Scientific publications of Rolf Sundberg in refereed journals since 1974
* Maximum likelihood theory for incomplete data from an exponential family. Scand. J. Statist. 1, 49-58, (1974). Download pdf
* On estimation and testing for the folded normal distribution. Comm. Statist. 3, 55-72 (1974).
* Some results about decomposable (or Markov-type) models for multidimensional contingency tables: Distribution of marginals and partitioning of tests. Scand. J. Statist. 2, 71-79, (1975). Download pdf
* An iterative method for solution of the likelihood equations for incomplete data from exponential families. Comm. Statist. B 5, 55-64 (1976).Download pdf
* Discussion of paper by Dempster, Laird & Rubin (invited contribution). J. Roy. Statist. Soc. B 39, 33 (1977).
* Maximum likelihood estimation of a linear functional relationship when one of the departure variances is known (L. Moberg & R.S.) Scand. J. Statist. 5, 61-64, (1978).
* Statistical precision in the calibration and use of sorting machines and other classifiers (A. Grassia & R.S.). Technometrics 24, 117-121 (1982).
* Discussion of paper by P.J. Brown (invited contribution). J. Roy. Statist. Soc. B 44, 317 (1982).
* The predictive approach and random population type models for finite population inference from two-stage samples. Scand. J. Statist. 10, 223-238 (1983).
* Discussion of paper by Diaconis & Efron (invited contribution). Annals of Statistics 13, 903-904 (1985).
* When is the inverse regression estimator MSE-superior to the standard regression estimator in multivariate controlled calibration situations? Statist. Prob. Letters 3, 75-79 (1985).
* Recent trends in stereology (E.V.B. Jensen, A. Baddeley, H.J. Gundersen & R.S.) Internat. Statist. Review 53, 99-108 (1985).
* Generalized associated point methods for sampling planar objects (E.V.B. Jensen & R.S.). J. Microscopy 144, 55-70 (1986).
* Statistical models for stereological inference about spatial structures: On the applicability of best linear unbiased estimators in stereology (E.V.B. Jensen & R.S). Biometrics 42, 735-751 (1986).
* Tests for underlying Markovian structure from panel data with partially aggregated states. Biometrika 73 717-721 (1986)
* Confidence and conflict in multivariate calibration (P.J. Brown & R.S.). J. Roy. Statist. Soc. B 49, 46-57 (1987).
* Interplay between chemistry and statistics, with special reference to calibration and the generalized standard addition method. Chemometrics & Intell. Lab. Systems 4, 299-305 (1988)
* Prediction diagnostics and updating in multivariate calibration (P.J. Brown & R.S.). Biometrika 76, 349-361 (1989)
* Multivariate calibration with more variables than observations (R.S. & P.J. Brown) Technometrics 31, 365-371 (1989)
* Invited contribution to Discussion of paper by T. Tjur on Analysis of variance and design of experiments. Scandinavian Journal of Statistics 18, 309-313 (1991)
* Continuum regression and ridge regression. J. Royal Statist. Soc. B 55, 653-659 (1993). Download pdf
* Interpretation of unreplicated two-level factorial experiments, by examples. Chemometrics & Intell. Lab. Systems 24, 1-17 (1994).
* Precision estimation in sample survey inference: A criterion for choice between variance estimators. Biometrika 81, 157-172 (1994). Download pdf
* Most modern calibration is multivariate. Book of Invited Papers, XVIIth Internat. Biom. Conf., 395-405 (1994).
* The precision of the estimated generalized least squares estimator in multivariate calibration. Scandinavian Journal of Statistics 23, 257-274 (1996).Download pdf
* Continuum regression is not always continuous (with Anders Björkström). J. Royal Statist. Soc. B 58, 703-710 (1996). Download pdf
* Invited contribution to Discussion of RSS read paper by L. Breiman & J.H. Friedman, J. Royal Statist. Soc. B 59, (1997).
* Statistical aspects on fitting the Arrhenius equation. Chemometrics & Intell. Lab. Systems 41, 249-252 (1998). Download pdf
* Second-order calibration: bilinear least squares regression and a simple alternative (with Marie Linder). Chemometrics & Intell. Lab. Systems 42, 159-178 (1998). Download pdf
* A generalized view on continuum regression (with A. Björkström). Scand. J. Statist. 26, 17-30 (1999). Download pdf
* Multivariate calibrationdirect and indirect regression methodology (with discussion). Scand. J. Statist 26, 161-207 (1999). Download pdf
* The distribution of the maximum likelihood estimator in up-and-down experiments for quantal dose-response data (with M. Vågerö). Journal of Biopharmaceutical Statistics, 9, 499-519 (1999).
* Aspects of statistical regression in sensometrics. Food Quality and Preference 11, 17-26 (2000). Download pdf
* Statistical modelling and saddle-point approximation of tail probabilities for accumulated splice loss in fibre-optic networks (with Joanna. Tyrcha, P. Lindskog & B. Sundström). J. Applied Statistics 27, 245-256 (2000).
* Comparison of confidence procedures for type I censored exponential lifetimes. Lifetime Data Analysis 7, 393-413 ( 2001). Download pdf
* Precision of prediction in second order calibration, with focus on bilinear regression methods (with Marie Linder). Journal of Chemometrics 16, 12-27 (2002). Download pdf
* Collinearity. Article published in the Encyclopedia of Environmetrics, Vol. 1, pp 365-366 (2002) Download pdf
* Shrinkage regression. Article published in the Encyclopedia of Environmetrics, Vol 4, pp 1994-1998 (2002) Download pdf
* The convergence rate of the TM algorithm of Edwards and Lauritzen. Biometrika 89, 478-483 (2002). Abstract, see below. Full paper: Download pdf
* Continuum regression. Article for 2nd ed. of Encyclopedia of Statistical Sciences (2002). Download pdf
* Conditional statistical inference and quantification of relevance. J. Royal Statist. Soc. B 65, 299-315 (2003). Download pdf
* Decrease of serotonin receptor 2C in schizophrenia brains identified by high-resolution mRNA expression analysis (with Anja Castensson, L. Emilsson & E. Jazin). Biological Psychiatry, 54, 1212-1221 (2003).
* Logistic regression in three-point designs (with M Vågerö). Article for Encyclopedia of Biopharmaceutical Statistics (2003).
* Sensor fusion as a tool to monitor dynamic dairy processes (with Marcus Henningsson, K. Östergren & P. Dejmek). J. Food Engineering 76 (2006), 154-162, Online publication 2005.
* Statistical modelling in casecontrol realtime RT-PCR assays, for identification of differentially expressed genes in schizophrenia (Rolf Sundberg, A Castensson & E. Jazin). Biostatistics, Vol 7, pp 130-144 (2006). See http://biostatistics.oxfordjournals.org/content/vol7/issue1/index.dtl Early version: Download pdf
Data used in the paper can be downloaded here as text-file: Schizodata
* Control of confounding through secondary samples (Li Yin, Rolf Sundberg, X. Wang & D.R. Rubin). Statistics in Medicine, Vol 25, pp 38143825 (2006).
* Small-sample and selection bias effects in multivariate calibration, exemplified for OLS and PLS regression. Presented at SSC9, Reykjavik, August 2005. Published in Chemometrics & Intell. Lab. Systems, Vol. 84, pp 2125 (2006). Data used in the paper can be downloaded here as text-file: Data_KC
* Small sample and selection bias effects in calibration under latent factor regression models. Dept Res. Report 2006:13, Dec. 2006, 31 pages. J Chemometrics, Vol. 21, pp 227-238 (2007). Download pdf
* A two-parametric family of predictors in multivariate regression (Anders Björkström & Rolf Sundberg). J. Chemometrics, Vol. 21, pp 215-226 (2007).
* Multigene analysis can discriminate between ulcerative colitis, Crohn's disease and irritable bowel syndrome (Petra von Stein et al). Gastroenterology, Vol 134:7, pp 1869-1881 (2008).
* A statistical methodology for drugdrug interaction surveillance (Niklas Norén, Rolf Sundberg, A. Bate & R. Edwards). Dept Res. Report 2007:6, Febr. 2007, 22 pages. Statistics in Medicine, Vol 27:16, pp 3057-3070 (2008).
* A classical dataset from Williams, and its role in the study of supersaturated designs. Dept Res. Report 2007:22. For abstract, see below. J. Chemometrics, Vol 22, pp 436-440 (2008). Download pdf
* Decomposition of time series of geological data into long- and short-timescale variations under non-Gaussian state space models (Jelena Bojarova & Rolf Sundberg). Environmetrics, Vol. 21, 562587 (2010).
* Flat and multimodal likelihoods and model lack of fit in curved exponential families. Scand. J. Statistics, Vol. 37, pp 632643 (2010),
* Three contributions to International Encyclopedia of Statistical Science (Springer, 2011),
Entries: Chemometrics; Exponential family models; Statistical consulting
* Statistical framework for evaluation of climate model simulations by use of climate proxy data from the last millennium Part 1: Theory (R Sundberg, A Moberg & A Hind) Climate of the past, Vol. 8, pp 13391353 (2012). (Open access, http://www.clim-past.net/8/1339/2012/cp-8-1339-2012.html)
* Statistical framework for evaluation of climate model simulations by use of climate proxy data from the last millennium Part 2: A pseudo-proxy study addressing the amplitude of solar forcing (A Hind, A Moberg & R Sundberg). Climate of the past, Vol. 8, pp 13551365 (2012) http://www.clim-past.net/8/1355/2012/cp-8-1355-2012.html
* Review of Brereton, R.G.: Applied Chemometrics for Scientists, Wiley 2007. Published in J. Amer. Statist. Assoc., Vol 103 (No 483), pp 1317-1318 (2008),
Other recent publications
* Kemometri (in Swedish). Qvartilen 11:1, pp 10-11 (1996).
* Save important installation time by setting appropriate maximum splice loss values. (with P. Lindskog, B.O. Sundström & J. Tyrcha). Presented at 47th Internat. Wire and Cable Symp. (IWCS), Philadelphia, 16-19 Nov. 1998.
* Looking at process capability indices and data: Graphical statistical aspects, for application in manufacturing control (with M. Linder). Dept Res. Report B:49, Febr. 1999, 11 pages. Presented at 1st Internat. Symp. on Industrial Statistics (First ISIS), Linköping, 19-21 August 1999.
* Ancillarity and conditional inference for ML and interval estimates in some classical genetic linkage trials. Dept Res. Report 2001:3, Febr. 2001, 17 pages. Abstract, see below. Full paper: Download pdf
* Separation between classes of multidimensional signals with an improved wavelet packet local discriminant basis algorithm (with H. Carlqvist & J-O Strömberg). Manucscript Febr. 2005, in Carlqvist's Ph.D. thesis, KTH
* Real-time RTPCR. Talk for "Statistical Methods in Gene and Protein Expression", Göteborg 9-12 May 2006, Download pdf, 22 pages.
* Random loss of genetic segments during skin differentiation indicated by analysis of single cells (Emilie Hultin, A. Asplund, L. Berggren, K. Edlund, A. Ahmadian 1, Rolf Sundberg, F. Pontén & J. Lundeberg). In Emilie Hultin's Ph.D. thesis, KTH, March 2007.
* Classification of SNP genotypes by a Gaussian mixture model in competitive enzymatic assays (Hedvig Norlén, E. Pettersson, A. Ahmadian, J. Lundeberg & Rolf Sundberg).Dept Res. Report 2008:3. Submitted. Similar paper found in Erik Pettersson's Ph.D. thesis at KTH, Dec. 2007
* Student's t 100 år (in Swedish). Artikel för Qvintensen 2008:4
* Diskussion av artikel om statistiska test, för Qvintensen 2011:1 (in Swedish)
Lecture notes (mostly in Swedish)
* Statistical Modelling by Exponential Families. Lecture notes and book manuscript, last version April 2012. 227 pages.
* Lineära statistiska modeller. För kursen med samma namn vid mat stat, SU.Senaste version från okt 2012. 200 sidor
* Urval ur ändliga populationer. För kursen Statistisk analys, vid mat stat SU. Senaste version från nov. 2010. 46 sidor.
* Statistiska metoder för planering och analys av experimentella undersökningar. För biomedicinlinjen, KI. (last ed. 2003)
* Mätprecision och felfortplantning, med aspekter på MichaelisMenten-modellering. För biomedicinar-utbildningen vid Karolinska Institutet. Januari 2002, 12 pages.
* Statistiska modeller. Anteckningar från en föreläsningsserie av Per Martin-Löf läsåret 196970. Use this link
Updated March 2013
Abstract of The convergence rate of the TM algorithm of Edwards and Lauritzen: Edwards & Lauritzen (Biometrika 88, 2001, pp 961-972) have recently proposed the TM algorithm for finding the maximum likelihood estimate when the likelihood can be truly or artificially regarded as a conditional likelihood, and the full likelihood is more easily maximised, . They have presented a proof of convergence, provided that the algorithm is supplemented by a line search. In this note a simple expression, in terms of observed information matrices, is given for the convergence rate of the algorithm per se, when it converges, and the result elucidates also in which situations the algorithm will require a line search. Essentially these are cases when the full model does not adequately fit the data.
Some key words: Conditional likelihood; exponential families; graphical chain models; iterative method; ML estimation.
Abstract Report 2001:3 Ancillarity and conditional inference for ML and interval estimates in some classical genetic linkage trials. The main object of study here is a classical example of linkage analysis, in which there are two separately but not jointly ancillary statistics, which are mutually exchangeable. In such cases it is not obvious how or even if the statistical inference about the parameter of interest (here the recombination probability) should be a conditional inference. We consider various precision measures, viz.\ the observed and the expected (Fisher) information quantities, and various conditional expected values in between, and we compare their ability to quantify the precision of the parameter estimate, as well as to quantify the confidence to be attached to interval estimates. The general conclusion drawn is that there is not much to be gained but much to be risked by conditional inference in this example.
Some key words: Confidence, precision, recombination probability, relevance.
Abstract Report 2007:22 A classical dataset from Williams, and its role in the study of supersaturated designs
A PlackettBurman type dataset from a paper by Williams (1968), with 28 observations and 24 two-level factors, has become a standard dataset for illustrating construction (by halving) of supersaturated designs (SSDs) and for a corresponding data analysis. The aim here is to point out that for several reasons this is an unfortunate situation. The original paper by Williams contains several errors and misprints. Some are in the design matrix, which will here be reconstructed, but worse is an outlier in the response values, which can be observed when data are plotted against the dominating factor. In addition, the data should better be analysed on log-scale than on original scale. The implications of the outlier for SSD analysis are drastic, and it will be concluded that the data should be used for this purpose only if the outlier is properly treated (omitted or modified).
Key words: half-fraction, log-transformation, outlier, PlackettBurman, SSD