CODEWITHSUNDEEP
pythonscipystatisticsconfidence-interval
SirPunch
SirPunch

Reputation: 95

interpreting alpha variable in scipy interval function

I am using scipy's interval function for a normal random variable to calculate the confidence interval. However, there seems be some misunderstanding about the significance level.

From scipy.stats.norm docs:
*Signature: stats.norm.interval(alpha, *args, **kwds)
Docstring:
Confidence interval with equal areas around the median.*

Parameters:
alpha : array_like of float
Probability that an rv will be drawn from the returned range.
Each value should be in the range [0, 1].

It seems that they are representing the alpha parameter to be the confidence level rather than the significance level. For example, in statistics an alpha value of 0.05 would mean 5% significance level and 95% confidence level. However, scipy expect to pass 0.95 for alpha variable's value. This is confusing as it should be 0.05 going by statistics terminology. Am I missing something here?

Upvotes: 0

Views: 732

Answers (1)

user6655984
user6655984

Reputation:

Yes, the quantity called alpha in scipy.stats.rv_continuous.interval would be called 1-alpha in statistics textbooks. You are not missing anything. It's just a suboptimal name choice.

The only discussion of the name of that parameter that I found concerned a different name collision:

Sigh. Given that both interval and levy_stable are quite esoteric,[...]

Upvotes: 1

Related Questions