ABStats.js / ABStats.php

A statistical library for A/B testing (Version 1.02 Updated April 24, 2015).

How to use it

You can embed abstats.js in your project or run commands right in the browser console. No programming knowledge required. To use it in the browser, just visit this calculator page, which has the library already loaded. All you need to do is open the browser console and run any of these commands.

Functions included

interval_binary()

Calculate Adjusted Wald Confidence Interval with

interval_effect_binary()

Calculates Adjusted Wald Confidence Interval around the relative % difference in proportions

interval1_effect_binary()

Calculates 1-tailed Adjusted Wald Confidence Interval around the relative % difference in proportions

interval_effect_expected_binary()

Calculates predicted interval for relative increase

p_effect_false_binary()

Calculates probability of getting an effect size by chance.

confidence_binary()

Gives confidence that the difference between proportions is greater than 0

significance_binary()

Calculate p-value

sampleSize_binary()

Get sample size for binomial proportion

effect_binary()

Get minimum effect for fixed sample size. It's the same as sampleSize_binary() but solved for effect size.

sensitivity_binary()

Get expected power given a fixed sample size and desired effect. It's the same as sampleSize_binary() but solved for power pct.

interval_continuous()

Get confidence intervals for continuous data

significance_continuous()

Get p-value given 2 data sets of continuous data (e.g., revenue, time)

sampleSize_continuous1

Estimate sample size PER VARIATION for continuous data using current sample of data

sampleSize_continuous2

Estimates sample size PER VARIATION for continuous data using average and variance

normalDist()

Returns percentage (y axis) on Standard Normal Curve given z (x axis)

normalDistInv()

Figures out the z value (x) that yields a given percentage (y). It's the inverse of normalDist()

normalAreaZToPct()

Gives p-value form z score: calculates area under Standard Normal Curve from -z to z

normalAreaPctToZ()

Gives the 2-tailed z value such that the area under the Standard Normal Curve between -z and z is pct%. It's the inverse of normalAreaZToPct()

normalAreaZToPct_left()

Gives the left-tailed z value such that the area under the Standard Normal Curve between -z and z is pct%

normalAreaPctToZ_left()

Gives the left-tailed z value such that the area under the Standard Normal Curve between -z and z is pct%. It's the inverse of normalAreaZToPct_left()