【作者】 汪颖；

【导师】 姜婷婷；

【作者基本信息】 武汉大学 ， 管理科学与工程， 2022， 硕士

【摘要】 许多职业内存在从业者性别失衡的问题,这不仅降低了从业者的多样性、不利于人力资源的充分挖掘,也阻碍了行业的长期健康发展。职位申请者的性别构成和最终从业者的性别构成密切相关,提升职位申请者的性别多样化程度有利于从业者性别构成的相对均衡。而招聘广告是应聘者了解职位和公司的重要方式,在应聘者的职位申请决策中扮演着重要角色。为此,本研究以招聘广告为切入点,选取了文本性别偏向作为研究变量,旨在探究中文招聘广告文本的性别偏向呈现何种特点、是否会对申请者的性别构成产生影响以及相关因素对于二者关系的调节作用。首先,本文基于性别刻板印象这一概念,计算了招聘广告的文本性别偏向。具体地,本文利用文本挖掘技术建立了性别偏向词表,并结合问卷调查和机器学习方法对词汇的性别偏向及程度进行了评价;参考基于词典的文本情感计算方法,进一步完成了对招聘广告文本的性别偏向计算。此外,本文还从职业、组织、地区、文本四个维度操作化了职位主导性别、企业声誉、工作城市等级、文本交互性等其它变量。然后,为了了解招聘广告文本性别偏向的特点,本文使用方差分析检验了招聘广告文本性别偏向在各职位维度内部的差异。最后,通过开展线性回归分析,本文探究了招聘广告文本性别偏向对职位申请者性别构成的影响,以及相关因素的调节作用。结果表明:(1)绝大多数中文招聘广告呈男性偏向,仅有少部分呈女性偏向;(2)招聘广告文本的男性偏向程度在职位主导性别/企业声誉/工作城市等级/文本交互性的不同水平上呈现显著差异;(3)对于呈男性偏向的招聘广告,其男性偏向程度越高,其申请者中男性占比越大、女性占比越小;对于呈女性偏向的招聘广告,其女性偏向程度越高,其申请者中女性占比越大、男性占比越小;(4)职位主导性别、文本交互性负向调节招聘广告性别偏向和申请者性别构成之间的关系。本研究成果有助于增加对职业内从业者性别失衡问题成因的理解,同时也拓宽了文本分析的思路。此外,本研究为招聘广告的撰写以及第三方招聘平台的日常管理提供了启示。具体地,在招聘广告的撰写中应减少性别偏向词汇都的使用,并可通过调整词汇和句法改变文本的交互性从而减轻由性别偏向词语带来的影响。第三方招聘平台可参考本文思路实现对其平台上所有招聘广告性别偏向的计算与监测。更多还原

【Abstract】 There is a gender imbalance in many occupations,which not only reduces the diversity of practitioners,conducive to the full exploitation of human resources,but also hinders the long-term healthy development of the industry.The gender composition of job applicants and the gender composition of the final practitioners are closely related,and increasing the gender diversity of job applicants is conducive to the relative balance of the gender composition of practitioners.Job advertisements are an important way for candidates to understand the position and the company,and play an important role in the candidate’s job application decision.To this end,this study takes the recruitment advertisement as the starting point and selects the gender bias of the text as the research variable,aiming to explore the characteristics of the gender bias of the Chinese the text of the recruitment advertisement,whether it will affect the gender composition of the applicant,and the regulatory effect of related factors on the relationship between the two.First of all,based on the concept of gender stereotypes,this paper calculates the text gender bias of job advertisements.Specifically,this paper uses text mining technology to establish a gender bias thesaurus,and evaluates the gender bias and degree of vocabulary in combination with questionnaire surveys and machine learning methods;refers to the dictionary-based text sentiment calculation method to further complete the gender bias calculation of recruitment advertisement texts.In addition,this paper also operates other variables such as job dominant gender,corporate reputation,work city level,and text interactivity from the four dimensions of industry,enterprise,position,and text.Then,to understand the characteristics of implicit sexism in job ads,this paper uses ANOVA to examine the differences in the gender bias of job ads within each job dimension.Finally,through linear regression analysis,this paper explores the influence of gender bias in job advertisement text on the gender composition of job applicants,and the moderating effect of related factors.The results showed that:(1)the vast majority of Chinese recruitment advertisements were male biased,and only a small number were female biased;(2)the male bias of the recruitment advertisement text showed significant differences in the different levels of job dominant gender/corporate reputation/work city level/text interactivity;(3)for male biased recruitment advertisements,the higher the male bias,the greater the proportion of males among the applicants and the smaller the proportion of women;for female-biased recruitment advertisements,the higher the degree of female bias.The larger the proportion of females and the smaller the proportion of males among its applicants;(4)the dominant gender of the position and the negative direction of text interactivity adjust the relationship between the gender bias of the recruitment advertisement and the gender composition of the applicant.The results of this research help to increase the understanding of the causes of the gender imbalance of practitioners within the occupation,and also broaden the thinking of text analysis.In addition,this study sheds light on the writing of job advertisements and the day-to-day management of third-party recruitment platforms.Specifically,the use of gender-biased words should be reduced in the writing of job advertisements,and the impact of gender-biased words can be mitigated by adjusting vocabulary and syntax to change the interactivity of the text.Third-party recruitment platforms can refer to the ideas in this article to calculate and monitor the gender bias of all recruitment advertisements on their platforms.更多还原