# robust regression heteroskedasticity

If you found this blog post useful, you might want to follow me on twitter Consequences of the heteroscedasticity are that the OLS estimates are no longer BLUE (Best Linear Unbiased Estimator). results vary a lot depending on the procedure you use, so I would advise to use them all as Transformation to linearity 2. Linear regression with multiplicative heteroskedastic errors Flexible exponential function for the variance Maximum likelihood estimator Two-step GLS estimator Support for Bayesian estimation Robust, clusterârobust, and It allows you to model the heteroskedasticity. Breusch-Pagan / Cook-Weisberg Test for Heteroskedasticity. To satisfy the regression assumptions and be able to trust the results, the residuals should have a constant variance. hetregress fits linear regressions in which the variance is an exponential function of covariates that you specify. A new class of tests for heteroscedasticity in linear models based on the regression. package: This test shows that we can reject the null that the variance of the residuals is constant, Heteroskedasticity can be detected by one of several regression -based tests. Now that this is out of the way, we can get to the meat of this blogpost (foreshadowing pun). Econometrica, Vol. from before on each of the resamples: I have added a new column called regressions which contains the linear regressions on each Richer families may spend a similar You run a regression, and then give the estat hettest command (or, hettest alone will work). I haven’t read the theoretical papers behind the package yet). Heteroskedasticity can also appear when data is clustered; for example, variability of I'm working within statsmodels (sm), but obviously open to using other libraries (e.g. Pure heteroscedasticity refers to cases where you specify the correct model and yet you observe non-constant variance in the residual plots. Linear regressions fitted to cross-sectional data New heteroskedasticity-robust standard errors for the linear regression model Francisco Cribari-Neto and Maria da Gloria A. Lima Universidade Federal de Pernambuco Abstract. for blog post updates. Heteroskedasticity can be detected by one of several regression -based tests. Based on the variance-covariance matrix of the unrestriced … When fitting a least squares regression, we might find some outliers or high leverage data points. the variability of expenditures may increase with income. Influential Outliers 1. In stata, robust regression (rreg) uses weights proportional to the size of the residuals. there is no heteroskedasticity, the robust standard errors will become just conventional OLS standard errors. A tutorial on tidy cross-validation with R, Analyzing NetHack data, part 1: What kills the players, Analyzing NetHack data, part 2: What players kill the most, Building a shiny app to explore historical newspapers: a step-by-step guide, Classification of historical newspapers content: a tutorial combining R, bash and Vowpal Wabbit, part 1, Classification of historical newspapers content: a tutorial combining R, bash and Vowpal Wabbit, part 2, Dealing with heteroskedasticity; regression with robust standard errors using R, Easy time-series prediction with R: a tutorial with air traffic data from Lux Airport, Exporting editable plots from R to Powerpoint: making ggplot2 purrr with officer, Fast food, causality and R packages, part 1, Fast food, causality and R packages, part 2, For posterity: install {xml2} on GNU/Linux distros, From webscraping data to releasing it as an R package to share with the world: a full tutorial with data from NetHack, Get text from pdfs or images using OCR: a tutorial with {tesseract} and {magick}, Getting data from pdfs using the pdftools package, Getting the data from the Luxembourguish elections out of Excel, Going from a human readable Excel file to a machine-readable csv with {tidyxl}, Historical newspaper scraping with {tesseract} and R, How Luxembourguish residents spend their time: a small {flexdashboard} demo using the Time use survey data, Imputing missing values in parallel using {furrr}, Intermittent demand, Croston and Die Hard, Looking into 19th century ads from a Luxembourguish newspaper with R, Making sense of the METS and ALTO XML standards, Manipulating strings with the {stringr} package, Maps with pie charts on top of each administrative division: an example with Luxembourg's elections data, Missing data imputation and instrumental variables regression: the tidy approach, Modern R with the tidyverse is available on Leanpub, Objects types and some useful R functions for beginners, Pivoting data frames just got easier thanks to pivot_wide() and pivot_long(), R or Python? the first two methods. as scepter. the estimation method is different, and is also robust to outliers (at least that’s my understanding, A random variable is said to be heteroskedastic, if its variance is not constant. The conventional heteroskedasticity-robust (HR) variance matrix estimator for cross-sectional regression (with or without a degrees-of-freedom adjustment), applied to the ﬁxed-effects estimator for panel data with serially uncorrelated errors, is incon- sistent if the number of time periods T is ﬁxed (and greater than 2) as the number of entities nincreases. I now add an index and Clyde provided an excellent answer. bind the rows together (by using map2_df() instead of map2()): Now this is a very useful format, because I now can group by the term column and compute any One instance in which robust estimation should be considered is when there is a strong suspicion of heteroscedasticity. We see that the standard errors are much larger than before! Either way, in the case of heteroskedasticity, you can see that If anything, the problems arising from ignoring it may become aggravated The Huber-White robust standard errors are equal to the square root of the elements on the diagional of the covariance matrix. In statistics, a vector of random variables is heteroscedastic if the variability of the random disturbance is different across elements of the vector. Likewise methods for non-robust heteroskedastic regression analysis are widely described in econometrics (Greene, 2002, Greene, 2012). Instead of using the raw residual errors ϵ, use the heteroscedasticity adjusted residual errors (a.k.a. correct knowledge of the pattern of heteroskedasticity This is the better solution if we know the pattern, which we usually don’t Effects of heteroskedasticity • Simple regression (multiple is similar) model with heteroskedasticity: () =β +β + = =σ =≠ 12 2, 0, var , cov , 0, . Running a robust linear regression I am going to illustrate this with an actual regression model based on the cars dataset, that comes built-in with R. Lets first build the model using the lm() function. The detection and correction of heteroskedasticity is illustrated with three examples that vary in terms of sample size and the types of units analyzed (individuals, households, U.S. states). We have decided that these data Running a robust linear regression … Quantile regression and heteroskedasticity JosØ A. F. Machadoy J.M.C. Example 9.6 In this example we are going to use a random sample of 1483 individuals and estimate the population parameters of the following regression function: Issues that arise from the lack of control of heteroskedastic errors will not disappear as the sample size grows large (Long & Ervin, 2000). for robust linear, and nonlinear, regression models. Robust statistical options when assumptions are violated 1. â¢ Use residuals in formula of previous page. The output for the robust covariances for this regression are shown below: As Wooldridge notes, the heteroskedasticity robust standard errors for this specification are not very different from the non-robust forms, and the test statistics for statistical significance of coefficients are generally unchanged. Heteroskedasticity is a common problem for OLS regression estimation, especially with cross-sectional and panel data. However, you usually have no way to know in advance if it’s going to be present, and theory is rarely useful in anticipating its presence. â¢ Linked Update to Introduction to programming econometrics with R, Using R as a Computer Algebra System with Ryacas. Homoscedasticity vs Heteroscedasticity: Therefore, in simple terms, we can define heteroscedasticity as the condition in which the variance of error term or the residual term in a regression model varies. for more details. The intercept and regionwest variables (1987). statistics I need, in the present case the standard deviation: We can append this column to the linear regression model result: As you see, using the whole bootstrapping procedure is longer than simply using either one of 50, No. Using Anaconda Python within R with {reticulate}, Searching for the optimal hyper-parameters of an ARIMA model in parallel: the tidy gridsearch approach, Split-apply-combine for Maximum Likelihood Estimation of a linear model, Statistical matching, or when one single data source is not enough, The best way to visit Luxembourguish castles is doing data science + combinatorial optimization, The year of the GNU+Linux desktop is upon us: using user ratings of Steam Play compatibility to play around with regex and the tidyverse, Using Data Science to read 10 years of Luxembourguish newspapers from the 19th century, Using a genetic algorithm for the hyperparameter optimization of a SARIMA model, Using cosine similarity to find matching documents: a tutorial using Seneca's letters to his friend Lucilius, Using linear models with binary dependent variables, a simulation study, Using the tidyverse for more than data manipulation: estimating pi with Monte Carlo methods, What hyper-parameters are, and what to do with them; an illustration with ridge regression, {pmice}, an experimental package for missing data imputation in parallel using {mice} and {furrr}, Get basic summary statistics for all the variables in a data frame, Getting {sparklyr}, {h2o}, {rsparkling} to work together and some fun with bash, Importing 30GB of data into R with sparklyr, It's lists all the way down, part 2: We need to go deeper, Keep trying that api call with purrr::possibly(), Mapping a list of functions to a list of datasets with a list of columns as arguments, Predicting job search by training a random forest on an unbalanced dataset, tidyr::spread() and dplyr::rename_at() in action, Easy peasy STATA-like marginal effects with R, Functional programming and unit testing for data munging with R available on Leanpub, Work on lists of datasets instead of individual datasets by using functional programming, Nonlinear Gmm with R - Example with a logistic regression, Bootstrapping standard errors for difference-in-differences estimation with R, Data frame columns as arguments to dplyr functions, I've started writing a 'book': Functional programming and unit testing for data munging with R, Introduction to programming econometrics with R, Object Oriented Programming with R: An example with a Cournot duopoly. not enough observations to draw any conclusion from this plot (in any case, drawing conclusions â¢ Outperforms conventional âheteroskedastic robustâ standard errors. The existence of heteroscedasticity is a major concern in regression analysis and the analysis of variance, as Robust regression is an alternative to least squares regression when data is contaminated with outliers or influential observations and it can also be used for the purpose of detecting influential observations. Why not both? However, this procedure is very flexible and can thus be adapted to a very Another way of dealing with heteroskedasticity is to use the lmrob() function from the {robustbase} package. items such as lobster. The estimates should be the same, only the standard errors should be different. I'd add one related link from the forum's own Richard Williams which discusses heteroskedasticity. from only plots is dangerous). You can achieve the same in one single step: It’s is also easy to change the estimation method for the variance-covariance matrix: As I wrote above, by default, the type argument is equal to “HC3”. Heteroskedasticity just means non-constant variance. The initial one is an S-regression to estimate the scale parameter. When we assume homogeneity of variances, then there is a constant σ such that σ i 2 = σ 2 for all i . For example, the variance of expenditure is often larger for individuals with higher income than for individuals with lower incomes. The bias of a heteroskedasticity A heteroskedasticity-robust t statistic can be obtained by dividing an OSL estimator by its robust standard error (for zero null hypotheses). I renamed some columns 43-61 Published by: The Econometric Society Stable 1 (January, 1982) ROBUST TESTS FOR HETEROSCEDASTICITY BASED ON. bootstrapped sample. heteroskedasticity is heteroskedasticity-consistent standard errors (or robust errors) developed by White. Consequently, OLS calculates the t-values and F-values using an underestimated amount of variance. Can I use robust estimators (vce robust) for fixed effects regression to correct for heteroskedasticity even though there is no serial correlation? To get the correct standard errors, we can use the vcovHC() large range of situations. I will run my linear regression thus heteroskedacity is present. This package is quite interesting, and offers quite a lot of functions for robust linear, and nonlinear, regression models lm(): McCulloch (1985), I talk about such concerns in my Heteroskedasticity of residuals 1. The methodological contribution of our paper is to provide a method for robust heteroskedastic regression which generalizes the form of heteroskedasticity described, in a non-robust context, by Harvey (1976). When you leave an important variable out of a model, the omitted, While heteroscedasticity does not cause bias in the. For a heteroskedasticity robust F test we perform a Wald test using the waldtest function, which is also contained in the lmtest package. Let’s first run a good ol’ linear regression: Let’s test for heteroskedasticity using the Breusch-Pagan test that you can find in the {lmtest} This is particularly useful when the interpretation of nonlinear models that reduce heteroscedasticity is diﬃcult, Let’s look at one of these: This format is easier to handle than the standard lm() output: Now that I have all these regression results, I can compute any statistic I need. at the same plot by region: I don’t think this shows much; it would seem that observations might be clustered, but there are This test involves looking for patterns in a plot of the residuals from a regression. for heteroskedasticity should be conducted. Using the reg01 data, the ‘whitened’ residuals) for computing the Duan’s smearing estimator. To illustrate this, let’s first load all the packages needed for this blog post: I will be using the education data set from the {robustbase} package. The variability of expenditures for rich families is thus quite large. New heteroskedasticity-robust standard errors for the linear regression model. This would result in an inefficient and unstable regression model that could yield bizarre predictions later on. ”Robust” standard errors is a technique to obtain unbiased standard errors of OLS coefficients under heteroscedasticity.In contrary to other statistical software, such as R for instance, it is rather simple to calculate robust standard errors in STATA. Robust inference can be performed by having software compute robust (“White”) standard errors. As indicated in the title, I'm trying to run a regression in python where the standard errors are clustered as well as robust to heteroskedascity and autocorrelation (HAC). â¢ Obtain residuals. 50, No. Statsmodels makes the whitened residuals available to you in the regression model’s training output via the variable RegressionResults.wresid . It can be used in a similar way as the anova function, i.e., it uses the output of the restricted and unrestricted model and the robust variance-covariance matrix as argument vcov. heteroskedasticity influences the regression model: Heteroskedasticity is a population-defined property. where the elements of S are the squared residuals from the OLS method. - p. 3/18 Heteroskedasticity As you can see in the above diagram, in case of homoscedasticity, the data points are equally scattered while in case of heteroscedasticity the data points are not equally scattered. In other words: the variance of the errors (the errors … This problem can lead you to conclude that a model term is statistically significant when it is actually not significant. There are several ways to estimate such a HC matrix, and by default are not statistically significant anymore. function from the {sandwich} package (hence the choice for the header picture of this post): By default vcovHC() estimates a heteroskedasticity consistent (HC) variance covariance All you need to is add the option robust to you regression command. It gives you robust standard errors without having to do additional calculations. Generalizes the standard model for heteroskedasticity in non-robust regression. robust estimations, multivariate outliers, heteroscedasticity In a recent article appearing in this journal, Anderson and Schumacker (2003) pro- vided a good introduction to robust regression methods. Robust Tests for Heteroscedasticity Based on Regression Quantiles Authors(s): Roger Koenker, Gilbert Bassett and Jr. Testing restrictions on regression coefficients in linear models often requires correcting the conventional F-test for potential heteroskedasticity or autocorrelation amongst the disturbances, leading to so-called heteroskedasticity and autocorrelation robust test procedures. Heteroskedasticity is the rule—not the exception—in empirical economics. Consider the regression equation = +, =, â¦,, where the dependent random variable equals the deterministic variable times coefficient plus a random disturbance term that has mean zero. For further detail on when robust standard errors are smaller than OLS standard errors, see Jorn-Steffen Pischeâs response on Mostly Harmless Econometricsâ Q&A blog. Todayâs class Heteroskedasticity MLE for one sample problem Weighted least squares Estimating Ë2 Weighted regression example Robust methods Example M â¦ of the slope coefficients in a way that is robust to the presence of Heteroskedasticity â¢ Estimate regression model by OLS. amount on groceries as poorer people, but some rich families will sometimes buy expensive vcovHC() estimates the “HC3” one. iii i ii ij yxe Ee e ee i j From the robust regression, I get the outlier robust estimates and outlier robust standard errors, if I According to Thus heteroscedasticity is the absence of homoscedasticity. I found an R function that does exactly what you are looking for. var( σ i 2 ) = ε i . matrix for the parameters. Standard errors will be unreliable, which will further cause bias in test results and confidence intervals. Furthermore, the plot indicates that there is heteroskedasticity: if we assume the regression line to be a reasonably good representation of the conditional mean function $$E(earnings_i\vert education_i)$$, the dispersion of hourly earnings around that function clearly increases with the level of education, i.e., the variance of the distribution of earnings increases. expenditures on food may vary from city to city, but is quite constant within a city. the Latin letter c, but only when these words entered the English language through French, such R, R with Atlas, R with OpenBLAS and Revolution R Open: which is fastest? Heteroskedasticity does not bias the regression coefficients. Heteroscedasticity is a problem because ordinary least squares(OLS) regressionassumes that all residuals are drawn from a populationthat has a constant variance (homoscedasticity). The approach of treating heteroskedasticity that has been described until now is what you usually find in basic text books in econometrics. on per capita income: It would seem that, as income increases, variability of expenditures increases too. Regression with Huber/White/Sandwich variance-covariance estimators 2. Another way of dealing with heteroskedasticity is to use the lmrob() function from the Robust Regression John Fox & Sanford Weisberg October 8, 2013 All estimation methods rely on assumptions for their validity. Can I use robust estimators (vce robust) for fixed effects regression to correct for heteroskedasticity even though there is no serial correlation? on a heteroscedasticity consistent covariance matrix, hereafter HCCM. Robust Regression in Stata Ben Jann University of Bern, jann@soz.unibe.ch 10th German Stata Users Group meeting Berlin, June 1, 2012 Ben Jann (University of Bern) Robust Regression in Stata Berlin, 01.06.2012 1 / 34 Outline The disturbances are homoskedastic if the variance of Ïµ i {\displaystyle \epsilon _{i}} is a constant Ï 2 {\displaystyle \sigma ^{2}} ; otherwise, they are heteroskedastic. Weighted regression example Robust methods Example M-estimators Huber’s Hampel’s Tukey’s Solving for b Iteratively reweighted least squares (IRLS) Robust estimate of scale Other resistant ﬁtting methods Why not always use robust regression? This effect occurs because heteroscedasticity increases the variance of the coefficient estimates but the OLS procedure does not detect this increase. For example, But first, calculating robust standard errors, or heteroskedasticity-consistent standard errors; and generalized least squares estimation coefficients and standard errors. WLS regression and heteroskedasticity Suppose the variances of the residuals of a OLS regression are known, i.e. We say that an estimator or statistical procedure is robust if it provides useful information even if When we fit models using ordinary least squares (regress), we assume that the variance of the residuals is constant.If it is not constant, regress reports biased standard errors, leading to incorrect inferences. â¢ Flexibility of the robust model shown on complex international trade data. I assume that you know that the presence of heteroskedastic standard errors renders OLS estimators of linear regression models inefficient (although they remain unbiased). {robustbase} package. We call these standard errors heteroskedasticity-consistent (HC) standard errors. quantile statistics of Koenker and Bassett [17] is introduced. Stack Exchange network consists of 176 Q&A communities including Stack Overflow, the largest, most trusted online community for developers to learn, share their knowledge, and build their careers. Heteroskedasticity robust standard errors. This is because For this I will use the is just the same as with lm(): This however, gives you different estimates than when fitting a linear regression model. If the form of heteroskedasticity is known, a weighted least squares estimator has â¦ The HCCM provides a consistent estimator of the covariance matrix of the regression coeﬃcients in the presence of heteroscedasticity of an unknown form. â¢ A similar procedure can be adapted for What's this about? You run summary () on an lm.object and if you set the parameter robust=T it gives you back Stata-like heteroscedasticity consistent standard errors. # compute heteroskedasticity-robust standard errors vcov <-vcovHC (linear_model, type = "HC1") vcov #> (Intercept) STR #> (Intercept) 107.419993 -5.3639114 #> STR -5.363911 0.2698692 The output of vcovHC() is the variance-covariance matrix of coefficient estimates. heteroskedasticity is the proper spelling, because when transliterating Greek words, scientists and changed the values of the Region column. The Breusch-Pagan test is designed to detect any linear form of heteroskedasticity. Now, I will create a list of tidied regression results: broom::tidy() creates a data frame of the regression results. Doing that I get the same coefficients and standard errors slightly different bootstrap() function from the {modelr} package: Let’s take a look at the boot_education object: The column strap contains resamples of the original data. First of all, is it heteroskedasticity or heteroscedasticity? κ sometimes is transliterated as Heteroskedasticity robust standard errors I assume that you know that the presence of heteroskedastic standard errors renders OLS estimators of linear regression models inefficient (although they â¦ However, the expenditures on food of poorer families, who cannot afford lobster, will not vary much. Heteroscedasticity tends to produce p-values that are smaller than they should be. Finally, it is also possible to bootstrap the standard errors. About the Book Author In practice, heteroskedasticity-robust and clustered standard errors are usually larger than standard errors from regular OLS â however, this is not always the case. In the homoscedasticmodel, it is assumed that the variance of the error term is constant for all values of x. Heteroscedasticity allows the variance to be dependent on x, which is more accurate for many real scenarios. 24 [2019], Art. REGRESSION QUANTILES. Nonlinear regression 2. Hence in the practical work of your own you should always use the robust standard errors when running regression models. 1 https://scholarworks.umass rectly report heteroskedasticity-robust standard errors. å½¢ã¢ãã«ã¨ã¯ä¸è¬ã«è¢«èª¬æå¤æ°y ãè¤æ°ã®èª¬æå¤æ°x ã«ãã£ã¦ èª¬æããy = Î±+xâ²Î²+Îµ ã®ãããªã¢ãã«ãæãã¦ããã Impure heteroscedasticity refers to cases where you incorrectly specify the model, and that causes the non-constant variance. For example, in logistic regression heteroskedasticity can produce biased and misleading parameter estimates. let’s transform the data even further: list_mods is a list of the tidy_lm data frames. Heteroskedasticity-Robust Standard Errors for Fixed Effects Panel Data Regression James H. Stock and Mark W. Watson NBER Technical Working Paper â¦ Therefore correct heteroscedasticity either by changing the functional form or by using a robust command in the regression. A typical example is the set of observations of income in different cities. Detected heteroskedasticity, I proceed regression using Robust Standard Errors, where Iâm supposed to get heteroskedasticity-consistent standard errors. Privacy Policy, How to Specify the Correct Regression Model, How to Interpret Regression Coefficients and P-values, https://stats.stackexchange.com/questions/488423/can-i-validate-a-residual-plot-although-it-has-residual-patterns-if-i-am-not-int, https://www.researchgate.net/publication/263036348_Properties_of_Weighted_Least_Squares_Regression_for_Cutoff_Sampling_in_Establishment_Surveys, https://www.researchgate.net/publication/261534907_WEIGHTED_MULTIPLE_REGRESSION_ESTIMATION_FOR_SURVEY_MODEL_SAMPLING, https://www.researchgate.net/publication/333642828_Estimating_the_Coefficient_of_Heteroscedasticity, How To Interpret R-squared in Regression Analysis, How to Interpret P-values and Coefficients in Regression Analysis, Measures of Central Tendency: Mean, Median, and Mode, Multicollinearity in Regression Analysis: Problems, Detection, and Solutions, Understanding Interaction Effects in Statistics, How to Interpret the F-test of Overall Significance in Regression Analysis, Assessing a COVID-19 Vaccination Experiment and Its Results, P-Values, Error Rates, and False Positives, How to Perform Regression Analysis using Excel, Independent and Dependent Samples in Statistics, Independent and Identically Distributed Data (IID), Using Moving Averages to Smooth Time Series Data, Assessing Normality: Histograms vs. Normal Probability Plots, Guidelines for Removing and Handling Outliers in Data. The conventional heteroskedasticity-robust (HR) variance matrix estimator for cross-sectional regression (with or without a degrees-of-freedom adjustment), applied to the ï¬xed-effects estimator for panel data with serially uncorrelated errors, is incon- You can refer to Zeileis (2004) This package is quite interesting, and offers quite a lot of functions Here, variability could be quantified by the variance or any other measure of statistical dispersion. ... References Chesher, A. and Jewitt, I. Thus, the robust standard errors are appropriate even under homoskedasticity. Yes, I am interested in estimates and standard errors which are both outlier robust AND heteroskedasticity consistent. Heteroskedasticity in the Linear Model 6 7 Estimation with GLS/WLS when is Known When is known, is e ciently estimated with generalized least squares (GLS) b GLS= X0 Cribari-Neto, Francisco; Lima, Maria da Glória A. Source: Econometrica, Vol. Braz. robustness tests and discuss the differences. Testing for Heteroscedasticity There are several methods of testing for the presence of heteroscedasticity. Nonlinearity 1. use the Latin letter k in place of the Greek letter κ (kappa). If the form of heteroskedasticity is known, a weighted least squares estimator has better How to detect heteroscedasticity? Heteroskedasticity-Robust Inference in Linear Regression Models Estimated by OLS This note does three things: 1. identifies the nature and consequences of heteroskedasticity for estimation and inference based on OLS 2. Robust regression with robust weight functions 2. rreg y x1 x2 3. Nothing within the definition of heteroskedasticity pertains to the 2 Practical Assessment, Research, and Evaluation, Vol. Heteroskedasticity-Robust Standard Errors for Fixed Effects Panel Data Regression May, 2006 This revision: July, 2007 James H. Stock Department of Economics, Harvard University and the NBER Mark W. Watson1 Department Software packages usually default to a hom… Santos Silvaz June 18, 2013 Abstract This note introduces a wrapper for qreg which reports standard errors and t statistics that are asymptotically valid under Let’s look Now, let’s do a scatterplot of per capita expenditures linearmodels). Another way of dealing with heteroskedasticity is to use the lmrob() function from the {robustbase} package. This package is quite interesting, and offers quite a lot of functions for robust linear, and nonlinear, regression models. The most commonly used is the Time-Honored Method of Inspection (THMI). BY ROGER KOENKER AND GILBERT BASSETT JR.'. Robust regression can be used in any situation in which you would use least squares regression. Is this conceptually the same as weighted OLS (weight by 1/variance)? â¢ We use OLS (inefficient but) consistent estimators, and calculate an alternative â¦ J. Probab. 1 (Jan., 1982), pp. : Heteroskedasticity can be very problematic with methods besides OLS.