bad graphs
1 check the y axis to see if it goes down to 0
  - it looks like a MUCH bigger difference if not
2 check that labels exist
  - without labels, it could mean anything
3 the bar size is discrepant
  - it needs to be legitimately proportional

## Statistics uses

Most [reality](reality.md) yields [statistical results](math-stat.md), so statistics has a vast range of applications. It can track how far someone can throw a ball, a rocket's likely trajectory, radioactive chemicals' half-life, and [social trends](trends.md). It can also often show how we change [habits](habits.md), [behave in groups](groups-member.md), and [make decisions](people-decisions.md).

Normally, we interpret [certainty](understanding-certainty.md) in our minds through convictions grounded in [experiences](understanding.md). Statistical systems, however, create an anomaly to that mode of reasoning:

1. Gathering statistics requires heaping up [stories](stories.md), but only for specific and measurable elements of those many, many stories, without any consideration of other things that aren't being measured.
2. All [analysis](logic.md) derived from the results can't naturally regard the qualitative experiences that were trimmed during collection.
3. Since we all need stories to [understand](understanding.md) reality, most statisticians easily interpret cause-and-effect through correlation.

Statistics are tracking reality, and reality is [uncertain](understanding-certainty.md), so all statistics are always a little uncertain. They're alarmingly accurate (when measured correctly) at determining correlation, but only in a broad sense.

Statistics, however, can *only* demonstrate correlation. Any causation only comes through how the information is [interpreted](stories.md):

- People in the 1980s who ate hot breakfasts more often than cold were dramatically more likely to have Alzheimer's Disease. We [tend](habits.md) to eat what we ate when we were kids, and cold cereal took off in [popularity](trends.md) in the 1950s.
- Crime statistics *always* go up when cities expand their police forces. Technically, crime statistics are only measuring *caught* crimes.
- The [culture](people-culture.md) of most of the world aligns with anthropology studies of the West, representative by data. Most anthropology data is gathered near Western universities.
- People who saw the 1984 Ghostbusters movie are more likely to die than people who saw the 2021 Ghostbusters movie. This is simply because of age-related facts.

APPLICATION: A statistical report is a statement of curated observations and is as trustworthy as the [group](groups-small.md) it came from. There's lots of [money and power](power.md) if someone can [prove](influence.md) they're not biased (which isn't [possible](mind-bias.md)), so the people who create statistics often [have something to lose](power-types.md) by stating that they're not precisely accurate.

When we hear a statistical result, we're incapable of *not* inserting our [explanation](logic.md) on the [causes or effects](stories.md) of the information. Our only hope to gain full [understanding](understanding.md) lies in forced [uncertainty](understanding-certainty.md), which requires directing our thoughts toward the specific [purpose](purpose.md) of suspending judgment, which requires tremendous effort to maintain.

Statistics don't fit our [intuition](mind-feelings.md) for a few core reasons:

- When there are different sample sizes from different groups (e.g., populations of various districts in a country), the less-populated groups are *guaranteed* to have more high/low extremes, simply from having fewer numbers.
- Observing statistical anomalies (e.g., very unusual height) creates more engaging [stories](stories.md) and spurs the [imagination](imagination.md), and it requires training to *not* [feel](mind-feelings.md) the outliers more than the norm.
- Possessing information on a chart doesn't explain anything directly. It indicates a connection, but that connection has to [stay unknown](unknown.md) for us to be accurate, and [we *hate* not knowing](purpose.md).
- When we experience the [possibility](imagination.md) of a [risk](safety.md), we only [feel](mind-feelings.md) comfortable when its chances become 0%. For that reason, we tend to obsess about the smaller risks (which can theoretically be eliminated) at the opportunity cost of managing the larger ones.
- Over time, statistical things [trend](trends.md) back to normal from extremes (e.g., high-performing athletes, extremely depressed people). Without measuring what normally happens, we can imagine that a statistical change was caused *by* a measured element included in it.

APPLICATION: We can never expect something to go *precisely* how we envision it because we always gain 1 sample of a statistical range at any given time. All we can do is [change our statistical likelihoods](success-2_attitude.md) whenever possible.

Very frequently, statistics [hide lies](image-distortion.md) by making the statistician appear more credible. The value of the statistical [analysis](logic.md) comes from the [value](creations.md) of the raw data, which comes from the [quality](values-quality.md) of the data collection.

APPLICATION: Claiming statistical analysis is immaculate requires faith in a [community's](groups-member.md) collective [understanding](understanding.md). Nothing is all/nothing, but statistics provide the opportunity to make a clear [story](stories.md) of reality, even if it's not precise. However, most [younger people](maturity.md) can be [influenced](influence.md) to [believe](understanding-certainty.md) statistics more than its truth.

---