Title Companion to Statistical Modelling in R
Author Murray Aitkin, Brian Francis, John Hinde, Ross Darnell <[email protected]>
Maintainer Ross Darnell <[email protected]>
Suggests lattice, foreign, gdata, car, dglm, gnm, MASS, npmlreg,survival
Description This package accompanies Aitkin et al, Statistical
Modelling in R, OUP, 2009. The package contains some functions and datasets used in the text.
betablok . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
bronchitis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
byssinosis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
cars . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
chd . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
claims . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
coxph.disparity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
disparity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
disparity.lm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . faults
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
feigl . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . gehan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ghq
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
hostility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . insult
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
lsat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . miners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . NPL.bands
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
poison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . prentice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . R2CV . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Rsq
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
solv . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . stackloss . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . stan . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . statlab . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . summary.treg . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . toxaemia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . toxoplas . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . trees . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . treg
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
trypanos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vaso . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vietnam . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . woolson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
A 22-centre clinical trial of beta-blockers for reducing mortality after myocardial infarction, de-scribed by Yusuf et. al. (1984). The important issue is the generalizability of the treatment effectacross different patient populations.
A dataframe with 44 obs. and 4 variables:
factor w /2 levels "C","T"
Yusuf, S., Peto, R., Lewis, J., Collins, R. and Sleight, P. (1984). "Beta blockade during and aftermyocardial infarction: an overview of the randomized trials", Progress in Cardiovascular Diseases,27, 335–371.
Chronic bronchitis in a sample of mean in Cardiff
The data consist of observations on three variables for each of 212 men in a sample of Cardiffenumeration districts.
integer, 1= respondent suffered from chronic bronchitis
numeric, the number of cigarettes per day
Jones, K. (1975), A geographical contribution to the aetiology of chronic bronchitis, UnpublishedBSc dissertation, University of Southampton. Published in Wrigley, N. (1976). Introduction tothe use of logit models in geography, Geo.Abstracts Ltd, CATMOG 10, University of East Anglia,Norwich.
The dataset contains the number of workers in a survey of the US cotton industry suffering andnot suffering from the lung disease byssinosis, together with the values of five cross-classifyingcategorical explanatory variables: the race, sex and smoking habit of the worker, the length ofemployment in three categories, and the dustiness of the workplace in three categories.
Factor w/ 3 levels "most","less"
Factor w/ 2 levels "white","non-white"
Factor w/ 2 levels "male","female"
Factor w/ 2 levels "smoker","non"
Factor w/ 3 levels "<10","10-20",.
int, Number of workers who suffered byssinosis
int, Number of workers who did not suffer from byssinosis
Higgins, J.E. and Koch, G.G. (1977), "Variable selection and generalized chi-square analysis ofcategorical data to a large cross-sectional occupational health survey", Int. Statist. Rev., 45, 51–62.
Performance data for cars from Motor Trend magazine
The data are quarter-mile acceleration time in seconds and fuel consumption in miles per (US) gal-lon for 32 cars tested by the US Motor Trend magazine in 1974. Nine explanatory variables aregiven: shape of engine, number of cylinders, transmission type, number of gears, engine displace-ment in cubic inches, horsepower, number of carburettor barrels, final drive ratio, and weight of thecar in thousands of pounds.
integer, shape of engine (straight = 1, vee = 0)
integer, transmission type, (automatic = 0, manual = 1)
numeric, engine displacement in cubic inches
numeric, weight of car in thousands of pounds
numeric, quarter-mile acceleration time in seconds
numeric, fuel consumption in miles per gallon
Factor w/ 32 levels "AMC Javelin".:18 19 .
Henderson, H.V. and Velleman, P.F. (1981), "Building multiple regression models interactively",Biometrics, 37, 391–411.
The file gives the number of men diagnosed as having coronary heart disease (CHD) in an Americanstudy of 1329 men (the data are presented and analysed in Ku and Kullback, 1974). The serumcholesterol level and blood pressure in mm mercury were recorded for each man, and are reportedin one of four categories, giving a 4X4 cross-classified in each cell of which the number of menwith CHD and the total number of men examined are given.
Factor w/ 4 levels "<200","200-219",.
Factor w/ 4 levels "<127","127-146",.
Ku, H.H. and Kullback, S. (1974), "Loglinear models in contingency table analysis", The AmericanStatistician, 28, 115–22.
The file gives the number of policyholders of an insurance company who were “exposed to risk”,and the number of car insurance claims made in the third quarter of 1973 by these policyholders,arranged as a contingency table cross-classified by three four-level factors: dist, the district inwhich the policyholder lived (1: rural, 2: small towns, 3: large towns, 4: major cities), car, theengine capacity of the car (1: $<$ 1 litre, 2: 1 – 1.5 litres, 3: 1.5 – 2 litres, 4: $>$ 2 litres), and age,the age of the policyholder (1: $<$ 25, 2: 25 – 29, 3: 30 – 35, 4: $>$ 35)
Factor w/ 4 levels "<25","25-29","30-35",">35"
Factor w/ 4 levels "rural","small towns","large towns", "major cities"
Factor w/ 4 levels "<1","1-1.5","1.5-2",">2"
Baxter, L.A., Coutts, S.M. and Ross, G.A.F., 1980, "Application of linear models in motor insur-ance", Proceedings of the 21st International Congress of Actuaries, Zurich, 11–29.
Check disparity in a Cox Proportional Hazard Model
The coxph.disparity() function returns the disparity from the piecewise exponential model, includ-ing all the terms in the likelihood, and is directly comparable to the disparity for the fit of othermodels used in this chapter.
This form of the likelihood, allows the Cox proportional hazards model to be compared directly tofully parametric models. (Note that log-likelihood value stored in coxph.object is not comparableas it is based on the proportional hazards function and does not include the baseline hazard, thiscancels out in the conditional probabilities that form the partial likelihood.)
Aitkin, M., Francis, B., Hinde, J. and Darnell, R. (2009). Statistical modelling in R, OUP.
require(survival)data(feigl)feigl <- within(feigl, {lwbc <- log(wbc)})feigl.cph <- coxph(Surv(time) ~ ag * lwbc, data = feigl,
disparity is a generic function used to produce the disparities of the results of various models.
Aitkin, M., Francis, B., Hinde, J. and Darnell, R. (2009). Statistical Modelling in R, OUP.
## The function is currently defined asfunction(model, .)
Disparities for Genralized Linear Model Fits
This function is a methods for class glm objects.
## S3 method for class ’glm’disparity(model)
disparity prints $-2 X $ log-likelihood.
This function is a method for class lm objects.
## S3 method for class ’lm’:## S3 method for class ’lm’disparity(model)
disparity prints $-2 X $ log-likelihood.
Bissell gives the numbers of yarn breaks observed in a roll of fabric whilst a textile process wasrunning, as well as the length of the roll of fabric.
Bissell, A. F. (1972),"A negative binomial model with varying elemental sizes", Biometrika, 59,435–441.
Leukaemia survival times — Feigl \& Zelen
The file contains the survival times in weeks of 33 patients suffering from acute myelogeneousleukaemia, and the values of two explanatory variables, white blood cell count in thousands anda positive or negative factor , positive values being defined by the presence of Auer rods and/orsignificant granulature of the leukaemic cells in the bone marrow at diagnosis, and negative valuesif both Auer rods and granulature are absent.
numeric, white blood cell count in thousands
Factor w/ 2 levels "+","-"
Feigl, P. and Zelen, M. (1965). "Estimation of exponential probabilities with concomitant informa-tion", Biometrics, 21, 826–38.
Remission times of acute leukemia patients — Gehan et al
Data from a clinical trial which compared 6-mercaptopurine (6-MP) to a placebo in the maintenanceof remissions in acute leukemia. The remission times in weeks one year after the start of the studywere recorded. Participants were paired according to remission status, an aspect not described inGehan (1965).
A dataframe containing 42 obs. of 5 variables:
numeric defining pair according to remission status
numeric time to remission available at the time the trial was stoppped
numeric "0" indicating censored ,"1" uncensored
factor w/ 2 levels "6-MP", "control"
Gehan, E. A. (1965), "A generalized Wilcoxon test for comparing arbitrarily singly-censored sam-ples", Biometrika, 52, 203–233.
These data were published by Silvapulle, and come from a psychiatric study of the relation betweenpsychiatric diagnosis (as case or non-case) and the value of the score on a 12-item General HealthQuestionnaire (GHQ), for 120 patients attending a general practitioner’s surgery. Each patient wasadministered the GHQ, resulting in a score between 0 and 12, (however there were no cases or non-cases with GHQ scores of 11 or 12) and was subsequently given a full psychiatric examination by apsychiatrist who did not know the patient’s GHQ score. The patient was classified by the psychiatristas either a “case”, requiring psychiatric treatment, or a “non-case”.
Factor w/ 2 levels "men","women"
integer, number of patients considered a "case"
integer, number of patients considered a "non-case"
Silvapulle, M. J. (1981), "On the existence of maximum likelihood estimators for the binomialresponse model", J. Roy. Statist. Soc. B., 43, 310–13.
A measure of hostility based on word use exhibiting hostility by husbands of wives who had beenadmitted to hospital after suicide attempts by taking drug overdoses compared to a “control” groupof husbands.
A data frame with 67 observations on the following 10 variables.
g a factor with levels overdoses F controls T controls
nation a factor with levels Australian British
Bennett, M. D. (1974). The Emotional Response of Husbands to Suicide Attempts by Their Wives,Sydney University, Unpublished MD thesis.
data(hostility)## maybe str(hostility)plot(hostility)
Ten male and nine female subjects were asked to fill out a questionnaire which mixed innocuousquestions with questions attempting to assess the subject’s self-reported hostility. A hostility scorefor each individual was calculated from these responses. After completing the questionnaire, thesubjects were then left waiting for a long time, and were subjected to insults and verbal abuse by theexperimenter when the questionnaire was eventually collected. All subjects were told that they hadfilled out the questionnaire incorrectly, and were instructed to fill it out again. A second hostilityscore was then calculated from these later responses.
integer, hostility score before verbal attack
integer, hostility score after verbal attack
Factor w/ 2 levels "female","male"
Erickson, B. H. and Nosanchuk, T. A., (1979). Understanding Data, Milton Keynes, UK, OpenUniversity Press, UK.
The original dataset consists of responses from 1,000 subjects to five dichotomous items from sec-tion 6 of the LSAT exam. The version here is presented as frequencies of unique patterns of re-sponses. The data is from Bock and Lieberman 1970.
A data frame with 32 observations on the following 7 variables. The variable wt7 represents thenumber with each pattern.
Bock, R. and Leiberman, M. (1970), "Fitting a response model for a $n$ dichotomously scoreditems." Psychometrika, 35, 179–197.
Bock, R. D. and Aitkin, M. (1981). "Marginal maximum likelihood estimation of item parameters:An application of an EM algorithm." Pyschometrika, 46, 443–459.
The file gives the numbers of coalminers classified by radiological examination into one of threecategories of pneumoconiosis, normal, mild pneumoconiosis and severe pneumoconiosis, and byyears spent working at the coalface (interval midpoint).
numeric, years (midpoint) of years spent at coalface
integer, number of miners classified as normal
integer, number of miners with mild pneumoconiosis
integer, number of miners with severe pneumoconiosis
Ashford, J. R. (1959), "An approach to the analysis of data from semi-quantal responses in biolog-ical response", Biometrics, 15,573–581.
Nonparametric likelihood confidence bands
Computes the confidence bands for the empirical distribution function as described by Owen, A. (1997) JASA 90:516–521.
### Empirical distribution of a gamma variable### and comparing to a normallibrary(lattice)y <- round(rgamma(100,shape=1.4,scale=20))meany <- mean(y)sdy <- sd(y)print(xyplot(qnorm(lower)+qnorm(upper)~x,data=NPL.bands(y),panel=function(x,y,.){panel.xyplot(x,y,.)panel.curve(qnorm(pnorm(x,mean=meany,sd=sdy)))}))### and for a larger sampleyy <- round(rgamma(1000,shape=1.4,scale=20))meanyy <- mean(yy)sdyy <- sd(yy)print(xyplot(qnorm(lower)+qnorm(upper)~x,data=NPL.bands(yy),panel=function(x,y,.){panel.xyplot(x,y,.)panel.curve(qnorm(pnorm(x,mean=meanyy,sd=sdyy)))}))###
and for a t-distributed variable with df=10
yyy <- round(rt(1000,df=10),1)meanyyy <- mean(yyy)sdyyy <- sd(yyy)print(xyplot(qnorm(lower)+qnorm(upper)~x,data=NPL.bands(yyy),panel=function(x,y,.){panel.xyplot(x,y,.)panel.curve(qnorm(pnorm(x,mean=meanyyy,sd=sdyyy)))}))###
and for a mixture of t-distributed variables with df=5
yyyy <- round(c(rt(100,df=5)*5+20,rt(100,df=5)*5+40))meanyyyy <- mean(yyyy)sdyyyy <- sd(yyyy)print(xyplot(qnorm(lower)+qnorm(upper)~x,data=NPL.bands(yyyy),panel=function(x,y,.){panel.xyplot(x,y,.)panel.curve(qnorm(pnorm(x,mean=meanyyyy,sd=sdyyyy)))}))#
Survival times (units 10 hrs) of animals in a 3 X 4 factorial experiment, the factors being (a) threepoisons and (b) four treatments given in Box and Cox (1964). Each combination of the two factorsis used for four animals, the allocation to animals being completely randomized.
A dataframe containing 48 observations for 2 factors type and treat and the vector time.
Factor w/ 3 levels "I","II","III"
Factor w/ 4 levels "A","B","C","D"
Box, G. E. P. and Cox, D. R. (1964). "An analysis of transformations", Journal of the RoyalStatistical Society, B, 26, 211–252.
Veteran’s Administration Lung Cancer Trial
The file consists of survival times in days of 137 lung cancer patients from a Veteran’s Adminis-tration Lung Cancer trial, together with explanatory variables: performance status, a measure ofgeneral medical status on a continuous scale 1–9.9, with 1–3 completely hospitalized, 4–6 partialconfinement to hospital, 7–9.9 able to care for self; age in years; time in months from diagnosis tostarting on the study; a factor prior therapy (1 no, 2 yes); a factor treatment (1 standard, 2 test) and afactor tumour type (1 squamous, 2 small, 3 adeno, 4 large). There are three censored observations.
integer, 1= squamous, 2 =small, 3= adeno, 4= large
integer, general medical status on a scale 1–9.9
Prentice, R. L. (1973), "Exponential survivals with censoring and explanatory variables", Biometrika,60, 279–88.
Coefficient of determination of linear models
This function provides the coefficient of determination for lm objects that may not have an intercept
Aitkin, M., Francis, B., Hinde, J. and Darnell, R. (2009). Statistical Modelling in R, UOP.
data(trees)R2(lm(v ~ d + h - 1, data=trees))
Cross-validated coefficient of determination
This function provides the leave-one-out crossvalidation version of the coefficient of determinationfor regression models
Aitkin, M., Francis, B., Hinde, J. and Darnell, R. (2009). Statistical Modelling in R, UOP.
data(trees)R2CV(lm(v ~ d + h, data=trees))
Calculates the coefficient of determination for any model of class ’lm’.
a model object list with argument y and method fitted
A sample of twenty-four children was randomly drawn from the population of fifth-grade childrenattending a state primary school in a Sydney suburb. Each child was assigned to one of two experi-mental groups, and given instructions by the experimenter on how to construct, from nine differentlycoloured blocks, one of the 3X3 square designs in the Block Design subtest of the Wechsler Intel-ligence Scale for Children (WISC). Children in the first group were told to construct the design bystarting with a row of three blocks (row group), and those in the second group were told to startwith a corner of three blocks (corner group). The total time in seconds to construct four differentdesigns was then measured for each child.
Before the experiment began, the extent of each child’s “field dependence” was tested by the Em-bedded Figures Test (EFT), which measures the extent to which subjects can abstract the essentiallogical structure of a problem from its context (high scores corresponding to high field dependenceand low ability).
Factor w/ 2 levels "corner","row"
Aitkin, M. Anderson, D., Francis, B. and Hinde, J. (1981), Statistical Modelling in GLIM, OxfordUniversity Press
The data consist of 21 observations on stack-loss (the loss of acid through the stack) in a chemicalplant for the conversion of ammonia to nitric acid, with three explanatory variables: air flow(x1),cooling water inlet temperature(x2) and acid concentration(x3).
Lange, K L and Little, R J A and Taylor, J M G, (1989). "Robust statistical modeling using the $t$distribution", J Amer Statist Assoc, 84, 881–896.
The file contains the data on 65 transplanted patients, consisting of the patient’s age at transplan-tation, prior open-heart surgery (1 = yes, 0 = no), a censoring indicator (1 = yes, 0 = no), thesurvival time in days after transplant , a score representing the mismatch between the patient’s andthe donor’s tissue type (values range from 0.00 to 3.05), and an indicator for death by rejection (1= yes, 0 = no). One zero survival time is recoded to 0.5. There are 41 deaths and 24 censoredsurvivals, with 39 distinct death times.
A data frame with 65 observations on the following 12 variables.
Crowley, J. and Hu, M. (1977), Covariance analysis of heart transplant survival data. Journal of theAmerican Statistical Association, 72, 27–36.
The STATLAB Census covers 1296 member families of the Kaiser Foundation Health Plan (a pre-paid medical care program) living in the San Francisco Bay Area during the years 1961 - 1972. These families were participating members of the Child Health and Development Study conceivedand directed by Jacob Yerushalmy, for many years Professor of Biostatistics in the School of PublicHealth, University of California, Berkeley.
On her first visit to the Oakland hospital of the Health Plan after pregnancy was diagnosed, eachwoman was interviewed intensively on a wide range of medical and socioeconomic matters relatingboth to herself and to her husband. In addition, various physical and physiological measures weremade. When her child was born, further data about her and her newborn baby were recorded.
Approximately 10 years later the child and mother were called in for follow-up testing, interviewing,and measurement. In some instances, the husband was also interviewed and measured.
The 1296 families of the STATLAB Census are divided into two equal subpopulations: 648 familiesconsisting of a mother, father, and female child; and 648 families of a mother, father, and male child. The children were all born in the Kaiser Foundation Hospital, Oakland, California, between 1 April1961 and 15 April 1963. The Census does not cover any other children who may also have existedin these families.
A data.frame of 1296 obs. of 34 variables:
Factor w/ 2 levels "boy","girl"
Hodges, J.L., Krech, D. and Crutchfield, R.S. (1975). StatLab: An Empirical Introduction to Statis-tics, McGraw-Hill Ryerson, Toronto
This function is a method for class treg.
## S3 method for class ’treg’summary(object, .)
further arguments passed to or from other methods.
The function summary.treg computes and prints statistics of "lm" class objects as well as the robustestimates of coefficients, the disparity and ’r’, the degrees of freedom.
Aitkin, M., Francis, B., Hinde, J. and Darnell, R. (2009). Statistical Modelling in R, OUP.
The number of women giving birth to their first child who showed toxaemic signs (hypertensionand/or proteinurea, classified as Yes or No) during pregnancy.
A data frame with 60 observations on the following 4 variables.
response a factor with levels HN HU NN NU
Brown, P.J., Stone, J., and Ord-Smith, C. (1983). Toxaemic signs during pregnancy. Applied Statis-tics, 32, 69–72.
data(toxaemia)tox.prop.table1 <- with(toxaemia, prop.table(tapply(count,
list(class = class, response = response, smoke = smoke),sum), c(1, 3))[, c(2, 1, 4, 3), 1:2])
tox.prop.table2 <- with(toxaemia, prop.table(tapply(count,
list(class = class, response = response, smoke = smoke),sum), c(1, 3))[, c(2, 1, 4, 3), 3, drop = FALSE])
The file shows the number of men tested and the number with a positive test for toxoplasmosis in34 cities in El Slavador, together with the annual rainfall in metres.
integer, number of men with a positive test
Efron, B., (1986), Double exponential families and their use in generalized linear regression, J AmerStatist Assoc, 81, 709–721.
The volume of usable wood v in cubic feet (1 foot \ = 30.48 cm) is given for each of a sample of31 black cherry trees, and the height h in feet and the diameter d in inches (1 inch = 2.54 cm) at aheight 4.5 feet above the ground.
Ryan, T. and Joiner, B. and Ryan, B., (1976). Minitab Students Handbook, Duxbury Press, NorthScituate, Mass
Robust regression by modelling errors as $t$-distributed with known degrees of freedom rather thannormal
TRUE prints estimates for $-2 X $ log likelihood, sigma, and r at each interation.
Fits the $t$ distribution for known degrees of freedom , $r$, and computes the profile likelihood andobtains the joint MLEs of the regression coefficients, sigma and disparity of a robust regression.
Aitkin, M., Francis, B., Hinde, J. and Darnell, R. (2008). Statistical modelling in R, OUP.
library(SMIR)data(stackloss)stackloss.lm <- lm(y ~ x1 + x2 + x3, data = stackloss)(stackloss.treg1.1 <- treg(stackloss.lm , r=1.1, verbose = FALSE) )
Follman and Lambert (1989) gave an example of a logistic regression with a varying intercept term. The data consist of numbers $y$ of trypanosomes killed out of $n$ treated at a treatment dose $x$.
A data frame with 8 observations on the following 3 variables.
Follman, D.A. and Lambert, D. (1989). Generalizing logistic regression by nonparametric mixing. Journal of the American Statistical Association, 84, 295–300.
data(trypanos)library(npmlreg)(trypanos.np1 <-
(trypanos.np2 <- update(trypanos.np1,k=2))
These data were obtained in a carefully controlled study of the effect of the rate and volume ofair inspired by human subjects on the occurrence or non-occurrence of a transient vasoconstrictionresponse in the skin of the fingers.
Factor w/ 3 levels "1","2","3", .
integer, transient vasoconstriction response, 1=yes, 0=no
Finney, D.J. (1947)., "The estimation from individual records of the relationship between dose andquantal response", Biometrika, 34, 320–34.
Gilliatt, R.W. (1948). "Vaso-constriction in the finger after deep inspiration", J. Physiol., 107, 76–88.
University of North Carolina Vietnam War Student Survey
A survey of student opinion on the Vietnam War was taken at the University of North Carolina inChapel Hill in May 1967 and published in the student newspaper. Students were asked to fill in“ballot papers”, available in the Student Council building, stating which policy out of A, B, C orD they supported. Responses were cross-classified by sex and by undergraduate year or graduatestatus. The policies were:
The US should defeat the power of North Vietnam by
widespread bombing of its industries, ports and
The US should follow the present policy in Vietnam.
The US should de-escalate its military activity,
stop bombing North Vietnam, and intensify its efforts
The US should withdraw its military forces from Vietnam
Factor w/ 4 levels "A","B","C","D"
Factor w/ 5 levels "1","2","3","4","5"
Factor w/ 2 levels "female","male"
integer , the number of students in each cell
Aitkin, M. (1996). "A short history of a Vietnam War attitude survey.", Stats, 17, 1–9.
This is a subset of the Obesity dataset. Binary indicators of obesity on 1014 children who were 7-9years old in 1977, and were followed up in 1979 and 1981. Children were classified as obese iftheir weights were more than 210% of the population median weight for their gender and height.
A data frame with 48 observations on the following 4 variables.
Woolson, R.F. and Clark, W.R.(1984). Analysis of categorical incomplete longitudinal data. JRSSA. 147, 87–99.
toxaemia, toxoplas, trees, treg, trypanos,

Chronicle of a Pandemic Foretold: Lessons from the 2009 Influenza Epidemic Available at SSRN: http://ssrn.com/abstract=1398445. Executive Summary The A(H1N1) influenza epidemic provided the first indication of the effectiveness of the pandemic preparations that countries and international organizations initiated in the wake of the 2003 SARS epidemic. In the case of SARS, China was

PHARMACY FORUM HELD ON 3RD DECEMBER 2010 AT THE HOLIDAY INN RUNCORN Attendees Jo Bateman, Countess of Chester Hospital (JB) Sarah Roden, Western Cheshire PCT (SR) Danny Forrest, Liverpool Heart and Chest Hospital (DF) Victoria Birchall, NHS CL WL Locality (VB) Diane Hornsby Western Cheshire PCT (DH) Dave Thornton, Aintree Hospital (DT) Michael Lloyd, Whiston Hosp