Questions: Comparative Methods in Historical Analysis

Question 1

A historian studying industrialization selects two countries that differ in colonial history, religion, language, geographic size, and political structure — yet both industrialized rapidly in the nineteenth century. Which research design does this represent, and what kind of question can it answer?

• A. A most-similar design; it isolates which specific variable accounts for their shared outcome of rapid industrialization
• B. A most-different design; it asks what common mechanism produced the same outcome despite radically different background conditions
• C. A controlled experiment; the variation in background factors serves as the experimental manipulation
• D. A case study; comparative designs require cases that differ only on a single variable

Answer: (B)

Selecting cases that differ on many background variables but share the outcome of interest is a most-different design. The logic is: if cases that look nothing alike on the surface all produced rapid industrialization, the explanation must be something that cuts across those surface differences — a common mechanism despite divergent contexts. Contrast this with most-similar design, where cases are alike in most respects but differ in outcome, which isolates the specific variable responsible for divergence. A most-different design cannot isolate individual causes but can establish that a candidate explanation must work independently of many background conditions.

Question 2

What is the core methodological challenge of 'commensurability' in comparative historical analysis, and why does ignoring it undermine the comparison?

• A. Commensurability refers to finding cases from the same time period — cross-period comparisons are methodologically invalid
• B. Commensurability is the question of whether two cases are genuinely the same kind of thing in ways that make comparison meaningful; premature comparison imports the assumptions of one case into the analysis of another
• C. Commensurability means that the statistical sample sizes must be equal — small-N comparisons are inherently incommensurable
• D. Commensurability requires that both cases use the same primary sources — different archival bases make comparison invalid

Answer: (B)

Commensurability asks: are we actually comparing the same thing? Concepts like 'feudalism,' 'revolution,' 'state,' or 'capitalism' carry assumptions derived from specific historical contexts. Applying them across cases without examining whether they fit can create the illusion of comparison while actually forcing one case into the conceptual mold of another. For example, comparing 'feudalism' in Japan and medieval Europe requires first examining whether the specific institutional arrangements in each case functioned similarly enough that the same concept illuminates rather than distorts both.

Question 3

In a most-similar comparative design, researchers select cases that are alike in most relevant respects but differ on the outcome to be explained, in order to isolate what accounts for the divergence.

• T. True
• F. False

Answer: True

True. Most-similar design is the historical analogue of a controlled comparison: by holding background conditions roughly constant, variation in the outcome should track variation in the key explanatory variable. For example, if two countries with similar levels of economic development, colonial history, and institutional structure produce different democratization outcomes, the comparison directs attention to whatever differs between them — the candidate cause. The logic mirrors the 'difference method' in Mill's canons of induction, adapted to cases where true experimental control is impossible.

Question 4

The choice of which cases to compare in historical research is a neutral methodological decision, guided mainly by analytical considerations about variation and outcome, and does not reflect the historian's theoretical assumptions or perspective.

• T. True
• F. False

Answer: False

False — case selection is never neutral. Choosing which cases to compare, what the baseline 'normal' case is, and which conceptual vocabulary to use reflects theoretical assumptions and the historian's location in global intellectual traditions. For much of the twentieth century, European and American historical trajectories served as the implicit standard against which non-Western cases were measured and found 'lacking' specific stages of development. Contemporary comparative historians must be explicit about these framing assumptions and increasingly work with frameworks that do not privilege any single case as the normative center.

Question 5

Why can't comparative historians simply 'control' for variables the way natural scientists run controlled experiments, and how do most-similar and most-different case selection strategies partially compensate for this limitation?

Model answer: Historians cannot assign cases randomly to conditions or hold variables constant by intervention — every historical case is a unique, complex configuration of factors that co-evolved together. Most-similar design compensates by selecting cases where many background factors are similar, so remaining differences are more likely to explain outcome divergence. Most-different design compensates by identifying what common factor produces the same outcome across cases with otherwise different profiles. Neither eliminates confounding, but both make the comparison more analytically tractable than random case selection.

The deep problem is that historical causation is typically conjunctural — multiple factors combine in specific configurations to produce outcomes, and the same factor may be causal in one context but not another. This is 'equifinality' (multiple paths to the same outcome) and 'multifinality' (same cause, different outcomes in different contexts). These features mean that even a well-designed comparative study cannot fully simulate experimental control. The most honest comparative historians acknowledge these limits explicitly, treat their comparisons as hypothesis-generating or hypothesis-testing rather than conclusive proof, and combine comparison with deep knowledge of individual cases.