How to study this econometrics deck

Econometrics questions often ask you to interpret an estimate, identify an assumption, or explain why a result may not support a causal claim. Flashcards work well for these tasks when each card tests one precise idea. They work less well when the answer is a vague prompt such as “explain regression”.

MySummaries can turn your own revision-board material into flashcards. This deck covers common introductory econometrics concepts; compare it with your course notes, since notation and the topics assessed vary by course.

Use the cards as a short retrieval session, not as a page to reread. For each one, read the question, formulate your answer before turning it over, then compare your wording with the answer. Grade it by how readily you recalled the full fact: use Again if you could not retrieve it, Hard if you needed substantial effort, Good if you recalled it correctly, and Easy if it came immediately and clearly. If you remembered the headline but missed a condition or interpretation, choose Hard rather than treating it as mastered.

A deck built around core econometric methods ends up looking like this:

Cards — Econometrics — core methods18 due

In a simple OLS regression, what does the slope coefficient measure?

The change in the conditional mean of Y associated with a one-unit increase in X.

All 18 cards
In a simple OLS regression, what does the slope coefficient measure?The change in the conditional mean of Y associated with a one-unit increase in X.
What is the OLS estimator in matrix form?β̂ = (X′X)⁻¹X′y, when X′X is invertible.
What is the zero conditional mean assumption?E(u | X) = 0: the error has conditional mean zero at every value of the regressors.
What is the omitted variable bias condition?An omitted variable biases an included coefficient when it affects Y and is correlated with the included regressor, conditional on other included variables.
What is the standard interpretation of a coefficient in a log-level model, ln(Y) = β₀ + β₁X + u?A one-unit increase in X is associated with approximately 100β₁ percent change in Y; the exact percentage change is 100(exp(β₁) − 1).
In a log-log model, ln(Y) = β₀ + β₁ ln(X) + u, how is β₁ interpreted?β₁ is the elasticity: a 1% increase in X is associated with approximately a β₁% change in Y.
What does a binary regressor coefficient measure in a linear model with an intercept?The conditional difference in mean Y between the group coded 1 and the reference group coded 0, holding other regressors fixed.
In Y = β₀ + β₁X + β₂D + β₃(X×D) + u, what is the slope of X when D = 1?β₁ + β₃. The interaction coefficient β₃ is the difference in the X slope between D = 1 and D = 0.
What does a heteroskedasticity-robust standard error change?The estimated uncertainty of a coefficient, allowing the error variance to differ across observations; it does not change the OLS coefficient itself.
What is the null hypothesis in a two-sided t-test of one regression coefficient?H₀: βⱼ = 0, against H₁: βⱼ ≠ 0.
What does a p-value of 0.03 mean for a specified test?If the null hypothesis and test assumptions were true, the probability of a test statistic at least as extreme as the observed one is 0.03.
What does an F-test of several coefficients jointly equal to zero test?The joint null that all specified coefficients are zero; rejection means the data are inconsistent with that joint restriction at the chosen significance level.
What two conditions are required for an instrument Z to identify the effect of endogenous X on Y?Relevance: Z predicts X. Exogeneity: Z is independent of the structural error, commonly expressed as Cov(Z,u) = 0; exclusion requires Z to affect Y only through X.
What is the first stage in two-stage least squares with instrument Z?Regress the endogenous regressor X on the instrument(s) and included exogenous controls, then use the fitted values of X in the outcome stage.
What does an individual fixed effect remove in a panel model?Time-invariant differences across individuals that may be correlated with the regressors.
What is the parallel-trends assumption in a basic difference-in-differences design?Without treatment, the treated and comparison groups would have had the same average outcome trend over time.
What is the difference-in-differences estimator for two groups and two periods?(Treated post − treated pre) − (Control post − control pre). Its causal interpretation requires assumptions including parallel trends.
What does R² measure in a regression with an intercept?The share of sample variation in the dependent variable explained by the fitted values relative to its sample mean; it does not establish causation.
An 18-card econometrics deck, with a prompt and a precise answer for each concept.

That is a MySummaries deck, filled with econometrics material. Yours is written from your own notes. Start free

Turn the cards, then diagnose the miss

The first card is face-up with its answer and Again, Hard, Good and Easy grading choices. Work through it, then continue to the remaining prompts. Say the answer aloud or write a short version before checking the back. In particular, do not stop at “it measures the effect”: specify the comparison, units, or condition the question asks for.

After each card, grade your recall rather than your confidence. If you said that OLS requires “no correlation” but could not state E(u | X) = 0, you have not recalled the zero conditional mean assumption precisely. If you remembered the difference-in-differences formula but forgot what parallel trends means, grade that card as a struggle and return to it soon.

The deck is deliberately built from single facts. That makes it possible to locate a gap: a wrong answer about an interaction is different from a wrong answer about instrument validity. When a course uses different notation, keep the concept and rewrite the card using the notation in your lecture notes. For example, an instrument may be denoted by Z or by another symbol; the relevance and exogeneity requirements remain the key distinction.

The must-not-miss core

Before you finish a session, check whether you can explain these four ideas without looking. They link many common econometrics problems: what OLS estimates, why omitted variables matter, how uncertainty is assessed, and what makes an instrumental-variable argument credible. If one is weak, revisit the related cards rather than trying to memorise every formula as an isolated symbol.

A board’s “Must not miss core” for these methods could look like this:

Key pointsEconometrics — core methods
OLS: β̂ = (X′X)⁻¹X′y when X′X is invertible; causal interpretation needs assumptions such as E(u | X) = 0.
Omitted variable bias requires both an effect on Y and correlation with an included regressor, conditional on other controls.
A heteroskedasticity-robust standard error changes estimated uncertainty, not the OLS coefficient; a p-value is conditional on the null and test assumptions.
IV requires relevance and exogeneity; a basic difference-in-differences causal interpretation requires parallel trends in the absence of treatment.
Four foundational econometrics checks distilled from the deck.

Use missed marks to make a better card

When you miss a card, record the exact confusion. “I forgot IV” is too broad to fix. “I treated instrument relevance as proof of exogeneity” identifies the misconception: a first stage can show that Z predicts X, but it does not establish that Z is independent of the structural error or affects Y only through X.

Use the same approach with written exercises. Compare your answer with your notes or the solution, identify the missing fact, then turn that fact into a question with one focused answer. Keep the original card if its wording is useful; add a new one only when it tests a genuinely different distinction. This avoids growing a deck full of duplicate prompts.

A card waiting after the same gap has been missed twice could read:

Remediation tray

You lost this mark twice: if an instrument predicts the endogenous regressor, which separate assumption is still needed for a causal IV interpretation, and what does it require?

Add cardDismiss
A focused reminder about the distinction between instrument relevance and instrument exogeneity.

A repeatable session

A practical session can take 15–25 minutes. Start with due cards and answer each before revealing it. Spend another few minutes on the cards you graded Again or Hard, stating the answer in your own words. Finish by applying one concept to a small example from your course: interpret a coefficient with its units, name the omitted assumption in a causal claim, or explain which groups and periods enter a difference-in-differences calculation.

Keep the answers tied to your course material. For instance, do not infer that a statistically significant coefficient is economically large, or that a high R² proves a model is correctly specified. Those conclusions require additional evidence and interpretation. When your lecturer defines a particular estimator, test, or assumption differently or adds conditions, prioritise that course-specific wording and adapt the cards accordingly.

How MySummaries helps

Build an econometrics revision board from your own notes, then use it to generate and review focused flashcards as your understanding develops. Open MySummaries to organise your study material and practise recall from it.