I’m working with ASEC files from 2000 to 2022. I’m coding first and second CPS spells (SPELL), as a way to generate new participant IDs. SPELL==1 for MISH of 1-4; SPELL==2 for MISH 5-8. I’m also then recoding MISH so that the range for all is 1-4. This way, CPSIDP + SPELL + MISH (recoded) uniquely identify observations (I did this for work with the BMS data). But I’m having issues generating SPELL, because it seems that for some participants of two consecutive ASECs, in 2021 and 2022, the MISH is the same. So, for example, person A participated in the 2021 ASEC, and it represented her 4th month in sample. Person A is then shown again, with a second ASEC record, for 2022, but it has her month in sample as 4, when it should be 8. Right? So I’m unable to distinguish CPS spells (SPELL) for certain recent longitudinal participants. I’m attaching a sample of observations with this problem. In these cases, race, sex, and age values match up across the two records.
It feels like someone would have noticed this by now. So I’m wondering if it’s just something I did wrong, or am misunderstanding.
asec_sample_obs_mish_error.xlsx (10.2 KB)