No abstract

Background: Despite a 2010 Kenyan constitutional amendment limiting members of elected public bodies to < two-thirds of the same gender, only 22 percent of the 12th Parliament members inaugurated in 2017 were women. Investigating gender bias in the media is a useful tool for understanding socio-cultural barriers to implementing legislation for gender equality. Natural language processing (NLP) methods, such as word embedding and sentiment analysis, can efficiently quantify media biases at a scope previously unavailable in the social sciences.Methods: We trained GloVe and word2vec word embeddings on text from 1998 to 2019 from Kenya’s Daily Nation newspaper. We measured gender bias in these embeddings and used sentiment analysis to predict quantitative sentiment scores for sentences surrounding female leader names compared to male leader names.Results: Bias in leadership words for men and women measured from Daily Nation word embeddings corresponded to temporal trends in men and women’s participation in political leadership (i.e., parliamentary seats) using GloVe (correlation 0.8936, p = 0.0067, r2 = 0.799) and word2vec (correlation 0.844, p = 0.0169, r2 = 0.712) algorithms. Women continue to be associated with domestic terms while men continue to be associated with influence terms, for both regular gender words and female and male political leaders’ names. Male words (e.g., he, him, man) were mentioned 1.84 million more times than female words from 1998 to 2019. Sentiment analysis showed an increase in relative negative sentiment associated with female leaders (p = 0.0152) and an increase in positive sentiment associated with male leaders over time (p = 0.0216).Conclusion: Natural language processing is a powerful method for gaining insights into and quantifying trends in gender biases and sentiment in news media. We found evidence of improvement in gender equality but also a backlash from increased female representation in high-level governmental leadership.

Background There is a scarcity of research that comprehensively examines programme impact from a context-specific perspective. We aimed to determine the conditions under which the Bihar Technical Support Programme led to more favourable outcomes for maternal and child health in Bihar. Methods We obtained block-level data on maternal and child health indicators during the state-wide scale-up of the pilot Ananya programme and data on health facility readiness, along with geographical and sociodemographic variables. We examined the associations of these factors with increases in the levels of indicators using multilevel logistic regression, and the associations with rates of change in the indicators using Bayesian Hierarchical modelling. Results Frontline worker (FLW) visits between 2014-2017 were more likely to increase in blocks with better night lighting (odds ratio (OR) = 1.23, 95% confidence interval (CI) = 1.01-1.51). Birth preparedness increased in blocks with increasing FLW visits (OR = 3.43, 95% CI = 1.15-10.21), while dry cord care practice increased in blocks where satisfaction with FLW visits was increasing (OR = 1.52, 95% CI = 1.10-2.11). Age-appropriate frequency of complementary feeding increased in blocks with higher development index (OR = 1.55, 95% CI = 1.16-2.06) and a higher percentage of scheduled caste or tribe (OR = 3.21, 95% CI = 1.13-9.09). An increase in most outcomes was more likely in areas with lower baseline levels. Conclusions Contextual factors (eg, night lighting and development) not targeted by the programme and FLW visits were associated with favourable programme outcomes. Intervention design, including intervention selection for a particular geography, should be modified to fit the local context in the short term. Expanding collaborations beyond the health sector to influence modifiable contextual factors in the long term can result in a higher magnitude and more sustainable impact. Registration ClinicalTrials.gov: NCT02726230.