McCullagh and Nelder discuss a log linear model to study the risks of damage in ships...
Fantastic news! We've Found the answer you've been seeking!
Question:
Transcribed Image Text:
McCullagh and Nelder discuss a log linear model to study the risks of damage in ships according to a number of factors. While the original dataset lists the damage counts as responses, let us assume instead that you collect a censored version in which you only observe if at least one ship had any damages in a certain number of months. There is one important predictor: the type of ships in each period, a factor with five levels, A through E. Based on the result from the previous question, we can estimate the effect of type on the ship damage rate per month by regressing on the censored response using a complementary log-log link. The results are summarized below; note that we use months, the observed period in months, as an exposure, and hence the offset. glm (formula = any.damage type, family = binomial (link = "cloglog"), data ship.censored, offset log (months)) Coefficients: = Estimate Std. Error z value Pr(>[z]) (Intercept) -5.98204 0.72518 -8.249 < 2e-16 typeB -3.34766 0.89391 -3.745 0.00018 typeC 0.01917 0.89407 0.98289 typeD -0.51598 0.021 0.95828 -0.538 0.59027 1.14325 0.848 0.39629 typeE 0.96979 (Dispersion parameter for binomial family taken to be 1) Null deviance: 58.678 on 33 Residual deviance: 19.671 on 29 degrees of freedom degrees of freedom (a) Interpret the coefficient for ship type E in the context of the original (assumed Poisson) response. How would you rank the ship types in increasing order of risk? (b) Conduct a likelihood ratio (x²) test to assess if type is a relevant predictor of damage rates. State the test statistic and its distribution under the null. (c) Is the model under- or over-dispersed? If so, consider a dispersed model with dispersion parameter ² and provide an estimate for this parameter. Now conduct a F-test for type being included in the model (that is, list the test statistic and its distribution under the null). (d) Give a rough 95% confidence interval for the damage rate per month of ships of type A under the dispersed model. McCullagh and Nelder discuss a log linear model to study the risks of damage in ships according to a number of factors. While the original dataset lists the damage counts as responses, let us assume instead that you collect a censored version in which you only observe if at least one ship had any damages in a certain number of months. There is one important predictor: the type of ships in each period, a factor with five levels, A through E. Based on the result from the previous question, we can estimate the effect of type on the ship damage rate per month by regressing on the censored response using a complementary log-log link. The results are summarized below; note that we use months, the observed period in months, as an exposure, and hence the offset. glm (formula = any.damage type, family = binomial (link = "cloglog"), data ship.censored, offset log (months)) Coefficients: = Estimate Std. Error z value Pr(>[z]) (Intercept) -5.98204 0.72518 -8.249 < 2e-16 typeB -3.34766 0.89391 -3.745 0.00018 typeC 0.01917 0.89407 0.98289 typeD -0.51598 0.021 0.95828 -0.538 0.59027 1.14325 0.848 0.39629 typeE 0.96979 (Dispersion parameter for binomial family taken to be 1) Null deviance: 58.678 on 33 Residual deviance: 19.671 on 29 degrees of freedom degrees of freedom (a) Interpret the coefficient for ship type E in the context of the original (assumed Poisson) response. How would you rank the ship types in increasing order of risk? (b) Conduct a likelihood ratio (x²) test to assess if type is a relevant predictor of damage rates. State the test statistic and its distribution under the null. (c) Is the model under- or over-dispersed? If so, consider a dispersed model with dispersion parameter ² and provide an estimate for this parameter. Now conduct a F-test for type being included in the model (that is, list the test statistic and its distribution under the null). (d) Give a rough 95% confidence interval for the damage rate per month of ships of type A under the dispersed model.
Expert Answer:
Answer rating: 100% (QA)
a The coefficient estimate for ship type E is 096979 This means that the expected log of the probability of at least one ship of type E being damaged in a given month is 096979 units higher than that ... View the full answer
Related Book For
Applied Regression Analysis and Other Multivariable Methods
ISBN: 978-1285051086
5th edition
Authors: David G. Kleinbaum, Lawrence L. Kupper, Azhar Nizam, Eli S. Rosenberg
Posted Date:
Students also viewed these accounting questions
-
We have used the specific-factors model to study the effects of both goods trade and international factor mobility. The following questions ask you to compare these effects. Use the version of the...
-
If at least one ij 0, show that the F, which is used to test that each interaction is equal to zero, has non-centrality parameter equal to c =1=11/02.
-
If there is at least one x value at which more than one observation has been made, there is a formal test procedure for testing H0: µY.x = β0 + β1x for some values...
-
Fill in each blank so that the resulting statement is true. I find (f g)(x) by replacing each occurrence of x in the equation for_______ with________ .
-
For each of the following transactions, note the fund(s) affected, and prepare appropriate journal entries. 1. General obligation bonds with a par value of $750,000 are issued at $769,000 to finance...
-
OSI Model Vs TCP/IP Model A small healthcare facility contacted you for help to provide them with an evaluation of how the OSI model setup differs from the TCP/IP model to help them have a better...
-
Which statement best describes which theoretical perspective nurse managers should use to motivate their employees? A. Each employee has individual needs, so no one theory applies to all employees....
-
Torrey Realty Co. pays weekly salaries of $9,375 on Friday for a five-day workweek ending on that day. Journalize the necessary adjusting entry at the end of the accounting period assuming that the...
-
Drill Problem 10-2 (Algo) [LU 10-1 (1)] Calculate the simple interest and maturity value. Note: Do not round intermediate calculations. Round your answers to the nearest cent. Principal Interest rate...
-
Use the pie chart of a person's monthly income. a. Suppose the total monthly income is $5,800. Determine the amounts in each income category. b. Suppose the income from the secondary job was $1,225....
-
Roma has a large tomato farm on Maryland's Eastern Shore. To meet the water rements of his carefully cultivated crop, Farmer Roma regularly irrigates his tomatoes. Prior the erowing season, Farmer...
-
In Figure P31.69, calculate (a) the equivalent resistance of the circuit (b) the magnitude of the current through each resistor. Use these values: \(R_{1}=1.2 \Omega, R_{2}=2.0 \Omega\), \(R_{3}=2.0...
-
In Figure P31.72, \(\mathscr{E}_{1}=5.0 \mathrm{~V}, \mathscr{E}_{2}=5.0 \mathrm{~V}, \mathscr{E}_{3}=1.5 \mathrm{~V}\), \(R_{1}=50 \Omega, R_{2}=50 \Omega\), and \(R_{3}=50 \Omega\). What are (a)...
-
(a) Determine the current through and the potential difference across each resistor in Figure P31.83. Assume that \(R_{1}=R_{2}=R_{3}=R_{4}=3.0 \Omega, \quad \mathscr{E}_{1}=9.0 \mathrm{~V}\), and...
-
Consider the block diagram of an antenna control system shown in Fig. P8.6-5. Let T = 0.05 T = 0.05 and the sensor gain be unity ( H = 1 ) ( H = 1 ) . Extra \left or missing ight The frequency...
-
At each moment a consumers income, the price of the first good and the price of the second good change according to equations: t = [0, 20]
-
1. Cost of goods manufactured equals $65,000 for 2013. Finished goods inventory is $2,000 at the beginning of the year and $5,500 at the end of the year. Beginning and ending work in process for 2013...
-
The comparative statements of financial position of Menachem NV at the beginning and end of the year 2019 appear below. Net income of ¬34,000 was reported, and dividends of ¬23,000 were paid...
-
In 1990, Business Week magazine compiled financial data on the 1,000 companies that had the biggest impact on the U.S. economy. Data sampled from the top 500 companies in Business Week's report are...
-
An educator has obtained class average scores on a standardized test for classes from each grade from first through seventh at each of four schools in a city. These are shown in the following table....
-
Answer the following questions using the accompanying SPSS output, which is based on data on the growth rates (Y) of depleted chicks at different (log) dosage levels (X) of vitamin B for males and...
-
The Arcadia Company bid on a multiyear contract to construct the new hockey arena for the Phoenix Coyotes. The contract required that construction begin not later than January 1, 2015, and be...
-
Forms are provided in the Working Papers. Your instructor will guide you through the following examples. 1. Create a schedule that shows the payment of an annuity over a four-year period. The initial...
-
The Hilton Company sells inexpensive furniture on installment plans to customers that have poor credit. Many of the companys customers are sometimes not able to make the required payments, forcing...
Study smarter with the SolutionInn App