Patrick Heuveline
Data on intergenerational relationships and support:The Mekong Integrated Population-Registration Areas of Cambodia (MIPRAoC)
Patrick Heuveline (University of California, Los Angeles)
- Home
- Entry to Cyberseminar 1
- Information for participants in cyberseminar 1
- Supporting organisations
- Cyberseminar series conference organisers
- IUSSP Panel on Family Demography in Developing Countries
- Other recommended reading
- University Homepage
Thank you for your comments and encouragements. The comparison with the approach used in ACDIS and Agincourt is indeed useful. Our DSS has been a full member of the INDEPTH network at some point, and if weāve now reduced our participation to associate membership, it is mostly for logistical reasons (full membership requires participation in the Annual General meetings, the majority of which taking place in Africa entails long journeys). While each project has its own idiosyncrasies linked to its genesis and current environment, there are undoubtedly much that can be learned from one another.
The choice of a light structure combined with rider surveys was certainly made under some constraints, financial ones mostly. Our DSS receives no institutional or governmental support, and its continued funding requires continued success submitting grant applications. The list of agencies that we have received support from over the years is acknowledged in the paper, but the National Institutes of Health (NIH), in the U.S., has been our main source of funding by far. NIH grants are limited in time (typically 5 years) and must be renewed, so there is always the possibility of a gap in funding. A light DSS structure allows us to continue functioning by stretching past resources or securing small alternative grants (bridge or seed grants).
This light approach also has its advantages. In spite of funding insecurity, our goal has always been to be present for the long haul. This requires considering survey fatigue as a potential threat to the validity of the DSS. Our DSS is very quick to administer, except for immigration forms which require new registration, and refusal to participate has remained minimal (in the order of a few percent). On surveys that can last between one and two hours, the refusals are much more common, and refusals on the DSS would likely increase as well if our DSS updates were longer to administer.
Our DSS began in 2000 with a periodicity of 6 months. This was chosen at the outset because the status of the ongoing fertility transition and levels of infant mortality were much debated and measuring pregnancy outcomes accurately was one of our main objectives. With a visit every 6 months, we thought we would be able to record most pregnancies, which we would then follow-up to register births and infant deaths that might have intervene before the next round. As the we observed the progress of the fertility transition and low level of infant mortality, our objectives evolved and we reduced our periodicity to 12 months apparently without appreciable losses in death or birth registration. We are currently operating on a 24-month periodicity caused by delays in the data entry and data cleaning processes that need to be complete before satisfactory rosters can be established for the next DSS update. As explained in the paper, our project has grown 7 folds between 2006 and 2008. This desirable growth for the standpoint of representativeness has presented a number of logistical, personnel and institutional challenges. An annual update would be preferable and we are exploring partnerships with professional firms that would speed up these processes.
Residential mobility has presented a challenge because, as noted in the paper, this is the most common event affecting household composition, and as noted above, immigration is the most time consuming event to record. We donātā believe it has led to double counts, as long as one is recorded moving out and also moving in. The main issue has been to determine return migrants and avoid that a person who already has been given an ID, is given a new one upon return. Unfortunately, for individuals who donāt come back to the same household exactly, making this determination is time consuming and error prone (in Cambodia, names do not have the lifetime endurance that they typically have in the U.S.). What we have been trying to limit the registration of short-term visitors that would be gone by the next round is to introduce a table of visitors for individuals who have been present for less than 6 months. For these individuals, we only register minimum information which is re-printed below the household roster at the next visit. If the āvisitorsā are still present at the next visit, then they require a full registration (immigration form). Else, we have potentially saved an immigration registration followed by an outmigration record at the next round.
On the issue you note on the age pyramid, I cannot answer offhand. I will need to look into this. For outmigration, yes, in the example of a widow and a child, we record the destination of each individual, and if within the catchment area (PRA), we record the name of the new head of household and relationships of each migrants to the new head.
For the orphan survey, yes, misclassification would be an issue and so would not knowing the survival status of parents. On the latter, we have an āunknownā orphan category. The social parent category is not determined by the fieldworker. S/he only records a history of living arrangements after parental death. The head of the households in which the orphan has lived, other than her own biological parents, are considered to be possible social parents.
Again, thank you for your attention to and detailed comments on the paper. Looking forward to continuing this conversation.
Dear Patrick, Thank you for explaining the background to your light DSS approach. It would seem that by anticipating the potential gaps in funding early on you tailored the design recognizing what you could or could not do well, or would need to put in place to ensure you could prospectively link individuals and households efficiently. This may be a firmer starting point for the conceptual and operational model than surveillance sites starting out as intensive household-based follow-up studies, becoming intensive DSS but moving to infrequent, unpredictable visits as funding stops and starts.
You mention an early motivation for making more frequent visit was to record pregnancies. In ACDIS, we opted for 3-monthly follow-up primarily for migration while also hoping that it would increase the number of observed pregnancies and ensure fewer missed pregnancy outcomes. While the frequent follow-up undoubtedly assists with tracking and follow-up, the impact of changing the frequency does not appear to have a marked impact on pregnancies reported.
I can see why survey fatigue and refusals might be less with fewer visits. However, on the other hand is there a possibility that infrequent visits might mean that the study community views the engagement of the research project differently to one where the frequent visits support other research and programmatic activities requiring contemporaneous sampling, tracking and additional information i.e. where the DSS is also a platform supporting support other research and programme activities. Does the MIPRAoC have linkages with service provides or intervention studies or are the inter-round surveys primarily observational in nature?
The visitors list is not one that we use in ACDIS where visitors (a person staying with the household but not belonging to it) are not recorded. But this is an approach used in other DSS. If I understand correctly, visitor is actually a state that all new residents are initially assigned irrespective of the circumstances under which they have moved into the household e.g. a visiting parent to a woman moving in on marriage. Is that correct? When thinking about the groups of people who are very mobile, is there a risk with this system that highly mobile individuals, for example people regularly moving between two households such as children between separated parents, might effectively remain on the visitors list of one or indeed several households. A kind of floating population?
You wrote in your reply: āThe social parent category is not determined by the fieldworker. S/he only records a history of living arrangements after parental death. The head of the households in which the orphan has lived, other than her own biological parents, are considered to be possible social parents.ā Could I clarify, is this a post-data collection assignment during analyses? In which non-household heads (and their spouses?) would not be assigned as social parents. We have been recently piloting questions asking respondents if there is anyone other than the biological parent is playing the role of parent to a particular child – this is irrespective of whether the childsā biological parents are alive or dead.
Sangeetha Madhavan and her colleagues at Agincourt have had some experience of asking about non-biological parents. In South Africa, this is a particularly challenging aspect in describing relationships and support given the prevalence of orphaning during childhood, and family dispersal with low rates of marriage, high residential mobility of parents both mothers and fathers.
The issue of social parents is something we pick up across the strands even if not articulated in the same way. The issue is also pertinent to studies by Estela Rivero and colleagues āWith One Child Here and One Child There: Is There Specialization and Complementarity in Childrenās Support Related to Their Place of Residence? The Mexican Caseā in strand 2, Sangeetha Madhavan and colleagues paper in Strand 1, the paper by Makandwe Nyirenda et al also in session 3 among others.
Just to clarify quickly, I donāt think that respondentsā fatigue is increased by frequent visits (and you might be right that actually frequent visit help in keeping commitment up), but possibly, by the length of the interview. This is why we keep our event registration fairly light (except for immigration records). In particular, we have been considering verbal autopsies for a while, but so far have not implemented them. We have started to ask more questions about other events (e.g., marriage as discussed in the paper).
The visitorsā table is intended mostly for people moving from outside of the area into the PRA. It is possible that a person splits time between households but it is a lot more rare than I understand it is in South Africa. For these people, immigration registration is not too much of an issue because we already have all their information from their previous household, and there would not be much gain in āholdingā them up in the visitorsā table (although, I can guarantee that field workers havenāt done so). The main issue for us would be the possibility of double counts (being registered as a resident at two places). We try to avoid this by asking field worker to use a āmajorityā rule, that is, register as a resident someone whoās there 4 nights or more a week, or six months or more of the year.
For social parents, you are correct that this is a post-data collection assignment. We are mostly interested, however, in the āsocial siblingsā so that we can compare orphans we non-orphans net of some of the childhood household characteristics. We thus include in the pool of potential social siblings all biological children of the head of the household (and/or spouse) in any of the households the orphan has lived as a child after the death of a parent (excluding her own full siblings). We then look for those who might still live in the PRA and randomly select one of those for interview. In other words, we are not really investigating which adult might have fulfilled that āsocial parentingā role in the way that Sangeetha might have.
I agree itās a fascinating issue though. In the upcoming round, we are actually trying to map out who the potential social parents were by asking about whether older siblings, uncle/aunts and grand-parents were married and still alive at the time of parental death. I worry that some of the causal claims about the dis/advantages of particular kin as social parents is affected by their availability in the first place.
Hi Patrick and Vicky, Thanks for the exchange. I am curious how innovations in technology are or are not facilitating the frequency and types of contacts with participants in DSS. Are there methodological innovations to be considered here?