Mean vs Median Causal Effect
<p>However, sometimes we might be interested in quantities different from the average, such as the <a href="https://en.wikipedia.org/wiki/Median" rel="noopener ugc nofollow" target="_blank"><strong>median</strong></a>. The median is an alternative measure of <em>central tendency</em> that is more robust to outliers and is often more informative with skewed distributions. More generally, we might want to estimate the effect for different <a href="https://en.wikipedia.org/wiki/Quantile" rel="noopener ugc nofollow" target="_blank">quantiles</a> of the outcome distribution. A <strong>common use case</strong> is studying the impact of a UI change on the loading time of a website: a slightly heavier website might translate into an imperceptible change for most users, but a big change for a few users with very slow connections. Another common use case is studying the impact of a product change on a product that few people buy: do existing customers buy it more or are we attracting new customers?</p>
<p><a href="https://towardsdatascience.com/mean-vs-median-causal-effect-37057a6c54c9"><strong>Click Here</strong></a></p>