Questions and Answers of Linear State Space Systems

Reanalyze the auto accident data of Example 4.8.1 without treating any of the factors as a response factor.
Reanalyze the abortion attitude data of Exercise 4.8.4 without treating any of the factors as a response.
Using our discussion of graphical models and collapsibility in Chapter 5, argue that when the true model is [123][24][456],the test of marginal association for the u123(ijk)’s does not ignore any
Reanalyze the Intelligence versus Clothing table of Exercise 2.6.3 using the methods of Section 1. Note that this ignores the potentially complicating factor Standard and the complex sampling scheme.
For the model log(mij ) = u + u1(i) + u2(j) + γ1 i j + γ2 i 2j , find the log odds ratio log[mijmi+1 j+1mi+1 jmi j+1]in terms of the model parameters.
Duncan, Schuman, and Duncan (1973) and Duncan and McRae (1979) present data on evaluations made in 1959 and 1971 of the performance of radio and TV networks. The data are given in Table 7.5.Use the
Assuming the use of consecutive integer scores, find the log odds ratios log[mijkmi+1 j+1 kmi+1 jkmi j+1 k]TABLE 7.5. Radio and Television Network Performance.Respondent’s Performance of Networks
Use the methods of Section 3 to analyze the Intelligence– Clothing table of Exercise 2.6.3.
Brown (1980) presents data that are reproduced in Table 8.5, on a cross-classification of 53 prostate cancer patients. The factors are acid phosphatase level in the blood serum, age, stage, grade,
Extend your analysis of the Berkeley graduate admissions data (cf. Exercise 3.8.4) by incorporating the method of partitioning polytomous factors from Example 8.1.2.
Partitioning Two-Way Tables.Lancaster (1949) and Irwin (1949) present a method of partitioning tables that was used in Exercise 2.7.4.We now establish the validity of this method. Consider a
The Bradley-Terry Model.“Let’s suppose, for a moment, that you have just been married and that you are given a choice of having, during your entire lifetime, either x or y children. Which would
Show that for the binomial model of Section 1, r(θ) =log(1 + eθ) and that ˙r(θ) = p and ¨r(θ) = p(1 − p).
For the gamma model of Section 1, use the definition of θ and r(θ) to show that the mean and variance are as given.
Show that the likelihood equations (9.2.2) have the form Qj (β)/φ = 0, j = 1,...,p.
Show that if f(yi|θ, φ; w) from (9.1.1) is the common density of independent observations yi, i = 1,...,n, then n i=1 yi has a density f(yi|θ∗, φ∗, w∗) for some θ∗, φ∗, and w∗.
Let Y = (y1,...yn). Show that a generalized linear model with canonical link has XY as a sufficient statistic.
Using the definitions of this chapter, find the Pearson statistic (defined in Section 3) for Poisson and binomial regression in terms of the yi’s and mi’s. Show that these are identical to the
Show that for a 3×3 table with n11 = n13, n31 = n33, and n·1 = n·3 that the ˆm’s for the independence model are also the ˆm’s for the uniform association model.
Waite (1911) reports data on classifications of general intelligence made for students from a secondary school in London. Classifications were made after two different school terms and each student
Use the saturated model and Theorem 10.2.1 to find the large sample distribution of a multinomial sample in terms of its category probabilities.
The Delta Method.Let vN be a sequence of q × 1 random vectors and suppose that√N(vN − θ) L→ N0, Σ(θ).Suppose that F(·) is a differentiable function taking q vectors into r vectors.Let dF
Use the delta method of the previous exercise to show that if√N(vN − θ) L→ N0, Σ(θ), then for any r × q matrix A,√N(AvN − Aθ) L→ N0, AΣ(θ)A.Show that if Cov(√NvN ) = Σ(θ),
Testing Marginal Homogeneity in a Square Table.For an I × I table, the hypothesis of marginal homogeneity is H0 : pi· = p·i , i = 1,...,I. A natural statistic for testing this hypothesis is the
Analyze the data of Exercise 8.4.1 as a logistic regression with nodal involvement as the response. Include the investigation of higher-order interactions in your analysis. The original investigator
Asymptotic Inference for the LD(50).In Exercise 4.8.9, models and methods for estimating the LD(50) were discussed. Use the delta method of Exercise 10.8.4 to obtain an asymptotic standard error for
Fieller’s Method for the LD(50).Fieller’s method is an alternative to the delta method for obtaining an asymptotic confidence interval for the LD(50), cf. Exercise 11.8.2.Fieller’s method is
Show that the conditional likelihood given by equation(11.7.1) and display (11.7.2) does not depend on the αi’s or the intercept terms γh1. Here, xi = (1, xi2,...,xik) and γ = (γh1,...,γhk).
Use the delta method of Exercise 10.8.4, the logistic transform, and (11.1.3) to show thatpˆij − pij pˆij (1 − pˆij )ˆaii/Ni∼ N(0, 1), where Ni = ni1 + ni2.
In a Newsweek article on “The Wisdom of Animals”(May 23, 1988), one of the key issues considered was whether animals (other than humans) understand relationships between symbols. Some animals can
Consider a 2×2 table of multinomial probabilities that models how subjects respond on two separate occasions.First Trial Second Trial A B A p11 p12 B p21 p22 Show that Pr(A Second Trial |B First
Weisberg (1975) reports the following data on the number of boys among the first seven children born to a collection of 1,334 Swedish ministers.Number of Boys 0 1 2 3 4 5 6 7 Frequency 6 57 206 362
The data given in the previous problem may be 1,334 independent observations from a Bin(7, p) distribution. If so, use the defining assumptions of the binomial distribution to show that this is the
Let y1, y2, y3, and y4 be random variables and let a1, a2, a3, and a4 be real numbers. Show that the following relationships hold for finite discrete distributions.(a) E(a1y1 + a2y2 + a3) = a1E(y1) +
Assume that(n1,...,nq) ∼ Mult(N, p1,...,pq)and let t be an integer less than q. Define y = n1 + ··· + nt and ˜p =p1 + ··· + pt. Show that(y, nt+1,...,nq) ∼ Mult(N, p, p ˜ t+1,...,pq)so
Suppose y ∼ Bin(N, p). Let ˆp = y/N. Show that E(ˆp) =p and that Var(ˆp) = p(1 − p)/N.
The data in Table 2.2 are on graduate admissions by sex at the University of California, Berkeley, and are given by Bickel et al. (1975) and Freedman et al. (1978). Test for independence, examine the
Cram´er (1946) presents data on the distribution of birth dates for males and females born in Sweden in 1935. The data given in Table 2.3 presume a natural ordering for the months of the year that
Gilby (1911) presents data on the relationships among instructor’s evaluation of general intelligence, quality of clothing, and school standard. General intelligence was classified using a system
Partitioning Tables.The examination of odds ratios and residuals provide two ways to investigate lack of independence in a two-way table. The partitioning methods of Irwin (1949) and Lancaster (1949)
Fisher’s Exact Test.Consider the problem of testing whether the probability of success is the same for two independent binomials. Let yi ∼ Bin(Ni, pi), i = 1, 2. Write the 2 × 2 table as TABLE
Yule’s Q.For 2×2 tables, a measure of association similar to a correlation coefficient is Yule’s Q, which is defined as Q = p11p22 − p12p21 p11p22 + p12p21.Find Q in terms of the odds ratio.
Freeman-Tukey Residuals.Freeman and Tukey (1950) suggest a variance stabilizing transformation for Poisson data that leads to using the quantities√nij + nij + 1 −4 ˆm(0)ij + 1 as residuals,
Power Divergence Statistics.Cressie and Read (1984) and Read and Cressie (1988) have introduced the power divergence family of test statistics 2Iλ = 2λ(λ + 1)ij nij⎡⎣nij mˆ (0)ij λ−
Compute the power divergence test statistics 2I−1/2 and 2I1/2 for the knee injury data of Example 2.3.1. Compare the results to G2 and X2. What conclusions can be reached about knee injuries?
Testing for Symmetry.Consider a multinomial sample arranged in an I ×I table. In square tables with similar categories for the two factors, it is sometimes of interest to test H0 : pij = pji for all
Correlated Data.There are actually 410 observations in Exercise 2.7.10 and Table 2.9. There are 205 men and 205 women. Why was Table 2.9 set up as a 3 × 3 table with only 205 observations rather
McNemar’s Test.McNemar (1947) proposes a method of testing for homogeneity of proportions among two binary populations when the data are correlated. (A binary population is one in which all members
Suppose the random variables nij , i = 1, 2, j =1,...,Ni, are independent Poisson(µi) random variables. Find the maximum likelihood estimates for µ1 and µ2 and find the generalized likelihood
Yule’s Q (cf. Exercise 2.7.6.) is one of many measures of association that have been proposed for 2 × 2 tables. Agresti (1984, Chapter 9) has a substantial discussion of measures of association.
Complete an analysis similar to that of Example 3.4.2 for the classroom behavior data of Example 3.0.1.
Complete an analysis similar to that of Example 3.4.2 for the auto accident data of Example 3.2.4.
Radelet (1981) gives data on the relationship between race and the imposition of the death penalty. The data are given in Table 3.2. Analyze the data.TABLE 3.2. Race and the Death Penalty
The data on graduate admissions at Berkeley given in Exercise 2.6.1 was actually collapsed over the six largest departments within the university. The possibility exists that the data may display
Discuss Simpson’s paradox in terms of the following probability inequalities.Pr(A|B and C) < Pr(A| not B and C),TABLE 3.3. Graduate Admissions at Berkeley Male Female Dept. Admitted Rejected
Reevaluate your analysis of the data discussed in Exercise 2.6.3 in light of Simpson’s paradox. Are there other factors that need to be accounted for in a correct analysis of these data?
For the data of Example 3.2.4, do the first step of the iterative proportional fitting algorithm for ˆm(7)ijk using a hand calculator.Use starting values of ˆm[0]ijk = 1. Compare the results after
Consider the model log(mijk) = u + u1(i) + u2(j) +u3(k) + u12(ij).(a) Show that the maximum likelihood estimate of u3(1) − u3(2) is log(n··1) − log(n··2).(b) Show that maximum likelihood
The Mantel-Haenszel Statistic.In biological and medical applications, it is not uncommon to be confronted with a series of 2 × 2 tables that examine the same effect under different conditions. If
Using the data of Table 3.1 fit the all main effects model, the all two-factor effects model, and the all three-factor effects model. Perform all of the tests possible among these three models.
With regard to Section 3, show that the ˆm[3t+2]ijk ’s and mˆ [3(t+1)]ijk ’s also satisfy M(7).
As can be seen from the iterative proportional fitting algorithm, the ˆm’s for the model of no three-factor interaction depend only on the 3 two-dimensional marginal tables. Discuss how this fact
The auto accident data of Example 3.2.4 was actually a subset of a four-dimensional table. The complete data are given in Table 4.15. Analyze the data treating severity of injury as a response
Breslow and Day (1980) present data on the occurrence of esophageal cancer in Frenchmen. Explanatory factors are age and alcohol consumption. High consumption was taken to be anything over the
The data in the previous experiment is a series of 2×2 tables collected under five different age conditions. This is the same situation as the Mantel-Haenszel setup of Exercise 3.8.9.The
Haberman (1978) reports data from the National Opinion Research Center on attitudes toward abortion (cf. Table 4.17). The data TABLE 4.16. Occurrence of Esophageal Cancer Alcohol Cancer Age
Feigl and Zelen (1965), Cook and Weisberg (1982), and Johnson (1985) give data on survival of 33 leukemia patients as a function of their white blood cell count and the existence of a certain
Finney (1941) and Pregibon (1981) present data on the occurrence of vasoconstriction in the skin of the fingers as a function of the rate and volume of air breathed. The data are reproduced in Table
Mosteller and Tukey (1977) reported data on verbal test scores for sixth graders. They used a sample of 20 Mid-Atlantic and New England schools taken from The Coleman Report. The dependent variable y
The Logistic Distribution.Show that F(x) = ex/(1 + ex) satisfies the properties of a cumulative distribution function (cdf). Any random variable with this cdf is said to have a logistic distribution.
Stimulus–Response Studies.The effects of a drug or other stimulus are often studied by choosing r doses of the drug (levels of the stimulus), say x1,...,xr, and giving the dose xj to each of Nj
Probit Analysis.An alternative to the logistic analysis of dose-response data is probit analysis. In probit analysis, the model isΦ−1(pj ) = α + β log(xj )where Φ(·) is the cumulative
Woodward et al. (1941) report several data sets, one of which examines the relationship between exposure to chloracetic acid and the death of mice. Ten mice were exposed at each dose level. The data
Consider a sample of j = 1,...,r independent binomials yj ∼ Bin(Nj , pj ), each with a covariate xj . Suppose that for some cumulative distribution function F(·), pj = F(xj ).Show that for some
The data of Exercise 4.8.2 are actually a retrospective study. A sample of cancer patients was compared to a sample of men drawn from the electoral lists of the department of Ille-et-Vilaine in
Any multinomial response model can be viewed as the model for an I × J table. Assume product-multinomial sampling from J independent multinomials each with I categories. Defineπij = pij Ih=i phj so
Give the log-linear model corresponding to (4.1.1).
Analyze the trauma data that are described in Example 13.2.2.
Using the methods of Section 5.1, discuss the independence relationships for all of the models given below.(a) [123][24][456](b) [12][13][23][24][456](c) [123][124][456](d) [123][24][456][15](e)
In the saturated log-linear model for a four-dimensional table, let u34 = 0 and let all of the corresponding higher-order terms also be zero, e.g., u134 = 0.(a) Based on this model, find a formula
The vertices for a five-factor model are given below. Connect the dots to give a graphical representation of the model[123][135][34][24]. Use the illustration to show that [123][135][34][24][25] is
Which of the models given below are graphical? Graph them. Which of these are decomposable? Discuss the independence relationships for all of the models. For each model, what marginal tables will
Consider all of the graphs in Example 5.4.2. Classify each as equivalent or not equivalent to a decomposable log-linear model. For those that are equivalent, prove the equivalence of the probability
Exercise 1.13.1. Rescale x3 to make its values lie between 0 and 1. Plot the data.Using least squares, fit models with p = 10 using polynomials and cosines. Plot the regression lines along with the
Exercise 1.13.2. Using s = 8, fit the Coleman Report data using Haar wavelets.How well does this do compared to the cosine and polynomial fits?
Exercise 1.13.3. Based on the s = 10 polynomial and cosine models fitted in
Exercise 1.13.1, useCp to determine a best submodel for each fit. Plot the regression lines for the best submodels. Which family works better on these data, cosines or polynomials? Use Cp to
Exercise 1.13.4. Investigate whether there is a need to consider heteroscedastic variances with the Coleman Report data. If appropriate, refit the data.
Exercise 1.13.5. Fit a cubic spline nonparametric regression to the Coleman Report data.
Exercise 1.13.6. Fit a regression tree to the Coleman Report data using just variable x4.
Exercise 1.13.7. In Sect. 1.7 we set up the interpolating cubic spline problem as one of fitting a saturated linear model that is forced be continuous, have continuous first and second derivatives,
Exercise 1.13.8. Fit a tree model to the battery data and compare the results to fitting Haar wavelets.
Exercise 6.9.1. Show that the following functions are nonnegative definite.(a)σ (k) =⎧⎪⎨⎪⎩1 ifk = 0 14 27 if k = ±1 427 if k = ±2 0 other k .(b)σ (k) =⎧⎪⎨⎪⎩1 ifk = 0ρ if k =
Exercise 6.9.4. Consider the simple linear regression model yt = α +β t +et , where et is a white noise process. Let wt be the symmetric moving average of order 5 introduced in Example 6.6.1 as
Exercise 6.9.5. Shumway (1988) reports data from Waldmeier (1960-1978, 1961) on the number of sunspots from 1748 to 1978. The data are collected monthly, and a symmetric moving average of length 12
Exercise 6.9.6. Box, Jenkins, and Reinsel (1994, p. 597) report data on international air travel. The values in Table 6.8 are the number of passengers in thousands.Do a frequency analysis of these
Exercise 6.9.7. Let et be a white noise process. The spectral density of et is fe(ν) =σ 2; see Exercise 6.2. Let |φ1| < 1 and |θ1| < 1.(a) Find the spectral density of yt = et −θ1et−1
Exercise 6.9.8. Let et and εt be uncorrelated white noise processes. Let yt =φ1yt−1+et and wt =φ1wt−1+yt +εt .Find the spectral density of wt .
Exercise 6.9.9. Let et be second-order stationary and define yt =∞Σs=−∞aset−s and wt =∞Σr=−∞bryt−r, where Σ∞s=−∞ |as| < ∞ and Σ∞r =−∞ |br| < ∞.(a) Show that fw(ν)
Exercise 6.9.10. Let yt and wt be two second-order stationary processes and suppose that fw(ν ) ≤ fy(ν ) for all ν ∈ [−π,π]. Show thatΣyy−Σww is nonnegative definite, where Σyy and

Showing 1 - 100 of 482