Kari Djerf - Academia.edu (original) (raw)
Papers by Kari Djerf
Wiley StatsRef: Statistics Reference Online, 2014
Encyclopedia of Statistics in Quality and Reliability, 2008
Journal of Applied Statistics, 2016
In this paper, we discuss the analysis of complex health survey data by using multivariate modeli... more In this paper, we discuss the analysis of complex health survey data by using multivariate modeling techniques. Our main interests are in design-based and model-based methods that aim at accounting for clustering, stratification and weighting effects. Our main interests are in clustering effects. Methods considered include generalized linear modeling with on pseudo-likelihood and generalized estimating equations, linear mixed models estimated by restricted maximum likelihood, and hierarchical Bayes techniques using Markov Chain Monte Carlo (MCMC) methods. The methods will be compared empirically, using data from a health interview and examination survey conducted in Finland in 2000 (Health-2000 Study). The data of the Health-2000 Study were collected using personal interviews, questionnaires and clinical examinations. A stratified two-stage cluster sampling design was used. The sampling design involved positive intra-cluster correlation for several study variables. We selected the study variables systolic blood pressure and chronic morbidity for a closer investigation. In many cases, the different methods produced similar numerical results and supported similar statistical conclusions. Methods that failed to account for the design complexities sometimes led to conflicting conclusions. We also discuss the application of the methods by using standard statistical software packages.
Le présent article traite de l'analyse des données d'enquêtes complexes sur la santé par des méth... more Le présent article traite de l'analyse des données d'enquêtes complexes sur la santé par des méthodes multivariées de modélisation. Nous nous concentrons sur les méthodes basées sur le plan d'échantillonnage et celles basées sur un modèle qui visent à tenir compte des effets de la mise en grappes, de la stratification et de la pondération. Nous nous intéressons avant tout aux effets de la mise en grappes. Les méthodes étudiées incluent la modélisation linéaire généralisée fondée sur la pseudo-vraisemblance et les équations d'estimation généralisées, les modèles linéaires mixtes estimés par le maximum de vraisemblance restreints et les techniques hiérarchiques bayésiennes basées sur les méthodes de Monte Carlo par chaîne de Markov (MCCM). Nous comparons empiriquement ces méthodes sur des données provenant d'une enquête comprenant une entrevue sur la santé et un examen physique réalisée en Finlande en 2000 (Health-2000 Study). Les données de la Health-2000 Study ont été recueillies au moyen d'entrevues personnelles, de questionnaires et d'examens cliniques. L'enquête a été réalisée auprès d'un échantillon en grappes stratifié à deux degrés. Le plan d'échantillonnage a produit des corrélations intra-grappes positives pour nombre de variables étudiées. Nous avons sélectionné en vue d'une étude plus approfondie les variables pression sanguine systolique et morbidité chronique, tirées des volets de l'entrevue sur la santé et de l'examen clinique. Dans de nombreux cas, les diverses méthodes ont produit des résultats numériques comparables et appuyé des conclusions statistiques similaires. Celles qui ne tenaient pas compte de la complexité du plan d'échantillonnage ont parfois produit des conclusions contradictoires. Nous discutons aussi de l'application des méthodes lors de l'utilisation de logiciels statistiques standards.
In this paper, we discuss the analysis of complex health survey data by using multivariate modeli... more In this paper, we discuss the analysis of complex health survey data by using multivariate modeling techniques. Our main interests are in design-based and model-based methods that aim at accounting for clustering, stratification and weighting effects. Our main interests are in clustering effects. Methods considered include generalized linear modeling with on pseudo-likelihood and generalized estimating equations, linear mixed models estimated by restricted maximum likelihood, and hierarchical Bayes techniques using Markov Chain Monte Carlo (MCMC) methods. The methods will be compared empirically, using data from a health interview and examination survey conducted in Finland in 2000 (Health-2000 Study). The data of the Health-2000 Study were collected using personal interviews, questionnaires and clinical examinations. A stratified two-stage cluster sampling design was used. The sampling design involved positive intra-cluster correlation for several study variables. We selected the s...
The Finnish Consumer Barometer was introduced in autumn 1987. Data were first collected twice a y... more The Finnish Consumer Barometer was introduced in autumn 1987. Data were first collected twice a year and from August 1991 until September 1995 quarterly. After Finland joined the European Union in 1995, the survey was adopted as one member of the Harmonised Consumer Survey of the European Communities. Since October 1995, data have been collected monthly. Performance of the Consumer Barometer has already been evaluated by means of descriptive studies (see Djerf 1990). As the survey matures, it becomes feasible to make a more thorough study on the usefulness of the survey. Here we are, for example, interested in investigating how consumers were able to predict the long-lasting recession of our economy. The consumer confidence index and the five questions used for calculating it are compared to various components of Finnish macroeconomic time series. Additionally, we analyse the coincidence of other common measures (unemployment expectations, inflation expectations, etc.) as well as ot...
Wiley StatsRef: Statistics Reference Online, 2014
Encyclopedia of Statistics in Quality and Reliability, 2008
Journal of Applied Statistics, 2016
In this paper, we discuss the analysis of complex health survey data by using multivariate modeli... more In this paper, we discuss the analysis of complex health survey data by using multivariate modeling techniques. Our main interests are in design-based and model-based methods that aim at accounting for clustering, stratification and weighting effects. Our main interests are in clustering effects. Methods considered include generalized linear modeling with on pseudo-likelihood and generalized estimating equations, linear mixed models estimated by restricted maximum likelihood, and hierarchical Bayes techniques using Markov Chain Monte Carlo (MCMC) methods. The methods will be compared empirically, using data from a health interview and examination survey conducted in Finland in 2000 (Health-2000 Study). The data of the Health-2000 Study were collected using personal interviews, questionnaires and clinical examinations. A stratified two-stage cluster sampling design was used. The sampling design involved positive intra-cluster correlation for several study variables. We selected the study variables systolic blood pressure and chronic morbidity for a closer investigation. In many cases, the different methods produced similar numerical results and supported similar statistical conclusions. Methods that failed to account for the design complexities sometimes led to conflicting conclusions. We also discuss the application of the methods by using standard statistical software packages.
Le présent article traite de l'analyse des données d'enquêtes complexes sur la santé par des méth... more Le présent article traite de l'analyse des données d'enquêtes complexes sur la santé par des méthodes multivariées de modélisation. Nous nous concentrons sur les méthodes basées sur le plan d'échantillonnage et celles basées sur un modèle qui visent à tenir compte des effets de la mise en grappes, de la stratification et de la pondération. Nous nous intéressons avant tout aux effets de la mise en grappes. Les méthodes étudiées incluent la modélisation linéaire généralisée fondée sur la pseudo-vraisemblance et les équations d'estimation généralisées, les modèles linéaires mixtes estimés par le maximum de vraisemblance restreints et les techniques hiérarchiques bayésiennes basées sur les méthodes de Monte Carlo par chaîne de Markov (MCCM). Nous comparons empiriquement ces méthodes sur des données provenant d'une enquête comprenant une entrevue sur la santé et un examen physique réalisée en Finlande en 2000 (Health-2000 Study). Les données de la Health-2000 Study ont été recueillies au moyen d'entrevues personnelles, de questionnaires et d'examens cliniques. L'enquête a été réalisée auprès d'un échantillon en grappes stratifié à deux degrés. Le plan d'échantillonnage a produit des corrélations intra-grappes positives pour nombre de variables étudiées. Nous avons sélectionné en vue d'une étude plus approfondie les variables pression sanguine systolique et morbidité chronique, tirées des volets de l'entrevue sur la santé et de l'examen clinique. Dans de nombreux cas, les diverses méthodes ont produit des résultats numériques comparables et appuyé des conclusions statistiques similaires. Celles qui ne tenaient pas compte de la complexité du plan d'échantillonnage ont parfois produit des conclusions contradictoires. Nous discutons aussi de l'application des méthodes lors de l'utilisation de logiciels statistiques standards.
In this paper, we discuss the analysis of complex health survey data by using multivariate modeli... more In this paper, we discuss the analysis of complex health survey data by using multivariate modeling techniques. Our main interests are in design-based and model-based methods that aim at accounting for clustering, stratification and weighting effects. Our main interests are in clustering effects. Methods considered include generalized linear modeling with on pseudo-likelihood and generalized estimating equations, linear mixed models estimated by restricted maximum likelihood, and hierarchical Bayes techniques using Markov Chain Monte Carlo (MCMC) methods. The methods will be compared empirically, using data from a health interview and examination survey conducted in Finland in 2000 (Health-2000 Study). The data of the Health-2000 Study were collected using personal interviews, questionnaires and clinical examinations. A stratified two-stage cluster sampling design was used. The sampling design involved positive intra-cluster correlation for several study variables. We selected the s...
The Finnish Consumer Barometer was introduced in autumn 1987. Data were first collected twice a y... more The Finnish Consumer Barometer was introduced in autumn 1987. Data were first collected twice a year and from August 1991 until September 1995 quarterly. After Finland joined the European Union in 1995, the survey was adopted as one member of the Harmonised Consumer Survey of the European Communities. Since October 1995, data have been collected monthly. Performance of the Consumer Barometer has already been evaluated by means of descriptive studies (see Djerf 1990). As the survey matures, it becomes feasible to make a more thorough study on the usefulness of the survey. Here we are, for example, interested in investigating how consumers were able to predict the long-lasting recession of our economy. The consumer confidence index and the five questions used for calculating it are compared to various components of Finnish macroeconomic time series. Additionally, we analyse the coincidence of other common measures (unemployment expectations, inflation expectations, etc.) as well as ot...