Eng Statistical Tactics

Cherry picking

happens when someone only shows the data that supports their point while ignoring data that contradicts it

Selection bias

when the people,cases, or data points included in a study are not representative of the group being discussed

Survivorship Bias

we focus only on the people or things that made it through a process and ignore those that failed or disappeared

Correlation vs. Causation

Correlation = 2 thing move together Causation= one thing directly causes the other

Truncated Y-axis

y axis starts above zero or cuts off parts of the scale, making small differences look much larger than they are

Misleading Pie charts

when the slices are hard to compare, do not add up, correctly or show too many categories

Misleading Bar Graphs

when the axis is distorted, the bars are not proportional, or the categories are arranged in a confusing way

Small Sample Sizes

the conclusion is based on too few observations to be reliable

Outdated data

information that may have been true in the past but not longer reflects the current situation

Relative risk vs. Absolute risk

relative: how much something changes proportionally, absolute: actual size of the change

Average that hide reality

hide important differences inside a group

Median vs. Mean Confusion

means: the total divided by the number of values, median: the middle value when all values are ordered

Missing context

a number is shown without the background needed to understand it

Biased Survey questions

questions that pushes people toward a particular answer

Framing effects

the same information creates a different reaction depending on how it is presented

Simpson’s Paradox

a trend appears in several separate groups but disappears or reverse when the groups are combined

Percentage Without Baselines

does not tell you what the percentage is based on

Misleading Rankings

when the criteria is unclear, weighted strangely or based on small differences

Cumulative Graph Tricks

add values over time, so they usually keep rising even when the current rate is slowing down

Unequal Scales

when two charts or axes use different intervals, making comparison unfair

Confusing Data Levels

unclear what the data actually measures

Overgeneralization

someone applies a conclusion to a much larger group than the data supports

Fake Precision

very exact numbers to create a false sense of certainty

Comparing Different Times periods

data frond different time period is compared as if the period are equivalent