday's all-hands mentor session and the buffer that Friday's show-and-tell creates. This is the magnitude the eight-week book was written to. ### Why the day is split this way Three pedagogical pressures shape the schedule. First, **morning curriculum, afternoon research** matches the cognitive science: declarative learning (reading a chapter, deriving an estimator) goes best when working memory is freshest, and applied work (debugging your own pipeline, refining a threats table) is more resilient to mid-day fatigue. Second, **the mentor research block is the camp's actual deliverable engine** — the Symposium poster, the JSSR abstract, and the 8-minute talk are all built in this block, not in the curriculum block; giving it three protected hours a day prevents the typical research-program failure where students "ran out of time" because the curriculum ate the project. Third, **the cohort sync is the remote-format anti-isolation mechanism** — fourteen students working alone in fourteen different homes is the worst-case dropout scenario; the 4:00 pm end-of-day check-in is what makes Tuesday morning's start-up reliable. --- ## 3. The eight-week schedule, week by week What follows is the mapping of the eight book-weeks onto the eight ASSIP program-weeks, with the weekday calendar and the holidays marked. Each week's curriculum block runs the book's five chapters + five notebooks + five problem sets across five days (one per day) and closes Friday with the end-of-week assessment graded against IM-2. | Program week | ASSIP dates | Book content | Mon AM mentor (Prof. Gao) | Friday deliverable | |---|---|---|---|---| | **W1** | Mon Jun 22 – Fri Jun 26 | **W1 — Probability & Inference** (ch1.1–1.5, nb1.1–1.5, ps1.1–1.5) | M1: *"What is a finding?"* | W1 mini-sim (size of a t-test) — Assessment W1 | | **W2** | Mon Jun 29 – **Thu Jul 2** (Fri Jul 3 OFF) | **W2 — OLS Engine** (ch2.1–2.5, nb2.1–2.5, ps2.1–2.5) | M2: *"Why your SE is the whole ballgame"* | W2 SE-choice replication — Assessment W2 (compressed to Thu; Fri off) | | **W3** | Mon Jul 6 – Fri Jul 10 | **W3 — Causal I (PO/matching/IV)** (ch3.1–3.5, nb3.1–3.5, ps3.1–3.5) | M3: *"Natural experiments"* (Deng, Gao & Kim 2020) | W3 weak-IV / AR-coverage task — Assessment W3 | | **W4** | Mon Jul 13 – Fri Jul 17 | **W4 — Causal II (DiD/RD/SC/shift-share)** (ch4.1–4.5, nb4.1–4.5, ps4.1–4.5) | M4: *"Detecting discrimination with a clean design"* (Sun & Gao 2019) | W4 staggered-DiD task **+ first capstone idea memo** — Assessment W4 | | **W5** | Mon Jul 20 – Fri Jul 24 | **W5 — Reading the Frontier I** (Reader's-Guide pack across five classics) | M5: *"Anatomy of a JF paper"* | W5 Reader's Guide on an unseen paper — Assessment W5 | | **W6** | Mon Jul 27 – Fri Jul 31 | **W6 — Reading the Frontier II + AI co-pilot** (ch6.1–6.5, nb6.1–6.5, ps6.1–6.5) | M6: *"Text as data, AI without fooling yourself"* | W6 OOS-validated classifier + leakage audit — Assessment W6 | | **W7** | Mon Aug 3 – Fri Aug 7 | **W7 — Question to Pre-Analysis Plan** (ch7.1–7.5, nb7.1–7.5, ps7.1–7.5; Lab 7 = stand up the capstone repo) | M7: *"How I pick a project — and kill one"* | **PAP + identification memo, filed as `pap-filed` tagged commit** — Assessment W7 | | **W8** | Mon Aug 10 – Tue Aug 11; **Wed Aug 12 = SYMPOSIUM** | **W8 — Execution, Robustness, Poster, Talk** (ch8.1–8.5 + poster + abstract + talk) | M8: *"Defending a result"* | **Symposium deliverables: poster (in-person + virtual breakout) + JSSR abstract + 8-min talk + repo** | Two structural points the table does not surface. First, **Week 2 is shortened by the Jul 3 holiday** — the cohort loses Friday — and the week is reshaped so that the W2 assessment lands on **Thursday Jul 2** rather than Friday, and the W2 cohort sync that would have been Friday's show-and-tell is folded into Thursday's 4:00 pm block. Second, **Week 8 is dominated by the Symposium**: Monday and Tuesday are the final execution-and-rehearsal sprint (dry-run defense on Tuesday afternoon, run by the sub-leaders and Prof. Gao), and **Wednesday August 12 is the Symposium itself** — an in-person poster session at GMU plus virtual breakouts for remote interns (this cohort), with audience including Mason faculty, mentors, families, College of Science leadership, and Mason News press. ### What happens after Wed Aug 12 The four-week paper-refinement phase that the book's W8 chapter (and the JSSR submission standard) imagines is **outside the formal ASSIP window** but inside the camp's deliverable arc. Students who elect to write the optional full paper for JSSR continue to work with their pod sub-leader and with Prof. Gao for four weeks after the Symposium, on a revise-and-resubmit cycle (referee report → R&R memo → revision → final reproducibility check). The Symposium poster and JSSR abstract are the required deliverables; the full paper is opt-in. Plan grading of the **200-point terminal capstone rubric** (IM-2 §4) at the end of that four-week refinement window for students who pursue it, and at the Symposium itself for those who do not. ### A worked Tuesday, end to end (W3 Tuesday as the example) To make the daily shape concrete, here is **Tuesday July 7** — the second day of the Causal-I week, when the cohort is meeting the IV estimator for the first time: - **9:00–9:10 — Daily kickoff.** Sub-leader rotates the lead; reads the day's "what we're building" in one sentence (today: *the first-stage F and why a plausible second stage is not enough*) and the rubric row this day's PS feeds. - **9:10–10:30 — Chapter 3.2 read-along + nb3.2 live work.** Cohort reads ch3.2 (the IV estimator and the two assumptions) under a shared screen, then live-runs nb3.2 — the worked 2SLS pipeline. Pod sub-leaders mute and watch their pod's chat; questions surface through chat-first to keep the shared room from fragmenting. - **10:30–10:40 — Break.** - **10:40–12:00 — ps3.2 clinic.** Students attempt the problem set's six problems in pods; the pod sub-leader floats. The two highest-signal problems are the **first-stage F diagnosis** and the **Anderson–Rubin coverage check** — IM-5 flags which ps3.2 problems separate strong from weak work. - **12:00–1:00 — Lunch.** - **1:00–4:00 — Mentor research block.** Students work on their own capstone projects. Most W3 students are still in the data-acquisition / replication phase; pod sub-leaders run drop-in office hours in their pod's Zoom rooms; Prof. Gao floats by rotation (today: Pod 2's afternoon). - **4:00–4:10 — Pod standups.** Each pod's sub-leader runs a ten-minute round: what built today, what is blocked, what tomorrow brings. - **4:10–4:50 — Tuesday peer-review pairs.** Students pair across pods; each pair walks each other through their morning nb3.2 work and one problem from ps3.2. This is the cohort-glue move — it forces students out of their pod silo and into a fresh perspective. - **4:50–5:00 — Cohort close.** Sub-leader on rotation summarizes; flags Wed's chapter (3.3, matching and propensity scores) and any wiki updates due. Every weekday follows this shape; only the chapter, the live demo, the PS clinic, and the cohort-sync element rotate. --- ## 4. The layered-mentorship model ASSIP 2026's cohort is 14 students — at the high end of typical ASSIP group sizes (2–15). One faculty mentor cannot give 14 remote students the same depth of weekly attention a 4-student in-person group would receive; the cohort is therefore organized as a **layered mentorship**: - **Prof. Lei Gao** is the faculty mentor of record. He runs the Monday all-hands mentor session (9:00–10:00, recorded), is in the room for the Tuesday dry-run defense in Week 8, sets the Symposium poster standards, and rotates through pod afternoon office hours across the week. - **Four PhD / RA sub-leaders** (one per pod) are the daily owners of their pod's progress. Each sub-leader: runs the 4:00 pm pod standup, holds the open drop-in office hour during the 1:00–4:00 mentor research block, grades their pod's daily problem sets against IM-2, escalates any pod concern to Prof. Gao the same day, and is the named co-mentor on each of their pod's three or four capstone posters. - **Four pods of three or four students each** are the working unit. Students stay in their pod Monday–Friday for standups and problem-set clinics; pods are reshuffled exactly once, at the start of W4, to break early cliques and to balance project topics — the second-half pod composition is the one that runs through the Symposium. The layered model is what makes a 14-student remote cohort work. It also makes IM-4 — the mentor-session facilitation chapter — load-bearing in a new way: **Prof. Gao runs the Monday all-hands session and is the escalation backstop, but the sub-leaders are the ones running daily research mentorship**, and IM-4 now documents both layers. --- ## 5. What gets cut, and what never does The ASSIP full-time calendar is generous enough that **nothing in the book gets cut for time**. What changes from a pure-residential design is: - **The W2 Friday holiday compression.** The W2 assessment moves to Thu Jul 2 and the W2 chapter load is front-loaded into Mon–Thu. No content drops; the rhythm tightens for one week only. - **The Symposium deliverable replaces the standalone capstone defense.** The 8-minute talk + poster + JSSR abstract at the Symposium on Wed Aug 12 *is* the capstone presentation; the rubric in IM-2 §5 grades it live. - **The optional full paper moves outside the program window.** Students who opt to write the full paper for JSSR refinement run a four-week R&R cycle with their sub-leader and Prof. Gao after the Symposium; this is graded at the end of that window, not during the camp. You **protect, without exception**: the dependency order of Weeks 1→4; the Monday all-hands mentor session every week; the W7 PAP-and-tag discipline (the pre-registration freeze is non-negotiable — breaking it voids the meaning of every Week-8 p-value); and the W8 reproducibility standard (`make clean && make all`). --- ## 6. A pacing checklist for the instructor Before orientation (Thu Jun 18): - Provision **WRDS seats** for every student via GMU's College of Science institutional subscription — coordinate with **Dr. Amanda Haymond Still / cosassip@gmu.edu** (IM-6). Lead time is the binding constraint; request the batch the week the cohort is announced. - Provision **Hopper accounts** for every student (Appendix B.4), and the camp's Azure OpenAI allocation for the Week-6 AI module (IM-6). - Confirm the **fee-waiver process** has reached every Pell-eligible / reduced-lunch admit before the \$25 application fee or the \$1,299 tuition is invoiced (IM-6). - Stand up the **cohort wiki** (Notion, Confluence, or GitHub-hosted MkDocs — whichever the cohort knows) with one page per student, one page per pod, the eight weekly chapter folders pre-seeded, and the daily-standup template pinned. This is the cohort's shared brain (IM-3). - Run **FM-4 (Prerequisite Self-Test)** the orientation afternoon (Thu Jun 18). Assign Appendix A/B remediation as async work for the orientation weekend to anyone who misses the calculus/probability/Python anchors. - Pin the environment (`python=3.11` + the CONVENTIONS §5 stack) and confirm every notebook runs on a fresh env on the Zoom shared display. - Stand up **GitHub Classroom** and the capstone repo template (Lab 7 / Appendix B.2) so the `pap-filed` tag workflow is ready by W7. Each day: - Confirm the prior day's problem set was graded by the pod sub-leader against its rubric (IM-2) before the 9:00 kickoff, so the day's chapter lands on a cohort already aware of yesterday's highest-signal pitfall (IM-3). - Run the 4:00 pm cohort sync ruthlessly — no day ends without standups, even on Friday with the show-and-tell. - Push every chapter recording, every Monday mentor recording, and every cohort-sync recording to the wiki the same evening. The recording is the equity guarantee for any student who lost bandwidth, had a family emergency, or had an accommodation that needed asynchronous review (IM-6). Each Monday: - Open with the all-hands mentor session (9:00–10:00); record; post-session reflection prompt assigned by 10:00 for Friday review. - Sub-leaders deliver one-sentence pod status to Prof. Gao by 9:00 — *who shipped Friday's deliverable, who needs intervention this week*. Each Friday: - Grade the Friday assessment against IM-2 by Saturday so the following Monday's all-hands can debrief from the cohort's own work, not generic examples. - 4:30–4:50 show-and-tell: one randomly drawn student demos the week's deliverable to the cohort. This is graded only as participation, but the cohort knows their name might be drawn — the *possibility* is what holds attention through Friday afternoon. Week 8 specifically: - Tue Aug 11: dry-run defense for every student (sub-leaders + Prof. Gao). IM-4 gives the question protocol. - Wed Aug 12: Symposium. Virtual breakouts open by 9:00 ET; in-person poster session per the College of Science schedule. - Final week of the four-week paper-refinement window (for students writing the full paper): run `make clean && make all` on every capstone repo. It is the single most objective row in the terminal rubric, and it is the JSSR submission standard. The clock is unforgiving precisely because the deliverable is real: these posters and abstracts head to a real Symposium audience and a real journal. Pace the camp so that the freeze holds, the code rebuilds, and every student can say the sentence the whole program exists to let them say truthfully — *here is what I found, here is why you should believe it, and here is the command that lets you check me.*