Robust statistics provide valid results across a vast selection of problems, consisting of assumption violations, the presence of outliers, and also miscellaneous various other problems. The term “robust statistic” applies both to a statistic (i.e., median) and also statistical analyses (i.e., hypothesis tests and regression).

You are watching: What does it mean when a sample is robust

Huber (1982) identified these statistics as being “distributionally robust and also outlier-resistant.”

Conversely, non-durable statistics are sensitive to to much less tha suitable conditions.


In this post, learn around robust statistics and also analyses.


Robust Statistics

The intend, median, traditional deviation, and interquartile range are sample statistics that estimate their corresponding population worths. Ideally, the sample values will certainly be fairly cshed to the population value and also will certainly not be systematically too high or too low (i.e., unbiased).

Unfortunately, outliers and also excessive values in the lengthy tail of a skewed circulation can cause some sample statistics to come to be biased, negative quality approximates. What does that mean? The sample statistics will be systematically as well high or as well low and relocate further away from the correct worth.


Conversely, a robust statistic will certainly be efficient, have only a slight predisposition, and also be asymptomatically unbiased as sample dimension rises once tbelow are outliers and also excessive worths in long-tails.

In plain English, once outliers and long-tails are present, durable statistics will be reasonably close to the correct value offered your sample dimension, and it will certainly not systematically over- or under-estimate the population worth. Furthermore, as the sample dimension boosts, the statistic viewpoints ending up being totally unbiased.

Robust statistics resist the influence of outliers and long-tails. They job-related well in a vast range of probcapacity distributions, especially non-normal distributions.


Related post: How to Identify the Distribution of Your Data

The Breakdown Point and also Robustness

An intuitive means to understand also the robustness of a statistic is to consider exactly how many data points in a sample you have the right to replace through fabricated outliers before the sample statistic becomes a bad estimate.

Statisticians describe this as the breakdown suggest. That’s the maximum percent of observations you deserve to rearea through outliers prior to resulting in boundless changes in the estimate. Higher breakdvery own points correspond to more durable statistics.


Let’s occupational with examples through the intend and median.

The calculations for the intend involve all data points. Consequently, a solitary outlier have the right to substantially affect the mean. Imagine we have the adhering to dataset: 50, 52, 55, 56, 59, 59, 60. If we change one of the worths to 1000, it’ll have actually a huge affect on the mean! Theoretically, the effect is unbounded because we might force the expect to be any kind of value we pick by adjusting one worth in the dataset. The breakdown suggest for the expect is 1/n. The intend is not a durable statistic.

Conversely, the median is a robust statistic bereason it has actually a malfunction point of 50%. You have the right to change up to 50% of the monitorings before creating unbounded changes. Using the exact same dataset: 50, 52, 55, 56, 59, 59, 60, if we adjusted the 60 to 1000, the median is totally uninfluenced. It’s still 56 in both instances.


Consequently, the median is a robust statistic for main tendency while the intend is not. In graph listed below, notification that the median is near the most common worths while the intend is gaining pulled ameans by the long tail of the skewed circulation.

*
*

Related posts: Measures of Central Tendency and Five Ways to Find Outliers


Robust Statistics for Variation

Tright here are several common steps of variability, including the conventional deviation, variety, and interquartile range. Which statistics are robust?

The conventional deviation is comparable to the mean bereason its calculations include all worths in the data collection. A single outlier can drastically influence this statistic. Because of this, it is not robust.

The variety is the difference in between the highest possible and lowest worth in the dataset. If you have a solitary uncommonly high or low worth, it can greatly influence the range. It’s also not robust.

The interquartile variety (IQR) is the middle half of your datacollection. It is equivalent to the median in that you have the right to relocation many type of worths without altering the IQR. It has actually a breakdown point of 25%. Consequently, of these three steps, the interquartile range is the most durable statistic.

Related post: Measures of Variability

What are Robust Statistical Analyses?

Robust statistical analyses have the right to develop valid results also once the appropriate problems perform not exist through real-world data. These analyses percreate well as soon as the sample information follow a range of distributions and have inexplicable values. In various other words, you deserve to trust the outcomes also once the presumptions are not completely satisfied.

For example, parametric hypothesis tests that assess the expect, such as t-tests and ANOVA, assume the data follow a normal circulation. However before, these tests are robust to deviations from the normal distribution when your sample size per team is sufficiently large, many thanks to the central limit theorem.

Similarly, nonparametric analyses assess the median and also are robust distributionally bereason they don’t assume the information follow any kind of certain distribution. In addition, favor the median, nonparametric analyses stand up to the results of outliers.

Related post: Nonparametric vs. Parametric Analyses

Robust regression is a type of regression analysis that statisticians designed to avoid troubles linked through plain least squares (OLS). Outliers can invalidate OLS results, while robust regression can handle them. It have the right to likewise resolve heteroscedasticity, which occurs as soon as the residuals have actually a non-continuous variance.

Related posts: OLS Assumptions and Heteroscedasticity in Regression

Be sure to understand for which properties each statistical evaluation is durable. For example, while conventional t-tests and also ANOVAs have the right to manage violations of the normality assumption, they cannot stand up to the impacts of outliers. Nonparametric tests don’t call for a specific distribution, yet the various groups in your evaluation should have the same dispersion. Hence, nonparametric tests are not durable to violations of the equal variances presumption.

See more: Politics: Why Have People Been Rethinking The Microsoft Model Is Done

Robust statistical analyses could be resistant to specific presumption violations and yet be sensitive to other breaches.