Online Supporting Information for

This paper measures turnout rates before and after a spousal death to estimate the effects of spousal loss on turnout, and analyzes variations in turnout changes to evaluate social explanations for the observed widowhood effects. The social explanation analysis is based on three assumptions: 1) persistent changes in turnout (changes not accompanied by a substantial gradual recovery for the year half of a spouse’s death) are directly attributable to the absence of a spouse, and not the loss and grieving process; 2) relative voting histories of spouses (whether one spouse votes more, the same amount, or less than the other) are indicators of political engagement, and observed variations in turnout changes by relative spousal voting histories are attributable to differences in political engagement; and 3) changes in turnout caused by the absence of a mobilizing spouse can be moderated (over time) by the presence of other electors in the same household (e.g. as they assume new household and social support roles). The tests of social explanations for drops in turnout following the death of a spouse are presented in the body of the article. We observe limited aggregate return to previous voting behavior in the year and a half following the death of a spouse and no aggregate return to previous voting behavior past the one year anniversary of spousal death, electors who voted less than the decedent spouse are substantially more affected by the spouse’s absence than electors who voted more, and widowed individuals who lived with other electors (likely family members) at the time of their spouse’s death gradually return to previous voting behavior past the one year mark. In this supporting information, we detail the structure of the dataset used in the above analysis, and test alternative explanations for observed turnout patterns.