Skip to main content

LUP Student Papers

LUND UNIVERSITY LIBRARIES

Effekter av kön, ålder och region på sjukpenningen i Sverige - en variansanalys

Berner, Rikke (2006)
Department of Statistics
Abstract
According to TCO (The Swedish Confederation for Professional Employees) the worst threat to the Swedish welfare is that so many people because of bad health and unemployment are not a part of the labour force. If Sweden could make the number of people with health and unemployment problems half the size as today, the GDP would increase with more than five percent. In real money this means 110 billion Swedish kronor a year. According to SCB (Statistics Sweden) the difference between women’s and men’s absence owing to illness should be investigated more closely in order to get at better understanding of the problem. A good way to study the absence owing to illness is to analyse the share of people with sick-benefit. The purpose of this essay... (More)
According to TCO (The Swedish Confederation for Professional Employees) the worst threat to the Swedish welfare is that so many people because of bad health and unemployment are not a part of the labour force. If Sweden could make the number of people with health and unemployment problems half the size as today, the GDP would increase with more than five percent. In real money this means 110 billion Swedish kronor a year. According to SCB (Statistics Sweden) the difference between women’s and men’s absence owing to illness should be investigated more closely in order to get at better understanding of the problem. A good way to study the absence owing to illness is to analyse the share of people with sick-benefit. The purpose of this essay is to analyse whether the sick-benefit differs due to the variation in the independent factors (age, gender, region and time). Furthermore, the interaction, if any, between the analysed factors will be studied and interpreted. To get a better understanding of how much every factor and interaction explains the differences in sick-benefit, the strength of the effect (eta square) is calculated. This measure together with significant differences is far more informative then just significant differences. This allows us to eliminate significant factors from our model because of the small contribution to the variation in the dependent factor. We have chosen to call the dependent factor “Proportion of the population with sick-benefit” and the independent factors in this essay are age, gender, region and time. The age-factor is divided into nine different groups, the region-factor is divided into three groups and the years included in this study are 2000-2005. The reason why a time factor is chosen for analyse is to study whether the absent owing to illness has changed over time and if so, in what way. A four-way analysis of variance with fixed effects is the best way to examine the differences in “Proportion of the population with sick-benefit”. We start out by eliminating the insignificant interactions and then according to interaction plots and the size of eta squared we reduce the model even more. In the final model we find significant interactions between gender and age, between gender and region and between region and age. There is a tendency for women in the northern parts of Sweden to have a higher proportion of sick-benefit. If we look at the interaction between gender and age it shows that the difference between men and women is not the same for those in the age group 20-24 years and 60-64 years, as for those in the middle age groups. The differences in the middle age groups are greater and the biggest difference is seen in the age-group 35-39 years where the proportion with sick-benefit is twice as big for women. The interaction between age and region indicates that people between 35 and 39 years, in the northern parts of Sweden, receive sick-benefits to a larger extent than those between 40 and 44 years. This is not the case in the rest of Sweden. Tendencies were shown for the “Proportion of the population with sick-benefit” to decrease after year 2002. This indicates that the Swedish goal of cutting the proportion of people absent owing to illness in half is going in the right direction. (Less)
Please use this url to cite or link to this publication:
author
Berner, Rikke
supervisor
organization
year
type
H1 - Master's Degree (One Year)
subject
keywords
Interaction, sick-benefit, analysis of variance, strength of the effect, Statistics, operations research, programming, actuarial mathematics, Statistik, operationsanalys, programmering, aktuariematematik
language
Swedish
id
1336992
date added to LUP
2006-11-12 00:00:00
date last changed
2010-08-03 10:49:28
@misc{1336992,
  abstract     = {{According to TCO (The Swedish Confederation for Professional Employees) the worst threat to the Swedish welfare is that so many people because of bad health and unemployment are not a part of the labour force. If Sweden could make the number of people with health and unemployment problems half the size as today, the GDP would increase with more than five percent. In real money this means 110 billion Swedish kronor a year. According to SCB (Statistics Sweden) the difference between women’s and men’s absence owing to illness should be investigated more closely in order to get at better understanding of the problem. A good way to study the absence owing to illness is to analyse the share of people with sick-benefit. The purpose of this essay is to analyse whether the sick-benefit differs due to the variation in the independent factors (age, gender, region and time). Furthermore, the interaction, if any, between the analysed factors will be studied and interpreted. To get a better understanding of how much every factor and interaction explains the differences in sick-benefit, the strength of the effect (eta square) is calculated. This measure together with significant differences is far more informative then just significant differences. This allows us to eliminate significant factors from our model because of the small contribution to the variation in the dependent factor. We have chosen to call the dependent factor “Proportion of the population with sick-benefit” and the independent factors in this essay are age, gender, region and time. The age-factor is divided into nine different groups, the region-factor is divided into three groups and the years included in this study are 2000-2005. The reason why a time factor is chosen for analyse is to study whether the absent owing to illness has changed over time and if so, in what way. A four-way analysis of variance with fixed effects is the best way to examine the differences in “Proportion of the population with sick-benefit”. We start out by eliminating the insignificant interactions and then according to interaction plots and the size of eta squared we reduce the model even more. In the final model we find significant interactions between gender and age, between gender and region and between region and age. There is a tendency for women in the northern parts of Sweden to have a higher proportion of sick-benefit. If we look at the interaction between gender and age it shows that the difference between men and women is not the same for those in the age group 20-24 years and 60-64 years, as for those in the middle age groups. The differences in the middle age groups are greater and the biggest difference is seen in the age-group 35-39 years where the proportion with sick-benefit is twice as big for women. The interaction between age and region indicates that people between 35 and 39 years, in the northern parts of Sweden, receive sick-benefits to a larger extent than those between 40 and 44 years. This is not the case in the rest of Sweden. Tendencies were shown for the “Proportion of the population with sick-benefit” to decrease after year 2002. This indicates that the Swedish goal of cutting the proportion of people absent owing to illness in half is going in the right direction.}},
  author       = {{Berner, Rikke}},
  language     = {{swe}},
  note         = {{Student Paper}},
  title        = {{Effekter av kön, ålder och region på sjukpenningen i Sverige - en variansanalys}},
  year         = {{2006}},
}