Calculate binary statistical significance (with 99% confidence)

script.js

// require('jStat');

function calculateSignificance(a_sessions, a_conversions, b_sessions, b_conversions) {
    a_cr = a_conversions / a_sessions;
    b_cr = b_conversions / b_sessions;
    a_stdErr = Math.sqrt(a_cr * (1 - a_cr) / a_sessions);
    b_stdErr = Math.sqrt(b_cr * (1 - b_cr) / b_sessions);

    // get_z_score()
    const z_score =
        (a_cr - b_cr) /
        Math.sqrt(Math.pow(a_stdErr, 2) + Math.pow(b_stdErr, 2));
    const p_value = NORMDIST(z_score, 0, 1, true);

    console.table({
        a_cr: _.round(a_cr, 2),
        b_cr: _.round(b_cr, 2),
        a_stdErr: _.round(a_stdErr, 2),
        b_stdErr: _.round(b_stdErr, 2),
        z_score: _.round(z_score, 3),
        p_value: _.round(p_value, 4),
    });

    return p_value > 0.99;

    function NORMDIST(x, mean, sd, cumulative) {
        // Check parameters
        if (isNaN(x) || isNaN(mean) || isNaN(sd)) return "#VALUE!";
        if (sd <= 0) return "#NUM!";

        // Return normal distribution computed by jStat [http://jstat.org]
        return cumulative
            ? jStat.normal.cdf(x, mean, sd)
            : jStat.normal.pdf(x, mean, sd);
    }
}