What 0.14 Means
In statistics, standard deviation measures how spread out values are from the mean. For context, the average standard deviation in IMDb ratings across all directors with 8+ films in our database is approximately 0.85. Martin Scorsese's is 0.61. Christopher Nolan's is 0.52. Ingmar Bergman's is 0.39.
Yasujirō Ozu's is 0.14.
That's not a rounding artifact. Across 10 films in our database, Ozu's IMDb ratings range from 7.8 to 8.2 — a total spread of 0.4 points. His average is 8.0. The coefficient of variation (standard deviation divided by mean) is 0.018, or 1.8%. By any statistical measure, this is the tightest clustering of quality scores for any director with a meaningful filmography.
The Films
Consider the uniformity: Tokyo Story (8.2), Late Spring (8.1), An Autumn Afternoon (8.1), Early Summer (7.9), Late Autumn (7.9), Floating Weeds (7.8), Good Morning (7.8), The Flavor of Green Tea Over Rice (7.8), Equinox Flower (7.8), Tokyo Twilight (7.8). Plot these on a number line and they form an almost impossibly tight cluster.
No spikes. No dips. No "experimental failure" or "late-career decline." Every film lands in the same narrow band of quality, decade after decade.
Master Score: 86.4
Ozu's average Master Score of 86.4 places him 5th among all directors in our database (behind Masaki Kobayashi at 88.9, Kenji Mizoguchi at 87.5, Satyajit Ray at 87.2, and Haruo Sotozaki at 86.9). But none of those directors match his consistency. Kobayashi's standard deviation is 0.40. Ray's is 0.26. Ozu's 0.14 remains unmatched.
The Methodology Behind the Magic
Ozu's consistency isn't accidental — it's methodological. He worked within deliberately narrow parameters: domestic family dramas, low camera angles, minimal camera movement, recurring actors, similar narrative structures. From a data science perspective, he controlled his variables. By minimizing the variance in his inputs (genre, style, subject matter), he minimized the variance in his outputs (audience and critical response).
This is the filmmaking equivalent of a controlled experiment: change one variable at a time, keep everything else constant, and the results become predictable.
The Counter-Argument to "Range"
Film criticism often valorizes range — the ability to jump between genres, tones, and scales. Our data suggests an alternative reading: that depth within a narrow domain can produce higher and more consistent quality scores than breadth across many domains.
Ozu never made an action film, never attempted a comedy, never chased a trend. And his average rating of 8.0 across 10 films — with a standard deviation so small it barely registers on a chart — is the mathematical proof that mastery doesn't require range. It requires relentless refinement of what you already do well.
σ = 0.14. The number speaks for itself.