New paper with Garmina Singh and Minna van Gerven, in Transfer: European Review of Labour and Research: https://journals.sagepub.com/doi/10.1177/10242589251385915

Background: The concept of resilience in familial and social contexts has gained prominence in academic and policy discussions. However, the interplay between family life and social inequalities, and how these relate to each other in the resilience literature, has yet to be documented.

Objective: This scoping review addresses this gap by analysing 250 articles published between 1998 and 2023. We compare the concept of resilience as applied in family and social resilience studies through four constitutive elements: (1) the unit of analysis, (2) definitions, (3) types, and (4) the risks, outcomes, and explanatory factors that are examined empirically.

Results: While both perspectives study individuals’ resilience, the emphasis in family resilience is on families, whereas social resilience studies focus more on communities and societies. Both perspectives emphasize the centrality of risks in defining resilience, yet family resilience scholarship seeks solutions within the family, while social resilience highlights community dynamics. Additionally, family resilience studies explore topics related to family-specific risks and resources, while social resilience studies examine external risks and resources.

Conclusions: The family resilience scholarship follows the clinical tradition in the resilience literature, viewing families as a separate entity that is resourceful and agentic. Socioeconomic risks are recurrent themes in social resilience literature, but not in family resilience.

Contribution: Understanding resilience through the lens of family inequalities in socioeconomic contexts can bridge these two perspectives. Incorporating factors such as labour market dynamics, family transitions, and educational attainment into definitions of risks, outcomes, and explanatory factors of resilience can enhance this integration.

Five blindspots in reform studies of early childhood education and care (ECEC) policy.

At breakneck speed, with a very short deadline, we delivered the work, and a report (available online). The report had some interesting findings, we think, and demonstrated again how important the accessibility, affordability and quality of childcare policies are.

However, our commissioned focus was very much on reform studies, and in particular impact assessments with very strong research designs aim to isolate the causal effect. The value of such studies is indisputable, but a tradeoff is that these reforms studies tend to be empirically narrow. We kept wondering: “What are we missing?”.

So, we reflected on five blindspots:

Blindspot A: The context-specific nature of reform studies.

The research design of reform studies explicitly seeks to exclude the influence of ‘confounding’ or contextual factors. Consequently, such studies generally focus on a single country, and are less able to explain how to increase effectiveness depends on its interplay with other institutional and structural conditions.

Blindspot B: The limited focus on reducing inequality in the use and benefits of ECEC.

It is well established that parents with higher levels of education and income are more likely to enrol their children in ECEC. Yet, reform studies are silent on the degree of cross-country inequality in ECEC use. Consequently, which aspects of ECEC policy lead to an increase or decrease in inequality in ECEC use remains unknown.

Blindspot C: The focus on short-term outcomes.

A substantial share of studies in the ECEC reform database examined the immediate and short-term effects of policy changes. Only a few reform studies included in the database have examined how long it takes for implemented policy reforms to have an effect.

Blindspot D: The focus on individual-level rather than macro-level outcomes.

Reform studies focus on individual-level outcomes, which facilitates causal inference but overlooks higher-order outcomes and thus the relationship between ECEC and important societal developments.

Blindspot E: Various forms of publication bias.

Very few studies reported on ineffective reforms, and only a handful of studies examined fathers.

Countries known for extensive provision of ECEC were overrepresented. This means that the evidence base for the revision of the Barcelona targets might be the weakest for those countries that might be furthest away from achieving them.

In the conclusion, we critically reflect on our own role in taking on this commissioned work, the importance that policy makers involve academics already at the design stage of the tender/commissioned work, and the importance of methodological pluralism.

Alm, S., Nelson, K., & Nieuwenhuis, R. (2019). The Diminishing Power of One? Welfare State Retrenchment and Rising Poverty of Single-Adult Households in Sweden 1988–2011. European Sociological Review, 0(0), 20.

This discussion paper provides an updated analysis of gendered economic inequality in high- and middle-income countries. A review of the literature demonstrates that such an analysis needs to explicitly recognize that gender, poverty, and (economic) inequality are intrinsically linked. Specifically, the paper addresses two sets of questions:

• First, how do intra-family resource allocation and distribution patterns both reflect and shape gender inequalities in power and well-being, and what factors—including policy-related ones—can mitigate these inequalities?
• Second, how do families as gendered institutions contribute to broader socio-economic inequalities, and what can be done to reduce/reverse these inequalities?

Using data from the LIS Database, this paper shows considerable differences among 42 countries with respect to how likely women were to have their own income. The period from 2000 to 2010/2014 saw increasing rates of own incomes as well as women’s incomes constituting larger shares in total household income. A key finding is that, in countries where many women have an income of their own, relative poverty rates are lower.

The comparative analyses, combined with a review of the literature, suggest that welfare state arrangements that support working women not only improve the overall employment rates of women but also help to prevent particularly women in low-income households from living in dependence and instead to have an income of their own—thus reinforcing the potential for poverty reduction. Moreover, institutional contexts that are generally conducive to women’s employment tend to be effective across family forms.

I was a great pleasure to work with the excellent LIS team on this project, and an honour to prepare this report commissioned by UN Women (the United Nations organization dedicated to gender equality and the empowerment of women). This paper was produced for UN Women’s flagship report, “Progress of the World’s Women 2019”.

Abstract

It is demonstrated that family policies are an important aspect of the institutional context of earnings inequality among coupled households. Although seldom integrated into prominent analyses of economic inequality, women’s earnings are consistently found to reduce relative inequality among households. This means that family policies, as well-known determinants of women’s employment and earnings, are important contextual determinants of economic inequality. Using Luxembourg Income Study data from 18 OECD countries in the period 1981–2008, this study demonstrates that women have higher earnings, and that their earnings reduce inequality among coupled households more in institutional contexts with generous paid leave and public childcare. We found no sizeable association between financial support policies, such as family allowances and tax benefits to families with children, and the degree to which women’s earnings contribute to inequality among coupled households. Family policy arrangements that facilitate women’s employment and earnings are associated with less economic inequality among coupled households.

Nieuwenhuis, R., Need, A. & Van der Kolk, H. (2018). Family policy as an institutional context of economic inequality. Acta Sociologica. Forthcoming, online first: http://journals.sagepub.com/doi/full/10.1177/0001699318760125

This is quite a lengthy process, and although I wrote my application before last summer, I thought it would be interesting to share the main body of my application. In true Swedish fashion where the application and the reviewer reports will be made public anyway. It describes four research lines that I have developed.

I am truly grateful of the support I continue to experience at the Swedish Institute for Social Research (SOFI), and the opportunity to collaborate with many great academics.

Application to associate professor

Rense Nieuwenhuis is a quantitative sociologist, interested in how social policy developments and demographic trends affect in economic inequality. Almost all of his studies are country-comparative in nature. With respect to social policy, he has mainly focused on family policies. Recently, more attention has been given to the European Commission’s social investment perspective, including the turn towards active labour market policies. In terms of demographic trends, his focus has been on trends in motherhood, educational expansion and homogamy, and single parenthood. Outcomes of interest were gendered differences in employment, poverty, and income inequality. Nieuwenhuis has developed three main research lines answering questions pertaining to the interplay between social policy developments and demographic trends, and one research line on methodological tools that help address the substantive questions. These research lines are detailed below.

The overview below includes a selection of 10 studies (referenced below). These include his dissertation, 7 studies in international peer-reviewed journals, and two chapters from a book he co-edited. The seven studies in his dissertation are discussed separately, with reference to the dissertation (#1). Most of these are now published in peer-reviewed journals.

In addition to developing these research lines, Nieuwenhuis is an active member of the academic community. He co-edited a forthcoming book on single parents, bringing together over 30 international scholars and coordinating their contributions to timely completion. He is often invited to give talks about his work, for research visits, and to teach at summer schools. He was commissioned to do work for the DG Employment of the European Commission, the European Institute for Gender Equality (EIGE), and the United Nations Entity for Gender Equality and the Empowerment of Women (UN-Women). He is also actively involved in the InGRID-2 project, which is a major EU Horizon 2020 funded initiative.

Combining institutional and demographic explanations

Nieuwenhuis’ first research line was developed in his PhD dissertation (#1), which was titled “Family Policy Outcomes: Combining institutional and demographic explanations of women’s employment and earnings inequality in OECD countries, 1975-2005″. In his dissertation, the argument was developed that the combination of institutional explanations and demographic explanations of women’s employment is more informative than a mere reference to determinants of either of these different strands of explanation. Institutional explanations have predominantly been invoked to help understand differences in women’s employment across countries, or trends in their employment, based on inter-country differences in the context in which women make their employment-related decisions. Demographic explanations have predominantly been invoked to understand differences in employment rates between women with a different demographic background within a single country. Combining these two strands of explanation not only provides a more detailed explanation of country-differences in economic inequality, it also provides the opportunity to address economic disparities between individuals with a different demographic, or socio-economic, background. In other words, it points towards institutional explanations of inequalities between demographic (and socio-economic) groups within societies, as well as to possible explanations of variation in the magnitude of such demographic inequalities across countries.

In the first empirical chapter of his dissertation (#1, also published), “Institutional and Demographic Explanations of Women’s Employment in 18 OECD countries, 1975-1999″, the first empirical support was provided. The article showed the importance of a wide range of institutional determinants, including several family policies and labour market characteristics such as unemployment and service sector size. Demographic determinants of women’s employment were covered by showing higher employment among women of younger generations, with higher levels of education, without a partner, and women without children. The multilevel design allowed to assess trends of the ‘motherhood-employment gap’ within countries over time, and to show how most work-family reconciliation policies (maternity leave, childcare leave, pay during leave) were beneficial to mothers but did not affect the employment of women without children. By differentiating between mothers and women without children, this was one of the first studies to convincingly demonstrate that financial support policies to families with children, measured as generous family allowances, are associated with reduced employment rates among women. The importance of combining institutional and demographic explanations was further demonstrated in Chapter 3 of Nieuwenhuis’ dissertation (#1) titled “Stratified Outcomes of Family Policies”, that found that work-family policies were particularly effective in closing the motherhood-employment gap (i.e. the difference in employment between mothers and women without children) among higher educated women. Such ‘Matthew effects’ could not have been found using institutional approaches alone. In chapter 4 of his dissertation (“Is these such a thing as too long childcare leave?”, (#1, also published), the question whether very long periods of childcare leave can have negative outcomes for women’s employment was revisited. In this study, it was argued and demonstrated that to correctly answer this question, one should combine both institutional and demographic indicators of women’s employment. Doing so, it was indeed found that very long periods of childcare leave reduce the employment of mothers but not of women without children, even while accounting for various demographic and institutional control variables.

Other studies in this research line include a study with Mark Levels and colleagues (#2) on how changes in Dutch legislation on contraceptives and induced abortion have affected prevalence of unintended pregnancies and women’s decisions regarding induced abortion by level of education. Unintended pregnancies were less likely when contraceptives were legal and available, and when abortion was legal and abortion services were available, women were more likely to abort an unintended pregnancy. The results further showed that higher educated women are less likely to experience an unintended pregnancy, but condition on having an unintended pregnancy higher educated women were more likely to choose for an abortion. Together with Andringa and Van Gerven (#3) it was shown how the absence of public childcare negatively affects maternal employment, and particularly so among women with traditional gender role attitudes.

Single-parent poverty and well-being

The second research line, developed after completing my dissertation, focuses on the interplay between social policy and a particular demographic trend: the rising number of single-parent families. This research line was developed together with Laurie C. Maldonado, starting with a study (#4) on how family policies might impact poverty risk of single-parent families differently from two-parent families. The study covered 18 OECD countries from 1978 to 2008. Along the line of Nieuwenhuis’ earlier research, a distinction was made between work-family reconciliation policies (e.g. paid leave) and financial support policies (e.g. family allowances). It was shown that both types of family policies reduce poverty among single-parent families, and that work-family reconciliation policies operate by facilitating employment particularly among single-parent families. Earnings from this employment helped to reduce poverty. In a follow-up article (#4), the importance and feasibility of stimulating employment among single-parent families was interpreted from the perspective of the European Social Investment paradigm. It was argued that policies that ‘prepare’ families to reduce poverty by means of employment can be beneficial for single-parent families, yet that many single-parent families live in poverty even while employed. This means that financial support policies that ‘repair’ poor economic outcomes, such as family allowances, cannot not be abandoned without increasing poverty risks for single parents.

In a forthcoming book chapter (#6) it was found that poverty among working single parents is very common. Three distinct patterns of performance were distinguished in how countries approach in-work poverty among single parents: A balanced approach of ensuring low inequality on the labour market combined with redistribution, an unbalanced approach of combating in-work poverty mostly through redistribution, and an approach in which high inequality on the labour market is compensated with redistributive policies only to a very limited extent. Countries that rely on a balanced approach to reduce inequality on the labour market, both with respect to class and gender, combined with an adequate level of redistribution, seem best situated for a durable reduction of poverty among working single parents.

Nieuwenhuis organised (with Maldonado) a total of five panels on single parents at three international conferences (Work and Family Researchers Network in New York, ESPAnet in Odense, and Policy & Politics in Bristol). Based on the contributions in these panels, as well as including contributions from other experts, Nieuwenhuis co-edited a book on single parents. In the introduction chapter of this book (#9), the theoretical concept of “the triple bind of single-parent families” is introduced: single parents and their families face higher risks of impaired socio-economic well-being, because of the combination of (1.) inadequate resources, (2.) inadequate employment, and (3.) inadequate policies. The importance and interplay of each of these three inadequacies was empirically demonstrated in the book. As an example, in an empirical chapter (#10) by Nieuwenhuis (with Tøge and Palme) the impact of employment and social policy on the health of single parents is analysed. The chapter finds that single parents are in better health when they are employed, and when their level of education is higher. Active labour market policies and public childcare benefit the health of single parents in two ways: by increasing their likelihood of being employment, and by benefitting the health of those who are employed (for instance, by reducing their work-family imbalance). Yet, the policies were also found to be associated with worse health among those who are not employed, suggesting selection effects. Financial transfer policies, measured as social assistance and financial supplements to families with children, were found to benefit the health of the non-employed.

The impact of rising women’s employment on inequality & poverty

The third research line addresses questions pertaining to how changes in female labour force participation have affected trends in economic inequality and poverty. It was argued (elsewhere) that an increase in women’s employment contributed to more income inequality between households, because spouses’ incomes (within households) are positively correlated. In other words, because higher earning men tend to be married to higher earning women, rising female labour force participation was argued to increase income inequality between households. Nieuwenhuis, however, showed with Van der Kolk and Need (Chapter “Earnings inequality within and between households” in dissertation #1, now published) that this is not the case. Indeed, with higher rates of women’s employment, the positive correlation between spouses’ incomes was more strongly positive, and this made a small contribution to higher income inequality between households. However, it was also found that rising employment rates of women resulted in a reduction of income inequality among women, which in turn strongly reduced income inequality between households. The latter effect was found to be stronger. Therefore, the net effect of increasing women’s employment rates, and thus of an (a.) increased correlation between spouses’ earnings and (b.) reduced income inequality among women, was found to be a sizeable reduction of income inequality between households. In a second study (Chapter “Family Policies and Earnings Inequality Between Households” in dissertation #1), it was found that women’s employment more strongly reduced income inequality between households in societies with generous work-family reconciliation policies, and with limited family allowances. This links this research line back to Nieuwenhuis’ first one on combining institutional and demographic explanations of inequality. The relevance of this finding is that it shows that, in contrast to various arguments in the literature, family policy arrangements that stimulate women’s employment are not at odds with low levels of income inequality between households. Currently, Nieuwenhuis (with Van Lancker, Collado & Cantillon, not included as appendix) works on study examining the impact of rising female labour force participation on trends in household poverty. The findings indicate that this impact, at best, has been moderate. Moreover, in many countries female labour force participation rates seem to have hit some form of a ceiling, not increasing any further. This has important implications for social policy goals set by the European Commission, as it limits the potential for reducing poverty by stimulating women’s employment.

Statistical Methods

Nieuwenhuis developed a methodological research line to support the three research lines above, by developing the statistical tools required to answer the country-comparative research questions central to his research.

Nieuwenhuis developed (with Te Grotenhuis and Pelzer) statistical software that provides statistical diagnostics with respect to influential cases in multilevel regression models, and published an article on this software package (chapter “Influence.ME: Tools for detecting influential data in multilevel regression models” in dissertation #1, also published). The importance of this highly cited article lies in the fact that multilevel models in country-comparative research often face the challenge of a relatively small number of countries included in the analyses. Such models can be severely biased because of a single country being a strong outlier / influential case. This software helps to assess and prevent such bias.

As much of the research is about comparing measures of economic inequality across countries, tools had to be developed to make such income datasets comparable across countries (in technical terms: measurement equivalent). Although the data in the LIS Database that he used are mostly comparable across countries, issues of comparability arose when comparing income of spouses in a country-comparative study. Tools had to be developed to allow for such comparisons, and an evaluation of these tools was accepted for publication (In Chapter “Comparative analyses of gross and net earnings” in dissertation #1, also published).

When applying regression models to categorical (independent) variables, dummy variables are often used. A disadvantage of the traditional way in which dummy are created, is that a specific reference category has to be chosen. An alternative was developed, called ‘weighted effect coding’. With weighted effect coding, the effect for each category represents the deviation of that category from the weighted mean (which corresponds to the sample mean). This technique has particularly attractive properties when analysing observational data, that commonly are unbalanced. This technique was described in publication #7, and Nieuwenhuis developed the statistical software to apply weighted effect coding in generalised linear regression models. In subsequent publications, the mathematical procedures were extended to allow for the interaction between categorical variables and for the interaction between a categorical variable and a continuous variable (#8). These are novel methods, and neither the mathematical procedure for the interactions, nor the application in statistical software were available.

References

1. Nieuwenhuis, R. (2014). Family Policy Outcomes: Combining Institutional and Demographic Explanations of Women’s Employment and Earnings Inequality in OECD Countries, 1975-2005. PhD Dissertation, University of Twente.
2. Levels, M., A. Need, R. Nieuwenhuis, R. Sluiter, and W. Ultee (2012). Unintended Pregnancy and Induced Abortion in the Netherlands 1954-2002. European Sociological Review 28, 301-318.
3. Andringa, W., R. Nieuwenhuis, and M. Van Gerven (2015). Women’s working hours: the interplay between gender role attitudes, motherhood, and public childcare support in 23 European countries. International Journal of Sociology and Social Policy 35(9/10), 582-599.
4. Maldonado, L. C. and R. Nieuwenhuis (2015). Family Policies and Single-Parent Poverty in 18 OECD Countries, 1978-2008. Community, Work and Family 18(4), 395-415.
5. Nieuwenhuis, R. and L. C. Maldonado (2015). Prepare versus Repair? Combining parental leave and family allowances for social investment against single-parent poverty. Belgisch Tijdschrift voor Sociale Zekerheid (Belgian Review of Social Security) 57(1), 115-123.
6. Nieuwenhuis, R. and L. C. Maldonado (2018). Single-Parent Families and In-Work Poverty. In: Handbook of Research on In-Work Poverty. Ed. by H. Lohmann and I. Marx. Edward Elgar.
7. Grotenhuis, M. te, B. Pelzer, A. Schmidt-Catran, R. Nieuwenhuis, R. Konig, and R. Eisinga (2017). When size matters: advantages of weighted effect coding in observational studies. International Journal of Public Health, 62(1), pp. 163-167.
8. Nieuwenhuis, R., M. Te Grotenhuis, and B. Pelzer (2017). Weighted Effect Coding for Observational Data with wec, R Journal, 9(1), pp. 477-485.
9. Nieuwenhuis, R. and L.C. Maldonado (2018). The Triple Bind of Single-Parent Families: resources, employment, and policies. In Nieuwenhuis, R. and L.C. Maldonado (eds). The Triple Bind of Single-Parent Families: resources, employment and policy to improve well-being. Bristol: Policy Press.
10. Nieuwenhuis, R., Tøge, A-G., Palme, J. (2018). The health penalty of single parents in institutional context. In Nieuwenhuis, R. and L.C. Maldonado (eds). The Triple Bind of Single-Parent Families: resources, employment and policy to improve well-being . Bristol: Policy Press.

Family policies are those public policies that directly affect families with children. Given the fact that definitions of what constitutes a family are constantly changing, and with changing goals of governments, the nature of family policies has been changing since their early inception (in their modern form) at the time of industrialization. Family policies are understood as having a variety of goals, including 1) poverty reduction and income maintenance, 2) direct compensation for the financial cost of raising children, 3) fostering employment, 4) improving gender equity, 5) support for early childhood development, and 6) raising birth rates (see Thévenon 2011 in Origin and Variety of Family Policies).

The available research on family policies is vast, and naturally difficult decisions had to be made to end up with the selection of studies presented here. Important works and topics had to be left out, although many of the topics that are not explicitly discussed emerge in one form or another in our selection of research. Although our selection seeks to cover a broad range of perspectives, we have focused on 1) empirical research, often (but not exclusively) quantitative in nature, 2) research on outcomes of family policies, 3) research on family policy outcomes that include employment, wages, poverty, and fertility, and 4) in addition to some classics, some recent works that point to current frontiers in family policy research.

The references are organized in six major categories, which necessarily show some overlap. We begin with several General subsections that cover the Origin and Variety of Family Policies, selected Classics, concepts, Research Overviews, discussions on Gender in Welfare State Regimes, and recent perspectives on family policies such as social investment. Next, we detail studies that examined various forms of family Policies as their object of study, covering parental leave, childcare, and cash support for families, including child support and alimonies. Third, we selected studies on Outcomes pertaining to employment, unpaid work, occupations and earnings, poverty, and fertility. Fourth, we highlight several Debates that are ongoing in the literature, including on the Matthew Effect and on using aggregated data to study the link between fertility and (women’s) employment. Fifth, we highlight several research Frontiers: areas of more recently developed topics that include the role of fathers, family policies outside the EU or OECD, company-level (rather than public) family policies, and the use of experimental approaches. We conclude by listing a range of Data Sources that are relevant for the continued study of family policies and their outcomes.

The chapter is available online

The attractive property of applying weighted effect coding to categorical (‘factor’) variables is that each category represents the deviation of that category from the sample mean. This is unlike the more commonly used treatment coding where each a specific category has to be selected as a reference. Weighted effect coding is a generalized form of effect coding that applies to both balanced and unbalanced data.

A form of weighted effect coding was already formulated in 1972 by Sweeney and Ulveling, but it seems to never have found its place in statistical repertoires. Weighted effect coding was not implemented in mainstream statistical software. In an ongoing project, we have now further developed weighted effect coding to also apply to interactions (with both categorical and continuous variables), and provide procedures for mainstream statistical software. For R, we developed the ‘wec’ package, and procedures for STATA and SPSS are available as well.

A key innovation in our article in the R Journal is the formulation of interactions between a categorical variable with a continuous variable. This is visualised in the Figure above. The benefit of estimating such an interaction with weighted effect coding is that upon entering the interaction terms the estimate for the continous variable (as well as the ‘main effects’ for the categorical variable) does not change. The ‘main’ continous term reflects the average effect in the sample, and the interaction terms represent the deviation of the effect size for each category.

References

Grotenhuis, Te, M, Pelzer, B., Eisinga, R., Nieuwenhuis, R., Schmidt-Catran, A., & Konig, R. (2017b). A novel method for modelling interaction between categorical variables. International Journal of Public Health, 62(3), 427–431. (open access!)

Grotenhuis, Manfred, Ben Pelzer, Eisinga, R., Nieuwenhuis, R., Schmidt-Catran, A., & Konig, R. (2017a). When size matters: advantages of weighted effect coding in observational studies. International Journal of Public Health, (62), 163–167. (open access!)

Nieuwenhuis, R., Grotenhuis, Te, M., & Pelzer, B. (2017). Weighted Effect Coding for Observational Data with wec. R Journal, 9(1), 477–485. (open access!)

Sweeney, R. E., & Ulveling, E. F. (1972). A transformation for simplifying the interpretation of coefficients of binary variables in regression analysis. The American Statistician.

To avoid choosing a reference category, weighted effect coding provides estimates representing deviations from the sample mean. This is particularly useful when the data are unbalanced (i.e., categories holding different numbers of observation). The basics of this technique, with applications in R, were detailed here.

In a new publication, available open access,, we show that weighted effect coding can also be applied to regression models with interaction effects (also commonly referred to as moderation). The weighted effect coded interactions represent the additional effects over and above the main effects obtained from the model without these interactions.

To apply the procedures introduced in these papers, called weighted effect coding, procedures are made available for R, SPSS, and Stata. For R, we created the ‘wec’ package which can be installed by typing:

install.packages(“wec”)

References (Open Access!)

Grotenhuis, M., Ben Pelzer, Eisinga, R., Nieuwenhuis, R., Schmidt-Catran, A., & Konig, R. (2017). A novel method for modelling interaction between categorical variables. International Journal of Public Health, 62(3), 427–431. http://link.springer.com/article/10.1007/s00038-016-0902-0

Grotenhuis, M., Ben Pelzer, Eisinga, R., Nieuwenhuis, R., Schmidt-Catran, A., & Konig, R. (2017). When size matters: advantages of weighted effect coding in observational studies. International Journal of Public Health, 62(1), 163–167. http://doi.org/10.1007/s00038-016-0901-1