Paper accepted to poster presentation at the IUSSP International - - PDF document

SLIDE 1

Umeå University September 2016 [2016-09-28] Authors (among whom the 1st is corresponding/presenting author):

• 1. Lotta Vikström, Professor of History | Centre for Demographic and Ageing Research (CEDAR)

Umeå University | SE-901 87 Umeå, Sweden | E-mail: lotta.vikstrom@umu.se

• 2. Helena Haage, Dr. in History | Centre for Demographic and Ageing Research (CEDAR) Umeå

University | SE-901 87 Umeå, Sweden | E-mail: helena.haage@umu.se

• 3. Erling Häggström Lundevaller, Dr. in Statistics | Centre for Demographic and Ageing Research

(CEDAR) | Umeå University | SE-901 87 Umeå, Sweden | E-mail: erling.lundevaller@umu.se

• 4. Sören Edvinsson, Professor of History | Centre for Demographic and Ageing Research (CEDAR) |

Umeå University | SE-901 87 Umeå, Sweden | E-mail: soren.edvinsson@umu.se

Paper accepted to poster presentation at the IUSSP International Population Conference, Cape Town, South Africa, October 28 to November 4, 2017

[NB. This paper is work in progress. Please, don’t quote without authors’ permission.]

Poster Session: Historical Demography

On how disability impede individuals’ life opportunities: Demographic life- course indicators of social exclusion in a Swedish 19th-century population

ABSTRACT

Contemporary disability studies indicate that impairments jeopardize people’s health and make them weaker positioned in the labor market than the ’able’ majority. Disabled individuals are also more likely to live alone as one. Historically, little is known about how disability impeded people’s

• lives. This study investigates the trajectories and death among disabled and non-disabled persons

in a population (N=35,109) followed across life in the 19th-century Sundsvall region, Sweden. Having access to micro-data that report disabilities among individuals from parish registers digitized by the Demographic Data Base, Umeå University, we employ sequence analysis on a series

• f events expected to occur in young adult life: work, marriage and parenthood, also accounting

for premature mortality risks by Cox regression models. Although some variations by type of disability and gender are found, the trajectories of the disabled did not include work nor family to the same extent as among the non-disabled. Disability further rendered a premature death but this was not as frequently caused by infectious diseases as among non-disabled. Difficulties in the labor and marriage markets and low survival are indicative for that disability promoted social exclusion in past populations. These findings are discussed from labeling and life course perspectives.

SLIDE 2

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

2

• 1. Introduction: aims and rationales of the study

From contemporary disability studies, we know that impairments tend to jeopardize individuals’ health status and that disabled people run the risk of constituting the ‘otherness’ in society (Mont 2007, Priestley 2004; Susman 1994; Jaeger & Bowman 2005; Kudlick 2003; Solvang 2000). In history, most disability studies concern institutionalized individuals for whom the sources tell little about their lives beyond institution; and if the sources do, they usually document only a few persons with disabilities (Förhammar and Nelson 2004). The research design and data of this paper enable us to analyze the life courses of a considerably large quantity of disabled persons and to compare the findings with a group of non-disabled. As a result, we can answer three major questions: 1) Did disability imply higher premature mortality risks for individuals and did these risks differ between type of disability and gender? 2) Did the death causes vary substantially if disability was present in life? 3) Do we find any explanation as to why individuals’ lives eventually ended in untimely death by comparing the life trajectories of disabled and non-disabled men and women? Answering these questions, our primary aim is to advance the knowledge on whether and how disabilities impeded people’s opportunities in history. Studying the life and death of disabled people are important as they have been a minority long hidden in history. Their demographic experiences reflect not only their living conditions but also the attitudes of the majority population in past societies. The results are obtained by employing two types of life course methods: event history analyses (EHA) and sequence analyses (SA). Being able to conduct such analyses of a population showing disabilities about 150-200 years ago, makes our investigation novel in its approach. We have used 19th-century parish registers from Sweden, which allow us to research disabled people’s demographic experiences. These registers are digitized by the Demographic Data Base (DDB) at Umeå University in Sweden, and have not yet been explored systematically from a disability perspective. As the registers are recorded longitudinally, we can follow disabled people

• ver time and obtain data on them. The disabilities chosen to identify disabled people in this study

are those labeled ‘blind’, ‘deaf mute’, ‘crippled’, ‘idiot’ and ‘insane’ by the ministers in the parishes. While these concepts may be offensive, due to the derogatory meaning they carry today, we have no intention to offend readers in the few cases we use them. The DDB digitization further enables us to supplement the dataset with a reference group of individuals who do not have any of these selected disabilities. The people in the reference group lived in the same time-space context as the disabled individuals, made up by the 19th-century Sundsvall region in Sweden.

• 2. Previous research: Mortality patterns and disability in the past

From the 18th century onward, mortality patterns have been investigated through both macro and micro studies, especially in the Western world (Bengtsson et al. 2004). These studies demonstrate gendered variations in mortality across different time-space contexts and age groups. Tabular Commission (Tabellverket) started with population statistics in 1749 (Sköld 2001), and since then we can see that the mortality among Swedish men has been higher than that among women, except for some brief time periods and mainly among young people (Willner 1999). This male excess in mortality persisted throughout the 19th century although the gap between the genders decreased. During the latter part of the century, improving social, working and housing conditions promoted individuals’ longevity, particularly the men’s, as the gendered gap in mortality continued to

• decrease. Explanations for this gap in life expectancy have been discussed among scholars, some

SLIDE 3

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

3

• f whom highlight biological or genetic differences inherent in the female sex (Willner 1999;

Fridlizius 1988). Rough working conditions associated with particularly men’s occupations in agricultural production or dirty industries, their lifestyle and the abuse of alcohol are among the major socio-cultural factors historians suggest as explanations for the male excess in mortality. Similar to other countries, the general decline in mortality in late 19th-century Sweden was due to advancements in nutrition and to towns becoming less unhealthy places for urban populations to live in as sanitary water supplies and systems gradually improved (Nilsson 1994; Edvinsson 1994; Söderberg et al. 1991). Whereas there are many studies analyzing mortality and its gendered variations in past societies, few, if any, concern how it looked among a larger number of disabled individuals and whether their mortality differed from general patterns. In her recent thesis on deaf people in East Flanders, Belgium, 1750–1950, Sofie De Veirman (2015) is one among the first historians to present statistical results indicating the risks of premature deaths that deaf individuals ran over their

• lifetime. Comparing their mortality risks with those of their hearing siblings, who constitute a

reference group, De Veirman cannot find that deafness significantly impacted on the survival

• chances. Ingrid Olsson (1999) provides some results in her study of disabled people in 19th-century

Linköping, a town in central Sweden. Average measures of their longevity demonstrate that disabled women grew older than their male counterparts, but this did not make their mortality patterns different from the gendered death differentials outlined above. However, Olsson (1999) finds that the gendered gap in the life expectancy of 234 disabled individuals at age 15 (men: 38 years; women: 48 years) was greater than the national average gap among men (41–45 years) and women (45–48 years) at the time. These results indicate that disabled men experienced a shorter life expectancy than their female peers. Of course, being ill or disabled limits people’s health and opportunities to find work and subsistence, which might cut their life short in terms of years. However, historical research shows that individuals with disabilities confronted difficulties in life, not only with respect to a possibly less healthy status but also due to discriminatory attitudes based on social classifications and prevailing norms in contemporary society, which made them represent the ‘otherness’ in society (Jaeger & Bowman 2005; Kudlick 2003; Rogers & Nelson 2003). According to Anne Borsay (2005), disabled people in past Britain were denied citizenship through policymaking, which denied them full rights and inclusion in society, causing them to differ even more from ‘abled’ and ‘normal’

• citizens. Studies further indicate that the economic modernization in terms of industrialization

promoted disabled individuals’ exclusion from society as agricultural work and handicraft were replaced by factory work. These economic developments particularly undermined the occupational

• ptions among those who suffered from disabilities and inflicted unemployment and poverty on

them, which jeopardized their chances to fit into society (Barnes et al. 2010). Catherine J. Kudlick (2003) argues that the attitudes towards disabled people were of a broad spectrum in which policy, religion and norms played an overall role. She concludes that this was further fueled during industrialization by the capitalism system, which celebrated the able-bodied ideals of independence, self-mastery and control, and held the male, non-impaired person as the most ‘normal’. Mike Oliver (2006) asserts that the industrial process stimulated the economic changes and ideals, and built on the negative view of the disabled as a ‘social problem’. The scientific development in medicine further contributed to this view, calling for a classification of those who were ill or disabled. According to Deborah Stone (1994), this was regarded as necessary for policy reasons, in order to determine whether these individuals were to enjoy certain governmental benefits. Disabled people

SLIDE 4

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

4

thus became subject to definitions depicting them as normal or abnormal, healthy or insane and sick, which partly formed the basis for whether they should be separated from society, institutionalized and possibly cured. This made them appear even odder in society, and increased the perception of them as both a social and a medical problem. However, in his study of disabled people in 19th-century Scotland, Ian Hutchison (2007) finds that they did not have a uniform experience of marginalization simply due to having a disability. The main issue was rather whether they possessed any ability to work and could support themselves, which of course varied given their different disabilities. This makes Hutchison conclude that individuals’ disability was above all an economic concern for society and themselves.

• 3. Theoretical notions: Life course perspectives and labeling theories

Glen H. Elder (1985) was among the first to define the life course in terms of trajectories (Giele and Elder 1998; Elder, Kirkpatrick Johnson & Crosnoe 2006). With the term ‘from cradle to grave’ he reveals a line of development that includes phases such as childhood, adolescence, education, career, adulthood and parenthood, as well as old age; phases that affect individuals’ status, behavior, identity, social activities and rights in society. Getting a job or having a child, marrying or moving away – all exemplify life events that impact this line of development. But being disabled might jeopardize one’s chances to take the trajectory one would otherwise have taken, or the trajectory

• f the non-disabled. Some disabilities more than others might impede people’s chances to work,
• r entail harsh living conditions resulting in increased risk of dying. All this makes the life course

approach constitute the of basis our analysis, as it reflects people’s living conditions and helps us trace their experiences in life. In disability studies life course analyses are still rarely employed, particularly with regards to the quantitative examination of the prevalence and impacts of disability (Priestley 2003; Siminiski 2003). The infrequency of such studies concerned with the past is greatly due to poor access to historical data documenting a comprehensive number of disabled individuals and their trajectory across lifetime. According to sociological theories, the labeling concepts refer to those aspects of a person’s behavior or attributes considered deviant by society (Barnes et al. 2010). Edwin M. Lemert (1967) divides the labeling effect into a primary and a secondary type of ‘deviance’. The former concerns the perceived social reaction in a variety of social, cultural and psychological contexts, which has minimal consequences for those labeled. Secondary deviance is when the perceived label gives rise to a new social role, status and/or self-identity for the person it afflicts. Lemert (1967) argues that secondary deviance is a societal reaction of the primary counterpart. Erving Goffman (1972) explored how society categorized people through social interaction, and concluded that the ‘differences’ from normality provide the basis for deviance, which he called the ‘stigma’. According to Goffman, the notion of what is a ‘normal’ human being is based on medical grounds and normative systems in society, and thus a stigmatized person could be seen as ‘not quite human’. In Joan Susman’s (1994) overview of the concepts of disability, deviance and stigma, she contends that Goffman’s theories about stigma and deviance have not escaped criticism from other social scientists, as he treats disabled people as passive and victimized. Susman (1994) concludes that perceptions of disability influence disabled people’s life experiences, and that deviance is caused by negative perceptions and therefore evokes a negative response implying stigma. Colin Barnes et al. (2010) argue that the social construction of deviances and individual experiences of stigma raise a great deal of questions in disability studies which need to be further explored. Yet most research

SLIDE 5

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

5

indicates that disabled individuals constitute a group who do not match perceptions of normalcy, and are thus highly likely to be subject to a stigma as the theory of ‘secondary deviance’ suggests. The disabled individuals in this study are recognized by the ministers’ marks of impairment

• f the parishioners, which signify the primary deviance (Haage 2012; Rogers & Nelson 2003;

Drugge 1988). However, we focus more on the secondary deviance, which concerns the major consequences a primary label can cause (Becker 1962; Goffman 1972). As discussed above, social exclusion is often the result of labeling and secondary deviance. This is difficult to overcome because it affects individuals’ social networks and occupational options negatively, which in the longer run risks ruining their health and living conditions and may end in untimely deaths (Vikström 2008, 2011). To determine whether higher mortality risks were associated with disabled people, we use statistical life course methods to compare their death risks with a reference group

• f non-disabled cases in the same age and time-space context. If the results show significant

evidence that the disabled group confronted markedly higher mortality risks, we find it likely that this was implied not only by a possibly poor health status but also because disabled individuals were subject to a stigma associated with secondary deviance.

• 4. Research design: area, data and methods

4.1. The Sundsvall region in the 19th century

As a research area, the Sundsvall region is chosen (Figure 1). It shows a fairly representative picture

• f the population makeup and the economic structure found elsewhere in 19th-century Sweden

and North Western Europe (Tedebrand 1997). The lion part of people depended on agricultural production, eight parishes of which are included in this study (Attmar, Hässjö, Indal, Ljustorp, Selånger, Sättna, Tuna and Tynderö). In another four parishes (Alnö, Skön, Njurunda and Timrå) the socio-economic and demographic structure transformed from the middle of the century

• nwards, where the production and labor market came to be primarily based on the sawmill
• industry. Its economic side effects for business and commerce were most evident in the only urban

setting of the region (Sundsvall town), which is also subject to our study. Beside the mortality decline typical for the 19th century (Edvinsson 1994), the large influx of migrants looking for better prospects in this expanding sawmill industry explain the rapid population growth during the latter half of the century (Vikström 2003). In 1840 there were 18,793 inhabitants, a number that had increased to 46,418 in 1880 (Alm Stenflo 2004). As these contextual characteristics were present in the Sundsvall region and because its parish registers are digitized and indicate disabilities, it constitutes a sufficient ‘laboratory’ for our analysis. FIGURE 1 about here

4.2. Disability in the data

Our sources consist of digitized parish registers stored at the Demographic Data Base (DDB) at Umeå University, Sweden. It provides linked parish records composed of original records of birth and baptism, marriage, migration, death and burial, as well as catechetical examination records. From selected parishes in Sweden from the 18th and 19th centuries, the DDB registers are linked

• n an individual level and thus provide demographic data summarized for each parishioner

(Vikström et al. 2006). Sweden’s catechetical examination records are exceptional, as they were constructed on a yearly basis when the ministers fulfilled their obligation to check the parishioners’

SLIDE 6

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

6

knowledge of the catechism and their reading ability, first stated in the Church law of 1686 (Nilsdotter Jeub 2009). Defining ‘disability’ is complex because definitions work as identity markers that are not clear-cut but rather socio-culturally constructed of their time and further are individually experienced (Grönvik 2009). Scholars in the field of disability conclude that the definitions used by society originate from some dysfunction inside or outside the body (Grönvik & Söder 2009; Kudlick 2003; McCall 2005). In the parish registers, particularly the catechetical examination records, the ministers made marks regarding parishioners’ impairments (lytesmarkeringar). Although it is difficult to interpret these marks to know how severe or painful the disabilities were, they show commonly used terms in this historical context to report limits in people’s physical and mental functions (Haage 2012; Rogers & Nelson 2003; Drugge 1988). We use the ministers’ marks of impairment to identify the disabled individuals, and view the concept of ‘disability’ as also socially constructed in relation to a physical or mental status that was perceived to be ‘normal’ or ‘healthy’. The parish registers reflect these circumstances and inform us about differences distinguishing ‘normal’ or ‘able’ parishioners from others who the ministers recognized as ‘disabled’ due to their behavior, ability or health status. This means that those who the ministers recognized as disabled in the 19th-century context are defined as such in our study. Governed by these considerations, 508 disabled individuals were categorized into disability groups (Table 1), which we modify into three groups: 1. Sensory disabilities (visual or hearing defects, being blind or deaf mute); 2. Physical disabilities (bodily defects, cripples); 3. Mental disabilities (due to ‘idiocy’, ‘insanity’).1 TABLE 1 about here In 1860 Statistics Sweden, to which the ministers annually reported demographic data on their parishioners (Sköld 2001), provided the first general guidelines for making the ministers’ notes

• n disabilities more consistent (SFS, No. 63 1860). This desire was based on the advancement in

medical knowledge and a desire to assist the government’s and authorities’ intention to trace the health status of the population in Sweden. Ministers were thus to account for parishioners who were ‘blind’, ‘deaf’, ‘idiots’, ‘insane’ or who suffered from ‘epilepsy’; but the ministers already did report information to Statistics Sweden about these disabilities and even more, such as whether someone was crippled (Rogers & Nelson 2003; Drugge 1988). However, due to the lack of instructions in the past the ministers had largely developed their own practices. Their individual attention to detail and the guidelines of 1860 affect the disability marks we came across in the parish registers, as likely did the number of parishioners each minister was to keep records on.

4.3. The dataset – the entire dataset and one subset of it

A dataset of 36,118 observations from 31,790 unique individuals was extracted to run event history analysis showing the premature mortality risks. Migrations between the parishes in the Sundsvall region explain why the number of observations outnumber the number of individuals. If people moved from one regional parish to another, a new observation of them starts. Out of all the unique cases, 508 individuals had marks of impairment in the parish registers.2 Both the disabled and non- disabled cases resided in the Sundsvall region and were 15–35 years old at observation start, out of whom the vast majority were under 20 years old. All individuals were followed over time in the parish registers to detect whether and when they died and from what, and for 18 years at the

• longest. Even though the majority did not die during the observation time, this method enables us

SLIDE 7

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

7

to detect the occurrence of untimely deaths. Hence, only death occurring between the age of 15 and 54 are accounted for in this study. Table 2 shows some descriptive statistics on distribution of the entire dataset according to the covariates (see section 4.4). TABLE 2 about here In order to prepare for the sequence analysis, we extracted a subset from the above dataset which targeted 15-year old individuals. It consists of observations of 8,874 unique cases, out of whom 117 had marks of impairments before the age of 15 or the 15th birthday at the latest (section 4.5.). The reason for selecting these young persons is that they were in the beginning of their transition to adulthood associated with central events in life, such as getting the first job, marrying for the first time and giving birth to the first child or migrating. These events are of interest to picture individuals’ life experiences that may help explain their untimely death or avoidance of it.

4.4. Event history analysis and the covariates under study

Event history analysis (EHA) is one of two methods we employ in this study to reveal variations in the individuals’ life courses over time. The method of sequence analysis (SA) is detailed below (section 4.5.). EHA models the time it takes before the event in question occurs, in this case death, and enables us to highlight distinctions and similarities to explain the death risks between disabled and non-disabled individuals on a longitudinal basis. As an analytical tool we use Cox regression, using the statistical computing environment of R (Broström 2012). Cox regression is a standard technique for modelling time to event data, such as the data at hand, and allows the study of the combinational effects of several variables, making it a suitable choice for this study (Kok 2007; Mayer & Tuma 1990; Vikström 2008, 2011). Hazard ratios estimate the covariates’ effects on the propensity to experience the event of death during the observation time. It starts when the individuals were 15–35 years old and ends when they died, migrated from the region of residence,

• r at the end of 18 years of observation time. Because the individuals included in the entire dataset

enter the study at different ages (left truncation) and the observation period is quite extended, using time to event as the time scale is not appropriate (Korn et al. 1997; Thiébaut & Bénichou 2004). To both control for the effect of age on mortality and handle the effect of left truncation, we use age as the time scale in the Cox regressions. This means that the individual’s age at observation start is the enter-value, and the exit-value is their age when observation stops due to death or right censoring. The covariate of major concern is whether the individuals in the dataset were recognized as disabled or not. To identify dissimilarities or similarities between them, they are grouped into the three categories of sensory, physical and mental disabilities. The last group also consists of cases having two or more disabilities, as at least one of the marks of impairment shows ‘idiocy’ or ‘insanity’. The non-disabled cases constitute a category of their own. The second covariate of major concern is gender, since previous research shows gendered differences in mortality. We also include another three covariates in the EHA models, which can impact on people’s mortality risks. The first one is socio-economic status, here manifested by the father’s occupation, which is grouped into three categories: lower strata, upper/middle strata and the unknown/undefined cases.3 This covariate serves as a proxy for the individuals’ socio-economic origin. Their father’s occupation was categorized into to occupational codes that researchers at the DDB have developed from the parish registers. These codes facilitate the construction of social groups ranging from higher to

SLIDE 8

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

8

lower strata, according to the classification scheme in Table 2.4 The second covariate concerns cohort by distinguishing between individuals depending on when in time they lived and are under

• bservation: during pre-industrial time (observation starts between 1835 and 1844), or during

industrial time (observation starts between 1865 and 1874). The third covariate shows the type of area the people lived in, considering whether they resided in a rural, urban or rural/industrial parish (see section 4.1.).

4.5. Sequence analysis methods

Whereas the EHA is run on the entire dataset, the SA is used for the subset. According to Abbott (1990), sequence analysis (SA) can be applied to identify: (1) typical trajectories; (2) factors contributing to shaping these trajectories; and (3) if different trajectories affect some outcome

• variable. This study refers to all these applications, more or less, as it concerns whether disabilities

(cf. factor) made people’s trajectories differ from the typical ones, and with what consequences for the mortality (cf. outcome). We start to observe all disabled and non-disabled individuals in the subset after their 15th birthday to investigate five events: first occupation, first marriage, first child,

• ut-migration (from parish) and death. The observation period is the same as for the entire dataset

(maximum 18 years), but as the individuals targeted for the SA are aged 15 at observation start we follow them at the longest until 33 years of age. All our analyses are performed in the statistical environment R, complemented with the package and toolbox TraMineR (Life Trajectory Miner for R) (Gabadinho et al. 2011; Gabadinho et al. 2015). This package makes it possible to analyze categorical sequential data to mine and visualize the life course sequences in our distinct dataset. A sequence is the chronological states that the individual in the subset holds over his/her

• lifetime. Any shift in state is due to an event or transition in life, here first job, marriage,

childbearing, out-migration or death. These events result in a sequence of different states called ‘life trajectories’. SA enables us not only to analyze single events and their transitions from one point in time to another isolated from other events, but also to study multiple events and their sequential continuity or change in the same analysis (Aisenbrey et al. 2010; Abbott et al. 2000). In this study, all the dates of events are discretized into ‘age years’ whereby the order and spacing of events within one age year is ignored. A ‘state’ describes a person that particular year of his/her

• life. The length of each state is consequently one year and the whole sequence is 18 years long

corresponding to the observation time. The first state starts at the 15th birthday and the second state starts at the 16th birthday etcetera. For example, the most common state at observation start is 0. This refers to ‘zero’ events, hence that the individuals had no occupation, were not married, and had no children. If an occupation is recorded at the age of 20, the state changes into 1 at the time point representing that age. During the individual sequences the states change if one event in the series of event we analyze occur.

• 5. Results on premature mortality risks and life trajectories

First, this section provides some descriptive statistics on the extent to which the individuals experienced a premature death during the observation time. Some recent results on the causes of death are also reported. Second, outcomes from event history analysis differentiate the mortality risks associated with in particular disability and the male gender. Third, sequence analysis helps to detect individual life trajectories that add explanation to the survival chances we find.

SLIDE 9

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

9

Table 4 shows that among the individuals labeled disabled the men have higher mortality rates than their female counterparts in all disability types. The mortality percentage is also higher for men than women among the reference group of non-disabled cases. These results echo the gendered death differentials found in Sweden at the time (see section 2). However, the mortality percentage is considerably higher among the disabled individuals and exceptionally high among disabled men compared to the non-disabled men in the reference group (6–8%). Depending on type of disability, 15–27% of the disabled men were dead at the end of observation, whereas the same percentages among disabled women varied between 8% and 17%. TABLE 4 about here

5.1. Event history analysis (EHA) of premature death risks

Results from three Cox regression models are presented in Table 4, which accounts for the entire dataset and covariates that might influence people’s propensity to die. Due to assumed death differences between the genders, Models 2 and 3 show separate outcomes for men and women, while Model 1 includes both genders. The individual’s age is compensated for, as age is used as a time scale in the regressions. A look at whether the control covariates in Table 4 influenced the propensity to die to any significant degree reveals that disability, gender, cohort and residence did, whereas socio-economic status did not. It is possible that the effect of disability differs for different categories of the other covariates. To check for this, we ran interactions between disability and all the other covariates. The only statistically significant effect found is between disability and residence; therefore, this interaction is kept in the regression models of Table 4. TABLE 4 about here

5.1.1. Disability and gender

Model 1 in Table 4, which accounts for the whole dataset, reveals that disabled people experienced significantly higher mortality risks than the non-disabled. Furthermore, the type of disability impacted on the mortality propensity. Those with mental disabilities were the most afflicted, as their hazard ratios of dying were 3.2 times higher than that of the non-disabled. Those with sensory (1.7) or physical (2.9) disabilities also show higher mortality risks than the non-disabled, but the hazard is statistically insignificant for the former cases.5 Previous research contends that men more than women suffered from an excess of mortality in Sweden (section 2), and the findings presented in Table 4 parallel this gendered pattern. Men show a significantly higher hazard ratio (1.3) to die during the observation time compared to

• women. Models 2 and 3 suggest that disability shaped the mortality risks differently within each

gender and particularly limited men’s survival. Men recognized as mentally disabled demonstrate the highest hazard ratios, 3.7 times higher than non-disabled men. Physical and sensory disabled men show 3.2 and 2.0 times higher hazard ratios, respectively, in comparison with non-disabled

• men. The results for women (Model 3) must be regarded with some caution as the model has some

proportionality problems. Yet it indicates that disability did not affect their mortality risks as much as it did for men. No matter the type of disability, disabled women show lower hazard ratios (in the range 1.2 to 2.4) than the disabled men.

5.1.2. The influence of residence, cohort and socio-economic status

SLIDE 10

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

10

Type of residence is a most obvious factor in shaping people’s mortality. Table 4 shows that this significantly impacted on mortality in all three models. Type of residence did not rule out the gender differences, however, as men but not women consistently show the highest mortality hazards no matter where they resided. Men living in urban or rural/industrial areas were 2.3 and 1.3 times as prone to die, respectively, compared to men in rural settings. The same pattern is found among their female counterparts, although they show lower hazard ratios (1.7 and 1.2). All results of the impact of residence were statistically significant. The outcome of the interaction between disability and residence shows a hazard ratio of close to 0.3 in both Models 1 and 2 for mentally disabled people residing in rural/industrial parishes. This means that men with mental disabilities in the coastal area, which underwent industrialization during the latter part of the 19th century, ran lower mortality risks than the main effects proposed by the model. For women (Model 3), there was no such statistically significant interaction effect. The covariate of cohort, indicating the time period in which the individuals lived, had a statistically significant effect on the mortality risk for men (Model 2) and women (Model 3), but was insignificant in Model 1 covering the whole dataset. A comparison of men (Model 2) and women (Model 3) suggests that there were reversed period effects as regard the mortality risks between the genders. During the pre-industrial period (Cohort 1) men experienced a higher risk of dying (1.1) compared to men living in industrial time (the reference). Women experienced a lower risk of dying (0.8) in the pre-industrial period compared to industrial time. These period effects in mortality hold true for both disabled and non-disabled individuals. Even though we found no statistically significant interaction effect between disability and cohort, this mortality result indicates that industrial production, here exemplified by the sawmill industry, implied difficulties for men with disabilities but not disabled women. The socio-economic status, represented by the father’s occupation, had insignificant effects

• n the death risks regardless of people’s disability, gender or any other demographic characteristics.

This parallels the findings of many mortality researchers, who found no correlation between social status and mortality until the late 19th century (Bengtsson & Dribe 2011; Edvinsson & Lindkvist 2011; Edvinsson & Broström 2012; Razzel & Spence 2006). In all, our EHA results suggest that both gender and disability affected the mortality to a substantial and significant extent. The highest risks of dying are found among male individuals with disabilities.

5.1.3. Death causes by disability

Now, what did those who passed away while we observed them die from? As the causes of death are in general under-reported in Swedish parish registers and often document symptoms rather than actual death causes, this information is difficult to interpret (Edvinsson 1994). However, it appears that this under-registration is less the case among the individuals we study, probably because their premature death was less expected than among infants and elderly, who made up the mortality majority. As for the entire dataset, we find that the ministers made notations that inform us about death causes in about 60% of the total number of death cases (N=2,702), albeit to a lower extent (50%) in the small group of disabled people (in 50 out of 99 cases). Table 5 shows the relative distribution of the most frequent death causes among them and the group of non-disabled according to the ICD-10 classification system (ICD 10 2004). Although there are no statistically significant differences between the two groups, a couple of things stand out. Apparently, the non- disabled people suffered more from infectious diseases (e.g. tuberculosis, typhoid fever, measles, diarrhea) than did the disabled (26% vs. 17%). Furthermore, the external death causes (e.g. murder,

SLIDE 11

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

11

suicide, violence, drowning, accidents, poisoned) reveals that disabled people were not subject to dramatic death causes to any exceptional extent. TABLE 5 about here

5.2. Sequence analysis of the life trajectories

The below sequence analysis concerns the subset (N=8,874) covering the 15-year-old individuals extracted from the whole dataset. The findings will show how the trajectories of young adults developed across their lifetime regarding the five life events we examine (work, marriage, parenthood, migration, death), which may add explanation to the above mortality risks. Figure 2 displays entire life sequences that indicate the occurrence and timing of these events during

• bservation by plotting the state distributions by time points (here age) per gender and disability

(Gabadinho et al. 2011). The graphs give an aggregated view of the general pattern for all life trajectories – or sequences – among the individuals by group. FIGURE 2 about here According to Figure 5 the share of men who did not get any job and remained unmarried and did not get any child (green-colored state) is considerably larger among the disabled men than their non-disabled peers. At the age of 28, the slope for disabled men levels off at about 20% while it for non-disabled men continues to decrease. That the share drops in a slightly slower pace and delayed in time for disabled men, suggests that men who left the ‘green’ state due to occupation, took up work later in life if being disabled. The proportion of men who held occupation but did not experience marriage or parenthood (yellow-colored state) is similar despite disability. Death (orange-colored state) is the only state where the disabled men were consistently proportionally greater than the non-disabled. The development over time for the state equal to occupation, marriage and family formation (purple-colored state) shows no large differences between men based on disability. However, this similarity is partly due to that a greater proportion of non- disabled men migrated from the parish (grey-colored state), which limit our longitudinal study of possible life events. While only about 24% of the disabled individuals, both men and women, departed from their parish during observation, the share of migrants were substantially larger among their non-disabled counterparts (53% of the women and 41% of the men). Figure 2 displays a more irregular trajectory pattern for disabled women, due to a smaller number of cases and higher number of states, because women and not men were associated with illegitimate offspring. The state equal to having a child without marriage and occupation (blue- and

• range-colored states) and the state representing death (red-colored state) show higher shares of

disabled than non-disabled women. In all, the graphical view of the state distributions clearly illustrates that both gender and disability influenced people’s life trajectories. FIGURE 3 about here

5.2.1. The occurrence and order of events by end of observation

Figure 2 clearly shows that the events of dying and moving out are special in that they break the completeness of the sequences possible to observe for us. We now turn to the occurrence and

• rder of events among the 15-year individuals possible to follow until age 33, which provides a

SLIDE 12

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

12

summarized picture of the events they experienced (Aisenbrey et al. 2010). This allow us to further answer the question whether disabled people differed from their non-disabled counterparts with regards to the three events of getting an occupation, marrying and/or having a child. Figure 3 shows results where the variations between groups are statistically significant.6 FIGURE 3 about here Figure 3 shows that about 30% of the disabled men and almost 40% of the disabled women did not experience any of the events we study (state 0). This means they did not hold any occupation during observation time, neither did they marry nor got a child. Only about 10% of the non- disabled individuals ended up in that state (0). Regardless of disability, considerably more men than women took up an occupation but stayed unmarried and childless (state 1) during observation. Getting a job, marrying also giving birth to a child (state 111) was the most common male trajectory, particularly among non-disabled men (58%) compared to disabled men (40%). An evident share of the women experienced marriage and childbearing without having any occupation reported while we observe them (state 110).7 However, this particular state (110) was twice as big among the disabled women (40%) than among their non-disabled sisters (20%). While only about 7% of the latter women gave birth to illegitimate offspring during observation (states 10 and 11), almost 20% of the disabled did.

• 6. Summary and concluding discussion

The overarching aim of our study was to advance the knowledge on whether and how disabilities impeded people’s opportunities in history. First, we investigated whether disability implied higher premature mortality risks for individuals and whether these risks differed across disabilities and between men and women. Second, we checked whether the death causes varied substantially if disability was present in life. Third, the life trajectories of disabled and non-disabled people were

• utlined to find clues as to why their lives eventually ended in untimely death or not. Making use
• f micro-level longitudinal parish registers digitized by the Demographic Data Base (DDB), Umeå

University, we employed event history analysis and sequence analysis on a 19th-century population comprising more than 30,000 individuals in the Sundsvall region, Sweden. This section sums up the major results on how disability shaped humans’ life and death, followed by a discussion on why we end up with these particular findings. The outcomes of the event history analysis (Cox regression models) demonstrate that disability significantly increased the risk of experiencing premature mortality and to a profound degree, no matter of gender. However, men did to a much higher degree than women if disability was part of life. We further found that type of disability shaped the mortality propensity. It was lower for persons with sensory disabilities and substantially higher for those with mental disabilities. This variety in survival suggest that disabled people comprised a heterogenous collection of people having different obstacles and opportunities in life. However, there were a great many similarities in the survival between them and their non-disabled peers. When accounting for residence, cohort and socio-economic status, both men and women living in the town of Sundsvall suffered from the ‘urban penalty’ resulting in higher mortality propensity, disabled or not (Kearns 1988). This ‘penalty’ was also at work in the parishes which underwent industrialization, probably because they echoed the environments of urban areas. When investigating the effect of socio-economic status, there was no impact on the death risk of neither the disabled nor the non-disabled men and women.

SLIDE 13

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

13

Our preliminary examination of the death causes showed that a lower share of the disabled people died from infectious diseases compared to their non-disabled peers. This may suggest that the former led lives in isolation and were lesser exposed to such diseases. This notion of isolation is further supported by the fact that few disabled people died from external death causes, perhaps due to low participation in the labor market, social life and society at large. Yet they ran far higher risks to meet an untimely death than did the non-disabled. To get further clues to this issue we made use of sequence analysis to investigate how the life trajectories of 15-year old individuals developed until age 33. The outcomes yielded a comparatively complete picture of how disabled individuals moved through life as young adults. With regards to work and family, we find substantial differences between their trajectories and those of the non- disabled individuals. Whereas the latter often experienced several of the events under study while we observe them (job, marriage, parenthood and migration), disabled people did to a lower extent. Even if it was not impossible for them to find occupation, considerably fewer of both the men and women worked or married a spouse as compared to their non-disabled peers; and if they did, they did later in life. In all, the life sequences clearly demonstrate that disability impeded individuals’

• pportunities in both the labor and marriage markets. These findings suggest why disability resulted

in premature mortality, but there are two more explanations with regards to gender. First, even if young disabled men find work and income, this did not render many of them earnings to eventually afford a family, it seems. Second, when comparing women’s trajectories one result is particularly indicative for that disability was associated with vulnerability in life. That no less than one disabled women in five gave birth to illegitimate offspring may be a consequence from them being sexually abused. Then, why do we come across the above results? Although impairment certainly brought difficulties in life, it must not per se have ruined people’s health or opportunities to find work and access to income or a spouse to marry. It takes more to explain why these typical events were less present in the life trajectories of disabled people and to their premature mortality. Beside the impairment, we argue that a stigma and secondary deviance associated with being labeled disabled adds to explain our findings. If individuals’ appearance, behavior or inability were perceived as deviant according to prevailing perceptions about normalcy in society, this would render exclusion from social and working life. We think that these labeling circumstances and negative attitudes contribute explanation to the limited opportunities to work and family and low survival among the disabled people we find. Less work or no job cut their income and subsequently their marital prospects, as we have seen from the sequence analysis. Consequently, disability accumulated disadvantages over lifetime that help explain the high mortality propensity. Interestingly, this reasoning appears to be more relevant for men than women, which makes us propose yet another explanatory factor to our findings, the breadwinner ideal (Janssens 1999). The socio-cultural expectations linked to the male gender presupposed men to be independent providers. The ideal man, husband or father should provide for himself and the family by performing manual and often rough jobs, whereas the woman was primarily expected to work with household tasks and care for the children and to depend on male relatives. The gendered ideals and sex-segmented labor market associated with 19th-century society, women may have hold an advantage against their male counterparts if disabilities interfered with life.

SLIDE 14

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

14

References

Abbott, A. (1995) ‘Sequence Analysis: New Methods for Old Ideas’, Annual Review of Sociology 21, 93-113. Abbott, A. and J. Forrest (1986) ‘Optimal Matching Methods for Historical Sequences’, The Journal of Interdisciplinary History 16(3), 471-494. DOI: 10.2307/204500 Abbott, A. and Tsay, A. (2000) ‘Sequence Analysis and Optimal Matching Methods in Sociology’. Review and prospect, Sociological Methods & Research 29(1), 3-33. DOI: 10.1177/0049124100029001001 Alm Stenflo, G. (1994) Demographic Description of the Skellefteå and Sundsvall Regions during the 19th century. Umeå: Demographic Data Base. Aisenbrey, A and Fasang, A. E. (2010) ‘New Life for Old Ideas: The “Second Wave” of Sequence Analysis Bringing the “Course” Back Into the Life Course’, Sociological Methods & Research 38(3), 420-462. DOI: 10.1177/00491241093557532 Barnes, C. and Mercer, G. (2010) Exploring Disability. A Sociological Introduction – Second Edition. Cambridge: Polity Press. Becker, H. S. (1962) Outsiders. Studies in the Sociology of Deviance. New York: The Free Press Bengtsson, T. (2004) ‘Living Standards and Economic Stress’. In Life under Pressure. Mortality and Living Standards in Europe and Asia, 1700–1900, edited by T. Bengtsson, C. Campbell, J.Z. Lee, et.al., 27–59. Cambridge: The MIT Press Bengtsson, T. and Dribe, M. (2011) ‘The Late Emergence of Socioeconomic Mortality Differentials: A Micro-level Study of Adult Mortality in Southern Sweden 1815–1968’. Explorations in Economic History 48: 389–400. doi: 10.1016/j.eeh.2011.05.005

• BiSOS. (1907) A. Befolknings-statistik. Statistiska Central-Byråns underdåniga berättelse för år 1900. Tredje
• afdelningen. Stockholm: SCB

Borsay, A. (2005) Disability and Social Policy in Britain since 1750. Hampshire: Palgrave Macmillan, Hampshire Broström, G. (2012) Event History Analysis with R, Boca Raton: CRC Press, 2012. De Veirman, S. (2015) Breaking the silence: the experiences of deaf people in East Flanders, 1750-1950: a life course

• approach. Dissertation. Ghent University.

Demographic Data Base (DDB), Umeå University, Sweden, Digitized parish registers and catechetical examination records from the parishes in the Sundsvall region Dribe, M. (2000) Leaving Home in a Peasant Society. Economic Fluctuations, Household Dynamics and Youth Migration in Southern Sweden, 1829-1866. Dissertation, Lund University. Drugge, U. (1988) Om husförhörslängder som medicinsk urkund. Psykisk sjukdom och förståndshandikapp i en historisk källa. Umeå: Scriptum nr 8. Rapportserie utgiven av forskningsarkivet vid Umeå universitet Edvinsson, S. (1994) Den osunda staden: Sociala skillnader i dödlighet i 1800-talets Sundsvall. PhD diss. Umeå: Umeå University Edvinsson, S. and Lindkvist. M. (2011) ‘Wealth and Health in 19th Century Sweden. A Study of Social Differences in Adult Mortality in the Sundsvall Region’. Explorations in Economic History 48: 376–388. doi: 10.1016/j.eeh.2011.05.007 Edvinsson, S. and Broström, G. (2012) ‘Old Age, Health, and Social Inequality: Exploring the Social Patterns of Mortality in 19th century Northern Sweden’. Demographic Research 26: 633–660. doi: 10.4054/DemRes.2012.26.23 Eggeby, E. (1993) ‘Avvita, galen, sinnessvag – något om synen på mentalsjukdomar och de mentalsjuka under 1700- och 1800-talet’. Historisk Tidskrift 4: 538–581. Elder, G.H. Jr., ed. (1985), Life course dynamics, Ithaka, NY: Cornell Univ Press Elder G.H. Jr, Kirkpatrick Johnson, M. and Crosnoe. R. (2006) ‘The Emergence and Development of Life Course Theory’. In Handbook of the Life Course, edited by J. T. Mortimer and M. J. Shanahan, 3–19. New York: Springer Science+Business Media. Fridlizius, G. (1988) ‘Sex-Differential Mortality and Socio Economic Change. Sweden 1750–1910’. In Society, Health and Population during the Demographic Transition, edited by A. Brändström and L. Tedebrand, 237–272. Stockholm: Almqvist & Wiksell International Förhammar, S. and Nelson, M. C., Eds (2004) Funktionshinder i ett historiskt perspektiv, Lund: Studentlitterature

SLIDE 15

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

15

Gabadinho, A., Ritschard, G., Müller, N. S. and Studer, M. (2011) ‘Analyzing and Visualizing State Sequences in R with TraMineR’, Journal of Statistical Software 40(4), 1-37. DOI: 10.18637/jss.v040.i04 Giele, J. Z. and Elder, G.H. Jr. (1998) ‘Life Course Research. Development of a Field’. In Methods of Life Course Research. Qualitative and Quantitative Approaches, edited by J. Z. Giele and G. H. Elder Jr. 5–27. Thousand Oaks California: SAGE Publications Inc. Goffman, E. (1972) Stigma. Den avvikandes roll och identitet (trans. Richard, M). Stockholm: Prisma. English edition: Goffman, E. (1963). Stigma, Notes on the management of spoiled identity. Engelwood Cliffs, New Jersey: Prentice Hall Inc. Grönvik, L. (2009), ‘Defining disability: Effects of disability concepts on research outcome’, International Journal of Social Research Methodology, 12, 1-18 Grönvik, L. and Söder, M., (2008) Bara funktionshindrad? Funktionshinder och intersektionalitet. [Just disabled? Disability and Intersectionality] Malmö: Gleerups Utbildning AB Haage, H. (2012) ‘Identifying Disability using Nineteenth Century Parish Registers’, Paper presented at the Workshop on Categories and Concepts in Health, Medicine and Society, Umeå, March 15-17 Hutchison, I. (2007) A History of Disability in Nineteenth-Century Scotland. New York: The Edwin Mellen Press Jaeger, P. T. and Bowman, C.A (2005) Understanding Disability. Inclusion, Access, Diversity, and Civil Rights. Westport, Connecticut: Praeger Publishers Janssens, A. (1998) ‘The Rise and Decline of the Male Breadwinner Family? An Overview of the Debate’. In International Review of Social History: The Rise and Decline of the Male Breadwinner Family? Supplement 5, edited by Angélique Janssens, 1–24. Kearns, G. (1988) ‘The Urban Penalty and the Population History of England’. In Society, Health and Population during the Demographic Transition, edited by A. Brändström and L. Tedebrand, 213–236. Stockholm: Almqvist & Wiksell International Kok, J. (2007) ‘Principles and prospects of the life course paradigm’. Annales de démographie historique, No. 1. Belin. Korn, E. L, Graubard, B. I and Midthune, D. (1997) ‘Time-to-Event Analysis of Longitudinal Follow-up

• f a Survey: Choice of the Time-scale’. American Journal of Epidemiology, 145: 72–80.

Kudlick, C. J. (2003) Disability History: Why we Need Another Other, The American Historical Review 108, 763-793. DOI: 10.1086/529597 Lemert, E. M. (1967) Human Deviance, Social Problems, & Social Control. New Jersey: Prentice-Hall Inc. Mayer, K. U. and Tuma, N. B. (1990) ‘Life Course Research and Event History Analysis: An Overview’. In Event History Analysis in Life Course Research, edited by K. U. Mayer and N. B. Brandon Tuma, 3–20. Madison Wisconsin: The University of Wisconsin Press McCall L. (2005) ‘The complexity of intersectionality’ Signs 30.3: 1771-1800 Mont, D. (2007) ‘Measuring Health and Disability’. The Lancet 369:1658–1663. doi: 10.1016/S0140- 6736(07)60752-1 Nilsdotter Jeub, U. (2009) Parish Records, Digitalized Material from the Demographic Data Base. http://www.cedar.umu.se/digitalAssets/81/81164_parishrecords.pdf. Accessed December 17, 2015. Nilsson, H. (1994) Mot bättre hälsa. Dödlighet och hälsoarbete i Linköping 1860–1894. PhD diss. Linköping: Linköping University Oliver, M. (1996) Understanding Disability. From Theory to Practice. New York: St. Martins Press cop. Olsson, I. (1999) Att leva som lytt. Handikappades levnadsvillkor i 1800-talets Linköping. [Life as a cripple. The Living Conditions of the Handicapped in 19th Century Linköping.] PhD diss. Linköping: Linköping University Priestley, M. (2003) Disability: A Life Course Approach. Cambridge: Polity Press. Razzel, P. & Spence, C. (2006) ‘The Hazards of Wealth: Adult Mortality in Pre-Twentieth-Century England’. Social History of Medicine 19: 381–405. doi: 10.1093/shm/hkl048 Ritschard, G. and Oris, M. (2005) ‘Life Course Data in Demography and the Social Sciences: Statistical and Data Mining Approaches’, in R. Lévy et al. (eds), Advances in Life Course Research, pp. 283-314, Oxford: Elsevier. Rogers, J. and Nelson, M.C. (2003) ‘Lapps, Finns, Gypsies, Jews and Idiots Modernity and the Use of Statistical Categories in Sweden’. Annales de démographie historique 1 no 105: 61–79. SF SFS (Svensk författningssamling) No. 64 1859, valid from January 1, 1860 Solvang, P. (2000) ‘The Emergence of an Us and Them Discourse in Disability Theory’. Scandinavian Journal of Disability Research 2: 3–20. doi: 10/1080/15017410009510749

SLIDE 16

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

16

Siminski, P. (2003) ‘Patterns of Disability and Norms of Participation through the Life Course: Empirical Support for a Social Model of Disability’. Disability & Society 18: 707–718. doi: 10.1080/0968759032000119479 Sköld, P. (2001) Kunskap och kontroll – Den svenska befolkningsstatistikens historia. Stockholm: Almqvist & Wiksell International Stone, D. A. (1994) The Disabled State. Philadelphia: Temple University Press Susman, J. (1994) ‘Disability, Stigma and Deviance’. Social Science and Medicine 38, 15-22. Söderberg, J., Jonsson U. and Persson, C. (1991) A Stagnating Metropolis. The Economy and Demography of Stockholm, 1750–1850. Cambridge: Cambridge University Press Tedebrand, L. (1997) ‘Gamla och nya stadsbor efter 1860’. In Sundsvalls Historia Del II, edited by L. Tedebrand, 101–136. Sundsvall: Stadshistoriska Kommittén Sundsvalls Kommun Thiébaut, A. C. M., Jacques Bénichou, J. (2004) ‘Choice of Time-scale in Cox’s Model Analysis of Epidemiologic Cohort Data: a Simulation Study’. Statistics in Medicine 23: 3803–3820. doi: 10.1002/sim.2098 Vikström, L. (2003) Gendered routes and courses: The socio-spatial mobility of migrants to nineteenth-century Sundsvall,

• Sweden. Diss: Umeå University, Demographic Data Base

Vikström, L. (2010) ‘Identifying dissonant and complementary data on women through the triangulation

• f historical sources’, International Journal of Social Research Methodology 13, 211-21.

Vikström, L. (2008) ‘Illuminating the Labeling Impact of Incarceration: Life-Course Perspectives of Young Offenders’ Pathways in Comparison to Non-Offenders in Nineteenth-Century Northern Sweden’. Crime, History and Societies 12: 81–117. doi: 10.4000/chs.361 Vikström, L. (2011) ‘Before and After Crime: Life-Course Analyses of Young Offenders Arrested in Nineteenth-Century Northern Sweden’. Journal of Social History 44: 861–888. doi: 10.1353/jsh.2011.0001 Vikström, P., Brändström, A. and Edvinsson, S. (2006) ‘Longitudinal Databases: Sources for Analyzing the Life Course’, History and Computing 14, 109-128. Willner, S. (1999) Det svaga könet? Kön och vuxendödlighet i 1800-talets Sverige. [The Weaker Sex? – Gender and Adult Mortality in 19th Century Sweden.] PhD diss. Linköping: Linköping University World Health Organization (2004) International statistical classification of diseases and health related problems (The) ICD-10. Diss. World Health Organization, 2004.

SLIDE 17

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

17

Tables and figures

TABLE 1: Disability types based on the marks of impairment reported in the parish register on the population under study in the Sundsvall region.

Blind Visual defects from weak-sighted, short-sighted to blind Deaf mute Hearing or communication dysfunctions, ranges from bad hearing to deaf and from difficulties to speak or stammer to mute Crippled Physical dysfunctions, e.g. lame, limping, walking on crutches, missing body parts, hare lipped, small in size or crippled Mental disabilities Mental dysfunctions, e.g. foolish, silly, less cognizant (Mindre vetande), insane, feeble-minded or crazy Multiple disabilities Combination of two or more of above disabilities

Source: Digitized parish registers, the Sundsvall region, Demographic Data Base (DDB), Umeå University, Sweden Comments: On the categorization of mental disabilities, see BiSOS. (1907) A. Befolknings-statistik. Statistiska Central-Byråns underdåniga berättelse för år 1900. Tredje afdelningen. Stockholm: SCB.

TABLE 2. Descriptive statistics of demographic characteristics concerning the disabled and non-disabled cases under observation in the Sundsvall region 1835–1892: frequencies and percentages.

Covariates Men N= 17,909 N (%) Women N=18,209 N (%) Total N=36,118 N (%) Disability Sensory 91 (0.5) 60 (0.3) 151 (0.4) Physical 110 (0.6) 57 (0.3) 167 (0.5) Mental 109 (0.6) 81 (0.4) 190 (0.5) Non-disabled 17,599 (98.3) 18,011 (98.9) 35,610 (98.6) Cohort Pre-industrial (1835–44) 6,485 (36.2) 6,803 (37.4) 13,288 (36.8) Industrial (1865–74) 11,424 (63.8) 11,406 (62.6) 22,830 (63.2) Socio-economic Lower strata 3,875 (21.6) 3,800 (20.9) 7,675 (21.2) status Upper/middle strata 5,747 (32.1) 5,775 (31.7) 11,522 (31.9) Unknown/undefined 8,287 (46.3) 8,634 (47.4) 16,921 (46.8) Residence Rural parish 8,755 (48.9) 8,968 (49.3) 17,723 (49.1) Urban parish 3,949 (22.1) 4,007 (22.0) 7,956 (22.0) Rural/industrial parish 5,205 (29.1) 5,234 (28.7) 10,439 (28.9)

Source: Digitized parish registers, the Sundsvall region, Demographic Data Base (DDB), Umeå University, Sweden Comments: Number (N) refers to total number of observations generated from 31,790 unique individuals aged 15–35 years old: 16,181 men and 15,609 women. The longitudinal study of the pre-industrial cohort starts between 1835 and 1844, and between 1865 and 1874 for the industrial cohort. Explanations: Socio-economic categories as below:

Upper strata

• 1. Large-scale business entrepreneurs
• 2. Higher civil officials

Middle strata

• 3. Small-scale entrepreneurs in trade and industry, master artisans

and craftsmen; farmers, tenant farmers

• 4. Lower civil officials

Lower strata

• 5. Skilled laborers, craftsmen and artisans below the rank of master
• 6. Unskilled laborers in trade and industry; farmhands, crofters,

maidservants

SLIDE 18

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

18

TABLE 3: Percentage experiencing death among the cases under observation for a maximum of 18 years in the Sundsvall region 1835–1892: comparisons between the genders and types of disability.

Death Characteris istic ics o

• f t

the in indiv ivid iduals Total Per ercen centage o e of a all wi withi hin e n each h ca categ egory y All Men Women N Sensory 12.6 15.4 8.3 151 Physical 23.4 27.3 15.8 167 Mental 21.6 24.8 17.3 190 Non-disabled 7.3 8.5 6.2 35,610

Source: Digitized parish registers, the Sundsvall region, Demographic Data Base (DDB), Umeå University, Sweden Comments: See Table 2.

TABLE 4: Cox regression of the propensity to die among the disabled and non-disabled cases under

• bservation in the region of Sundsvall 1835–1892: three Cox regression models.

Covariates s sho howi wing ng disabi bility a and nd d demographi phic ch charact cter eristics cs o

• f t

the e in indiv ivid iduals Cox

• x R

Regression

• n

Mod

• del 1

1 Both genders N=36,118 Cox

• x R

Regression

• n

Mod

• del 2

2 Men N=17,909 Cox

• x R

Regression

• n

Mod

• del 3

3 Women N=18,209 Hazard ratio P-value Hazard ratio P-value Hazard ratio P-value Disability – Non-disabled (ref.) – Sensory – Physical – Mental 1 1.696 2.864 3.200

• 0.097

0.000 0.000 1 1.950 3.232 3.724

• 0.079

0.000 0.000 1 1.232 1.953 2.447

• 0.719

0.183 0.019 Gender – Women (ref.) – Men 1 1.343

• 0.000
• Socio-economic status

– Lower strata (ref.) – Upper/middle strata – Unknown/undefined 1 0.973 0.909

• 0.596

0.079 1 1.032 0.858

• 0.644

0.036 1 0.886 0.961

• 0.140

0.632 Cohort – Industrial (ref.) – Pre-industrial 1 0.997

• 0.943

1 1.131

• 0.023

1 0.833

• 0.005

Residence – Rural parish (ref.) – Urban parish (Sundsvall) – Rural/industrial parish 1 1.978 1.274

• 0.000

0.000 1 2.276 1.334

• 0.000

0.000 1 1.663 1.209

• 0.000

0.009 Disability*Residence – Sensory: Urban – Physical: Urban – Mental: Urban – Sensory: Rural/ind. – Physical: Rural/industrial – Mental: Rural/industrial 0.239 1.608 1.245 0.836 0.718 0.386 0.173 0.314 0.636 0.706 0.351 0.007 0.289 1.486 1.151 0.838 0.654 0.292 0.246 0.486 0.804 0.752 0.291 0.005 0.000 1.705 1.595 0.829 0.876 0.649 0.960 0.504 0.562 0.838 0.864 0.464 Overall Global proportionality test

• 0.000

0.087

• 0.000

0.250

• 2.29e-11

0.090

Source: Digitized parish registers, the Sundsvall region, Demographic Data Base (DDB), Umeå University, Sweden Comments: See Table 2. The rural/industrial parishes are Alnö, Skön, Timrå, Njurunda. Socio-economic status is based on the father’s

• ccupation. The hazard ratio column shows estimates of the relationship between the levels of the categorical variable with regard

to the instantaneous risk of the event. The hazard ratio of the reference category is always set to 1. If the estimated hazard ratio of

• ne category is above 1, its impact is higher than that of the reference category, e.g. 1.5 means 50% higher hazard. If the p-value is

below 0.05, it demonstrates that the impact is statistically significant with a 5%-level of significance.

SLIDE 19

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

19

TABLE 5: Relative distribution of frequent death causes by disease groups among individuals who died during observation: comparison between the disabled and non-disabled group.

Characteris istic ics o

• f t

the in indiv ivid iduals Total Disea ease ca e categ egories es Di Disabled (%) Non

• n-disable

led (%) N Infectious diseases 17.2 26.4 705 Respiratory diseases 8.1 7.9 214 External death causes 7.1 10.0 266 Other diseases 17.2 17.2 465 Unknown/unspecified 50.5 38.5 1,052 Tot

• tal %

% Total ( (N) 100 (99) 100 (2,603) 100 (2,702)

Source: Digitized parish registers, the Sundsvall region, Demographic Data Base (DDB), Umeå University, Sweden Comments: Disease categories according to the ICD-10-Code Classification where only the most frequent types of disease causes reported for individuals are accounted for.

FIGURE 1: Map of Sweden and the Sundsvall region showing the parishes included in the study.

Source: Demographic Data Base (DDB), Umeå University, Sweden

SLIDE 20

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

20

FIGURE 2: Relative distribution of states by time points per gender and disability during observation (from 15 to 33 years of age).

Source: Digitized parish registers, the Sundsvall region, Demographic Data Base (DDB), Umeå University Explanations:

SLIDE 21

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

21

FIGURE 3: Relative distribution of states at end of observation by gender and disability after 18 years of

• bservation (from 15 to 33 years of age).

Source: Digitized parish registers, the Sundsvall region, Demographic Data Base (DDB), Umeå University Explanations: 0 = No occupation/No marriage/No child 100 = No occupation/Marriage/No child 1 = Occupation/No marriage/No child 101 = Occupation/Marriage /No child 10 = No occupation/No marriage/Child 110 = No occupation/Marriage/Child 11 = Occupation/No marriage/Child 111 = Occupation/Marriage/Child

SLIDE 22

Vikström et al, ‘On how disabilities impede individuals’ life opportunities’

22

1 The concepts of disabilities used in this paper are those commonly used by 19th-century society. While some

concepts may be offensive to some readers due to the derogatory meaning they carry today, we have no intention to

• ffend. The problem of using concepts which can be apprehended as offensive has been discussed (Eggeby 1993).

2 This means that the non-disabled people in our study were non-blind, non-deaf mute, non-crippled, non-idiot and

non-insane. Non-disabled individuals could have other marks in the registers, such as ‘sick’.

3 We selected the father’s occupation at the start of the observation of the individuals under study or immediately

before the start. Socio-economic status is divided into three categories because of small numbers in some of the groups.

4 The DDB classification does not completely correspond to the two commonly used classification schemes in

historical studies, SOCPO and HISCLASS, but there are many similarities between them; for a comparison between the schemes, see the Appendix in Edvinsson and Broström (2012).

5 With a significance level of 5%. 6 Pearson’s Chi-squared tests show P-value=3.63e-07 for men and P-value=9.15e-12 for women. The groups in the

test are defined by a contingency table relating disability status (disabled or non-disabled) to the eight possible end states.

7 The absence of occupation among the women must not imply that they did not work, but is primarily due to poor

documentation of their actual work in the sources. Similar to historical population registers in general, Sweden’s parish registers under-report women’s occupations (e.g. Vikström 2010).