Social Sciences Set

What is social science?

Systematic empirical study of social behavior, institutions, and relationships.

What is a research question?

A focused question that guides what evidence and comparisons are needed.

What is a theory?

An organized explanation linking concepts and generating observable implications.

What is a hypothesis?

A specific, testable expectation derived from a theory.

What is a variable?

A characteristic that can take different values across units or times.

What is an independent variable (X)?

A proposed cause or explanatory predictor.

What is a dependent variable (Y)?

The outcome the researcher seeks to describe or explain.

What is a unit of analysis?

The entity being studied, such as a person, county, state, or election.

What is an observation?

A recorded value or set of values for a unit at a particular time.

What is a population?

The full group of units about which a researcher wants to infer.

What is a sample?

The subset of a population actually observed.

What is descriptive inference?

Using observed evidence to infer unobserved facts or patterns.

What is causal inference?

Inferring how changing X would change Y relative to a counterfactual.

What is systematic empirical falsification?

Testing explicit predictions against observations that could disconfirm a claim.

What is the file-drawer problem?

Unpublished null results can distort the apparent balance of evidence.

What is publication bias?

Publication probability depends on results, producing a misleading evidence base.

What is a rival explanation?

An alternative mechanism that could produce the same observed outcome.

What is an observable implication?

A predicted pattern that should be present if a theory is correct.

What is research leverage?

Evidence or comparisons that help discriminate among explanations.

What is a cross-sectional dataset?

Many units observed at one time or period.

What is a time-series dataset?

One unit observed repeatedly over time.

How do you calculate the mean?

x̄ = Σxᵢ/n; add observations and divide by their count.

What is the median?

The middle ordered value, or mean of the two middle values.

What is the mode?

The most frequently observed value.

When is the median preferable to the mean?

With strong skew or extreme outliers, because it is less sensitive.

What is the range?

Maximum minus minimum.

What is an outlier?

An observation unusually far from the bulk of the distribution.

What is sample variance?

s² = Σ(xᵢ−x̄)²/(n−1).

What is standard deviation?

s = √s²; typical spread of individual observations in original units.

What does a histogram display?

The distribution of a numerical variable using frequency bins.

What does a scatterplot display?

The relationship between two quantitative variables.

What does a boxplot summarize?

Median, quartiles, spread, and possible outliers.

What is skewness?

Asymmetry in a distribution; the long tail indicates skew direction.

How does an outlier affect a mean?

It can pull the mean toward the extreme value.

What is a misleading graph axis?

A scale choice, such as truncation, that exaggerates or hides differences.

What is a proportion?

A part divided by the whole, between 0 and 1.

What is a percentage point change?

The arithmetic difference between two percentages.

How does a bar chart differ from a histogram?

Bars represent categories; histogram bins represent numeric intervals.

What is a frequency distribution?

Counts or shares of observations across values or categories.

What is a z-score?

z=(x−μ)/σ; number of population standard deviations from the mean.

What is probability?

A numerical representation of uncertainty from 0 to 1.

What does independence mean in probability?

P(A|B)=P(A); learning B does not change the probability of A.

What is the law of large numbers?

Sample averages converge toward expected values under suitable conditions.

What is the central limit theorem?

Under suitable conditions, standardized sample means approach a normal distribution.

What is sampling error?

Random difference between a sample estimate and the population value.

What is selection or sampling bias?

A systematic sampling process makes the observed sample unrepresentative.

What is standard error?

The standard deviation of an estimator across repeated samples.

What is the standard error of the sample mean?

SE(x̄)=s/√n for independent observations.

If n quadruples, what happens to SE of a mean?

It approximately halves if variability and assumptions stay the same.

What is a sampling distribution?

Distribution of a statistic across hypothetical repeated samples.

What is a parameter?

A fixed but often unknown numerical property of a population.

What is a statistic?

A quantity calculated from sample data.

What is the null hypothesis H₀?

A specific claim tested against the evidence, often no effect.

What is the alternative hypothesis H₁?

The competing claim, such as a nonzero effect.

What is a p-value?

Probability of results at least as extreme under H₀ and model assumptions.

What does p=.02 NOT mean?

It does not mean a 2% probability the null is true.

What is alpha (α)?

A prespecified significance threshold, often 0.05.

What is a Type I error?

Rejecting a true null hypothesis; false positive.

What is a Type II error?

Failing to reject a false null hypothesis; false negative.

What is statistical power?

Probability of rejecting H₀ under a specified alternative.

What is a confidence interval?

An interval procedure with a stated long-run coverage rate.

How is a basic 95% CI for a mean approximated?

Estimate ± 1.96×SE for an approximately normal estimator.

What is a t-statistic?

t=(estimate−null value)/SE.

What does statistical significance NOT establish?

Causation, practical importance, or certainty that a hypothesis is true.

What is Pearson correlation r?

A standardized measure of linear association from −1 to +1.

What does r=−0.8 indicate?

A strong negative linear association, not necessarily causation.

What does r=0 imply?

No linear association; a nonlinear relationship may still exist.

What is simple linear regression?

A fitted relation Y=β₀+β₁X+ε.

What is a regression intercept?

Predicted Y when all included predictors equal zero.

What is a regression slope?

Predicted Y difference for a one-unit X difference, other included Xs held fixed.

What is a regression residual?

Observed Y minus fitted Y.

What is omitted-variable bias?

An excluded cause related to X distorts the estimated X coefficient.

What is endogeneity in regression?

A predictor is correlated with the model error term.

What is reverse causality?

Y influences X, complicating claims that X causes Y.

What is multicollinearity?

Predictors are strongly correlated, making separate coefficients imprecise.

What is a confounder?

A variable associated with X and causally related to Y, biasing naive comparisons.

What is measurement error?

Recorded values differ from the underlying concept or quantity.

Why inspect residual plots?

To detect patterns, nonlinearity, unusual observations, or changing error spread.

What does holding variables constant mean?

Comparing observations at the same values of other included predictors.

What is a regression coefficient standard error?

An estimate of the sampling variability of that coefficient.

What is the fundamental problem of causal inference?

The same unit cannot reveal both treated and untreated potential outcomes at once.

What is an individual treatment effect?

Yᵢ(1)−Yᵢ(0), usually unobservable for a single unit.

What is a counterfactual?

The outcome that would occur under the alternative treatment condition.

What does random assignment do?

Makes treatment independent of potential outcomes in expectation.

Why is random sampling different from random assignment?

Sampling supports population generalization; assignment supports causal comparisons.

What is a randomized controlled trial?

A design allocating treatment by chance and comparing outcomes across groups.

Does random assignment guarantee identical groups?

No; chance imbalances can remain, especially in small samples.

What is a control group?

A comparison group not receiving the studied treatment.

What is treatment-group selection bias?

Preexisting differences between treated and untreated groups distort effects.

What is a pretest-posttest design weakness?

Other changes over time can mimic a treatment effect.

What is the STAR class-size experiment?

Random assignment to class-size conditions to study student outcomes.

Why might a tutoring program comparison be biased?

Students self-selecting into tutoring may differ in motivation or prior performance.

What is a causal mechanism?

The process through which X produces an effect on Y.

What is KKV’s central methodological ambition?

Use common standards of scientific inference in qualitative and quantitative research.

What is indeterminacy?

Available observations cannot distinguish among multiple competing explanations.

What is intentional selection of cases?

Choosing informative cases to test or distinguish explanations.

What is selection on the dependent variable?

Choosing cases based on their outcomes, often restricting informative variation.

Why is studying only successful revolutions risky?

It omits failed cases needed to explain why revolutions succeed.

What is survivorship bias?

Drawing conclusions only from cases that remain observable or survive.

What is unit homogeneity?

Comparable units are assumed to share a common causal response under stated conditions.