← Reddit Interview Insights

Reddit·Data Scientist·Technical Phone Screen·Senior

SeniorPrefer not to say
Jun 2026Remote

Summary

Reddit data scientist interview that went deep on causal inference, specifically synthetic control methods. One long technical question that covered a lot of ground, felt more like a dissertation defense than a typical DS screen.

Questions Asked (4)

Q1

Walk through the core identification assumptions behind synthetic control, and explain how each one, if violated, would distort your treatment effect estimates.

A/B Testing & ExperimentationTechnical Trade-offs
Author's notes

I knew convex hull and no interference but fumbled explaining why pre-period fit matters as a proxy for unobserved confounders.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by briefly defining synthetic control and its purpose, then systematically list each core assumption (no interference, parallel trends, convex hull, no anticipation, and donor pool validity). For each, explain the assumption, how it can be tested or assessed, and the specific direction of bias if violated. Conclude by discussing practical implications and robustness checks.

Pro tip: Emphasize that synthetic control is most credible when you can show pre-treatment fit and placebo tests; if an assumption is violated, quantify the potential bias rather than just stating it.

1. Define synthetic control and its goal

Briefly explain that synthetic control constructs a weighted combination of control units to mimic the treated unit's pre-treatment outcomes, then estimates the treatment effect as the post-treatment difference.

2. List core identification assumptions

Enumerate the key assumptions: no interference (SUTVA), parallel trends (or that the synthetic control would have followed the treated unit's counterfactual), convex hull condition, no anticipation, and valid donor pool.

3. Explain each assumption and its violation consequences

For each assumption, describe what it means, how it might be violated, and the specific distortion to the treatment effect estimate (e.g., bias direction, attenuation, exaggeration).

4. Discuss tests and robustness checks

Mention how to assess assumptions: pre-treatment fit, placebo tests, leave-one-out, and sensitivity analyses. Explain what to do if assumptions are violated.

5. Connect to practical implications

Summarize how violations affect decision-making and what alternative methods or adjustments could be used.

Key Points to Mention

  • No interference (SUTVA): treated unit's treatment does not affect control units; violation leads to biased counterfactual and over/underestimation.
  • Parallel trends / no differential shocks: synthetic control must track treated unit pre-treatment; violation causes biased post-treatment estimates.
  • Convex hull condition: treated unit must lie within the convex hull of donor units; violation means poor pre-treatment fit and unreliable estimates.
  • No anticipation: treatment effect should not appear before intervention; violation leads to underestimation of effect.
  • Donor pool validity: donors should be unaffected by treatment and similar in relevant characteristics; violation introduces bias.
  • Placebo tests and pre-treatment fit: assess assumption validity and quantify uncertainty.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q2

How do you select predictor variables and lag periods for synthetic control without accidentally leaking post-treatment information into your model?

A/B Testing & ExperimentationTechnical Trade-offs
Author's notes

This one I actually felt okay about.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Emphasize that synthetic control is a pre-treatment design: all predictor and lag choices must be justified using only pre-intervention data and domain knowledge. Describe a principled process that separates design decisions from outcome evaluation, and highlight how you would validate the absence of leakage through placebo tests and temporal cross-validation.

Pro tip: Frame the problem as 'designing a pre-treatment-only pipeline' and explicitly state that you never look at post-treatment outcomes when selecting predictors or lags—this signals deep understanding of causal inference and avoids the most common pitfall.

1. Define the pre-treatment window and outcome

Clearly specify the intervention date and restrict all data used for model selection to the period before that date. Identify the outcome variable and ensure it is not used in any form during predictor/lag selection.

2. Select predictors based on domain knowledge and pre-treatment fit

Choose predictors that are theoretically related to the outcome and measured pre-treatment. Use pre-treatment fit metrics (e.g., RMSE on pre-period) to compare candidate sets, but avoid using post-treatment outcomes to guide selection.

3. Determine lag periods via temporal validation

Test lag lengths using only pre-treatment data with time-series cross-validation (e.g., rolling-origin). Select lags that minimize pre-treatment prediction error without peeking at post-treatment performance.

4. Validate with placebo tests and sensitivity analysis

Run placebo interventions in the pre-period to check that the method does not produce false positives. Assess sensitivity of results to predictor and lag choices to ensure robustness.

5. Document and pre-register choices

Record all decisions (predictors, lags, validation metrics) before examining post-treatment outcomes. This prevents unconscious leakage and increases credibility of the analysis.

Key Points to Mention

  • Strict separation of pre-treatment and post-treatment data; never use post-treatment outcomes for model selection.
  • Use of domain knowledge and theory to guide predictor selection, not just statistical fit.
  • Temporal cross-validation (e.g., rolling-origin) for lag selection to avoid look-ahead bias.
  • Placebo tests and sensitivity analysis to validate the synthetic control model.
  • Pre-registration or documentation of design choices to prevent data leakage.
  • Avoiding overfitting by limiting the number of predictors and lags relative to pre-treatment periods.

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q3

Describe your full inference strategy for synthetic control: how do you construct treatment effects, generate p-values, and build uncertainty bands?

A/B Testing & ExperimentationProduct Analytics & Metrics
Author's notes

Talked through pointwise vs cumulative effects and the placebo distribution logic using donor units.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Structure your answer around the three core components: treatment effect estimation, p-value generation, and uncertainty quantification. Emphasize the synthetic control method's reliance on a donor pool and placebo tests, and connect it to Reddit's experimentation context by highlighting how it handles non-randomized settings. Be concrete about algorithms (e.g., constrained regression, permutation tests) and validation steps.

Pro tip: Mention that you validate the synthetic control by checking pre-treatment fit and running in-time placebo tests; this shows you understand the method's assumptions and pitfalls. Also, note that p-values from permutation tests are exact but limited by the number of donors, so you might use additional robustness checks like leave-one-out.

1. Construct the synthetic control

Explain how you build a weighted combination of donor units that mimics the treated unit's pre-treatment outcomes. Use constrained optimization (e.g., nested optimization or quadratic programming) to minimize pre-treatment RMSE, possibly including covariates.

2. Estimate treatment effects

Compute the difference between the treated unit's post-treatment outcomes and the synthetic control's post-treatment outcomes. This difference is the estimated treatment effect over time.

3. Generate p-values via placebo tests

Perform in-space placebo tests: iteratively assign treatment to each donor unit, compute the effect, and compare the treated unit's effect to the distribution of placebo effects. The p-value is the proportion of placebo effects as extreme as the treated effect.

4. Build uncertainty bands

Use the placebo distribution to construct confidence intervals or uncertainty bands around the treatment effect. Alternatively, use bootstrap or Bayesian methods if assumptions allow, but emphasize permutation-based intervals for robustness.

5. Validate and interpret

Check pre-treatment fit and placebo tests in time to ensure validity. Discuss limitations (e.g., donor pool size, spillovers) and how you would communicate uncertainty to stakeholders.

Key Points to Mention

  • Donor pool selection and pre-treatment fit metrics (e.g., RMSE, R-squared)
  • Constrained optimization for weights (e.g., simplex constraint, non-negative weights summing to 1)
  • In-space placebo tests for p-values and the importance of the rank of the treated unit's effect
  • In-time placebo tests to validate the method's assumptions
  • Uncertainty quantification via placebo distributions or bootstrap
  • Comparison to difference-in-differences and when synthetic control is preferred

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.

Q4

What diagnostics do you run when pre-period fit is poor, and what are your options for fixing it or switching to a different method entirely?

A/B Testing & ExperimentationRoot Cause AnalysisTechnical Trade-offs
Author's notes

Probably my strongest answer of the session.

Create a free account to read the full note

AI HintsAI Generated

Suggested Approach

Start by explaining that poor pre-period fit indicates a violation of the parallel trends assumption, which is critical for valid causal inference. Then outline a systematic diagnostic process: check for outliers, seasonality, and structural breaks; try transformations or alternative control groups; and if all else fails, consider switching to a method like synthetic control or matched cohorts. Emphasize that the choice depends on the trade-offs between bias and variance, and the specific context of the experiment.

Pro tip: Always quantify the impact of poor pre-period fit on your effect estimate—e.g., by simulating a placebo test—to decide whether the violation is fatal or manageable. This shows you're pragmatic and focused on decision-making, not just diagnostics.

1. Diagnose the nature of the poor fit

Plot pre-period trends for treatment and control groups, and compute metrics like RMSE or placebo tests. Identify whether the issue is due to outliers, seasonality, level shifts, or divergent trends.

2. Check for data quality and external factors

Investigate data collection issues, missing data, or external events (e.g., holidays, product changes) that could cause divergence. Ensure the pre-period is long enough and stable.

3. Attempt to fix the fit

Consider transformations (e.g., log, differencing), adding covariates, using propensity score matching, or trimming outliers. If seasonality is present, adjust with seasonal decomposition or include time fixed effects.

4. Evaluate alternative methods

If fit remains poor, switch to methods that are more robust to parallel trends violations, such as synthetic control, difference-in-differences with staggered adoption, or causal impact using Bayesian structural time series.

5. Assess trade-offs and make a decision

Weigh the bias-variance trade-off, interpretability, and business constraints. Run sensitivity analyses to see how results change under different assumptions, and communicate uncertainty clearly.

Key Points to Mention

  • Parallel trends assumption and its importance in difference-in-differences
  • Placebo tests and pre-period fit metrics (e.g., RMSE, R-squared)
  • Common causes of poor fit: outliers, seasonality, structural breaks, data quality issues
  • Remedies: transformations, covariate adjustment, matching, trimming
  • Alternative methods: synthetic control, Bayesian structural time series, matched cohorts
  • Trade-offs: bias vs. variance, interpretability, and business impact

AI-generated suggestions, not part of the candidate's original notes. May be inaccurate — verify before relying on them.