DOCSLIB.ORG

Hypothesis Testing I

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Hypothesis Testing I Introduction Wald test p-values t-test Quiz Hypothesis testing I. Botond Szabo Leiden University Leiden, 26 March 2018 Introduction Wald test p-values t-test Quiz Outline 1 Introduction 2 Wald test 3 p-values 4 t-test 5 Quiz Introduction Wald test p-values t-test Quiz Example Suppose we want to know if exposure to asbestos is associated with lung disease. We take some rats and randomly divide them into two groups. We expose one group to asbestos and then compare the disease rate in two groups. We consider the following two hypotheses: 1 The null hypothesis: the disease rate is the same in both groups. 2 Alternative hypothesis: The disease rate is not the same in two groups. If the exposed group has a much higher disease rate, we will reject the null hypothesis and conclude that the evidence favours the alternative hypothesis. Introduction Wald test p-values t-test Quiz Null and alternative hypotheses We partition the parameter space Θ into sets Θ0 and Θ1 and wish to test H0 : θ 2 Θ0 versus H1 : θ 2 Θ1: We call H0 the null hypothesis and H1 the alternative hypothesis. In our previous example θ is the difference between the disease rates in the exposed and not exposed groups, Θ0 = f0g; and Θ1 = (0; 1): Introduction Wald test p-values t-test Quiz Rejection region Let T be a random variable and let T be the range of X : We test the hypothesis by finding a subset of outcomes R ⊂ X ; called the rejection region. If T 2 R; we reject the null hypothesis, otherwise we retain it: T 2 R ) reject H0 T 2= R ) retain H0: In our previous example T is an estimate of the difference between the disease rates in the exposed and not exposed groups. We reject the null hypothesis if T is large. Introduction Wald test p-values t-test Quiz Test statistic and critical value Usually, the rejection region R is of the form R = fx : T (x) > cg; where T is a test statistic and c is critical value. The problem in hypothesis testing is to find an appropriate test statistic T and an appropriate critical value c: Introduction Wald test p-values t-test Quiz Warning There is a tendency to use hypothesis testing even when they are not appropriate. Point estimates and confidence intervals are at times a better option. Use testing only with well-defined hypotheses and in general take care of formulating them: things can go wrong when testing ill-defined or overly complex hypotheses. Introduction Wald test p-values t-test Quiz Type I and type II errors By rejecting H0 when H0 is true we commit type I error. By retaining H0 when H1 is true we commit type II error. Retain Null Reject Null H0 true X type I error H1 true type II error X Introduction Wald test p-values t-test Quiz Power function Definition The power function of a test T with rejection region R is defined by β(θ) = Pθ(T 2 R): The size of a test is defined to be α = sup β(θ): θ2Θ0 A test is said to be of level α; if its size is less or equal to α: Introduction Wald test p-values t-test Quiz Comments Let Θ0 = fθ0g and Θ1 = fθ1g: In this case β(θ0) is the probability of type I error. β(θ1) gives 1 minus the probability of type II error. In the ideal case we would like the probabilities of both types of errors be equal to zero. Except trivial cases, this is not possible. So we concentrate on tests of size or level α; where α is taken to be small, e.g. α = 0:05: Introduction Wald test p-values t-test Quiz Hypotheses and tests A hypothesis of the form θ = θ0 is called a simple hypothesis. A hypothesis of the form θ > θ0 or θ < θ0 (or both) is called a composite hypothesis. A test of the form H0 : θ = θ0 versus H1 : θ 6= θ0 is called a two-sided test. A test of the form H0 : θ ≤ θ0 versus H1 : θ > θ0 or H0 : θ ≥ θ0 versus H1 : θ < θ0 is called a one-sided test. Introduction Wald test p-values t-test Quiz Example Example Let X1;:::; Xn ∼ N(µ, σ); with σ known. We want to test H0 : µ ≤ 0 versus µ > 0: Hence Θ0 = (−∞; 0] and Θ1 = (0; 1): Let T = X n and conisder the test reject H0 if T > c: The rejection region is R = f(x1;:::; xn): T (x1;:::; xn) > cg : Introduction Wald test p-values t-test Quiz Example (continued) Example The power function is β(µ) = Pµ(X n > c) p p n(X − µ) n(c − µ) = n > Pµ σ σ p n(c − µ) = Z > P σ p n(c − µ) = 1 − Φ : σ The power function is increasing in µ. Introduction Wald test p-values t-test Quiz Example (continued) Example Hence p nc size = sup β(µ) = β(0) = 1 − Φ : µ≤0 σ For size α test we set this equal to α and solve for c to get σΦ−1(1 − α) c = p : n We reject the null hypothesis when X n > c: Equivalently, when p nX n > z = Φ−1(1 − α): σ α Introduction Wald test p-values t-test Quiz Example 2 Example Let X be an observation from a Uniform(0; θ) distribution. Suppose one wants to test H0 : θ = 2 versus H1 : θ 6= 2. Suppose one rejects H0 if X ≥ 1:9. Compute the probability of type 1 error. Introduction Wald test p-values t-test Quiz Most powerful test It would be desirable to find the test with highest power under H1 among all size α tests. Such a test, if it exists, is called the most powerful test. Finding such tests is not easy, and often they don't even exist. We shall instead consider several widely used tests. Introduction Wald test p-values t-test Quiz Wald test Definition Let θ be a scalar parameter and θ^ an estimate of θ with ^se its estimated standard error. Consider the test H0 : θ = θ0 versus H1 : θ 6= θ0: Assume that θ^ is asymptotically normal: θ^ − θ 0 N(0; 1): ^se The size α Wald test is: reject H0 when jW j > zα/2; where θ^ − θ W = 0 : ^se Introduction Wald test p-values t-test Quiz Size of the Wald test Theorem Asymptotically, the Wald test has size α; i.e. Pθ0 (jW j > zα/2) ! α as n ! 1: Proof. We have ! jθ^ − θ j (jW j > z ) = 0 > z Pθ0 α/2 Pθ0 ^se α/2 ! P(jZj > zα/2) = α: Introduction Wald test p-values t-test Quiz Remark An alternative version of the Wald test statistic is θ^ − θ W = 0 ; se0 where se0 is the standard error of θ^ computed at θ0: Both versions of the test are valid. Introduction Wald test p-values t-test Quiz Power of the Wald test Theorem ∗ ∗ Suppose the true value of θ is θ 6= θ0: The power β(θ ) (the probability of correctly rejecting the null hypothesis) is approximately given by ! ! θ^ − θ∗ θ^ − θ∗ 1 − Φ + z + Φ − z : ^se α/2 ^se α/2 Introduction Wald test p-values t-test Quiz Example Example Let X1;:::; Xm and Y1;:::; Yn be two independent samples from distributions with means µ1 and µ2; respectively. Let us test the null hypothesis that µ1 = µ2: Write this as H0 : δ = 0 versus H1 : δ 6= 0; where δ = µ1 − µ2: The plug-in estimate of δ is δ^ = X m − Y n with q 2 2 s1 s2 2 2 estimated standard error ^se= m + n ; where s1 ; s2 are sample variances. The size α Wald test rejects H0 when jW j > zα/2; where δ^ − 0 X − Y W = = m n : q 2 2 ^se s1 s2 m + n Introduction Wald test p-values t-test Quiz Example 2 Example Let us assume that we flip a coin 100 times and 25 out of the 100 we got head. Do we have strong enough evidence to say that the coin is not fair? Introduction Wald test p-values t-test Quiz Relation to confidence intervals Theorem The size α Wald test rejects H0 : θ = θ0 versus H1 : θ 6= θ0 if and only if θ0 2= C; where ^ ^ C = (θ − ^sezα/2; θ + ^sezα/2): In words, testing the hypothesis with Wald test isequivalent to checking whether the null value θ0 is in the confidence interval. Introduction Wald test p-values t-test Quiz Warning When we reject H0; we say the result is statistically significant. The result may be statistically significant without the effect size being large. In such a case we have a result that is statistically significant, but perhaps not scientifically or practically significant. Any confidence interval that excludes θ0 corresponds to rejecting θ0: But θ0 can be close to an endpoint of the confidence interval (not scientifically significant), or far away (scientifically significant). Introduction Wald test p-values t-test Quiz Heuristics Consider a hypothesis test with test statistic T and the rejection region Rα at level α: Fix α: Upon observing the data X we evaluate the statistic T (X ) and ask whether the test rejects the null hypothesis at level α: If yes, we gradually decrease α; while asking the same question. Eventually we arrive at α at which the test does not reject the null hypothesis.
Load more

Recommended publications
  • Three Statistical Testing Procedures in Logistic Regression: Their Performance in Differential Item Functioning (DIF) Investigation
    Research Report Three Statistical Testing Procedures in Logistic Regression: Their Performance in Differential Item Functioning (DIF) Investigation Insu Paek December 2009 ETS RR-09-35 Listening. Learning. Leading.® Three Statistical Testing Procedures in Logistic Regression: Their Performance in Differential Item Functioning (DIF) Investigation Insu Paek ETS, Princeton, New Jersey December 2009 As part of its nonprofit mission, ETS conducts and disseminates the results of research to advance quality and equity in education and assessment for the benefit of ETS’s constituents and the field. To obtain a PDF or a print copy of a report, please visit: http://www.ets.org/research/contact.html Copyright © 2009 by Educational Testing Service. All rights reserved. ETS, the ETS logo, GRE, and LISTENING. LEARNING. LEADING. are registered trademarks of Educational Testing Service (ETS). SAT is a registered trademark of the College Board. Abstract Three statistical testing procedures well-known in the maximum likelihood approach are the Wald, likelihood ratio (LR), and score tests. Although well-known, the application of these three testing procedures in the logistic regression method to investigate differential item function (DIF) has not been rigorously made yet. Employing a variety of simulation conditions, this research (a) assessed the three tests’ performance for DIF detection and (b) compared DIF detection in different DIF testing modes (targeted vs. general DIF testing). Simulation results showed small differences between the three tests and different testing modes. However, targeted DIF testing consistently performed better than general DIF testing; the three tests differed more in performance in general DIF testing and nonuniform DIF conditions than in targeted DIF testing and uniform DIF conditions; and the LR and score tests consistently performed better than the Wald test.
    [Show full text]
  • Wald (And Score) Tests
    Wald (and Score) Tests 1 / 18 Vector of MLEs is Asymptotically Normal That is, Multivariate Normal This yields I Confidence intervals I Z-tests of H0 : θj = θ0 I Wald tests I Score Tests I Indirectly, the Likelihood Ratio tests 2 / 18 Under Regularity Conditions (Thank you, Mr. Wald) a.s. I θbn → θ √ d −1 I n(θbn − θ) → T ∼ Nk 0, I(θ) 1 −1 I So we say that θbn is asymptotically Nk θ, n I(θ) . I I(θ) is the Fisher Information in one observation. I A k × k matrix ∂2 I(θ) = E[− log f(Y ; θ)] ∂θi∂θj I The Fisher Information in the whole sample is nI(θ) 3 / 18 H0 : Cθ = h Suppose θ = (θ1, . θ7), and the null hypothesis is I θ1 = θ2 I θ6 = θ7 1 1 I 3 (θ1 + θ2 + θ3) = 3 (θ4 + θ5 + θ6) We can write null hypothesis in matrix form as θ1 θ2 1 −1 0 0 0 0 0 θ3 0 0 0 0 0 0 1 −1 θ4 = 0 1 1 1 −1 −1 −1 0 θ5 0 θ6 θ7 4 / 18 p Suppose H0 : Cθ = h is True, and Id(θ)n → I(θ) By Slutsky 6a (Continuous mapping), √ √ d −1 0 n(Cθbn − Cθ) = n(Cθbn − h) → CT ∼ Nk 0, CI(θ) C and −1 p −1 Id(θ)n → I(θ) . Then by Slutsky’s (6c) Stack Theorem, √ ! n(Cθbn − h) d CT → . −1 I(θ)−1 Id(θ)n Finally, by Slutsky 6a again, −1 0 0 −1 Wn = n(Cθb − h) (CId(θ)n C ) (Cθb − h) →d W = (CT − 0)0(CI(θ)−1C0)−1(CT − 0) ∼ χ2(r) 5 / 18 The Wald Test Statistic −1 0 0 −1 Wn = n(Cθbn − h) (CId(θ)n C ) (Cθbn − h) I Again, null hypothesis is H0 : Cθ = h I Matrix C is r × k, r ≤ k, rank r I All we need is a consistent estimator of I(θ) I I(θb) would do I But it’s inconvenient I Need to compute partial derivatives and expected values in ∂2 I(θ) = E[− log f(Y ; θ)] ∂θi∂θj 6 / 18 Observed Fisher Information I To find θbn, minimize the minus log likelihood.
    [Show full text]
  • Comparison of Wald, Score, and Likelihood Ratio Tests for Response Adaptive Designs
    Journal of Statistical Theory and Applications Volume 10, Number 4, 2011, pp. 553-569 ISSN 1538-7887 Comparison of Wald, Score, and Likelihood Ratio Tests for Response Adaptive Designs Yanqing Yi1∗and Xikui Wang2 1 Division of Community Health and Humanities, Faculty of Medicine, Memorial University of Newfoundland, St. Johns, Newfoundland, Canada A1B 3V6 2 Department of Statistics, University of Manitoba, Winnipeg, Manitoba, Canada R3T 2N2 Abstract Data collected from response adaptive designs are dependent. Traditional statistical methods need to be justified for the use in response adaptive designs. This paper gener- alizes the Rao's score test to response adaptive designs and introduces a generalized score statistic. Simulation is conducted to compare the statistical powers of the Wald, the score, the generalized score and the likelihood ratio statistics. The overall statistical power of the Wald statistic is better than the score, the generalized score and the likelihood ratio statistics for small to medium sample sizes. The score statistic does not show good sample properties for adaptive designs and the generalized score statistic is better than the score statistic under the adaptive designs considered. When the sample size becomes large, the statistical power is similar for the Wald, the sore, the generalized score and the likelihood ratio test statistics. MSC: 62L05, 62F03 Keywords and Phrases: Response adaptive design, likelihood ratio test, maximum likelihood estimation, Rao's score test, statistical power, the Wald test ∗Corresponding author. Fax: 1-709-777-7382. E-mail addresses: [email protected] (Yanqing Yi), xikui [email protected] (Xikui Wang) Y. Yi and X.
    [Show full text]
  • Statistical Asymptotics Part II: First-Order Theory
    First-Order Asymptotic Theory Statistical Asymptotics Part II: First-Order Theory Andrew Wood School of Mathematical Sciences University of Nottingham APTS, April 15-19 2013 Andrew Wood Statistical Asymptotics Part II: First-Order Theory First-Order Asymptotic Theory Structure of the Chapter This chapter covers asymptotic normality and related results. Topics: MLEs, log-likelihood ratio statistics and their asymptotic distributions; M-estimators and their first-order asymptotic theory. Initially we focus on the case of the MLE of a scalar parameter θ. Then we study the case of the MLE of a vector θ, first without and then with nuisance parameters. Finally, we consider the more general setting of M-estimators. Andrew Wood Statistical Asymptotics Part II: First-Order Theory First-Order Asymptotic Theory Motivation Statistical inference typically requires approximations because exact answers are usually not available. Asymptotic theory provides useful approximations to densities or distribution functions. These approximations are based on results from probability theory. The theory underlying these approximation techniques is valid as some quantity, typically the sample size n [or more generally some measure of information], goes to infinity, but the approximations obtained are often accurate even for small sample sizes. Andrew Wood Statistical Asymptotics Part II: First-Order Theory First-Order Asymptotic Theory Test statistics Consider testing the null hypothesis H0 : θ = θ0, where θ0 is an arbitrary specified point in Ωθ. If desired, we may
    [Show full text]
  • Econometrics-I-11.Pdf
    Econometrics I Professor William Greene Stern School of Business Department of Economics 11-1/78 Part 11: Hypothesis Testing - 2 Econometrics I Part 11 – Hypothesis Testing 11-2/78 Part 11: Hypothesis Testing - 2 Classical Hypothesis Testing We are interested in using the linear regression to support or cast doubt on the validity of a theory about the real world counterpart to our statistical model. The model is used to test hypotheses about the underlying data generating process. 11-3/78 Part 11: Hypothesis Testing - 2 Types of Tests Nested Models: Restriction on the parameters of a particular model y = 1 + 2x + 3T + , 3 = 0 (The “treatment” works; 3 0 .) Nonnested models: E.g., different RHS variables yt = 1 + 2xt + 3xt-1 + t yt = 1 + 2xt + 3yt-1 + wt (Lagged effects occur immediately or spread over time.) Specification tests: ~ N[0,2] vs. some other distribution (The “null” spec. is true or some other spec. is true.) 11-4/78 Part 11: Hypothesis Testing - 2 Hypothesis Testing Nested vs. nonnested specifications y=b1x+e vs. y=b1x+b2z+e: Nested y=bx+e vs. y=cz+u: Not nested y=bx+e vs. logy=clogx: Not nested y=bx+e; e ~ Normal vs. e ~ t[.]: Not nested Fixed vs. random effects: Not nested Logit vs. probit: Not nested x is (not) endogenous: Maybe nested. We’ll see … Parametric restrictions Linear: R-q = 0, R is JxK, J < K, full row rank General: r(,q) = 0, r = a vector of J functions, R(,q) = r(,q)/’. Use r(,q)=0 for linear and nonlinear cases 11-5/78 Part 11: Hypothesis Testing - 2 Broad Approaches Bayesian: Does not reach a firm conclusion.
    [Show full text]
  • Package 'Lmtest'
    Package ‘lmtest’ September 9, 2020 Title Testing Linear Regression Models Version 0.9-38 Date 2020-09-09 Description A collection of tests, data sets, and examples for diagnostic checking in linear regression models. Furthermore, some generic tools for inference in parametric models are provided. LazyData yes Depends R (>= 3.0.0), stats, zoo Suggests car, strucchange, sandwich, dynlm, stats4, survival, AER Imports graphics License GPL-2 | GPL-3 NeedsCompilation yes Author Torsten Hothorn [aut] (<https://orcid.org/0000-0001-8301-0471>), Achim Zeileis [aut, cre] (<https://orcid.org/0000-0003-0918-3766>), Richard W. Farebrother [aut] (pan.f), Clint Cummins [aut] (pan.f), Giovanni Millo [ctb], David Mitchell [ctb] Maintainer Achim Zeileis <[email protected]> Repository CRAN Date/Publication 2020-09-09 05:30:11 UTC R topics documented: bgtest . .2 bondyield . .4 bptest . .6 ChickEgg . .8 coeftest . .9 coxtest . 11 currencysubstitution . 13 1 2 bgtest dwtest . 14 encomptest . 16 ftemp . 18 gqtest . 19 grangertest . 21 growthofmoney . 22 harvtest . 23 hmctest . 25 jocci . 26 jtest . 28 lrtest . 29 Mandible . 31 moneydemand . 31 petest . 33 raintest . 35 resettest . 36 unemployment . 38 USDistLag . 39 valueofstocks . 40 wages ............................................ 41 waldtest . 42 Index 46 bgtest Breusch-Godfrey Test Description bgtest performs the Breusch-Godfrey test for higher-order serial correlation. Usage bgtest(formula, order = 1, order.by = NULL, type = c("Chisq", "F"), data = list(), fill = 0) Arguments formula a symbolic description for the model to be tested (or a fitted "lm" object). order integer. maximal order of serial correlation to be tested. order.by Either a vector z or a formula with a single explanatory variable like ~ z.
    [Show full text]
  • Wald (And Score) Tests
    Wald (and score) tests • MLEs have an approximate multivariate normal sampling distribution for large samples (Thanks Mr. Wald.) • Approximate mean vector = vector of true parameter values for large samples • Asymptotic variance-covariance matrix is easy to estimate • H0: Cθ = h (Linear hypothesis) H0 : Cθ = h Cθ h is multivariate normal as n − → ∞ Leads! to a straightforward chisquare test • Called a Wald test • Based on the full model (unrestricted by any null hypothesis) • Asymptotically equivalent to the LR test • Not as good as LR for smaller samples • Very convenient sometimes Example of H0: Cθ=h Suppose θ = (θ1, . θ7), with 1 1 H : θ = θ , θ = θ , (θ + θ + θ ) = (θ + θ + θ ) 0 1 2 6 7 3 1 2 3 3 4 5 6 θ1 θ2 1 1 0 0 0 0 0 θ 0 − 3 0 0 0 0 0 1 1 θ = 0 − 4 1 1 1 1 1 1 0 θ 0 − − − 5 θ6 θ 7 Multivariate Normal Univariate Normal • 1 1 (x µ)2 – f(x) = exp − σ√2π − 2 σ2 (x µ)2 ! " – −σ2 is the squared Euclidian distance between x and µ, in a space that is stretched by σ2. 2 (X µ) 2 – − χ (1) σ2 ∼ Multivariate Normal • 1 1 1 – f(x) = 1 k exp 2 (x µ)#Σ− (x µ) Σ 2 (2π) 2 − − − | | 1 # $ – (x µ)#Σ− (x µ) is the squared Euclidian distance between x and µ, −in a space that− is warped and stretched by Σ. 1 2 – (X µ)#Σ− (X µ) χ (k) − − ∼ Distance: Suppose Σ = I2 2 1 d = (X µ)!Σ− (X µ) − − 1 0 x1 µ1 = x1 µ1, x2 µ2 − − − 0 1 x2 µ2 # $ # − $ ! " x1 µ1 = x1 µ1, x2 µ2 − − − x2 µ2 # − $ = !(x µ )2 + (x µ ")2 1 − 1 2 − 2 2 2 d = (x1 µ1) + (x2 µ2) − − (x1, x2) % x µ 2 − 2 (µ1, µ2) x µ 1 − 1 86 APPENDIX A.
    [Show full text]
  • Week 4: Simple Linear Regression II
    Week 4: Simple Linear Regression II Marcelo Coca Perraillon University of Colorado Anschutz Medical Campus Health Services Research Methods I HSMP 7607 2019 These slides are part of a forthcoming book to be published by Cambridge University Press. For more information, go to perraillon.com/PLH. c This material is copyrighted. Please see the entire copyright notice on the book's website. Updated notes are here: https://clas.ucdenver.edu/marcelo-perraillon/ teaching/health-services-research-methods-i-hsmp-7607 1 Outline Algebraic properties of OLS Reminder on hypothesis testing The Wald test Examples Another way at looking at causal inference 2 Big picture We used the method of least squares to find the line that minimizes the sum of square errors (SSE) We made NO assumptions about the distribution of or Y We saw that the mean of the predicted values is the same as the mean of the observed values and that implies that predictions regress towards the mean Today, we will assume that distributes N(0; σ2) and are independent (iid). We need that assumption for inference (not to find the best βj ) 3 Algebraic properties of OLS Pn 1) The sum of the residuals is zero: i=1 ^i = 0. One implication of this is that if the residuals add up to zero, then we will always get the mean of Y right. Confusion alert: is the error term; ^ is the residual (more on this later) Recall the first order conditions: @SSE = Pn (y − β^ − β^ x ) = 0 @β0 i=1 i 0 1 i @SSE = Pn x (y − β^ − β^ x ) = 0 @β1 i=1 i i 0 1 i From the first one, it's obvious that we choose β^0; β^1 that satisfy this Pn ^ ^ Pn ¯ property: i=1(yi − β0 − β1xi ) = i=1 ^i = 0, so ^ = 0 In words: On average, the residuals or predicted errors are zero, so on average the predicted outcomey ^ is the same as the observed y.
    [Show full text]
  • A Warning About Wald Tests David A
    Paper 5088 -2020 A Warning about Wald Tests David A. Dickey, NC State University ABSTRACT In mixed models, tests for variance components can be of interest. In some instances, several tests are available. For example, in standard balanced experiments like blocked designs, split plots and other nested designs, and random effect factorials, an F test for variance components is available along with the Wald test, Wald being a test based on large sample theory. In some cases, only the Wald test is available, so it is the default output when the COVTEST option is invoked in analyzing mixed models. However, one must be very careful in using this test. Because the Wald test is a large-sample test, it is important to know exactly what is meant by large sample. Does a field trial with 4 blocks and 80 entries (genetic crosses) in each block satisfy the "large sample" criterion? The answer is no, because, for testing the random block effects, it is the number of blocks (4) that needs to be large, not the overall sample size (320). Surprisingly it is not even possible to find significant block effects with the Wald test in this example, no matter how large the true block variance is. This problem is not shared by the F test when it is available as it is in this example. A careful look at the relationship between the F test and the Wald test is shown in this paper, through which the details of the above phenomenon is made clear. The exact nature of this problem, while important to practitioners is apparently not well known.
    [Show full text]
  • Fundamentals of Data Science the F Test
    Fundamentals of Data Science The F test Ramesh Johari 1 / 12 The recipe 2 / 12 The hypothesis testing recipe In this lecture we repeatedly apply the following approach. I If the true parameter was θ0, then the test statistic T (Y) should look like it would when the data comes from f(Y jθ0). I We compare the observed test statistic Tobs to the sampling distribution under θ0. I If the observed Tobs is unlikely under the sampling distribution given θ0, we reject the null hypothesis that θ = θ0. The theory of hypothesis testing relies on finding test statistics T (Y) for which this procedure yields as high a power as possible, given a particular size. 3 / 12 The F test in linear regression 4 / 12 What if we want to know if multiple regression coefficients are zero? This is equivalent to asking: would a simpler model suffice? For this purpose we use the F test. Multiple regression coefficients We again assume we are in the linear normal model: 2 Yi = Xiβ + "i with i.i.d. N (0; σ ) errors "i. The Wald (or t) test lets us test whether one regression coefficient is zero. 5 / 12 Multiple regression coefficients We again assume we are in the linear normal model: 2 Yi = Xiβ + "i with i.i.d. N (0; σ ) errors "i. The Wald (or t) test lets us test whether one regression coefficient is zero. What if we want to know if multiple regression coefficients are zero? This is equivalent to asking: would a simpler model suffice? For this purpose we use the F test.
    [Show full text]
  • Section 5 Inference in the Multiple-Regression Model
    Section 5 Inference in the Multiple-Regression Model Kinds of hypothesis tests in a multiple regression There are several distinct kinds of hypothesis tests we can run in a multiple regression. Suppose that among the regressors in a Reed Econ 201 grade regression are variables for SAT-math and SAT-verbal: giiii=β01 +βSATM +β 2 SATV +… + u • We might want to know if math SAT matters for Econ 201: H 01:0β = o Would it make sense for this to be a one-tailed or a two-tailed test? o Is it plausible that a higher math SAT would lower Econ 201 performance? o Probably a one-tailed test makes more sense. • We might want to know if either SAT matters. This is a joint test of two simultaneous hypotheses: H 01:0,0.β= β= 2 o The alternative hypothesis is that one or both parts of the null hypothesis fails to hold. If β1 = 0 but β2 ≠ 0, then the null is false and we want to reject it. o The joint test is not the same as separate individual tests on the two coefficients. In general, the two variables are correlated, which means that their coefficient estimators are correlated. That means that eliminating one of the variables from the equation affects the significance of the other. The joint test tests whether we can delete both variables at once, rather than testing whether we can delete one variable given that the other is in (or out of) the equation. A common example is the situation where the two variables are highly and positively correlated (imperfect but high multicollinearity).
    [Show full text]
  • Chapter 1 Introduction
    Outline Chapter 1 Introduction • Variable types • Review of binomial and multinomial distributions • Likelihood and maximum likelihood method Yibi Huang • Inference for a binomial proportion (Wald, score and likelihood Department of Statistics ratio tests and confidence intervals) University of Chicago • Sample sample inference 1 Regression methods are used to analyze data when the response variable is numerical • e.g., temperature, blood pressure, heights, speeds, income • Stat 22200, Stat 22400 Methods in categorical data analysis are used when the response variable takes categorical (or qualitative) values Variable Types • e.g., • gender (male, female), • political philosophy (liberal, moderate, conservative), • region (metropolitan, urban, suburban, rural) • Stat 22600 In either case, the explanatory variables can be numerical or categorical. 2 Two Types of Categorical Variables Nominal : unordered categories, e.g., • transport to work (car, bus, bicycle, walk, other) • favorite music (rock, hiphop, pop, classical, jazz, country, folk) Review of Binomial and Ordinal : ordered categories Multinomial Distributions • patient condition (excellent, good, fair, poor) • government spending (too high, about right, too low) We pay special attention to — binary variables: success or failure for which nominal-ordinal distinction is unimportant. 3 Binomial Distributions (Review) Example If n Bernoulli trials are performed: Vote (Dem, Rep). Suppose π = Pr(Dem) = 0:4: • only two possible outcomes for each trial (success, failure) Sample n = 3 voters,
    [Show full text]