feat(script): Add first draft Discussion and Conclusion

manuscript/article.qmd

@@ -961,7 +961,7 @@ The identified literature rises in volume over time between 2000 and 2023,
 with first larger outputs identified from 2014 onwards,
 as can be seen in @fig-publications-per-year.
 While fluctuating overall, with a significantly smaller outputs 2017 and in turn significantly higher in 2021,
-the overall output volume strongly increased during this period.
+the overall output volume increased throughout this period.
 
 ```{python}
 #| label: fig-publications-per-year
@@ -1161,8 +1161,180 @@ Another reason could be the actual implementation of different policy programmes
 
 # Discussion
 
+```{python}
+#| label: discussion-prep-inequality-df
+#| echo: false
+# dataframe containing each intervention inequality pair
+df_inequality = (
+    df[["region", "intervention", "inequality"]]
+    .assign(
+        Intervention = lambda _df: (_df["intervention"]
+            .str.replace(r"\(.+\)", "", regex=True)
+            .str.replace(r" ?; ?", ";", regex=True)
+            .str.strip()
+            .str.split(";")
+        ),
+        inequality = lambda _df: (_df["inequality"]
+            .str.replace(r"\(.+\)", "", regex=True)
+            .str.replace(r" ?; ?", ";", regex=True)
+            .str.strip()
+            .str.split(";")
+        )
+    )
+    .explode("Intervention")
+    .explode("inequality")
+    .reset_index(drop=True)
+)
+```
+
+Turning to the available studies from a perspective of inequalities,
+@tbl-inequality-crosstab breaks down the individually targeted inequalities per intervention type.
+
+```{python}
+#| label: tbl-inequality-crosstab
+#| tbl-cap: Intervention types by the inequalities targeted
+
+df_temp = df_inequality.loc[
+    (df_inequality["inequality"] == "income")
+    | (df_inequality["inequality"] == "gender")
+    | (df_inequality["inequality"] == "spatial")
+    | (df_inequality["inequality"] == "disability")
+    | (df_inequality["inequality"] == "ethnicity")
+    ]
+df_temp = df_temp.rename(columns={"inequality": "Inequality"})
+tab = pd.crosstab(df_temp["Intervention"], df_temp["Inequality"],
+                  margins=True).reindex(["income", "gender", "spatial", "ethnicity", "disability"], axis="columns").sort_values("income", ascending=False)
+del df_temp
+tab
+```
+
+Most studies focus on some indicator of income inequality within national or regional contexts.
+The second most analysed inequality is that of gender,
+followed by spatial inequalities, disabilities, ethnicities, age, inequalities of migration, education and intergenerational issues.
+
+Only a small amount of studies carry analysis of inequalities in the world of work surrounding migration,
+generational connections, age and education into the world of work.
+Age-related inequalities prominently factor into studies as an intersection with disability,
+in focusing on the effects of older people with disabilities on the labour market [@Kirsh2016].
+Studies that solely or mainly target age-related inequalities themselves often do so with a stronger focus on the effects on seniors' health outcomes and long-term activation measures,
+with some extending into the effects of differentiated pension systems.[^pension-studies]
+Equally, for migration few studies can strictly delineate it from racial inequalities or considerations of ethnicity.
+
+[^pension-studies]: Studies focusing on the effects of pensions themselves often do not intersect back into outcomes within the world of work. For an overview on pensions and health effects, see @VanDerHeide2013; for pensions and other intersectional inequalities, see for example @Zantinge2014.
+
+Thus, for the current state of the literature on analyses of policy interventions through the lens of inequality reduction within the world of work,
+there are remaining gaps of academic lenses for generational inequalities, age-related inequalities, educational inequalities and inequalities of migration processes when looking at the quantity of output.
+Care should be taken not to overestimate the decisiveness of merely quantified outputs ---
+multiple studies with strong risk of bias may produce less reliable outcomes than fewer studies with stronger evidence bases ---
+however, it does provide an overview of the size of evidence base in the first place.
+
+<!-- crosstab prevalences -->
+<!-- gender -->
+Looking into the prevalence of individual interventions from the gender inequality dimension,
+the crosstab shows that interventions on paid leave, subsidies, collective bargaining, and education received the most attention.
+Thus a slight preference towards institutional and structural interventions is visible,
+though the dimension seems to be viewed from angles of strengthening individual agency just as well,
+with subsidies often seeking to nourish this approach,
+and training, and interventions towards financial agency also being represented in the interventions.
+
+<!-- spatial -->
+Interventions affecting spatial inequalities are often also primarily viewed through indicators of income.
+Interventions aiming at reducing spatial inequalities primarily base themselves on infrastructural changes,
+which aligns with expectations of the infrastructural schism between urban and rural regions.
+Additionally, education interventions target spatial inequalities,
+with the effects of minimum wage, work programmes, interventions strengthening financial agency, trade liberalization and training playing a reduced role.
+
+This can pose a problem, as even non-spatial policies will almost invariably have spatially divergent effects which should be taken into account to avoid worsening issues:
+such as was seen in the further exclusion of already disadvantaged women from employment, infrastructure and training opportunities in India under bad targeting and elite capture [@Stock2021],
+or further deprivation of already disadvantaged regions in the UK work programme [@Whitworth2021].
+Rural communities relying on agricultural economies in particular may be vulnerable to exogenous structural shock events such as climate change, which may thus need to be a focal point for future structural interventions [@Salvati2014].
+
+The results agree with the systematic review of income, employment and poverty correlations by [@Perez2022],
+in that employment plays a significant in spatial disadvantages,
+however with different primary barriers for different spatial contexts.[^perez-interventions]
+On the other hand, as the results by @Hunt2004 have also shown,
+individual measures on their own such as commuting subsidies in this case, while having positive results,
+may not provide enough lasting impact over the long term and may need embedding in a more holistic approach,
+combining multiple policy packages.
+
+[^perez-interventions]: The identified interventions largely overlap with the identified pertinent interventions in this review: credit programmes, institutional support for childcare, guaranteed minimum income/universal basic income or the provision of living wages, commuting subsidies, and housing mobility programs. However, due to their focus on urban contexts, the identified barriers differ.
+
+<!-- disability -->
+Few studies approach disability inequalities primarily through the prism of income inequality,
+preferring return to work, employment rates or amount of hours worked as indicators.
+Only when looking at the intersection of disability and gender is income the more utilized indicator,
+through measuring female income ratios compared to those of males.
+
+Here, a split between frameworks that favour agency-based approaches,
+putting organisational barriers and environmental activation, as well as individual (re-)integration within the world of work into focus,
+and frameworks which focus on the institutional-structural component,
+with a focus on educational inclusion, and selection and eligibility criteria for benefit or vocational programmes.[^disability-approaches]
+In addition to employment or return to work based indicators it might thus be pertinent to include a focus on decent work and meaningful achievement as additional indications of successful programmes.
+Taken together, these results especially reinforce the results of @Poppen2017 and @Thoresen2021,
+for the importance of correct targeting to avoid unintended negative outcomes,
+while the evidence base also highlights research gaps in contexts and regions other than developed high-income countries.
+
+[^disability-approaches]: For exemplary frameworks in the agency perspective, see @Martin2020 and @Lindsay2015; for the latter see @Lindsay2015a and @Gruber2014.
+
+<!-- migration/ethnicity -->
+Studies on migration- or ethnicity-based inequalities predominantly focus on structural interventions such as education, fiscal policies or infrastructure,
+or the effects of institutional contexts such as collective bargaining, unionisation or universal incomes.
+The primary indicators are mixed between indicators of income inequality and others such as employment probability,
+though with a focus on absolute measures such as poverty, hours worked or debt.
+While some frameworks do put agency-driven necessities to the foreground,
+there is a consensus for structural approaches required to enable this agency.[^migration-frameworks]
+
+[^migration-frameworks]: For an agency-focused approach, see @Siebers2015; for an example of structural requirements, see @Goodburn2020 or @Samaluk2014 for a discussion of structural power dynamics; for an institutional focus, see @Clibborn2022.
+
 # Conclusions
 
+The preceding study undertook a systematic scoping review of the literature on inequalities in the world of work.
+It focused on the variety of approaches to policy interventions, from institutional to structural to more agency-driven programmes,
+and highlighted the inequalities targeted, analysed in subsequent study, their methods and limitations,
+to arrive at a picture of which lays out the strengths and weaknesses of current approaches.
+
+Wide gaps exist between the research on existing topics within the areas and intersections of inequalities in the world of work.
+First, while regionally research on such inequalities seems relatively evenly distributed,
+focus prevalence on individual inequalities varies widely.
+
+Research into interventions preventing income inequality are still the dominant form of measured outcomes,
+which makes sense for its prevailing usefulness through a variety of indicators and its use to investigate both vertical and horizontal inequalities.
+However, care should be taken not to over-emphasize the reliance on income inequality outcomes:
+they can obscure intersections with other inequalities,
+or diminish the perceived importance of tackling other inequalities themselves, if not directly measurable through income.
+Thus, while interventions attempt to approach the inequality from a variety of institutional, structural and agency-oriented approaches already,
+this could be further enhanced by putting a continuous focus on the closely intertwined intersectional nature of the issue.
+
+Gender inequality is an almost equally considered dimension in the interventions,
+a reasonable conclusion due to the inequality's global ubiquity and persistence.
+Most gender-oriented policy approaches tackle it directly alongside income inequality outcomes,
+especially viewed through gender pay gaps and economic (dis-)empowerment,
+approaching it from backgrounds of structural or agency-driven interventions.
+While both approaches seem fruitful in different contexts, few interventions strive to provide a holistic approach which combines the individual-level with macro-impacts,
+tackling both institutional-structural issues while driving concerns of agency simultaneously.
+
+Spatial inequalities are primarily viewed through rural-urban divides,
+concerning welfare, opportunities and employment probabilities.
+Spatially focused interventions primarily tackle infrastructural issues which should be an effective avenue since most positive interventions are focused on the structural dimension of the inequality.
+However, too many interventions, especially focused on reducing income inequalities,
+still do not take spatial components fully into view,
+potentially leading to worse outcomes for inequalities along the spatial dimension.
+
+Disabilities are rarely viewed through lenses other than employment opportunities.
+While most interventions already focus on dimensions of strengthening agency and improved integration or reintegration of individuals with disabilities into the world of work,
+a wider net needs to be cast with future research focusing on developing regions and the effects of more institutional-structural approaches before clearer recommendations can be given based on existing evidence.
+
+Ethnicity and migration provide dimensions of inequalities which are, while more evenly distributed regionally,
+still equally underdeveloped in research on evidence-based intervention impacts.
+Currently, there is a strong focus on institutional-structural approaches,
+which seems to follow the literature in what is required for effective interventions.
+However, similarly to research on inequalities based on disability, there are clear gaps in research
+on ethnicity and especially migration, before clearer pictures of what works can develop.
+
+The intertwined nature of inequalities, once recognized, requires intervention approaches which heed multi-dimensional issues and can flexibly intervene and subsequently correctly measure their relative effectiveness.
+To do so, perspectives need to shift and align towards a new, more intersectional approach which can incorporate both a wider array of methodological approaches between purely quantitative and qualitative research,
+while relying on indicators for measurement which are flexible yet overlapping enough to encompass such a broadened perspective.
+
 # Bibliography
 
 ::: {#refs}