Start with your syllabus, not a textbook
Econometrics combines statistical reasoning with economic questions and data. A useful study plan therefore needs three kinds of practice: explaining what a model assumes, carrying out calculations or interpreting software output, and deciding what a result does—and does not—allow you to conclude.
Before you schedule topics, collect your module outline, lecture notes, problem sets and any past questions your course provides. Mark each topic as required, optional or not covered. The plan below covers common introductory and intermediate topics, but courses differ. Follow your own syllabus: for example, study instrumental variables, panel methods or time-series models only if they are in scope.
Set aside about seven hours a week for six weeks: three two-hour sessions and one one-hour review. If you have less time, keep the session order and reduce the number of practice questions; do not skip retrieval and error review. Use the software required by your course—such as R, Stata or Python—and keep your notes and worked examples on one revision board.
A due queue for an econometrics study set might look like this:
The queue is a sample snapshot: it directs the first review session towards cards already scheduled for retrieval, rather than letting you spend the hour rereading familiar material.
Week 1: Find the gaps and rebuild OLS
Session 1 — 2 hours: diagnose, then sort. Spend 25 minutes attempting a small selection of problems from your course without notes. Include one interpretation question, one calculation or output-reading question, and one question about assumptions. Use the remaining time to identify why each error happened: missing definition, algebra, wrong standard error, or a conclusion that overstates the evidence. Put the topics in syllabus order on your board.
Session 2 — 2 hours: simple and multiple regression. Review the population regression model and the meaning of the intercept and slope. Then work through an ordinary least squares example by hand or from course material. Practise interpreting a coefficient while holding other included regressors fixed. In your software, fit the model and explain the coefficient, standard error, test statistic and confidence interval in words.
Session 3 — 2 hours: assumptions and interpretation. Make a short checklist for the assumptions your course uses. Distinguish a condition needed for unbiasedness from one needed for the usual standard-error formula. Solve questions that ask what happens when a relevant variable is omitted or a regressor is measured imperfectly; state the direction of bias only when the information supports it.
Session 4 — 1 hour: closed-note retrieval. Answer your new cards, then revisit each Week 1 error. Add only the facts or distinctions you could not recall. A working board should hold explanations you can use, not copied lecture pages.
A board for the first week should connect model, inference and interpretation rather than keeping those ideas in separate sets of notes:
- Population model: yᵢ = β₀ + β₁xᵢ + uᵢ
- Slope: expected change in y for a one-unit increase in x, holding included regressors fixed in a multiple regression
- Report units and the comparison implied by the regressor
- Statistical significance is not the same as economic importance
- Association alone does not establish causation
OLS chooses coefficients to minimise the sum of squared residuals. A residual is yᵢ − ŷᵢ; it is not the unobservable error uᵢ.
- For a single coefficient, t = (estimate − null value) / standard error
- A confidence interval combines the estimate with a critical value and its standard error
Week 2: Inference and model specification
Session 1 — 2 hours: tests and intervals. Practise stating a null and alternative hypothesis, choosing the relevant test, and interpreting the result at the significance level specified in the question. Work from the standard errors and critical values supplied by your course or question; do not memorise a universal cutoff as if it applied to every test. For each answer, state the estimate, uncertainty and conclusion in context.
Session 2 — 2 hours: transformations and categorical variables. Review logarithmic specifications and translate coefficients into the interpretation required by the model and the course convention. For dummy variables, identify the reference category and interpret the coefficient relative to it. Practise an interaction term by writing down the relevant conditional effect rather than reading one coefficient in isolation.
Session 3 — 2 hours: specification decisions. Work through questions about omitted variables, irrelevant regressors, functional form and multicollinearity. For each, distinguish a problem with the coefficient’s interpretation from a problem with its precision. Use an output table to explain what changes when you add a control variable, and avoid claiming that a changed coefficient proves a particular cause.
Session 4 — 1 hour: mixed retrieval. Review Week 1 cards and test yourself on Week 2 interpretations. For every missed question, add a concise card that asks for one decision or distinction. Avoid cards that simply ask you to reproduce a whole lecture.
Week 3: Diagnostics and robust reasoning
Session 1 — 2 hours: heteroskedasticity. Learn how heteroskedasticity affects the usual OLS standard errors and what a heteroskedasticity-robust standard error changes. Practise interpreting the same coefficient with conventional and robust standard errors. Be precise: changing the estimated standard error does not, by itself, change the OLS coefficient estimate.
Session 2 — 2 hours: serial correlation and other diagnostics. If your course covers time-dependent errors, review the assumptions behind its tests and the consequences for inference. If it does not, use this session for the diagnostics included in your syllabus. For each method, record the question it answers, the assumptions it relies on and the appropriate response to the result.
Session 3 — 2 hours: timed problem set. Choose questions from Weeks 1–3 and work under a fixed two-hour limit. Include interpretation of regression output and an explanation of a diagnostic result. Mark against your notes or the solutions supplied by your course, and label each lost point by topic and error type.
Session 4 — 1 hour: repair the weakest topic. Use your error log to choose one topic, then explain it aloud without notes and solve one new problem. A topic is not repaired simply because the worked solution now looks familiar.
A topic picker helps turn that error log into a short, targeted session:
Robust standard errors · Struggling — getting 4 of 9 cards wrong, scored 58% when last examined
Use the suggested topic as a starting point: spend the ten minutes retrieving key ideas, then use the rest of the session on a worked question. A warm section needs another check; a section with no heat still needs occasional review, but does not have to take priority today.
Week 4: Identification and instrumental variables
Session 1 — 2 hours: endogeneity and identification. If these topics are on your syllabus, learn the difference between correlation and a causal effect, and identify why a regressor might be correlated with the error term. Work through omitted variables, reverse causality or measurement error only to the depth required by your course. State which assumption is in question before choosing a method.
Session 2 — 2 hours: instrumental variables. For each example, explain instrument relevance and the exclusion restriction in the context of the question. Practise describing the first stage and the outcome equation if these are part of your course. Do not call an instrument valid just because it predicts the endogenous regressor; discuss why it would not affect the outcome through another route.
Session 3 — 2 hours: application and critique. Interpret a supplied IV result and compare it with an OLS estimate where appropriate. Identify the population or variation the method may speak to, following the terminology taught in your module. If IV is not assessed, spend this session on the next advanced topic in your outline.
Session 4 — 1 hour: cumulative cards. Mix OLS, inference and identification cards. Before checking each answer, say the relevant assumption and what would fail if it did not hold.
Week 5: Panel data or time series, according to your course
Choose the branch that matches your syllabus; do not try to learn both in one week if neither is familiar.
Panel-data branch. In Session 1, compare pooled regression and fixed effects, and practise explaining how within-unit variation is used. In Session 2, study random effects and the assumptions that distinguish it from fixed effects; use the comparison taught in your course. In Session 3, interpret a panel output and explain what unit and time effects account for. In Session 4, retrieve definitions and redo one problem you previously missed.
Time-series branch. In Session 1, review trends, stationarity and the consequences of modelling non-stationary series, as covered in your module. In Session 2, practise interpreting the lag structure of the models on your syllabus, such as an AR model. In Session 3, work through a time-series question involving diagnostics or forecasting, using the definitions and tests taught in your course. In Session 4, retrieve those definitions and redo one missed problem.
Each session is two hours except the final one-hour retrieval session. If your module covers both branches, divide the topic across two weeks by extending the plan rather than rushing through both.
Week 6: Consolidate and practise under time limits
Session 1 — 2 hours: build a mixed paper. Select questions across the topics you have studied, using your course’s past questions or problem sets. Match the time allowed to the question weights or instructions provided by your course; do not assume another institution’s format applies.
Session 2 — 2 hours: complete and mark. Work without notes. Mark your reasoning, not just the final number: did you state the estimand, identify the assumption, interpret the uncertainty and answer the question asked? Put every lost point into the error log.
Session 3 — 2 hours: targeted repair. Rank the recurring errors. Spend the first hour on the weakest topic and the second on a fresh problem that tests it in context. If the same error appears twice, make a card that forces you to choose the right interpretation or assumption.
Session 4 — 1 hour: final retrieval and next steps. Review due cards and explain the main methods aloud from memory. Use the course outline to check what remains uncovered. If there is more material, extend the schedule; a six-week plan is a framework, not evidence that every course has the same scope.
A weekly error review can make the next study decision concrete:
These are sample practice results, not an exam grade. Look first at the weakest area, but also notice frequency: a recurring mistake in a commonly practised topic may deserve more attention than one isolated miss. After a focused session, test yourself with a new question and see whether the error rate changes.
Keep the plan usable
At the end of each week, spend five minutes updating three things: topics covered, errors that repeated, and sessions available next week. If you miss a session, resume with the next planned activity rather than trying to cram every skipped hour into one sitting. Keep a small formula sheet only for expressions your course expects you to use; pair each formula with an explanation of its assumptions and interpretation.
Your main evidence of progress is whether you can explain a result and select a method on an unfamiliar question, not how many pages you have reviewed. Keep worked answers short enough to retrieve, and return to difficult cards at spaced intervals. For details about assessed content, software conventions or question formats, use your own module materials and ask your course team rather than relying on a generic plan.
How MySummaries helps
MySummaries can turn your own econometrics notes, slides and problem material into a revision board, then create spaced-repetition cards and written practice from that board. Use the board to organise the topics in your syllabus, review due cards during the short sessions, and direct longer sessions towards areas where your practice shows repeated errors. Start at portal.mysummaries.app.