Confidence Intervals in Ridge Regression using Jackknife and Bootstrap Methods

Size: px
Start display at page:

Download "Confidence Intervals in Ridge Regression using Jackknife and Bootstrap Methods"

Transcription

1 Chapter 4 Confidence Intervals in Ridge Regression using Jackknife and Bootstrap Methods 4.1 Introduction It is now explicable that ridge regression estimator (here we take ordinary ridge estimator (ORE) in particular) can be of great use for estimating the unknown regression coefficients in the presence of multicollinearity. But apart from its ability to create good parameter estimates with smaller MSE than the OLSE, it must also provide fine solutions when dealing with more intricate inference problems like obtaining confidence intervals. The problem with the ORE is that its sampling distribution is unknown, hence we make use of the resampling methods to obtain the asymptotic confidence intervals for the regression coefficients based on the ORE. Recently, Firinguetti and Bobadilla [40] developed asymptotic confidence intervals for the regression coefficients based on ORE and Edgeworth expansion. Crivelli et. al. [17] proposed the use of a technique that combines the bootstrap and the Edgeworth expansion to obtain an approximation to the distribution of some ridge regression estimators and carried out some simulation experiments. Alheety and Ramanathan [4] proposed a data dependent confidence interval for the ridge parameter k. Confidence intervals have become very popular in the applied statistician s col- 73

2 lection of data-analytic tools. They combine point estimation and hypothesis testing into a single inferential statement. Recent advances in statistical methods and fast computing allow the construction of highly accurate approximate confidence intervals. Confidence intervals for a given population parameter θ are sample based range [ˆθ 1, ˆθ 2 ] given out for the unknown θ. The range possesses the property that θ would lie within its bounds with a high specified probability. Of course this probability is with respect to all possible samples, each sample giving rise to a confidence interval which depends on the chance mechanism involved in drawing the samples. Two approaches for confidence intervals includes the construction of exact intervals for special cases like the ratio of normal means or a single binomial parameter and others are the most commonly used, approximate confidence intervals which are also known as standard intervals (or normal theory intervals) having the following form ˆθ ± z (α)ˆσ, (4.1.1) where ˆθ is a point estimate of the parameter θ, ˆσ is an estimate of standard deviation of ˆθ, and z (α) is the 100α th percentile of a normal variate, (for example z (0.95) = etc.). The main drawback of standard intervals is that they are based on an asymptotic approximation that may not be accurate in practice. There has been considerable progress on developing better confidence intervals techniques for improving the standard interval, involving bias corrections and parameter transformations etc.. Numerous methods have been proposed to improve upon standard intervals. These methods produce approximate confidence intervals that have better coverage accuracy than the standard one. Some references under this area includes McCullagh [67], Cox and Reid [16], Efron [34], Hall [49], DiCiccio and Tibshirani [24] etc.. The confidence intervals based on the resampling methods like bootstrap and jackknife can be seen as automatic algorithms for carrying out these improvements. We have already discussed that bootstrap and jackknife are two powerful methods for variance estimation which is why we can use them to produce confidence intervals. We discuss the different methods for construction of the bootstrap and jackknife confidence intervals in Section (4.2) and (4.3) respectively, Section (4.4) contains the model and the estimators, Section (4.5) gives a simulation study for the comparison of coverage probabilities and average confidence widths based on ORE and OLSE and 74

3 Section (4.6) consists of some concluding remarks. 4.2 Bootstrap Confidence Intervals Suppose that we draw a sample S = (x 1,x 2,...,x n ) of size n from a population of size N and we are interested in some statistic ˆθ as an estimate of the corresponding population parameter θ. The classical inference, where we derive the asymptotic distribution of sampling distribution of ˆθ when its exact distribution is intractable or has some drawbacks. The corresponding sampling distribution of statistic ˆθ may be inaccurate if the assumptions about the population are wrong or the sample size is relatively small. On the other hand, as discussed earlier also, nonparametric bootstrap enables us to estimate the sampling distribution of a statistic empirically without making any assumptions about the parent population and without deriving sampling distribution explicitly. In this method, we draw a sample of size n from the elements of sample S and the sampling is done with replacement. This is the first bootstrap resample S 1 = (x 11,x 12,...,x 1n). Each element of the resample is selected from the original sample with probability 1/n. We repeat this step of resampling a large number of times say B so that the r th bootstrap resample is given by S r = (x r1,x r2,...,x rn). Next, we compute the statistic ˆθ for each of the bootstrap resample i.e. ˆθ r. Now, the distribution of ˆθ r around the original statistic ˆθ is similar to the sampling distribution of ˆθ around population parameter θ. The average of the statistics computed based on bootstrap resamples estimates the expectation of the bootstrapped statistic. It is given as ˆθ = B r=1 ˆθ r B. Then ˆB = ˆθ ˆθ gives an estimate of bias of ˆθ which is ˆθ θ. The estimate of the bootstrap variance of ˆθ is ˆV (ˆθ ) = B r=1 (ˆθ r ˆθ ) 2, B 1 which estimates the sampling variance of ˆθ. With the help of this variance estimate, we can easily produce bootstrap confidence intervals. There are several 75

4 methods for constructing bootstrap confidence intervals, each of them are briefly discussed below Normal Theory Method The first method for constructing bootstrap confidence interval is based on the assumption that the sampling distribution of ˆθ is normal. This uses the bootstrap estimate of sampling variance to construct a 100(1 α)% confidence interval of the following form where ˆ SE(ˆθ ) = (2ˆθ ˆθ ) ± z (1 α/2) ˆ SE(ˆθ ), ˆV (ˆθ ) is the bootstrap estimate of the standard error of ˆθ and z (1 α/2) is the (1 α/2) quantile of the standard normal distribution Bootstrap Percentile and Centered Bootstrap Percentile Method Another method for the construction of bootstrap confidence intervals is the bootstrap percentile method which is the most popular among all primarily due to its simplicity and natural appeal. In this method, we use the empirical quantiles of ˆθ to form the confidence interval for θ. The bootstrap percentile confidence interval has the following form ˆθ (lower) < θ < ˆθ (upper), where ˆθ (1), ˆθ (2),..., ˆθ (B) are the ordered bootstrap replicates of ˆθ with lower confidence limit=[(b + 1)α/2] and upper confidence limit=[(b + 1)(1 α/2)]. The square brackets indicates rounding to the nearest integer. For example, there are 1000 bootstrap replications (i.e. B = 1000) for ˆθ, denoted by (ˆθ 1, ˆθ 2,..., ˆθ 1000). After sorting them from bottom to top, let us denote these bootstrap values as (ˆθ (1), ˆθ (2),..., ˆθ (1000) ). Then the bootstrap percentile confidence interval at 95% level would be [ˆθ (25), ˆθ (975) ]. Here, it is implicitly assumed that the distribution 76

5 of ˆθ is symmetric around θ as the method approximates the sampling distribution of (ˆθ θ) by the bootstrap distribution of (ˆθ ˆθ ) which is contrary to the bootstrap theory. The coverage error is substantial if the distribution of ˆθ is not nearly symmetric. Now, suppose the sampling distribution of (ˆθ θ) is approximated by the bootstrap distribution of (ˆθ ˆθ) which is what bootstrap theory explains. Then the statement that (ˆθ θ) lies within the range(b ˆθ,B ˆθ) would carry a probability of 0.95 where B s denotes the 100s th percentile of bootstrap distribution of ˆθ. In case of 1000 bootstrap replications B = ˆθ (25) and B = ˆθ (975), we see that the above statement reduces to the statement that θ lies in the following confidence interval (2ˆθ B ) < θ < (2ˆθ B ). The above mentioned range is known as centered bootstrap percentile confidence interval. This method can be applied to any statistic and works well in cases where the bootstrap distribution is symmetrical and centered around the observed statistic Bootstrap Studentized-t Method Another method for constructing the bootstrap confidence intervals is the studentized bootstrap, also called the bootstrap-t method. The studentized-t bootstrap confidence interval takes the same form as the normal confidence interval except that instead of using the quantiles from a normal distribution, a bootstrapped t-distribution is constructed from which the quantiles are computed (see Davison and Hinkley [21] and Efron and Tibshirani [35]). Bootstrapping a statistical function of the form t = (ˆθ θ)/(se), where SE is the sample estimate of the standard error of ˆθ brings extra accuracy. This additional accuracy is due to the so called one-term Edgeworth correction by the bootstrap (see Hall [50]). The basic example of it is the standard t statistics t = ( x µ)/(s/ n), which is a special case with θ = µ (the population mean), ˆθ = x (the sample mean) and s being the sample standard deviation. The bootstrap counterpart 77

6 of this is t = (ˆθ ˆθ)/(SE), where SE is the standard error based on bootstrap distribution. Denote the 100s th bootstrap percentile of t by b s and consider the statement that (b < t < b ) and after substituting t = (ˆθ θ)/se, we get confidence limits for θ as ˆθ (SE)b < θ < ˆθ (SE)b This range is known as bootstrap-t based confidence interval for θ at 95% confidence level. Like the normal approximation, this confidence interval leads to interval estimates that are symmetric about the original point estimator, which may not be appropriate. It is best suited for bootstrapping a location statistic. The use of the studentized bootstrap is controversial to some extent. In some cases, the endpoints of the intervals seem to be too wide and the method seem to be sensitive to outliers. Also, this method is computationally very intensive if (SE) is calculated using a double bootstrap but if (SE) is easily available, the method performs reliably in many examples in both its standard and variance stabilized forms (see Davison and Hinkley [21]). Percentile bootstrap endpoints are simple to calculate and can work well, especially if the sampling distribution is symmetrical. It may not have the correct coverage when the sampling distribution of the statistic is skewed. Coverage of the percentile bootstrap can be improved by adjusting the endpoints for bias and nonconstant variance. This method is known as BCa method which we discuss in the next subsection Bias corrected and Accelerated (BCa) Method If we have a distribution which has skew or bias, we need to do some adjustments. One method which is proved to be reliable in such cases is BCa method, this method will tend to be closer to the true confidence interval than the percentile method. This is an automatic algorithm for producing highly accurate 78

7 confidence limits from a bootstrap distribution. For a detailed review on the same, see Efron [34], Hall [49], DiCiccio [23], Efron and Tibshirani [35]. The BCa procedure is a method of setting approximate confidence intervals for θ from the percentiles of the bootstrap histogram. The parameter of interest being θ only, ˆθ is an estimate of θ based on the observed data and ˆθ is a bootstrap replication of ˆθ obtained by resampling. Let Ĝ(c) be the cumulative distribution function of B bootstrap replications of ˆθ, Ĝ(c) = #{ˆθ < c}/b. (4.2.1) The upper endpoint, ˆθBCa [α] of one-sided BCa interval at α level i.e. θ (, ˆθ BCa [α]) is defined in terms of Ĝ and two parameters z 0, the bias correction and a, the acceleration. That is how BCa stands for bias corrected and accelerated. The BCa endpoint is given by ( ) z 0 + z ˆθ (α) BCa [α] = Ĝ 1 Φ z 0 +. (4.2.2) 1 a(z 0 + z (α) ) Here Φ is the standard normal c.d.f., with z (α) = Φ 1 α. The central 90% BCa confidence interval is given by (ˆθ BCa [0.05], ˆθ BCa [.95]). In (4.2.2), if a and z 0 are zero, then ˆθ BCa [α] = Ĝ 1 (α), the 100α th percentile of the bootstrap replications. Also, if Ĝ is normal, then ˆθ BCa [α] = ˆθ + z (α)ˆσ, the standard interval endpoint. In general, (4.2.2) makes three different corrections to the standard intervals, improving their coverage accuracy. Suppose that there exists a monotone increasing transformation φ = m(θ) such that ˆφ = m(ˆθ) is normally distributed for every choice of θ, but with a bias and a non constant variance, The BCa algorithm estimates z 0 by ˆφ N(φ z 0 σ φ,σ 2 φ), σ φ = 1 + aφ. (4.2.3) { ẑ 0 = Φ 1 #{ˆθ (b) < ˆθ} }, B Φ 1 of the proportion of the bootstrap replications less than ˆθ. The acceleration 79

8 a measures how quickly the standard error is changing on the normalized scale. The acceleration a is estimated using the empirical influence function of the statistic ˆθ = t( ˆF), t((1 ǫ) U i = lim ˆF + ǫδ i ) ǫ 0,i = 1, 2,...,n. (4.2.4) ǫ Here δ i is a point mass on x i, so (1 ǫ) ˆF + ǫδ i is a version of ˆF putting extra weight on x i and less weight on the other points. The estimate of a is â = 1 6 n i=1 U3 i ( n (4.2.5) i=1 U2 i )3/2. There is a simpler way of calculating U i and â. Instead of (4.2.4), we can use the following jackknife influence function (see Hinkley [51]) in (4.2.5) U i = (n 1)(ˆθ ˆθ (i) ), (4.2.6) where ˆθ (i) is the estimate of θ based on the reduced data set x (i) = (x 1,x 2,...,x i 1, x i+1,...,x n ). In the next section, we discuss about the confidence intervals based on jackknife method. 4.3 Jackknife Confidence Intervals Jackknife technique is generally used to reduce the bias of parameter estimates and to estimate the variance. Let n be the total sample size and the procedure is to estimate a parameter θ for n times, each time deleting one sample data point. The resulting estimator denoted by ˆθ i where the i th data point has been excluded before calculating the estimate. The so-called pseudo values are computed as θ i = nˆθ (n 1)ˆθ i, here ˆθ is the value of the parameter estimated from the entire data set. These pseudo values act as independent and identical normal random variables. The average of the above gives the jackknife estimator of θ as 80

9 θ = 1 n n θ i = nˆθ i=1 (n 1) n n ˆθ i. i=1 Also, the variance of these pseudo values is the estimate of the variance of ˆθ which is given by V = 1 n(n 1) n ( θ i θ)( θ i θ). (4.3.1) i=1 Apart from reducing bias and estimating variance, the jackknife technique also offers a very simple method to construct confidence interval for θ (see Miller [71]). Now, using the consistent estimator for variance given in (4.3.1), we get the confidence interval for θ (see Hinkley [51]) as ) where t (1 α2 ;n 1 is the upper α 2 ( θ t 1 α ) 2 ;n 1 vii, (4.3.2) 100% point of the Students s t- distribution with n 1 degrees of freedom, which is suited even for small samples. The next section contains the model and the estimators and confidence intervals in terms of the unknown regression coefficients. 4.4 The Model and the Estimators Let us consider the following multiple linear regression model y = Xβ + u (4.4.1) where y is an (n 1) vector of observations on the variable to be explained, X is an (n p) matrix of n observations on p explanatory variables assumed to be of full column rank, β is a (p 1) vector of regression coefficients associated with them and u is an (n 1) vector of disturbances, the elements of which are assumed to be independently and identically normally distributed with E(u) = 0; V ar(u) = σ 2 I. 81

10 The OLSE for β in model (4.4.1) is given by ˆβ OLSE = (X X) 1 X y. (4.4.2) Now, suppose if the data has the problem of multicollinearity, we use the ORE given by Hoerl and Kennard [52] as ˆβ ORE = (X X + ki) 1 X y = (I A 1 ki)ˆβ, (4.4.3) where k > 0 and A = X X + ki. The jackknife ridge estimator JRE of ORE is calculated using the formula ˆβ JRE = [I (A 1 ki) 2 ]ˆβ (4.4.4) Now for bootstrapping, firstly, for the model defined in (4.4.1), we fit the least squares regression equation for full sample and calculate the standardized residuals u i and then draw an n sized bootstrap sample with replacement (u (b) 1,u (b) 2,...,u (b) n ) from the residuals u i s giving 1/n probability to each u i. After this, we obtain the bootstrap y values using the resampled residuals keeping the design matrix fixed as shown below y (b) = X ˆβ OLSE + u (b). We then regress these bootstrapped y values on the fixed X to obtain the bootstrap estimates of the regression coefficients. So, the ORE from the first bootstrap sample is ˆβ ORE(b1) = (X X + ki) 1 X y (b1). Repeat the above steps B times where B is the number of bootstrap resamples. Based on these, the bootstrap estimator for β is given by B ˆβ ORE = ˆβ ORE(br)/B, (4.4.5) r=1 where r = 1,...,B. The estimated bias is given by Bias est = ˆβ ORE ˆβ OLSE. 82

11 The estimated variance of ridge estimator through bootstrap is given as V ar est = B (ˆβ ORE(br) ˆβ ORE ) 2 /(B 1). (4.4.6) r=1 Now, based on these estimators, we construct the confidence intervals for the regression coefficient β in following subsection Confidence Intervals for Regression Coefficients using ORE The different confidence intervals discussed in last section gives the following forms for the confidence intervals in terms of the regression coefficient β. 1. Normal theory method A 95% confidence interval for β based on ORE is (2ˆβ ORE ˆβ ORE ) z (1 α/2) V arest < β < (2ˆβ ORE ˆβ ORE ) + z (1 α/2) V arest, where α = 0.05, V ar est is the bootstrap estimate of the variance of ˆβ ORE as defined in (4.4.6) and z (1 α/2) is the (1 α/2) quantile of the standard normal distribution. 2. Percentile Method A 95% confidence interval for 1000 bootstrap resamples would be ˆβ ORE(25) < β < ˆβ ORE(975), where ˆβ ORE(r) is the rth observation in the ordered bootstrap replicates of ˆβ ORE. 3. Studentized t Method A 95% studentized t-interval would be ˆβ ORE (SE)b < β < ˆβ ORE (SE)b 0.025, 83

12 where b s is the 100s th bootstrap percentile of t and t is as defined in Section (4.2.3). 4. BCa Method A 95% confidence interval for β based on ORE is ˆβ BCa [0.025] < β < ˆβ BCa [0.975], where ˆθ BCa [α] is as defined in (4.2.2). Another method known as the ABC (approximate bootstrap confidence intervals) method has been proposed by Efron[34], that gives analytic adjustment to BCa method for smoothly defined parameters in exponential families. They are touted in the literature as improvements for common parametric and nonparametric BCa procedures, and may be preferred in order to avoid the BCa s Monte Carlo calculations. [see avoidhese intervals are the approximations to the BCa intervals of Efron[34], using analytic methods to avoid the BCa s Monte Carlo calculations [see DiCiccio and Efron [25];Efron and Tibshirani [35]; DiCiccio and Efron [26]]. We have adopted this method in the linear model setup, however we did not see significant improvement in its performance over the BCa method. Hence, this method is not pursued in the numerical investigations carried out later. 5. Jackknife Method A 95% jackknife confidence interval for β based on ORE is ( ˆβ JRE t 1 α ) ( 2 ;n p vii < β < ˆβ JRE + t 1 α ) 2 ;n p vii, (4.4.7) ) where α = 0.05, t (1 α2 ;n p is the upper α 100% point of the Students s 2 t-distribution with (n p) degrees of freedom and v ii is as given in (4.3.1) with (n 1) in the denominator being replaced by (n p). In order to compare these methods of constructing asymptotic confidence inter- 84

13 vals based on ORE and OLSE, we calculate coverage probabilities which is defined as the proportion that the confidence interval includes the true parameter, under repeated sampling from the same underlying population and confidence width which is the difference between the upper and lower confidence endpoints. In the next section, we carry out a simulation study to obtain the confidence widths and coverage probabilities based on the confidence intervals developed using ORE and OLSE. 4.5 A Simulation Study After getting into some theoretical aspects of each method to construct confidence intervals, we now apply the methods on simulated data and compare the performance ORE and OLSE based on their coverage probabilities and confidence widths. The model is y = Xβ + u (4.5.1) where u N(0, 1). Here β is taken as the normalized eigen vector corresponding to the largest eigen value of X X. The explanatory variables are generated from the following equation x ij = (1 ρ 2 ) 1 2 wij + ρw ip, i = 1, 2...,n; j = 1, 2,...,p. where w ij are independent standard normal pseudo-random numbers and ρ 2 is the correlation between x ij and x ij for j,j < p and j j. When j or j = p, the correlation will be ρ. We have taken ρ = 0.9 and 0.99 to investigate the effects of different degrees of collinearity with sample sizes n = 25, 50 and 100. The feasible value of k is obtained by the optimal formula k = pσ2 β β Hoerl et al. [53], so that ˆk = pˆσ2 ˆβ ˆβ, as given by where ˆσ 2 = (y X ˆβ) (y X ˆβ). n p 85

14 For calculating different bootstrap confidence intervals like Normal, Percentile, Studentized and BCa, we have used the function called boot.ci in R. The confidence limits through jackknife are calculated using (4.4.7) and variance of jackknifed estimator is calculated using bootstrap method for variance estimation. We have calculated the coverage probabilities and average confidence width using 1999 bootstrap resamples and the experiment is repeated 500 times. The coverage probability, say CP is calculated using the following formula CP = #(ˆβ L < β < ˆβ U ), N and the average confidence width, say CW is calculated by CW = N i=1 (ˆβ U ˆβ L ), N where N is the simulation size, ˆβ L and ˆβ U are lower and upper confidence interval endpoints respectively. The results for coverage probability and average confidence width at 95% and 99% confidence levels with different values of n and ρ are summarized in Table (4.1)-Table (4.4). Note that in all the tables, column namely OLSE gives the coverage probability and confidence width based on the confidence intervals through ˆβ OLSE using t-distribution. From Tables (4.1) and (4.2), we find that the coverage probabilities of all the intervals improve with the increasing value of n and become close to each other which is due to the consistency of the estimators. It is evident from Tables (4.3) and (4.4) that the confidence intervals based on ORE have shorter widths in comparison to the width of the interval based on OLSE. It is interesting to note that the coverage probabilities and confidence width through OLSE and through jackknife are very close to each other. Also, from Tables (4.3) and (4.4), we find that with the increasing collinearity between the dependent variables, the difference between the confidence width of interval based on OLSE and intervals based on ORE is increasing. Also, with the increasing value of sample size, the confidence width of all the intervals is decreasing. 86

15 According to Tables (4.1) and (4.2), all the bootstrap methods are generally conservative in terms coverage probabilities, however jackknife method seems to give coverage probabilities closer to the target. In terms of confidence width, resampling methods have smaller confidence width than the OLSE; jackknife method having larger confidence width than bootstrap methods. In noting that bootstrap methods are conservative with smaller confidence width, they seem to have an advantage over the jackknife method. 4.6 Concluding Remarks In the present chapter, we illustrated the use of different confidence intervals based on bootstrap and jackknife resampling methods. We obtained the coverage probabilities and confidence width based on ORE using different bootstrap and jackknife method and compared it with that of OLSE which we computed using t-intervals. The shorter confidence widths obtained through ORE shows its superiority over OLSE in the case of multicollinearity. The next chapter deals with jackknifing and bootstrapping another biased estimator given by Liu [64] to overcome the drawback of ORE. 87

16 Table 4.1: Coverage Probabilities through different methods at 95% confidence level n ρ OLSE N ormal P ercentile Studentized BCa Jackknif e

17 Table 4.2: Coverage Probabilities through different methods at 99% confidence level n ρ OLSE N ormal P ercentile Studentized BCa Jackknif e

18 Table 4.3: Average confidence width through different methods at 95% confidence level using OLSE and ORE n ρ OLSE N ormal P ercentile Studentized BCa Jackknif e

19 Table 4.4: Average confidence width through different methods at 99% confidence level using ORE n ρ OLSE N ormal P ercentile Studentized BCa Jackknif e

Bootstrap Confidence Intervals

Bootstrap Confidence Intervals Bootstrap Confidence Intervals Patrick Breheny September 18 Patrick Breheny STA 621: Nonparametric Statistics 1/22 Introduction Bootstrap confidence intervals So far, we have discussed the idea behind

More information

The Nonparametric Bootstrap

The Nonparametric Bootstrap The Nonparametric Bootstrap The nonparametric bootstrap may involve inferences about a parameter, but we use a nonparametric procedure in approximating the parametric distribution using the ECDF. We use

More information

Better Bootstrap Confidence Intervals

Better Bootstrap Confidence Intervals by Bradley Efron University of Washington, Department of Statistics April 12, 2012 An example Suppose we wish to make inference on some parameter θ T (F ) (e.g. θ = E F X ), based on data We might suppose

More information

Finite Population Correction Methods

Finite Population Correction Methods Finite Population Correction Methods Moses Obiri May 5, 2017 Contents 1 Introduction 1 2 Normal-based Confidence Interval 2 3 Bootstrap Confidence Interval 3 4 Finite Population Bootstrap Sampling 5 4.1

More information

Bootstrap (Part 3) Christof Seiler. Stanford University, Spring 2016, Stats 205

Bootstrap (Part 3) Christof Seiler. Stanford University, Spring 2016, Stats 205 Bootstrap (Part 3) Christof Seiler Stanford University, Spring 2016, Stats 205 Overview So far we used three different bootstraps: Nonparametric bootstrap on the rows (e.g. regression, PCA with random

More information

The automatic construction of bootstrap confidence intervals

The automatic construction of bootstrap confidence intervals The automatic construction of bootstrap confidence intervals Bradley Efron Balasubramanian Narasimhan Abstract The standard intervals, e.g., ˆθ ± 1.96ˆσ for nominal 95% two-sided coverage, are familiar

More information

4 Resampling Methods: The Bootstrap

4 Resampling Methods: The Bootstrap 4 Resampling Methods: The Bootstrap Situation: Let x 1, x 2,..., x n be a SRS of size n taken from a distribution that is unknown. Let θ be a parameter of interest associated with this distribution and

More information

Improved Ridge Estimator in Linear Regression with Multicollinearity, Heteroscedastic Errors and Outliers

Improved Ridge Estimator in Linear Regression with Multicollinearity, Heteroscedastic Errors and Outliers Journal of Modern Applied Statistical Methods Volume 15 Issue 2 Article 23 11-1-2016 Improved Ridge Estimator in Linear Regression with Multicollinearity, Heteroscedastic Errors and Outliers Ashok Vithoba

More information

On the Accuracy of Bootstrap Confidence Intervals for Efficiency Levels in Stochastic Frontier Models with Panel Data

On the Accuracy of Bootstrap Confidence Intervals for Efficiency Levels in Stochastic Frontier Models with Panel Data On the Accuracy of Bootstrap Confidence Intervals for Efficiency Levels in Stochastic Frontier Models with Panel Data Myungsup Kim University of North Texas Yangseon Kim East-West Center Peter Schmidt

More information

MA 575 Linear Models: Cedric E. Ginestet, Boston University Non-parametric Inference, Polynomial Regression Week 9, Lecture 2

MA 575 Linear Models: Cedric E. Ginestet, Boston University Non-parametric Inference, Polynomial Regression Week 9, Lecture 2 MA 575 Linear Models: Cedric E. Ginestet, Boston University Non-parametric Inference, Polynomial Regression Week 9, Lecture 2 1 Bootstrapped Bias and CIs Given a multiple regression model with mean and

More information

Double Bootstrap Confidence Interval Estimates with Censored and Truncated Data

Double Bootstrap Confidence Interval Estimates with Censored and Truncated Data Journal of Modern Applied Statistical Methods Volume 13 Issue 2 Article 22 11-2014 Double Bootstrap Confidence Interval Estimates with Censored and Truncated Data Jayanthi Arasan University Putra Malaysia,

More information

Bootstrap Testing in Econometrics

Bootstrap Testing in Econometrics Presented May 29, 1999 at the CEA Annual Meeting Bootstrap Testing in Econometrics James G MacKinnon Queen s University at Kingston Introduction: Economists routinely compute test statistics of which the

More information

Model Selection, Estimation, and Bootstrap Smoothing. Bradley Efron Stanford University

Model Selection, Estimation, and Bootstrap Smoothing. Bradley Efron Stanford University Model Selection, Estimation, and Bootstrap Smoothing Bradley Efron Stanford University Estimation After Model Selection Usually: (a) look at data (b) choose model (linear, quad, cubic...?) (c) fit estimates

More information

Resampling and the Bootstrap

Resampling and the Bootstrap Resampling and the Bootstrap Axel Benner Biostatistics, German Cancer Research Center INF 280, D-69120 Heidelberg benner@dkfz.de Resampling and the Bootstrap 2 Topics Estimation and Statistical Testing

More information

The bootstrap. Patrick Breheny. December 6. The empirical distribution function The bootstrap

The bootstrap. Patrick Breheny. December 6. The empirical distribution function The bootstrap Patrick Breheny December 6 Patrick Breheny BST 764: Applied Statistical Modeling 1/21 The empirical distribution function Suppose X F, where F (x) = Pr(X x) is a distribution function, and we wish to estimate

More information

UNIVERSITÄT POTSDAM Institut für Mathematik

UNIVERSITÄT POTSDAM Institut für Mathematik UNIVERSITÄT POTSDAM Institut für Mathematik Testing the Acceleration Function in Life Time Models Hannelore Liero Matthias Liero Mathematische Statistik und Wahrscheinlichkeitstheorie Universität Potsdam

More information

ITERATED BOOTSTRAP PREDICTION INTERVALS

ITERATED BOOTSTRAP PREDICTION INTERVALS Statistica Sinica 8(1998), 489-504 ITERATED BOOTSTRAP PREDICTION INTERVALS Majid Mojirsheibani Carleton University Abstract: In this article we study the effects of bootstrap iteration as a method to improve

More information

COMBINING THE LIU-TYPE ESTIMATOR AND THE PRINCIPAL COMPONENT REGRESSION ESTIMATOR

COMBINING THE LIU-TYPE ESTIMATOR AND THE PRINCIPAL COMPONENT REGRESSION ESTIMATOR Noname manuscript No. (will be inserted by the editor) COMBINING THE LIU-TYPE ESTIMATOR AND THE PRINCIPAL COMPONENT REGRESSION ESTIMATOR Deniz Inan Received: date / Accepted: date Abstract In this study

More information

11. Bootstrap Methods

11. Bootstrap Methods 11. Bootstrap Methods c A. Colin Cameron & Pravin K. Trivedi 2006 These transparencies were prepared in 20043. They can be used as an adjunct to Chapter 11 of our subsequent book Microeconometrics: Methods

More information

Improved Liu Estimators for the Poisson Regression Model

Improved Liu Estimators for the Poisson Regression Model www.ccsenet.org/isp International Journal of Statistics and Probability Vol., No. ; May 202 Improved Liu Estimators for the Poisson Regression Model Kristofer Mansson B. M. Golam Kibria Corresponding author

More information

Bootstrap inference for the finite population total under complex sampling designs

Bootstrap inference for the finite population total under complex sampling designs Bootstrap inference for the finite population total under complex sampling designs Zhonglei Wang (Joint work with Dr. Jae Kwang Kim) Center for Survey Statistics and Methodology Iowa State University Jan.

More information

Statistics - Lecture One. Outline. Charlotte Wickham 1. Basic ideas about estimation

Statistics - Lecture One. Outline. Charlotte Wickham  1. Basic ideas about estimation Statistics - Lecture One Charlotte Wickham wickham@stat.berkeley.edu http://www.stat.berkeley.edu/~wickham/ Outline 1. Basic ideas about estimation 2. Method of Moments 3. Maximum Likelihood 4. Confidence

More information

Constructing Prediction Intervals for Random Forests

Constructing Prediction Intervals for Random Forests Senior Thesis in Mathematics Constructing Prediction Intervals for Random Forests Author: Benjamin Lu Advisor: Dr. Jo Hardin Submitted to Pomona College in Partial Fulfillment of the Degree of Bachelor

More information

Bootstrap, Jackknife and other resampling methods

Bootstrap, Jackknife and other resampling methods Bootstrap, Jackknife and other resampling methods Part III: Parametric Bootstrap Rozenn Dahyot Room 128, Department of Statistics Trinity College Dublin, Ireland dahyot@mee.tcd.ie 2005 R. Dahyot (TCD)

More information

A Non-parametric bootstrap for multilevel models

A Non-parametric bootstrap for multilevel models A Non-parametric bootstrap for multilevel models By James Carpenter London School of Hygiene and ropical Medicine Harvey Goldstein and Jon asbash Institute of Education 1. Introduction Bootstrapping is

More information

Nonparametric Methods II

Nonparametric Methods II Nonparametric Methods II Henry Horng-Shing Lu Institute of Statistics National Chiao Tung University hslu@stat.nctu.edu.tw http://tigpbp.iis.sinica.edu.tw/courses.htm 1 PART 3: Statistical Inference by

More information

Finding a Better Confidence Interval for a Single Regression Changepoint Using Different Bootstrap Confidence Interval Procedures

Finding a Better Confidence Interval for a Single Regression Changepoint Using Different Bootstrap Confidence Interval Procedures Georgia Southern University Digital Commons@Georgia Southern Electronic Theses & Dissertations COGS- Jack N. Averitt College of Graduate Studies Fall 2012 Finding a Better Confidence Interval for a Single

More information

Resampling and the Bootstrap

Resampling and the Bootstrap Resampling and the Bootstrap Axel Benner Biostatistics, German Cancer Research Center INF 280, D-69120 Heidelberg benner@dkfz.de Resampling and the Bootstrap 2 Topics Estimation and Statistical Testing

More information

STAT440/840: Statistical Computing

STAT440/840: Statistical Computing First Prev Next Last STAT440/840: Statistical Computing Paul Marriott pmarriott@math.uwaterloo.ca MC 6096 February 2, 2005 Page 1 of 41 First Prev Next Last Page 2 of 41 Chapter 3: Data resampling: the

More information

A Practitioner s Guide to Cluster-Robust Inference

A Practitioner s Guide to Cluster-Robust Inference A Practitioner s Guide to Cluster-Robust Inference A. C. Cameron and D. L. Miller presented by Federico Curci March 4, 2015 Cameron Miller Cluster Clinic II March 4, 2015 1 / 20 In the previous episode

More information

A better way to bootstrap pairs

A better way to bootstrap pairs A better way to bootstrap pairs Emmanuel Flachaire GREQAM - Université de la Méditerranée CORE - Université Catholique de Louvain April 999 Abstract In this paper we are interested in heteroskedastic regression

More information

Heteroskedasticity-Robust Inference in Finite Samples

Heteroskedasticity-Robust Inference in Finite Samples Heteroskedasticity-Robust Inference in Finite Samples Jerry Hausman and Christopher Palmer Massachusetts Institute of Technology December 011 Abstract Since the advent of heteroskedasticity-robust standard

More information

One-Sample Numerical Data

One-Sample Numerical Data One-Sample Numerical Data quantiles, boxplot, histogram, bootstrap confidence intervals, goodness-of-fit tests University of California, San Diego Instructor: Ery Arias-Castro http://math.ucsd.edu/~eariasca/teaching.html

More information

Bootstrapping Heteroskedasticity Consistent Covariance Matrix Estimator

Bootstrapping Heteroskedasticity Consistent Covariance Matrix Estimator Bootstrapping Heteroskedasticity Consistent Covariance Matrix Estimator by Emmanuel Flachaire Eurequa, University Paris I Panthéon-Sorbonne December 2001 Abstract Recent results of Cribari-Neto and Zarkos

More information

Multicollinearity and A Ridge Parameter Estimation Approach

Multicollinearity and A Ridge Parameter Estimation Approach Journal of Modern Applied Statistical Methods Volume 15 Issue Article 5 11-1-016 Multicollinearity and A Ridge Parameter Estimation Approach Ghadban Khalaf King Khalid University, albadran50@yahoo.com

More information

interval forecasting

interval forecasting Interval Forecasting Based on Chapter 7 of the Time Series Forecasting by Chatfield Econometric Forecasting, January 2008 Outline 1 2 3 4 5 Terminology Interval Forecasts Density Forecast Fan Chart Most

More information

Econometrics Summary Algebraic and Statistical Preliminaries

Econometrics Summary Algebraic and Statistical Preliminaries Econometrics Summary Algebraic and Statistical Preliminaries Elasticity: The point elasticity of Y with respect to L is given by α = ( Y/ L)/(Y/L). The arc elasticity is given by ( Y/ L)/(Y/L), when L

More information

Exact Inference for the Two-Parameter Exponential Distribution Under Type-II Hybrid Censoring

Exact Inference for the Two-Parameter Exponential Distribution Under Type-II Hybrid Censoring Exact Inference for the Two-Parameter Exponential Distribution Under Type-II Hybrid Censoring A. Ganguly, S. Mitra, D. Samanta, D. Kundu,2 Abstract Epstein [9] introduced the Type-I hybrid censoring scheme

More information

A Simulation Comparison Study for Estimating the Process Capability Index C pm with Asymmetric Tolerances

A Simulation Comparison Study for Estimating the Process Capability Index C pm with Asymmetric Tolerances Available online at ijims.ms.tku.edu.tw/list.asp International Journal of Information and Management Sciences 20 (2009), 243-253 A Simulation Comparison Study for Estimating the Process Capability Index

More information

401 Review. 6. Power analysis for one/two-sample hypothesis tests and for correlation analysis.

401 Review. 6. Power analysis for one/two-sample hypothesis tests and for correlation analysis. 401 Review Major topics of the course 1. Univariate analysis 2. Bivariate analysis 3. Simple linear regression 4. Linear algebra 5. Multiple regression analysis Major analysis methods 1. Graphical analysis

More information

The assumptions are needed to give us... valid standard errors valid confidence intervals valid hypothesis tests and p-values

The assumptions are needed to give us... valid standard errors valid confidence intervals valid hypothesis tests and p-values Statistical Consulting Topics The Bootstrap... The bootstrap is a computer-based method for assigning measures of accuracy to statistical estimates. (Efron and Tibshrani, 1998.) What do we do when our

More information

Fall 2017 STAT 532 Homework Peter Hoff. 1. Let P be a probability measure on a collection of sets A.

Fall 2017 STAT 532 Homework Peter Hoff. 1. Let P be a probability measure on a collection of sets A. 1. Let P be a probability measure on a collection of sets A. (a) For each n N, let H n be a set in A such that H n H n+1. Show that P (H n ) monotonically converges to P ( k=1 H k) as n. (b) For each n

More information

Higher-Order Confidence Intervals for Stochastic Programming using Bootstrapping

Higher-Order Confidence Intervals for Stochastic Programming using Bootstrapping Preprint ANL/MCS-P1964-1011 Noname manuscript No. (will be inserted by the editor) Higher-Order Confidence Intervals for Stochastic Programming using Bootstrapping Mihai Anitescu Cosmin G. Petra Received:

More information

Analysis of Type-II Progressively Hybrid Censored Data

Analysis of Type-II Progressively Hybrid Censored Data Analysis of Type-II Progressively Hybrid Censored Data Debasis Kundu & Avijit Joarder Abstract The mixture of Type-I and Type-II censoring schemes, called the hybrid censoring scheme is quite common in

More information

Bootstrapping, Randomization, 2B-PLS

Bootstrapping, Randomization, 2B-PLS Bootstrapping, Randomization, 2B-PLS Statistics, Tests, and Bootstrapping Statistic a measure that summarizes some feature of a set of data (e.g., mean, standard deviation, skew, coefficient of variation,

More information

The exact bootstrap method shown on the example of the mean and variance estimation

The exact bootstrap method shown on the example of the mean and variance estimation Comput Stat (2013) 28:1061 1077 DOI 10.1007/s00180-012-0350-0 ORIGINAL PAPER The exact bootstrap method shown on the example of the mean and variance estimation Joanna Kisielinska Received: 21 May 2011

More information

Jackknife Empirical Likelihood for the Variance in the Linear Regression Model

Jackknife Empirical Likelihood for the Variance in the Linear Regression Model Georgia State University ScholarWorks @ Georgia State University Mathematics Theses Department of Mathematics and Statistics Summer 7-25-2013 Jackknife Empirical Likelihood for the Variance in the Linear

More information

Advanced Statistics II: Non Parametric Tests

Advanced Statistics II: Non Parametric Tests Advanced Statistics II: Non Parametric Tests Aurélien Garivier ParisTech February 27, 2011 Outline Fitting a distribution Rank Tests for the comparison of two samples Two unrelated samples: Mann-Whitney

More information

STAT 830 Non-parametric Inference Basics

STAT 830 Non-parametric Inference Basics STAT 830 Non-parametric Inference Basics Richard Lockhart Simon Fraser University STAT 801=830 Fall 2012 Richard Lockhart (Simon Fraser University)STAT 830 Non-parametric Inference Basics STAT 801=830

More information

36. Multisample U-statistics and jointly distributed U-statistics Lehmann 6.1

36. Multisample U-statistics and jointly distributed U-statistics Lehmann 6.1 36. Multisample U-statistics jointly distributed U-statistics Lehmann 6.1 In this topic, we generalize the idea of U-statistics in two different directions. First, we consider single U-statistics for situations

More information

MA 575 Linear Models: Cedric E. Ginestet, Boston University Bootstrap for Regression Week 9, Lecture 1

MA 575 Linear Models: Cedric E. Ginestet, Boston University Bootstrap for Regression Week 9, Lecture 1 MA 575 Linear Models: Cedric E. Ginestet, Boston University Bootstrap for Regression Week 9, Lecture 1 1 The General Bootstrap This is a computer-intensive resampling algorithm for estimating the empirical

More information

Inferences for the Ratio: Fieller s Interval, Log Ratio, and Large Sample Based Confidence Intervals

Inferences for the Ratio: Fieller s Interval, Log Ratio, and Large Sample Based Confidence Intervals Inferences for the Ratio: Fieller s Interval, Log Ratio, and Large Sample Based Confidence Intervals Michael Sherman Department of Statistics, 3143 TAMU, Texas A&M University, College Station, Texas 77843,

More information

Indian Agricultural Statistics Research Institute, New Delhi, India b Indian Institute of Pulses Research, Kanpur, India

Indian Agricultural Statistics Research Institute, New Delhi, India b Indian Institute of Pulses Research, Kanpur, India This article was downloaded by: [Indian Agricultural Statistics Research Institute] On: 1 February 2011 Access details: Access Details: [subscription number 919046794] Publisher Taylor & Francis Informa

More information

Bootstrap. Director of Center for Astrostatistics. G. Jogesh Babu. Penn State University babu.

Bootstrap. Director of Center for Astrostatistics. G. Jogesh Babu. Penn State University  babu. Bootstrap G. Jogesh Babu Penn State University http://www.stat.psu.edu/ babu Director of Center for Astrostatistics http://astrostatistics.psu.edu Outline 1 Motivation 2 Simple statistical problem 3 Resampling

More information

Analytical Bootstrap Methods for Censored Data

Analytical Bootstrap Methods for Censored Data JOURNAL OF APPLIED MATHEMATICS AND DECISION SCIENCES, 6(2, 129 141 Copyright c 2002, Lawrence Erlbaum Associates, Inc. Analytical Bootstrap Methods for Censored Data ALAN D. HUTSON Division of Biostatistics,

More information

Contents. 1 Review of Residuals. 2 Detecting Outliers. 3 Influential Observations. 4 Multicollinearity and its Effects

Contents. 1 Review of Residuals. 2 Detecting Outliers. 3 Influential Observations. 4 Multicollinearity and its Effects Contents 1 Review of Residuals 2 Detecting Outliers 3 Influential Observations 4 Multicollinearity and its Effects W. Zhou (Colorado State University) STAT 540 July 6th, 2015 1 / 32 Model Diagnostics:

More information

Math 494: Mathematical Statistics

Math 494: Mathematical Statistics Math 494: Mathematical Statistics Instructor: Jimin Ding jmding@wustl.edu Department of Mathematics Washington University in St. Louis Class materials are available on course website (www.math.wustl.edu/

More information

STATS 200: Introduction to Statistical Inference. Lecture 29: Course review

STATS 200: Introduction to Statistical Inference. Lecture 29: Course review STATS 200: Introduction to Statistical Inference Lecture 29: Course review Course review We started in Lecture 1 with a fundamental assumption: Data is a realization of a random process. The goal throughout

More information

Chapter 1 Statistical Inference

Chapter 1 Statistical Inference Chapter 1 Statistical Inference causal inference To infer causality, you need a randomized experiment (or a huge observational study and lots of outside information). inference to populations Generalizations

More information

Supplement to Quantile-Based Nonparametric Inference for First-Price Auctions

Supplement to Quantile-Based Nonparametric Inference for First-Price Auctions Supplement to Quantile-Based Nonparametric Inference for First-Price Auctions Vadim Marmer University of British Columbia Artyom Shneyerov CIRANO, CIREQ, and Concordia University August 30, 2010 Abstract

More information

QED. Queen s Economics Department Working Paper No The Size Distortion of Bootstrap Tests

QED. Queen s Economics Department Working Paper No The Size Distortion of Bootstrap Tests QED Queen s Economics Department Working Paper No. 936 The Size Distortion of Bootstrap Tests Russell Davidson GREQAM, Queen s University James G. MacKinnon Queen s University Department of Economics Queen

More information

Chapter 1 Likelihood-Based Inference and Finite-Sample Corrections: A Brief Overview

Chapter 1 Likelihood-Based Inference and Finite-Sample Corrections: A Brief Overview Chapter 1 Likelihood-Based Inference and Finite-Sample Corrections: A Brief Overview Abstract This chapter introduces the likelihood function and estimation by maximum likelihood. Some important properties

More information

Contents. Acknowledgments. xix

Contents. Acknowledgments. xix Table of Preface Acknowledgments page xv xix 1 Introduction 1 The Role of the Computer in Data Analysis 1 Statistics: Descriptive and Inferential 2 Variables and Constants 3 The Measurement of Variables

More information

Statistical Inference

Statistical Inference Statistical Inference Liu Yang Florida State University October 27, 2016 Liu Yang, Libo Wang (Florida State University) Statistical Inference October 27, 2016 1 / 27 Outline The Bayesian Lasso Trevor Park

More information

Bootstrap Confidence Intervals for the Correlation Coefficient

Bootstrap Confidence Intervals for the Correlation Coefficient Bootstrap Confidence Intervals for the Correlation Coefficient Homer S. White 12 November, 1993 1 Introduction We will discuss the general principles of the bootstrap, and describe two applications of

More information

Outline of GLMs. Definitions

Outline of GLMs. Definitions Outline of GLMs Definitions This is a short outline of GLM details, adapted from the book Nonparametric Regression and Generalized Linear Models, by Green and Silverman. The responses Y i have density

More information

Spatially Smoothed Kernel Density Estimation via Generalized Empirical Likelihood

Spatially Smoothed Kernel Density Estimation via Generalized Empirical Likelihood Spatially Smoothed Kernel Density Estimation via Generalized Empirical Likelihood Kuangyu Wen & Ximing Wu Texas A&M University Info-Metrics Institute Conference: Recent Innovations in Info-Metrics October

More information

Monte Carlo Study on the Successive Difference Replication Method for Non-Linear Statistics

Monte Carlo Study on the Successive Difference Replication Method for Non-Linear Statistics Monte Carlo Study on the Successive Difference Replication Method for Non-Linear Statistics Amang S. Sukasih, Mathematica Policy Research, Inc. Donsig Jang, Mathematica Policy Research, Inc. Amang S. Sukasih,

More information

A note on multiple imputation for general purpose estimation

A note on multiple imputation for general purpose estimation A note on multiple imputation for general purpose estimation Shu Yang Jae Kwang Kim SSC meeting June 16, 2015 Shu Yang, Jae Kwang Kim Multiple Imputation June 16, 2015 1 / 32 Introduction Basic Setup Assume

More information

On Modifications to Linking Variance Estimators in the Fay-Herriot Model that Induce Robustness

On Modifications to Linking Variance Estimators in the Fay-Herriot Model that Induce Robustness Statistics and Applications {ISSN 2452-7395 (online)} Volume 16 No. 1, 2018 (New Series), pp 289-303 On Modifications to Linking Variance Estimators in the Fay-Herriot Model that Induce Robustness Snigdhansu

More information

Regression, Ridge Regression, Lasso

Regression, Ridge Regression, Lasso Regression, Ridge Regression, Lasso Fabio G. Cozman - fgcozman@usp.br October 2, 2018 A general definition Regression studies the relationship between a response variable Y and covariates X 1,..., X n.

More information

A Simulation Study on Confidence Interval Procedures of Some Mean Cumulative Function Estimators

A Simulation Study on Confidence Interval Procedures of Some Mean Cumulative Function Estimators Statistics Preprints Statistics -00 A Simulation Study on Confidence Interval Procedures of Some Mean Cumulative Function Estimators Jianying Zuo Iowa State University, jiyizu@iastate.edu William Q. Meeker

More information

Double Bootstrap Confidence Intervals in the Two Stage DEA approach. Essex Business School University of Essex

Double Bootstrap Confidence Intervals in the Two Stage DEA approach. Essex Business School University of Essex Double Bootstrap Confidence Intervals in the Two Stage DEA approach D.K. Chronopoulos, C. Girardone and J.C. Nankervis Essex Business School University of Essex 1 Determinants of efficiency DEA can be

More information

Small area prediction based on unit level models when the covariate mean is measured with error

Small area prediction based on unit level models when the covariate mean is measured with error Graduate Theses and Dissertations Iowa State University Capstones, Theses and Dissertations 2015 Small area prediction based on unit level models when the covariate mean is measured with error Andreea

More information

IEOR E4703: Monte-Carlo Simulation

IEOR E4703: Monte-Carlo Simulation IEOR E4703: Monte-Carlo Simulation Output Analysis for Monte-Carlo Martin Haugh Department of Industrial Engineering and Operations Research Columbia University Email: martin.b.haugh@gmail.com Output Analysis

More information

Slack and Net Technical Efficiency Measurement: A Bootstrap Approach

Slack and Net Technical Efficiency Measurement: A Bootstrap Approach Slack and Net Technical Efficiency Measurement: A Bootstrap Approach J. Richmond Department of Economics University of Essex Colchester CO4 3SQ U. K. Version 1.0: September 2001 JEL Classification Code:

More information

Contents 1. Contents

Contents 1. Contents Contents 1 Contents 1 One-Sample Methods 3 1.1 Parametric Methods.................... 4 1.1.1 One-sample Z-test (see Chapter 0.3.1)...... 4 1.1.2 One-sample t-test................. 6 1.1.3 Large sample

More information

Preliminaries The bootstrap Bias reduction Hypothesis tests Regression Confidence intervals Time series Final remark. Bootstrap inference

Preliminaries The bootstrap Bias reduction Hypothesis tests Regression Confidence intervals Time series Final remark. Bootstrap inference 1 / 171 Bootstrap inference Francisco Cribari-Neto Departamento de Estatística Universidade Federal de Pernambuco Recife / PE, Brazil email: cribari@gmail.com October 2013 2 / 171 Unpaid advertisement

More information

Relationship between ridge regression estimator and sample size when multicollinearity present among regressors

Relationship between ridge regression estimator and sample size when multicollinearity present among regressors Available online at www.worldscientificnews.com WSN 59 (016) 1-3 EISSN 39-19 elationship between ridge regression estimator and sample size when multicollinearity present among regressors ABSTACT M. C.

More information

Supporting Information for Estimating restricted mean. treatment effects with stacked survival models

Supporting Information for Estimating restricted mean. treatment effects with stacked survival models Supporting Information for Estimating restricted mean treatment effects with stacked survival models Andrew Wey, David Vock, John Connett, and Kyle Rudser Section 1 presents several extensions to the simulation

More information

Bootstrapping Dependent Data in Ecology

Bootstrapping Dependent Data in Ecology Bootstrapping Dependent Data in Ecology Mark L Taper Department of Ecology 310 Lewis Hall Montana State University/ Bozeman Bozeman, MT 59717 < taper@ rapid.msu.montana.edu> Introduction I have two goals

More information

Comparison of Some Improved Estimators for Linear Regression Model under Different Conditions

Comparison of Some Improved Estimators for Linear Regression Model under Different Conditions Florida International University FIU Digital Commons FIU Electronic Theses and Dissertations University Graduate School 3-24-2015 Comparison of Some Improved Estimators for Linear Regression Model under

More information

Preliminaries The bootstrap Bias reduction Hypothesis tests Regression Confidence intervals Time series Final remark. Bootstrap inference

Preliminaries The bootstrap Bias reduction Hypothesis tests Regression Confidence intervals Time series Final remark. Bootstrap inference 1 / 172 Bootstrap inference Francisco Cribari-Neto Departamento de Estatística Universidade Federal de Pernambuco Recife / PE, Brazil email: cribari@gmail.com October 2014 2 / 172 Unpaid advertisement

More information

Econometrics I KS. Module 2: Multivariate Linear Regression. Alexander Ahammer. This version: April 16, 2018

Econometrics I KS. Module 2: Multivariate Linear Regression. Alexander Ahammer. This version: April 16, 2018 Econometrics I KS Module 2: Multivariate Linear Regression Alexander Ahammer Department of Economics Johannes Kepler University of Linz This version: April 16, 2018 Alexander Ahammer (JKU) Module 2: Multivariate

More information

THE EFFECTS OF NONNORMAL DISTRIBUTIONS ON CONFIDENCE INTERVALS AROUND THE STANDARDIZED MEAN DIFFERENCE: BOOTSTRAP AND PARAMETRIC CONFIDENCE INTERVALS

THE EFFECTS OF NONNORMAL DISTRIBUTIONS ON CONFIDENCE INTERVALS AROUND THE STANDARDIZED MEAN DIFFERENCE: BOOTSTRAP AND PARAMETRIC CONFIDENCE INTERVALS EDUCATIONAL AND PSYCHOLOGICAL MEASUREMENT 10.1177/0013164404264850 KELLEY THE EFFECTS OF NONNORMAL DISTRIBUTIONS ON CONFIDENCE INTERVALS AROUND THE STANDARDIZED MEAN DIFFERENCE: BOOTSTRAP AND PARAMETRIC

More information

Multiple Testing of One-Sided Hypotheses: Combining Bonferroni and the Bootstrap

Multiple Testing of One-Sided Hypotheses: Combining Bonferroni and the Bootstrap University of Zurich Department of Economics Working Paper Series ISSN 1664-7041 (print) ISSN 1664-705X (online) Working Paper No. 254 Multiple Testing of One-Sided Hypotheses: Combining Bonferroni and

More information

Integrated likelihoods in survival models for highlystratified

Integrated likelihoods in survival models for highlystratified Working Paper Series, N. 1, January 2014 Integrated likelihoods in survival models for highlystratified censored data Giuliana Cortese Department of Statistical Sciences University of Padua Italy Nicola

More information

Simulating Uniform- and Triangular- Based Double Power Method Distributions

Simulating Uniform- and Triangular- Based Double Power Method Distributions Journal of Statistical and Econometric Methods, vol.6, no.1, 2017, 1-44 ISSN: 1792-6602 (print), 1792-6939 (online) Scienpress Ltd, 2017 Simulating Uniform- and Triangular- Based Double Power Method Distributions

More information

Bahadur representations for bootstrap quantiles 1

Bahadur representations for bootstrap quantiles 1 Bahadur representations for bootstrap quantiles 1 Yijun Zuo Department of Statistics and Probability, Michigan State University East Lansing, MI 48824, USA zuo@msu.edu 1 Research partially supported by

More information

Lifetime prediction and confidence bounds in accelerated degradation testing for lognormal response distributions with an Arrhenius rate relationship

Lifetime prediction and confidence bounds in accelerated degradation testing for lognormal response distributions with an Arrhenius rate relationship Scholars' Mine Doctoral Dissertations Student Research & Creative Works Spring 01 Lifetime prediction and confidence bounds in accelerated degradation testing for lognormal response distributions with

More information

Inference via Kernel Smoothing of Bootstrap P Values

Inference via Kernel Smoothing of Bootstrap P Values Queen s Economics Department Working Paper No. 1054 Inference via Kernel Smoothing of Bootstrap P Values Jeff Racine McMaster University James G. MacKinnon Queen s University Department of Economics Queen

More information

The Bootstrap Small Sample Properties. F.W. Scholz Boeing Computer Services Research and Technology

The Bootstrap Small Sample Properties. F.W. Scholz Boeing Computer Services Research and Technology The Bootstrap Small Sample Properties F.W. Scholz Boeing Computer Services Research and Technology March 15, 1993 Abstract This report reviews several bootstrap methods with special emphasis on small sample

More information

Least Absolute Value vs. Least Squares Estimation and Inference Procedures in Regression Models with Asymmetric Error Distributions

Least Absolute Value vs. Least Squares Estimation and Inference Procedures in Regression Models with Asymmetric Error Distributions Journal of Modern Applied Statistical Methods Volume 8 Issue 1 Article 13 5-1-2009 Least Absolute Value vs. Least Squares Estimation and Inference Procedures in Regression Models with Asymmetric Error

More information

Bootstrap & Confidence/Prediction intervals

Bootstrap & Confidence/Prediction intervals Bootstrap & Confidence/Prediction intervals Olivier Roustant Mines Saint-Étienne 2017/11 Olivier Roustant (EMSE) Bootstrap & Confidence/Prediction intervals 2017/11 1 / 9 Framework Consider a model with

More information

A Resampling Method on Pivotal Estimating Functions

A Resampling Method on Pivotal Estimating Functions A Resampling Method on Pivotal Estimating Functions Kun Nie Biostat 277,Winter 2004 March 17, 2004 Outline Introduction A General Resampling Method Examples - Quantile Regression -Rank Regression -Simulation

More information

Bootstrap, Jackknife and other resampling methods

Bootstrap, Jackknife and other resampling methods Bootstrap, Jackknife and other resampling methods Part VI: Cross-validation Rozenn Dahyot Room 128, Department of Statistics Trinity College Dublin, Ireland dahyot@mee.tcd.ie 2005 R. Dahyot (TCD) 453 Modern

More information

University of California San Diego and Stanford University and

University of California San Diego and Stanford University and First International Workshop on Functional and Operatorial Statistics. Toulouse, June 19-21, 2008 K-sample Subsampling Dimitris N. olitis andjoseph.romano University of California San Diego and Stanford

More information

STATISTICAL INFERENCE IN ACCELERATED LIFE TESTING WITH GEOMETRIC PROCESS MODEL. A Thesis. Presented to the. Faculty of. San Diego State University

STATISTICAL INFERENCE IN ACCELERATED LIFE TESTING WITH GEOMETRIC PROCESS MODEL. A Thesis. Presented to the. Faculty of. San Diego State University STATISTICAL INFERENCE IN ACCELERATED LIFE TESTING WITH GEOMETRIC PROCESS MODEL A Thesis Presented to the Faculty of San Diego State University In Partial Fulfillment of the Requirements for the Degree

More information

Technical Report No. 1/12, February 2012 JACKKNIFING THE RIDGE REGRESSION ESTIMATOR: A REVISIT Mansi Khurana, Yogendra P. Chaubey and Shalini Chandra

Technical Report No. 1/12, February 2012 JACKKNIFING THE RIDGE REGRESSION ESTIMATOR: A REVISIT Mansi Khurana, Yogendra P. Chaubey and Shalini Chandra Technical Report No. 1/12, February 2012 JACKKNIFING THE RIDGE REGRESSION ESTIMATOR: A REVISIT Mansi Khurana, Yogendra P. Chaubey and Shalini Chandra Jackknifing the Ridge Regression Estimator: A Revisit

More information