Sell Randall, Goldberg Shoshana, Conron Kerith
Department of Community Health and Prevention, School of Public Health, Drexel University, Philadelphia, Pennsylvania, United States of America.
University of North Carolina, Chapel Hill, North Carolina, United States of America.
PLoS One. 2015 Dec 7;10(12):e0144011. doi: 10.1371/journal.pone.0144011. eCollection 2015.
Gaps in data collection systems, as well as challenges associated with gathering data from rare and dispersed populations, render current health surveillance systems inadequate to identify and monitor efforts to reduce health disparities. Using sexual and gender minorities we investigated the utility of using a large nonprobability online panel to conduct rapid population assessments of such populations using brief surveys. Surveys of the Google Android Panel (four assessing sexual orientation and one assessing gender identity and sex assigned at birth) were conducted resulting in invitation of 53,739 application users (37,505 of whom viewed the invitation) to generate a total of 34,759 who completed screening questions indicating their sexual orientation, or gender identity and sex at birth. Where possible we make comparisons to similar data from two population-based surveys (NHIS and NESARC). We found that 99.4% to 100.0% of respondents across our Google Android panel samples completed the screening questions and 97.8% to 99.2% of those that consented to participate in our surveys indicated they were "OK" with the content of surveys that assessed sexual orientation and sex/gender. In our Google Android panel samples there was a higher percentage of sexual minority respondents than in either NHIS or NESARC with 7.4% of men and 12.4% of women reporting gay, lesbian or bisexual identities. The proportion sexual minority was 2.8 to 5.6 times higher in the Google Android panel samples than was found in the 2012 NHIS sample, for men and women, respectively. The percentage of "transgender" identified individuals in the Google sample was 0.7%, which is similar to 0.5% transgender identified through the Massachusetts BRFSS, and using a transgender status item we found that 2.0% of the overall sample fit could be classified as transgender. The Google samples sometimes more closely approximated national averages for ethnicity and race than NHIS.
数据收集系统存在的漏洞,以及从稀少且分散的人群中收集数据所面临的挑战,使得当前的健康监测系统不足以识别和监测为减少健康差距所做的努力。我们以性少数群体和性别少数群体为研究对象,调查了使用大型非概率在线样本库,通过简短调查对这类人群进行快速人口评估的效用。我们对谷歌安卓样本库进行了调查(四项评估性取向,一项评估性别认同及出生时被指定的性别),邀请了53739名应用程序用户(其中37505人查看了邀请),最终共有34759人完成了筛查问题,表明了他们的性取向、性别认同或出生时的性别。我们尽可能地将这些数据与两项基于人群的调查(美国国家健康访谈调查和全国酒精及相关疾病流行病学调查)中的类似数据进行比较。我们发现,在我们的谷歌安卓样本库中,99.4%至100.0%的受访者完成了筛查问题,在同意参与我们调查的人中,97.8%至99.2%的人表示他们对评估性取向和性/性别的调查内容“没问题”。在我们谷歌安卓样本库中,性少数群体受访者的比例高于美国国家健康访谈调查或全国酒精及相关疾病流行病学调查中的比例,7.4%的男性和12.4%的女性报告自己为同性恋、女同性恋或双性恋身份。在谷歌安卓样本库中,性少数群体的比例分别是2012年美国国家健康访谈调查样本中男性和女性比例的2.8至5.6倍。谷歌样本中“跨性别”身份的个体比例为0.7%,这与通过马萨诸塞州行为风险因素监测系统识别出的0.5%的跨性别者比例相似,并且使用一个跨性别身份项目,我们发现总体样本中有2.0%的人可被归类为跨性别者。与美国国家健康访谈调查相比,谷歌样本在种族和民族方面有时更接近全国平均水平。