Statistically significant
Said of a difference too large to come from chance in which answers the sample happened to collect. The Index claims an order between companies only where their difference is significant.
What it tells a buyer or a marketer
Two companies rarely score the same. The question is whether the gap between them is real or an accident of which answers the sample happened to collect. A difference is statistically significant when it is too large to put down to chance.
For a marketer, this is the line between a finding and a coincidence. A company a few points ahead of a rival may have no real lead at all. A company well ahead, on a large sample, does. Knowing which is which stops a team celebrating, or panicking, over a gap the next round of answers could reverse.
How it appears on the AI Authority Index™
The Index claims an order between companies only where their difference is significant. Where the sample cannot separate two companies, it lists them as level and says so. That is why a reader will see companies grouped by where they stand on the Star Chart, with no numbered places.
Each reading prints with its confidence interval and the number of answers behind it, so a reader can see for themselves how firm a difference is.
What moves it
Two things move it: the size of the gap and the size of the sample. A wide gap on a large sample is significant. A narrow gap, or a gap measured on few answers, usually is not. As the engines name a company in more answers, its readings firm up, and real differences between it and its rivals become easier to show.
Why Third Hemisphere® defines it
The term keeps its standard meaning in statistics. Third Hemisphere applies it as a rule on the AI Authority Index™: it states an order between two companies where the sample supports one, and it lists them as level elsewhere.
Read the CRISP™ methodology, or see how the Index groups the companies in your market.
Related terms
Whether a difference is statistically significant depends on each reading's confidence interval and on N, the sample size. It decides how companies on the Leaderboard are listed, and the Star Chart groups them by cell.
See also
Questions
What does statistically significant mean on the Index?
It is said of a difference too large to come from chance in which answers the sample happened to collect.
Why does the AI Authority Index™ print no numbered places?
It claims an order between companies only where their difference is significant. Where the sample cannot separate them, it lists them as level.
How can a reader judge a difference between two companies?
Each reading prints with its confidence interval and the number of answers behind it, which show how firm the difference is.