A fundamental principle of data analysis is that a statistic is an estimate of a parameter for the population. A statistic is calculated from a random sample. This leads to uncertainty in the estimate: a different random sample would have produced a different statistic. To quantify the uncertainty, SAS procedures
I recently read an article that describes ways to compute confidence intervals for the difference in a percentile between two groups. In Eaton, Moore, and MacKenzie (2019), the authors describe a problem in hydrology. The data are the sizes of pebbles (grains) in rivers at two different sites. The authors
At SAS Global Forum 2019, Daymond Ling presented an interesting discussion of binary classifiers in the financial industry. The discussion is motivated by a practical question: If you deploy a predictive model, how can you assess whether the model is no longer working well and needs to be replaced? Daymond
Many SAS procedures support the BY statement, which enables you to perform an analysis for subgroups of the data set. Although the SAS/IML language does not have a built-in "BY statement," there are various techniques that enable you to perform a BY-group analysis. The two I use most often are
When I run a bootstrap analysis, I create graphs to visualize the distribution of the bootstrap statistics. For example, in my article about how to bootstrap the difference of means in a two-sample t test, I included a histogram of the bootstrap distribution and added reference lines to indicate a
This article describes best practices and techniques that every data analyst should know before bootstrapping in SAS. The bootstrap method is a powerful statistical technique, but it can be a challenge to implement it efficiently. An inefficient bootstrap program can take hours to run, whereas a well-written program can give
If you want to bootstrap the parameters in a statistical regression model, you have two primary choices. The first, case resampling, is discussed in a previous article. This article describes the second choice, which is resampling residuals (also called model-based resampling). This article shows how to implement residual resampling in
If you want to bootstrap the parameters in a statistical regression model, you have two primary choices. The first is case resampling, which is also called resampling observations or resampling pairs. In case resampling, you create the bootstrap sample by randomly selecting observations (with replacement) from the original data. The
Since the late 1990s, SAS has supplied macros for basic bootstrap and jackknife analyses. This article provides an example that shows how to use the %BOOT and %BOOTCI macros. The %BOOT macro generates a bootstrap distribution and computes basic statistics about the bootstrap distribution, including estimates of bias, standard error,
This article shows how to implement balanced bootstrap sampling in SAS. The basic bootstrap samples with replacement from the original data (N observations) to obtain B new samples. This is called "uniform" resampling because each observation has a uniform probability of 1/N of being selected at each step of the
A previous article provides an example of using the BOOTSTRAP statement in PROC TTEST to compute bootstrap estimates of statistics in a two-sample t test. The BOOTSTRAP statement is new in SAS/STAT 14.3 (SAS 9.4M5). However, you can perform the same bootstrap analysis in earlier releases of SAS by using
Bootstrap resampling is a powerful way to estimate the standard error for a statistic without making any parametric assumptions about its sampling distribution. The bootstrap method is often implemented by using a sequence of calls to resample from the data, compute a statistic on each sample, and analyze the bootstrap
The SURVEYSELECT procedure in SAS 9.4M5 supports the OUTRANDOM option, which causes the selected items in a simple random sample to be randomly permuted after they are selected. This article describes several statistical tasks that benefit from this option, including simulating card games, randomly permuting observations in a DATA step,
I wrote more than 100 posts for The DO Loop blog in 2017. The most popular articles were about SAS programming tips, statistical data analysis, and simulation and bootstrap methods. Here are the most popular articles from 2017 in each category. General SAS programming techniques INTCK and INTNX: Do you
I recently showed how to compute a bootstrap percentile confidence interval in SAS. The percentile interval is a simple "first-order" interval that is formed from quantiles of the bootstrap distribution. However, it has two limitations. First, it does not use the estimate for the original data; it is based only
I previously wrote about how to compute a bootstrap confidence interval in Base SAS. As a reminder, the bootstrap method consists of the following steps: Compute the statistic of interest for the original data Resample B times from the data to form B bootstrap samples. B is usually a large
One way to assess the precision of a statistic (a point estimate) is to compute the standard error, which is the standard deviation of the statistic's sampling distribution. A relatively large standard error indicates that the point estimate should be viewed with skepticism, either because the sample size is small
Last week I showed how to use the simple bootstrap to randomly resample from the data to create B bootstrap samples, each containing N observations. The simple bootstrap is equivalent to sampling from the empirical cumulative distribution function (ECDF) of the data. An alternative bootstrap technique is called the smooth
A common question is "how do I compute a bootstrap confidence interval in SAS?" As a reminder, the bootstrap method consists of the following steps: Compute the statistic of interest for the original data Resample B times from the data to form B bootstrap samples. How you resample depends on
Many simulation and resampling tasks use one of four sampling methods. When you draw a random sample from a population, you can sample with or without replacement. At the same time, all individuals in the population might have equal probability of being selected, or some individuals might be more likely
How do you sample with replacement in SAS when the probability of choosing each observation varies? I was asked this question recently. The programmer thought he could use PROC SURVEYSELECT to generate the samples, but he wasn't sure which sampling technique he should use to sample with unequal probability. This
My colleagues at the SAS & R blog recently posted an example of how to program a permutation test in SAS and R. Their SAS implementation used Base SAS and was "relatively cumbersome" (their words) when compared with the R code. In today's post I implement the permutation test in
Bootstrap methods and permutation tests are popular and powerful nonparametric methods for testing hypotheses and approximating the sampling distribution of a statistic. I have described a SAS/IML implementation of a bootstrap permutation test for matched pairs of data (an alternative to a matched-pair t test) in my paper "Modern Data
Last week I showed three ways to sample with replacement in SAS. You can use the SAMPLE function in SAS/IML 12.1 to sample from a finite set or you can use the DATA step or PROC SURVEYSELECT to extract a random sample from a SAS data set. Sampling without replacement
Randomly choosing a subset of elements is a fundamental operation in statistics and probability. Simple random sampling with replacement is used in bootstrap methods (where the technique is called resampling), permutation tests and simulation. Last week I showed how to use the SAMPLE function in SAS/IML software to sample with
With each release of SAS/IML software, the language provides simple ways to carry out tasks that previously required more effort. In 2010 I blogged about a SAS/IML module that appeared in my book Statistical Programming with SAS/IML Software, which was written by using the SAS/IML 9.2. The blog post showed
A challenge for statistical programmers is getting data into the right form for analysis. For graphing or analyzing data, sometimes the "wide format" (each subject is represented by one row and many variables) is required, but other times the "long format" (observations for each subject span multiple rows) is more
I was recently asked the following question: I am using bootstrap simulations to compute critical values for a statistical test. Suppose I have test statistic for which I want a p-value. How do I compute this? The answer to this question doesn't require knowing anything about bootstrap methods. An equivalent
In a previous post, I described how to compute means and standard errors for data that I want to rank. The example data (which are available for download) are mean daily delays for 20 US airlines in 2007. The previous post carried out steps 1 and 2 of the method
I recently posted an article about representing uncertainty in rankings on the blog of the ASA Section for Statistical Programmers and Analysts (SSPA). The posting discusses the importance of including confidence intervals or other indicators of uncertainty when you display rankings. Today's article complements the SSPA post by showing how
