We adopt stratified randomisation based on ethnic groups and locations.
(1) We collected a trial survey among 170 key informants out of Yunnan and Sichuan provinces in March 2020 for livelihood arrangements of ethnic groups and various (informal) institutions guiding succession such as marriage, decision making in family, and inheritance norms. We constructed indices and rank ethnic groups by their extent of male-biasedness.
(2) The data are also matched to (formal) institutional categories proxied by SIGI variables (https://www.genderindex.org/methodology/). Using the information of the Ethnographic ATLAS on social structure, which gives an indication of the extend of “egalitarian” structure, we selected matrilineal and patrilineal ethnic groups and within each category there are ethnic groups of different extents of “egalitarian” structures.
(3) Given that many ethnic groups are not covered in ATLAS, we also referred to ethnography and cultural anthropology for the origins of ethnic ancestors, the linguistic system, culture and norms. We selected ethnic groups having not been covered by ATLAS or our trial interview but practising apparently matrilineal and patrilineal norms. The proposed set of sample ethnicities consists of Mosuo, Jinuo, Bulang, Jingpo, Dong, Dulong, Dai, Deang (i.e., Benglong in ATLAS), Yao, Lisu, Miao, Yi, Achang, Bai, Nu, Lahu, Tujia, Hani, Tibetan, Mongol, and Han.
(4) For each ethnic group, we consult the China Ethnicity Statistical Yearbook 2019 and use the list of main residential provinces for ethnic groups and the lists of ethnic autonomous towns, counties and prefectures in each province. Based on these lists, we select the main residential province and prefecture for each ethnic group and within the prefecture, randomly select the ethnic autonomous towns and counties according to the population and economic development. Particularly for each ethnic group, we sample at least two different counties, in order to capture any within-group variations as a result of decades of inter-ethnic mixing due to different residential locations.