1 Text summary of review search and extraction process
1.1 WHAT we search for and WHY
We are undertaking a systematic scoping review mapping out the current academic state of the art for policies explicitly or implicitly aimed at reducing inequalities in the world of work. To arrive at a mapping which is as unbiased as possible, we closely follow the scoping review methodology proposed by Arksey and O’Malley (2005) and extended by Pham et al. (2014), as well as those points from the Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA, 2020) guidelines that are applicable to scoping reviews.
The goal of this review strategy is to capture any possible coherent mechanisms of policy-making which positively influence the reduction of inequalities in the world of work and reach some measure of clarity on the extent and robustness of evidence in these areas.
ACCOMPANYING SLIDE Section 2.1
To achieve this, we follow the general typologies formulated by ILO for the world of work to find a definition of work, various forms of work and labour market adjacent outcomes to be measured. These provide one ‘cluster’ of terms for the world of work which will later be used for the actual search strategy. A second cluster is provided by terms considering a variety of terms posing a definition of a policy, as well as possible implementations of it in an institutional perspective, a structural perspective and the perspective of personal or collective agency. The last cluster is made up of terms defining the concept of inequality, and terms describing dimensions of vertical and horizontal inequalities.
These three clusters, taken together, then describe all domains which are of special interest in finding policies to reduce inequalities in the world of work.
ACCOMPANYING SLIDE Section 2.2
They, together with specific inclusion criteria, provide the semantic baseline for the search: the terms and concepts for which studies will first be searched and later included or excluded from consideration in the actual search implementation.
1.2 HOW we search
ACCOMPANYING SLIDE Section 2.4
The scoping review process itself consists of three consecutive steps: identification, screening and extraction.
The goal of the identification process is to allow a wide-breadth net to be cast, including all possibly relevant material published on the topic to be identified and included in an initial sample pool of sources. To facilitate this process, two search methods are used:
ACCOMPANYING SLIDE Section 2.5
Boolean-based searches of the extended World of Science corpus is the primary source of database literature. Then, relevant existing reviews are identified within and outside of this sample and used for a ‘snowballing’ technique, which adds all sources mentioned within those reviews, and any overlapping citations from those sources in turn. Through this identification technique, a wide breadth of sources are identified, generally covering all relevant literature making up the current state of the art.
ACCOMPANYING SLIDE Section 2.6
In the second scoping review step of screening, sources are systematically considered for their relevance from far-reading to close-reading techniques: first, sources are considered only based title; next, based on abstract; and only those sources remaining are screened for full-text relevance for the final extraction.
The criteria applied within each screening step are the aforementioned inclusion/exclusion criteria, which are repeatedly applied in each step. A source which can not unambiguously be assigned to any one of the criteria which would exclude it will be left included in the sample pool for another closer-reading consideration at a later step, until the full-text review.
Alongside the final full-text screening step, sources are assigned specific keyword ‘tags’ to ease their later organisation into domains of policies, inequalities, as well as countries and regions with the help of the internal keyword abilities of the Zotero reference manager.
1.3 HOW we extract from it
Extraction with the help of the extraction tool follows a strict grid of relevant data to extract from each source identified and screened as relevant. The extraction data to be pulled from each relevant source can be categorized into 4 overall dimensions: publication data, contextual data, results, and statistical data.
ACCOMPANYING SLIDE Section 2.7
Publication data captures the relevant information to uniquely identify the study under review, as well as identify its publication type and location. Results capture the primary of findings of a study, along with any suggested channels or mechanisms of operation, the theory they are basing them on if provided, and any limitations. They also capture the type of intervention under review as well as the types of inequalities, as well as any specific outcome measures.
Contextual data represents all information given within the source as to the intervention’s respective required contexts: what the primary country (or countries) of analysis were, which world region and country income class they represent, which target group was targeted by the intervention if any, which dataset the study at hand made use of if provided and importantly when and for how long the intervention and its study were undertaken.
Statistical data captures all study findings of statistical relevance: its sample size, level of representativeness, significance, but and if it is using relative or absolute indicators. It also captures the design (whether it was undertaken experimentally or observationally) and the methods used by the study.
2 Bullet-points / Possible Slides
2.1 Term cluster areas
- world of work cluster:
- Definition of work
- Forms of work
- Labour market adjacent outcomes
- policy cluster:
- definition of policy
- institutional implementations
- structural implementations
- implementations of personal/collective agency
- inequality cluster:
- definition of inequalities
- vertical inequalities
- horizontal inequalities
2.2 Inequalities term cluster example
| Defintion | Vertical inequality | Horizontal inequality |
|---|---|---|
| inequality | income | identity |
| barrier | class | demographic |
| advantaged | Gini | gender |
| disadvantaged | Palma | colour |
| discriminated | Theil | beliefs |
| disparity | Atkinson | racial |
| horizontal inequality | log deviation | ethnic |
| vertical inequality | fertility | migrant |
| bottom percentile | spatial | |
| top percentile | rural | |
| urban | ||
| small cities | ||
| peripherial cities | ||
| age | ||
| nationality | ||
| ethnicity | ||
| health status | ||
| disability | ||
| characteristics |
2.3 Inclusion criteria
| Parameter | Inclusion criteria | Exclusion criteria |
|---|---|---|
| Time frame | study published in or after 2000 | study published before 2000 |
| Study type | primary research | opinion piece, editorial, commentary, news article, literature review |
| Study recency | most recent publication of study | gray literature superseded by white literature publication |
| Study focus | inequality or labour market outcomes as primary outcome (dependent variable) | neither inequality nor labour market outcomes as dependent variable |
| policy measure or strategy as primary intervention (independent variable) | no policy measure/strategy as intervention or relationship unclear | |
| specifically relates to some dimension of world of work | exists outside world of work for both independent and dependent variables | |
| focus on dimension of inequality in analysis | no focus on mention of inequality in analysis |
2.4 The Search Process
- identification: create a sample of all possibly relevant sources from current literature
- screening: separate irrelevant from relevant sources and map source characteristics
- extraction: pull out the relevant outcomes for which data were sought based on extraction tool from relevant studies only
2.5 Identification
- Database sources:
- World of Science (Extended Corpus): white literature
- Google Scholar: possibly relevant grey literature
- Snowballing:
- starting from existing relevant reviews contained in initial database results & additional relevant ones
- all contained citations extracted and added to identified sources
- Deduplication:
- automated deduplication for exact source matches
- manual deduplication for inexact matches or superseding literature
2.6 Screening
- Repeated sorting out of irrelevant literature:
- repeated process from far-reading to close-reading (title, abstract, full-text)
- using inclusion/exclusion criteria for each round of separations
- Pre-sorting of literature with keyword tagging:
- reason for exclusion (title, abstract, superseded, or note for full-text reason)
- types of inequalities, types of policies, types of outcomes measured
2.7 Extraction
- Extraction of relevant data for current review:
- from all sources identified as relevant during full-text screening
- using extraction tool to unify extracted data
- Relevant data:
- Publication information (author, year, title, publisher, publication type)
- Contextual data (country, region, income class, period & length of analysis, target group, dataset)
- Results data (intervention, inequalities, outcome measures, main findings, channels, limitations)
- Statistical data (sample size, representativeness, methods, significance, absolute indicator)