Family planning · PLoS Med · 2019
Opening summary: vouchers multiplied the odds of starting birth control 2.51 times. The paper’s table: 5.02.
How much vouchers multiplied the odds of starting birth control by 14 weeks
The paper says
New mothers with HIV in Uganda were randomly given birth-control vouchers or not. ‘Odds’ means starters per non-starter. The paper’s opening summary (not its ‘Author summary’) says that by 14 weeks after birth, vouchers multiplied those odds 2.51 times (likely between 1.31 and 4.79).
Its own data show
The paper’s own table, at 14 weeks: 149 of 158 voucher women had started (94%), against 122 of 159 others (77%). That is 1.2 times as many starters, but odds of 149 to 9 (16.6 to 1) against 122 to 37 (3.3 to 1): 5.02 times, as the table prints. These counts cannot give 2.51.
Odds hinge on the few who hadn’t started (6% with vouchers, 23% without), so they move far more than the share who started. At 8, 12, 16 and 20 weeks the summary’s multipliers match the table’s. At 14 weeks the summary, the part most people read, gives one half the table’s: 2.51 instead of 5.02. The vouchers still clearly helped.