Livelihood of Street Children and the Role of Social Intervention: Insights from Literature Using Meta-Analysis

As studies done by different scholars indicate that the present status of street children is remarkably insightful, this invites us to systematically review the existing literature by using meta-analysis. In this paper 31 studies were reviewed by applying a predetermined set of inclusion and exclusion criteria as part of meta-analysis. These studies were compiled mainly from three continents (Africa, Asia, and Latin America), which are often observed to be diversified economically, politically, socially, and environmentally. Empirical evidences based ondata generated from reviewed studies provide a holistic picture on the predominance of male street children among a total sample size of 68014 street children. Working as a daily labourer is considered as the most predominant informal occupation for street children. Empirical evidences suggest that majority of street dwellers were categorized into children working on the street in Africa, while in Asia a sizable proportion of them were abandoned from their house. Interestingly, it suggests that children coming to a street may be due to push factors like coercion by family, lack of access to education, and the existence of displeasing life in Africa, while in Asia children were pushed by family to beg and act as a day labourer and street vendor to assure the livelihood of their abandoned families. Statistical evidence based on odds ratio suggested no association between location of the study region and the characteristics of street children. Analysis of variance results showed that there exists a significant variation within a continent for all four variables (living condition, education, gender, and livelihood strategies). In fact, daily labour was the most acceptable means of livelihood earning and it is followed by street vendor and others. The present condition of street children necessitates social intervention to address the present problems of street dwellers by ensuring sustainable livelihood options among them.


Introduction
The enormous problems faced by the street children in the last several years were stated by different scholars in different times and topic of diverse philosophy [1,2].The problems are widely dispersed (psychological, physical, and sexual abuses), but for several years they were not in societies' programme [3].They are excluded from economic, social, and political process.However, only few institutions are involving in helping them to sustain their livelihood.This may lead the future of hopeful children as well as the coming generation to uncontrollable bad habits, which have an impact on political (peace), social (norms, culture), and economy (on GDP, GNP) conditions of these countries.
Even though the problem of street children is understood as an urban phenomenon, the factors exacerbating the problem are originated, by and large, in the rural villages.Rural-urban migration to urban towns across the countries of the world (a brief overview of migration of children has been documented in a report prepared by Whitehead and Hashim [4]) is not dominated by a single factor but caused by a combination of multiple interrelated push and pull factors [5,6].It is usually in response to the deterioration of the living conditions in rural villages.Most of these children are compelled to work or live on the street in order to sustain themselves and their families.For many of them, the perceptions that larger towns offer greater economic opportunities make the street a more attractive destination compared to a poverty stricken rural economy [7].However, the life waiting in the city is often difficult.They often do not have education and basic skills necessary to deal with the risk factors and cope with adversity.In this context, it is expected that social intervention by the government or nongovernment organizations would address the problem of livelihood opportunities of the street children.
It is well known that the NGOs in every sphere of the world seek to provide a reasonable solution for the economic, social, and environmental problems of the societies and thereby try to fulfil the gap that could not be served by local government or state.Tracing back to the historical importance of NGOs, Epstein [8] found that the origins of NGOs were northern and southern part of USA.The northern NGOs were established after the end of First World War and at the begging of Second World War they extended their scopes throughout the world.de Benitez [9] illustrated the role of NGO Committee on UNICEF (the committee comprises a worldwide network of 125 international NGOs working on behalf of children in more than 110 countries of the world) in providing basic needs of street children throughout the world.While evaluating "broad based initiative" implemented by UNICEF for destitute children, it was found that this programme encourages children to achieve their primary school education, endorses physical and psychological development of children, protects against endemic and chronic disease, and last of all serves children to have a positive impact on their life.Besides these international organizations, some other local organizations are also working for the interest of street children in different countries of the world [3,10].A social intervention group in Mexico, EDNICA, is working not only for providing them with basic needs but also for imparting different training and economic support to street children [10].In a similar intervention in Ethiopia, NGOs in cooperation with the Bureau of Labour and Social Affairs address the socioeconomic problems of street children in Hawassa City of Ethiopia [3].Even though empirical results suggested that majority of children were excluded from the purview of NGOs working in the area, but it has been expected that these organizations can promote the welfare of disadvantaged children by planning, financing, managing, and providing advice and counselling services.The majorities of these programmes were designed on the basis of the principle of "curative approach" in short run and thereby neglect "preventive approach" based on long term solution.Under this backdrop, this paper uses meta-analysis to explore the livelihood options available to the street children and the role of social intervention in protecting their livelihood opportunities across countries of the world.

Conceptual Framework
In this twenty-first century, it is well known that the street children are facing a lot of physical and mental problems throughout the world.In this context, it is desirable to design an appropriate road map to address such problems.The conceptual frame work is designed to identify the causes of the problems of street children and the role of social enterprise intervention in overcoming such problems (Figure 1).
Several causes of becoming street children, such as the interplay of push and pull factors (environmental, social, and health factors) and the lack of educational access, have been identified in the existing literature [3,24,34].Social interventions are expected to promote wellbeing of the children through job skills, service linkages to local resources, income generating activities, and engagement.Unlike the earlier outreach model, this alternative model of social intervention has certain far-reaching implications, namely, socioeconomic (increased employment, income, savings, job skills, service utilization, housing stability, and social and labour networks) and physical or mental health related outcomes (increased self-esteem, motivation, and quality of life and decreased high risk behaviours) [46].

Collection of Data.
In this paper, a systematic attempt has been made to review the existing empirical literature from countries of Africa, Asia, and Latin America through the lens of meta-analysis.The search was done as follows.Scholarly written published documents in English were searched from Google Scholar and archive (like JSTOR) during September 2015-December 2015.During the searching time, phrases like "street children"; "homeless children"; "children on, of, abandoned on the street"; and "social intervention" were commonly used.
In the process of executing meta-analysis as presented in Figure 2, the study collected altogether 158 documents on street children.Inclusion and exclusion of studies were performed on the basis of some predetermined criterion in the studies: (i) age of the children being below 24; (ii) appropriate representation of sample size, which provides information on gender composition, educational background, livelihood sources, type of street stay, and average age; (iii) the studies written and published in English language; (iv) the study providing amenable data; and (v) the study employing quantitative or mixed research method.
Then from 158 collected documents, in the first step 19 studies were excluded due to redundancy, 42 articles because of having qualitative nature only, and 10 articles due to being written in regional languages.Finally 87 articles were screened.But still there was also vagueness in defining the age of street children in 22 studies and ultimately 65 potential articles were identified.By applying other exclusion criterion, 34 studies were further excluded: 9 studies were excluded due to lack of data accuracy and 25 of them do not have the required variables.Finally 31 studies were considered for our study on this meta-analysis.The data compiled from 31 articles were written in the context of three different continents (Asia, Africa, and Latin America) of the world.Among 31 articles, a majority of 18 studies explored the situation of street children and their problems of livelihood in the context of east, south, and western portions of Africa; 10 studies were in the context of southern part of Asia; 3 studies were from the southern part of America.
In the selected list of 31 articles, three articles dealt with the role of social intervention in addressing the problems of street children.To provide a meaningful analysis on social intervention, we have collected eight new studies, which are exclusively based on social intervention.Thus altogether we have eleven studies to explore the role of social intervention in addressing the problems of street children.

Data Analysis.
Historically meta-analysis was known and applied in health studies.But nowadays it is being familiar in many fields in social sciences like economics and business, as well.In this paper we have used meta-analysis, which is considered as the application of a set of statistical tools and a process used to combine the results of different studies [47,48].The following statistical tools have been employed to examine cross-country variations in the salient characteristics of the street children.
Odds Ratio and Funnel Plot.In a systematically review process using meta-analysis, we are often familiar with odds ratio (OR) and its diagrammatic presentation, funnel plot, to measure the association between two nominal variables.In our study we have defined odds ratio as follows: Odds ratio = Odds of street children characteristics for Group I (say, Africa)/Odds of street children characteristics for Group II (say, Rest of the world).
The confidence interval of the odds ratio has been calculated by following the same procedure as that used in the confidence interval for a population mean or population proportion, that is, Point estimate (ln(OR) in original scale) ± Margin of errors, where Margin of error = Critical value of the statistic (standard normal distribution at 95% confidence interval, i.e., 1.96) * Standard error.
In order to calculate ln(OR), we have converted ln(OR) value measured in the natural log scale into original scale by using exponential function.If the 95% confidence interval for the OR does not contain 1.0, then it can be concluded that there is a statistically significant association between the characteristics of street children and the location of the study area (i.e., continent).
Funnel plot is carried out to test the presence of publication bias by considering confidence interval of the odds ratio.It usually represents confidence interval of the odds ratio on the horizontal axis and category variable chosen from review studies on vertical axis.

Analysis of Variation (ANOVA).
ANOVA is one such statistical tool used to partition the total variation present in the sample data (collected from different studies across three continents) into a number of components associated with the nature of classification of data.For simplicity, we have classified sample data from different studies into two broad categories: the studies relating to Africa and the studies relating to rest of the world.In fact, data arising from a series of  independent studies are divided into  disjoint groups (here two groups: Africa and rest of the world, i.e.,  = 2) on the basis of predefined independent studies in Africa and rest of the world.The number of estimates on the characteristics of street children in the groups is denoted by   , where  = 1, 2, . . ., .An estimate calculated from each study is denoted by   with variance V  .The group mean effect estimate for the th group ( 0 ) can be written by where the weight   is the reciprocal of the variance of   ; that is, In most of the underlying studies, variance is not reported.As an alternative, we have started with our analysis by considering equal weight to all studies under consideration.Then the paper assigned weights to the studies.In this analysis, we have conveniently replaced the unknown variance by the estimated variance, which is a function of the within-study sample size and the effect size estimate.However, we have carried out the analysis by considering both equal and unequal variances.
The grand mean is given by To test that there is no variation of the population estimates on the characteristics of street children within the groups of studies in a continent, we have used  test statistic.The calculated value of  test statistic within a group is given by Under the null hypothesis of no variation of the population estimates,   test follows  2 distribution with (  − ) degrees of freedom.
Similarly we have carried out a test on the variation between groups, that is, variation on the characteristics of street children between continents.The test statistic   is given by Under the null hypothesis   test statistic follows  2 distribution with ( − 1) degrees of freedom.
Finally, total variation   can be computed as

Results and Discussion
4.1.Demographic Characteristics.Altogether, our ultimate sample of 31 studies exposed the presence of 68014 street children in three different continents.Table 1 briefly presented the demographic profile of the street children.It has been seen that 71.84 percent (48863) of them were boys and the remaining 28.16 percent (19151) were girls in the composition of street dwellers.In this composition a less percentage of young women were observed.This may be the reflection of the nonresponse bias of the hidden nature of street girls' activity for sustaining livelihood in the street [3].
In regard to the educational background of children, no variation across continents (specially in Africa and Asia) was noted.The reason for high record of poor educational background of the street children was that they often got used to the street life in their early years of life [35].In Africa among 25755 children in the sample size studied, 23.28 percent of them were illiterate and the residual 50.65 percent were literate.And we observed 27.37 percentage gap in the literacy level of the street dwellers in Africa.A similar gap in literacy (26.02) was revealed in Asia.But studies surveyed in Latin America exhibited the lowest gap in literacy level (6.73).This gap in literacy specially in Latin America necessitates a social intervention in promoting education status of the street children; otherwise the problem of illiteracy may drastically increase in the future.

Classification of Street Children and Living Condition.
Following UNICEF classification, street children can be categorized into children working on the street (on the street), children living on the street (of the street), and children from street families (abandoned) [49].Accordingly data collected from reviewed literature as shown in Table 2 showed that, out of 25755 street children in Africa, 37.55 percent of them work daytime on the street and return to their home in the evening; 11.10 percent were living on the street on a whole time basis.The number of abandoned children was extremely low.Among those children living on the street, 38.15 percent live with their family and 11.96 percent live alone or with their friends.
Ten selected studies on Asia indicated that a majority 60 percent of street children (out of 41650 street children) were living on the street with their family.Unlike other continents, Asian studies showed that parents in low economic status migrated to urban street corners with their children [1,32] and thereby faced several challenges on street due to unsafe environment in which they live [36,50].Due to insufficient number of studies in Latin America, no systematic conclusions can be drawn.However, no significant variation exists in the percentage share of the children working on the street vis-a-vis living on the street.
A comparison on the natures of street children in three continents suggested that in Africa majority of street dwellers were categorized as living on the street due to the existence of push factors like motivation of family members, unfavourable educational environment, and lack of basic needs, while in Asia children were pushed to begging and act as daily     labourer and street vender to ensure the livelihood of their abandoned families.Thus, in both the continents, children often sacrificed to secure the life of their hidden parents living in their home in Africa and unhidden parents living in the street in Asia.

Livelihood of the Street Children.
To examine the diversity of livelihood options exercised by the street children, existing literature provided information on three broad categories of informal activity: street beggar, street daily labourer, and street vendor.In this approach, even though we have reviewed 31 studies from three continents, no information on the livelihood diversification was observed for as many as 15 studies.Most specifically, studies in Africa and Latin America reported no quantitative information on livelihood.As stated in Table 3, out of the total sample size of street children in African countries, few of them supported themselves and their families through begging (1.06%) and street vending (4.13%) and the other 7.68% engaged on daily labour.In Latin America only 69 respondents among a sample size of 609 street children reported their alternative means of livelihood opportunities.Considering total sample size of the street children, it has been exhibited that 0.66% of them were involved in begging, 2.13% were in daily labour, and 8.54% perform different kinds of other duties.
Interestingly, 99% of the respondents in Asia participating in 10 different studies reported their various means of livelihood options; 8.18% of street children were involved in begging, 30.18% were on daily labour, and 11.06% of them occupied street vending, and a majority of 49.55% engaged in different types of other duties.In particular, majority of street children working as daily labourer were involved in hardship duties in industries.They were compelled to work as daily labourers in hazardous environment of industries.A study done in Kolkata showed that nearly 38% of a total of 468 street children work as daily labourers in a hazardous environment [31].

Social Intervention on the Street Children.
We have reviewed eleven articles to identify the role of social intervention in addressing the problems of street children.Among the selected articles, eight articles have newly been incorporated to explore the role of social intervention.In addition, we have made a sincere effort in searching number of street children served by these organizations from their respective websites.However, from the reviewed articles, it has been seen that a total of 18 NGOs and one community service association were working in these three continents (Table 4).In Africa, particularly in Ethiopia, there are a number of NGOs working for street children.Among them, 12 NGOs were working exclusively in Ethiopia.According to data revealed from four articles in Ethiopia 359,960 children were supported by these 12 NGOs.It has been found that Abebech Gobena Orphan and School, a local NGO in Ethiopia, supported 2,50,832 street children through provision of education.

Confidence Interval of Odds Ratio: Funnel Plot.
In the study we have chosen a priori classification of street children in the existing literature, that is, studies relating to Africa and studies relating to the rest of the world.Depending on four individual characteristics of street children (namely, living condition, literacy level, gender composition, and livelihood opportunity), case group and control group have been identified for Africa and the rest of the world.Accordingly we have computed odds ratio for the case group and control group and presented them in Table 5.For example, children not living in the street fall in the category of control group, while those living in the street were included in the case group.Empirical evidence suggests that the odds ratio of living condition for Africa compared to the rest of the world is 1.75, thereby indicating almost equal odds of living condition in Asia and the rest of the world.Moreover it is observed that the confidence interval (95%) of the odds ratio does include the value 1.0.Figure 3 shows the diagrammatic presentation of the confidence interval in a funnel plot.Thus, it tells us of no significant association between living condition of the street children and the continent considered in the studies.In other words, the location of the study has no significant effect on the living condition of the street children.Interestingly, two other criteria of street children (namely, gender and literacy) are not found to be associated with the location of the study area.However, the higher value of odds ratio in the category of livelihood condition (i.e., 2.957) suggests the increasing odds of livelihood condition in Africa.In other words, the probability of daily labour as the predominant source of livelihood condition in Africa is found to be higher compared to the rest of the world.The association of location of the study region and the livelihood condition of the street children is statistically significant.

Variation in the Characteristics of Street Children across
Continents: ANOVA.The study attempts to examine the variation within a group and between the groups (Africa and rest of the world) in the framework of ANOVA.Based on four criterions (living condition, education, gender, and livelihood strategies) of the street children the variations within a continent and between continents are identified.Cross-country variation on the characteristics of street children in a world map is presented in Figure 4.In the framework of ANOVA, at first, we have started with assigning equal weights to all studies (Table 6) and then presented results with unequal weights to studies based on the observed standard deviation (Table 7).Accordingly, the results show that there exists a significant variation within a continent for all four criterions in both the unweighted and weighted analysis.But there is no significant mean difference between two groups (Africa and rest of the world) except in education.This indicates that average percentage of literate street children in Africa is higher compared to the rest of the world.This may be possible due to the active involvement of Africa NGOs in the promotion of literacy campaign for the

Conclusions
This paper attempts to systematically review the existing literature by using meta-analysis on the problems of livelihoods of street children across countries of the world and the role of social intervention in addressing such problems.
In the process of executing meta-analysis, the study collected altogether 158 papers through searching at Google Scholar and JSTOR.Then, from those collected articles, finally 31 studies were considered for this meta-analysis study.By applying inclusion and exclusion criteria, we have selected a total of 31 articles for the purpose of collecting information on the salient characteristics of street children (e.g., demographic profile, livelihood options, living condition, and types of social intervention) from three continents (namely, Africa, Asia, and Latin America) of the world.In fact, these 31 studies covered 68014 street children in three different continents.The sample data from these three continents were processed to provide meaningful insights on the livelihood opportunities of three categories of street children (children on, of, and abandoned on the street).The sample data were 0.001276 SE 0.035719 Upper limit in the confidence interval in LN scale 1.15432 Lower limit in the confidence interval in LN scale −2.26119 Upper limit in the confidence interval in original scale 3.171867 Lower limit in the confidence interval in original scale 0.104226 then analyzed by using statistical methods like descriptive statistics, odds ratio, and ANOVA.Empirical evidences based on meta analysis provides a clear trend of male dominance in the gender composition of street children.A sizable number of street children are categorized as children on the street (rather than children of the street or children abandoned from their family) and they work as a daily labourers for their livelihood.However, the results of ANOVA show the presence of significant variation within a group (i.e., continent) in regard to the different characteristics of street children.Interestingly, the study does not support the presence of significant variation on gender, livelihood, and living condition between the continents (i.e., Africa and rest of the world).However, literacy is observed as

Figure 1 :
Figure 1: Conceptual framework on the problems of street children and the role of social intervention.

= 10 Figure 2 :
Figure 2: The flow diagram on selection of studies for meta-analysis.

Figure 3 :
Figure 3: Funnel plot on the confidence interval of different characteristics of street children.
Percentage of literate street children Percentage of girls children Percentage of street children living on the street Percentage of street children working as daily labourer

Figure 4 :
Figure 4: Cross-country variation in the characteristics of street children.

Table 1 :
Demographic characteristics of the street children identified from studies for review.

Table 2 :
Category of street children and their living condition identified from studies for review.

Table 3 :
Livelihood opportunities of the street children identified from studies for review.

Table 4 :
Profile of Social Interventions in the Literature on Street Children.

Table 5 :
Odds ratio on the different characteristics of the street children.

Table 6 :
ANOVA results for the different criteria of street children from 31 studies (unweighted).

Table 7 :
ANOVA results for the different criteria of street children from 31 studies (weighted). 1 (Critical Value) *  2 (Critical Value) *   (Critical Value) * *Note.* Significant at the 0.05 level of significance.