1Following women’s enfranchisement in New Zealand (1893), Australia (1902), and Finland (1906), a growing number of countries allowed women to vote on equal terms with men (Orman, 1918; Smith, 2008). The gradual expansion of women’s enfranchisement was considered a landmark in the political democratization of Europe, but less attention has been paid to the voting restrictions that remained, especially those that affected older people. Although no countries explicitly excluded individuals on the grounds of a maximum voting age (Blais et al., 2001), other limitations may have served to restrict suffrage for older citizens. As legal limitations tell us surprisingly little about their practical significance at the population level, population studies are needed to assess the proportion disqualified from voting for various reasons.
2Our study examines disenfranchisement attributable to poor relief that occurred in certain countries, including Finland and Sweden. This is one of the less studied restrictions to voting (Central Statistical Office, 1909; Harjula, 2009; Swedish Parliament, 2015). In principle, Finnish men and women from all social classes were granted the right to vote in 1906. However, regular recipients of poor relief were disenfranchised for almost 40 years, gaining the right to vote as late as 1944 (Harjula, 2009). The extent to which the poor-relief restriction caused disqualification from voting among various population subgroups has not been examined before.
3Using internationally unique microdata [1] on over 19,000 poor-relief recipients covering two historical provinces, we estimate the proportion of voting-age citizens disqualified from voting in the 1911 Finnish parliamentary elections due to receipt of poor relief. Furthermore, the data allow us to examine differences in the prevalence of disenfranchisement between men and women, and across age groups.
I – Voting restrictions
4The election system, including stating who has the right to vote, is a key democratic institution. It has implications for the political structure of the elected representatives and their policy choices (Engerman and Sokoloff, 2005). While universal suffrage is currently considered as an undeniable standard for an inclusive democracy (Young, 2010), certain voting restrictions still remain. For example, convicted felons and mentally deficient persons are excluded from voting in many countries today (Blais et al., 2001).
5Three common justifications for disenfranchisement are non-membership, incompetence, and lack of autonomy. The first criterion refers to membership of the political community and a personal stake in the election outcome. It may be related to qualities such as citizenship, residency, religion, tax-payment, or property ownership (Katz, 1997). The disqualification of convicted criminals is sometimes also included under this category (Blais et al., 2001). The second criterion of “incompetence” encompasses restrictions based on low age, mental health, literacy, and gender. The third criterion refers to the exclusion of individuals who are considered to have a limited will of their own, such as persons in custodial care and those dependent on public relief (Katz, 1997).
6Prior literature has indicated that the rationale behind the exclusion of Finnish poor-relief recipients was based on on their expected inability to make an independent voting choice due to their dependency on public welfare (Harjula, 2010). This restriction was criticized because of its unfairness to citizens in the later stages of their lives who were the most concerned by poor relief. Nevertheless, the leading legal scholars of the time argued that the universal nature of suffrage was intact, even if some citizens were disenfranchised for specific reasons (Harjula, 2009).
7The practice of excluding poor-relief recipients illustrates the belief that when the need for basic maintenance could not be met within the private sphere by individuals or their families and thus became the responsibility of a public institution, a citizen’s status was reduced. Harjula (2010) has noted that the Finnish case of poor relief portrayed the complex relationship between social and political citizenship, as persons who received social assistance were denied their political rights.
II – Institutional context
8Finland was an autonomous part of the Russian Empire in 1911 (until 1917), but only Finnish citizens were granted universal suffrage in Finland (Harjula, 2009). [2] The minimum voting age was 24 for both men and women (Tarasti, 2006).
9Regular poor relief was not the only grounds for parliamentary disenfranchisement among voting-age Finns. Other grounds included exclusion from the official population registry (henkikirja), loss of civic confidence, being under guardianship, being in regular military service, bankruptcy, non-payment of taxes, and election crime. Furthermore, citizens who were sentenced to the workhouse because of vagrancy were barred from voting for three years (Harjula, 2009). However, these other causes of disqualification fall beyond the scope of this study.
10The poor-relief system was designed to help individuals who could not support themselves or their families, but were not directly responsible for their troubled situation. Legally, poor relief was restricted to children with no one to provide for them and to adults who could not work because of disability, illness, or old-age frailty. However, the primary responsibility for helping these individuals rested with the family. If no family support was available, the public sector could provide poor relief (Jaakkola et al., 1994; Pitkänen, 1994). Regular reliance upon poor relief was often the result of an inability to work and sustain basic maintenance, which led to the loss of civic autonomy, which in turn caused disfranchisement.
III – Data and methods
1 – Historical microdata on poor-relief recipients
11This study is based on the original poor-relief microdata of 31 December 1910, located in the National Archives of Finland. The historical files are organized alphabetically within each province. We identified the year of birth, gender, the type of poor relief received, and the location of residence for almost all persons who received poor relief in the province of Turku and Pori, and the province of Häme. This information had been collected for administrative purposes since 1892.
12Given the enormous amount of work that would have been needed to assess demographic profiles for the entire country, we limited our study to two out of eight provinces. The province of Turku and Pori was chosen because of its size, and Häme because of the presumed quality of its poor-relief data. Furthermore, we wanted to avoid using information on provinces that were considered more international, such as the province of Uusimaa, including the capital, and that of Viipuri, near Saint Petersburg, in order to minimize the risk of including foreign citizens in our poor-relief data (Central Statistical Office, 1914). This was important in order to avoid numerator-denominator bias. The age-sex structures in both provinces were similar, and the population of the chosen provinces was quite representative of the demographic structure of the Finnish population as a whole (Central Statistical Office, 1915).
13The microdata include 19,184 poor-relief clients, a number which corresponds well with the 19,259 persons in the published tables of the time (Central Statistical Office, 1913). The published tables could not be used in our study, as they do not give the exact age of poor-relief recipients or the type of relief received. This information is essential because regular poor relief led to exclusion from suffrage, whereas temporary poor relief did not (Central Statistical Office, 1909; Harjula, 2010). Furthermore, published poor-relief statistics classify persons aged 15 and over as adults, while the minimum voting age was 24 (Central Statistical Office, 1913; Tarasti, 2006). For these reasons, disenfranchisement could only be studied using the original forms completed by hand. Permission to use them was obtained from the National Archives of Finland (permission no. AL/13780/07.01.03/2012).
14Of the 19,184 poor-relief recipients in our microdata, 19.8% were categorized as non-regular or temporary poor-relief recipients, and these were excluded from our analyses. We also excluded ten individuals whose poor-relief type was unknown. This brought the number down to 15,382, of whom 28.2% were below the minimum voting age of 24, and 1.5% had an unknown year of birth. Of the remaining 10,807, approximately 35.3% were men and 64.7% women. The mean age was 60.6, and its standard deviation 17.4. We further aggregated the microdata of poor relief by sex, province, and age group (24, 25-29, …, 80-84, 85+) in order to assess its magnitude by proportion of the population.
15In our data, the four types of regular poor relief were the following: i) residence in almshouses, ii) maintenance in a private person’s home, iii) maintenance by turns in freeholders’ farms, and iv) regular poor relief in one’s own home, for example, in the form of money or grain (Central Statistical Office, 1913). The four categories were mutually exclusive. The official policy favoured the establishment of almshouses (Helsingius, 1897; Pitkänen, 1994), and maintenance by turns in freeholders’ farms had become rare by 1910 (Central Statistical Office, 1913).
2 – Population size
16We linked the aggregated poor-relief microdata with a published parish registry of a corresponding population size to calculate the proportion of those who were disenfranchised because they were regular poor-relief recipients. The published (de jure) population sizes in the historical provinces on 31 December 1910 are categorized by five-year age groups (Central Statistical Office, 1915). We therefore had to assume that the number of 24-year-olds was one-fifth of that of the 20-24 age group. This assumption is unlikely to bias our results, since for the Finnish population as a whole, approximately 1 in 5 people in the 20-24 age group was aged 24 (Central Statistical Office, 1915). The total size of the voting-age population in the provinces of Turku and Pori, and Häme was approximately 410,794, of which 48% were men and 52% women. The voting-age population covered in the study represented 26.5% of the total voting-age population of Finland (Central Statistical Office, 1911; Central Statistical Office, 1915).
3 – Dates of the elections
17The Finnish parliamentary elections were held on 2 and 3 January 1911, and practically all the regular poor-relief recipients who were alive on 31 December 1910 must have been alive on the voting days. The closeness of the dates is important, as the mortality rate of poor-relief recipients is known to be higher than that of the total population (Statistics Finland, 1920). If the time between the gathered information and the voting day is long, the relationship between old age and poor-relief disenfranchisement could become overestimated, because the oldest recipients may not have survived until the elections. The 1911 election was the first to allow analysis, as previous elections since 1906 were not held in January (Tarasti, 2006).
4 – Statistical methods
18We used grouped data, Pearson’s chi-square test, and Poisson models to examine disqualification from voting. The data were grouped by sex, five-year age groups, and province of residence (Table 1). We employed Poisson models because they are commonly used for analysis of grouped data on the number of events in relation to the total number of persons or person-years (Loomis et al. 2005; Schmidt and Kohlmann, 2008). In cross-sectional studies, Poisson models can be used to provide prevalences, prevalence ratios (PRs), and 95% confidence intervals (CIs) (Behrens et al., 2004).
19We first assessed the prevalence ratio of poor-relief disenfranchisement by sex, independent of age and province. Men were used as the reference group. Second, we examined disenfranchisement by age, separately for men and women. Poisson modelling was used to test whether the prevalence of disenfranchisement in one age group was significantly different from that of the previous age group and whether it differed by sex at various ages. The results are presented as prevalences (Figure 1). The parameter estimates are shown in Appendix Tables A.1 and A.2. [3]
Characteristics of the voting-age population in the Finnish parliamentary elections of January 1911, the provinces of Turku and Pori, and Häme, Finland

Characteristics of the voting-age population in the Finnish parliamentary elections of January 1911, the provinces of Turku and Pori, and Häme, Finland
IV – Older people and women more likely to be disenfranchised
20Approximately 2.6% of the voting-age population was disenfranchised in the Finnish parliamentary elections of 1911 because they received regular poor relief (1.9% of men and 3.3% of women, Table 1). Women were more likely to be disenfranchised for this reason than men, independent of age and province (PR=1.56, 95% CI: 1.49, 1.62).
21Although the proportion of disenfranchised individuals was relatively modest in the voting-age population as a whole, its magnitude increased substantially with age (Table 1). Pearson’s chi-square test shows that such observed differences are unlikely to have arisen by chance since the differences are statistically significant. Exclusion from suffrage due to poor relief was only between 0.4% and 2.2% among 24-59-year-olds, versus 5% at ages 65-69 and as much as 15% at ages 75-79. By age 80, almost one in four citizens had lost their right to vote in parliamentary elections because they received regular poor relief.
22Figure 1 shows the prevalence of disenfranchisement by sex and age. With few exceptions, the exclusion from suffrage due to poor relief increased substantially with age among both men and women. Approximately 4% of men and 6% of women were excluded at ages 65-69, rising to 13% of men and almost 17% of women at ages 75-79. By age 80, almost 20% of men and over 25% of women were disenfranchised due to receipt of poor relief. Women were significantly more likely to be excluded at all ages, except at age 24.
Prevalence (%) and 95% confidence intervals of poor-relief disenfranchisement by sex and age in the Finnish parliamentary elections of January 1911

Prevalence (%) and 95% confidence intervals of poor-relief disenfranchisement by sex and age in the Finnish parliamentary elections of January 1911
* Relative difference from previous age group at the 5% significance level.V – Discussion
1 – Summary and interpretation of the results
23This is the first study to assess the relationship between sex, age, and parliamentary disenfranchisement due to receipt of poor relief in a European nation with universal suffrage. Using internationally unique microdata, we showed that poor-relief disenfranchisement increased substantially with age. Furthermore, the findings indicate that prevalence of poor-relief disenfranchisement was higher among women than men at all ages, except at the youngest age of 24. While the poor-relief disenfranchisement law may have seemed gender neutral, its impact on disqualifying citizens from voting was stronger among women than men. Poor relief was a concrete mechanism that transformed economic and social disadvantage into a form of exclusion from political life. In that respect, the voting restriction could be regarded as an indicator of social discrimination.
24There are several possible reasons why women were more likely to receive regular poor relief than men. Although all explanations go beyond the evidence of our historical microdata, it seems plausible that women received poor relief more often because they were more likely to have major financial difficulties, arising in some cases from single parenthood for example. Women became single parents if they bore an illegitimate child – approximately 6-7% of Finnish children were born out of wedlock between 1900 and 1910 (Pitkänen and Jalovaara, 2007) – or were widowed. Indeed, historical statistics show that women were more likely to be widowed than men in all age groups (Central Statistical Office, 1915), and widowhood has been linked to the use of poor relief (Smith, 1984). Among contemporary populations likewise, widowhood has been found to contribute to female over-representation in care homes (Einiö et al., 2012). In addition, despite their higher life expectancy (Koskinen and Martelin, 1997), older women may have experienced more health impairments than men. This argument is in line with a historical British study in which elderly women aged 65 and over were more often visually impaired than elderly men in 1891, 1901, and 1911 (Grundy, 1997), and with Finnish studies which found that visual impairment was more common among women than men in 1901 (Harjula, 1996). Contemporary studies have also shown that older women are more likely to experience functional impairments than men of the same age (Arber and Cooper, 1999).
25It seems reasonable to assume that the findings on age relate to the physiological processes of ageing and their detrimental effect on the capacity of older adults to work and sustain basic maintenance. Historical statistics from England and Wales show that certain impairments such as blindness, due in many cases to cataracts, increased substantially with age (Grundy, 1997). The findings on age also might reflect the fact that many adult children were incapable of providing sufficient care and financial support for their elderly parents, who were therefore forced to rely on poor relief. Although residual life expectancy at age 65 was already almost 11 years for Finnish men and 12 years for women at the beginning of the twentieth century (Koskinen and Martelin, 1997), old-age pensions and health insurance were practically non-existent (Harjula, 2009; Kuhnle, 1981).
26During the early twentieth century, Finland lagged behind other Nordic countries, including Sweden and Denmark, in terms of industrialization and the development of insurance and pension schemes (Kuhnle, 1981), and poor relief was practically the only source of social security for older adults before both the First and Second World Wars (Niemelä and Salminen, 2006). Sweden, for its part, introduced a universal old-age pension scheme in 1913 that covered almost the entire population (Kuhnle, 1981) by the time full suffrage was introduced in 1921 (Smith, 2008), but continued to disenfranchise poor-relief recipients until the mid-1940s (Berling Åselius, 2005; Swedish Parliament, 2015). The practical significance of the poor-relief disenfranchisement law among older citizens has not been internationally documented, however. It is reasonable to assume that its impact may have been greater among older people in Finland than in Sweden. Future studies should empirically test this hypothesis, as similar types of legal restrictions on suffrage can imply disenfranchisement among different proportions of population subgroups depending on other welfare options.
2 – Contribution and limitations
27The originality of this study is that the voting-age population was identified using parish registers which covered all citizens, regardless of their economic activity. It is noteworthy that poor relief data were originally recorded for other administrative purposes (e.g., evaluating trends in poor relief) and not for assessing restrictions to voting rights. Harjula (2009, 2010) has suggested that neither the categories of poor relief nor the practice of exclusion from voting were definitive or coherent at the local level. As a result, even those receiving temporary poor relief could occasionally lose the right to vote, despite the fact that only regular poor relief was defined as a criterion for exclusion in the legislation. Our microdata therefore provide an approximate picture of poor-relief exclusion. Another limitation to our analysis concerns variables that were not available, such as marital or disability status.
Conclusion
28Using unique historical microdata, we have shown that the prevalence of poor-relief disenfranchisement increased with age. It was also more common among women at all ages. While the poor-relief disenfranchisement law may have seemed demographically neutral, its impact was especially strong among older men and women. The key issue here is not only that these people were old, but that they also lacked financial resources and family support. Without adequate pension systems to provide financial security, they were denied their previous right to vote. To summarize, our empirical findings clearly verify the age-discriminatory character of Europe’s first implementation of universal suffrage in an era of limited pensions and point to the existence of gender inequality in disenfranchisement attributable to the receipt of public welfare support.
Acknowledgements
This study was supported by the Kone Foundation and the Academy of Finland (grant 273433).Appendix tables
Parameter estimates for Poisson model of poor-relief disenfranchisement by sex, age, and province, Finnish parliamentary election of January 1911

Parameter estimates, prevalences and prevalence ratios from Poisson models of poor-relief disenfranchisement by sex and age, Finnish parliamentary election of January 1911

Parameter estimates, prevalences and prevalence ratios from Poisson models of poor-relief disenfranchisement by sex and age, Finnish parliamentary election of January 1911
(a) Women aged 60-64 years are used as the reference group.(b) Previous age group of the same sex is used as the reference group.
(c) Men of the same age are used as the reference group.
* Relative difference from the reference group at the 5% significance level.
Interpretation: prevalence (%) for 80-84-year-old women: exp{-3.2525+1.8983}*100=25.8% (as in Figure 1).
Notes
-
[1]
To our knowledge, this is the only pre-WWI dataset on poor-relief recipients in a country with universal suffrage.
-
[2]
Russian citizens were not allowed to vote in the 1911 Finnish elections, but the Russian emperor had the right to dissolve Parliament (Meinander, 2010).
-
[3]
Stata software was used for the analyses.