Why Aggregates Are Conservative

Aggregates are conservative because they dilute extremes. A single data point, no matter how dramatic, has very little power to change the average of a large group. That's why the restaurant's 4.5-star rating barely budged from a single 1-star review; the new review was absorbed by the weight of the previous 500. This principle shows that the story of a group is often more moderate and stable than the story of any individual within it.

Notice this: how an entire country's average life expectancy changes very slowly, even with major events.

The conservative nature of aggregates comes down to simple math: each new data point is just one voice among many. The larger the existing group, the less weight any single new voice carries. An extreme event is effectively muffled by the collective normalcy of the majority.

This phenomenon is visible everywhere. In finance, an individual stock can be incredibly volatile, soaring or crashing in a single day. But a diversified index fund like the S&P 500, which aggregates the performance of 500 companies, moves much more slowly and predictably. The wild success of one company is balanced by the modest performance or failure of others.

In politics, a national poll for a candidate is far more stable than a poll of a single small town. The small town's opinion might swing wildly based on a local issue, but when aggregated with thousands of other towns, its dramatic shift becomes a tiny ripple in a large ocean.

This is why focusing on single, vivid data points can be misleading. An alarming news story about a specific crime doesn't necessarily mean the city's overall crime rate—a large aggregate—has changed. The stability of the aggregate provides a more reliable, though less exciting, picture of reality.

The conservative nature of aggregates is a direct consequence of fundamental statistical laws. As we collect more data, the average of that data becomes a more stable and reliable estimate of the true underlying average. This effect is not just an interesting quirk; it's the foundation of insurance, polling, and scientific measurement, allowing us to make surprisingly accurate predictions about the whole by observing a part.

• The Law of Large Numbers states that as a sample size grows, its mean gets closer to the average of the whole population, demonstrating how larger aggregates become more stable and predictable. [1]
• The Central Limit Theorem explains that the distribution of sample means will approximate a normal distribution (a bell curve) around the true population mean. This shows that averages naturally tend to cluster around a central, moderate value, making extreme averages very unlikely. [2]
• Research by Kahneman and Tversky identified the "belief in the law of small numbers," a cognitive bias where people wrongly expect small samples to be as representative and stable as large ones, leading to overreactions to limited data. [3]

This principle primarily applies when data points are independent. If data points are highly correlated (e.g., all companies in a sector failing due to the same shock), the aggregate is not as conservative. Furthermore, an aggregate like an 'average income' can hide extreme inequality; a country can have a moderate average while having vast numbers of very poor and very rich people. Aggregates reveal the central tendency but conceal the distribution's shape.