attrition in longitudinal studies

[Alcock, Katie]

This is something I know a reasonable amount about, having worked on two
longitudinal studies among possibly the lowest SES group possible - rural
children in Tanzania, one of the poorest countries in the world.  Both
studies were looking at school aged children so, for this group, attrition
may be slightly lower because parents are less likely to move the whole
family if a child is in school.  We found that some children stayed in the
same school but their family care arrangements changed.  In the second study
we did however also look at children who were not in school and the dropout
rate was higher.  In that study it was 15-20% (over 2 years) and in the
first study the figures (over 16 months) were as follows:

The total drop out rate was around 7% for educational achievement tests, 10%
for tests of cognitive function and 15% for biometric measures. Dropouts
were due to families moving away from the area (44% of all dropouts),
dropping out of school (13%), withdrawal of parental consent (12%), chronic
absenteeism (10%) and children refusing to participate (4%).

Strategies for following up are possibly a little different in this setting
- no family has a telephone or a postal address and to find children's
houses requires at least directions, if not accompaniment by another child
or adult.  We did have the advantage that in the first study (with 1000
children) all of them were in school at the start of the study and so we
could send messages via their classmates if they did not attend school on
the first day we were in their school for testing.  This study took place in
about 8 schools and the testing team was only in each school for a limited
amount of time, and we had a very small window for repeat testing (about 2
weeks) so we may have been able to find more children if the window had been
larger. Also, unless the child moved to another school in the study area,
although we could have found more children as family and friends knew where
they lived, and most moved a short distance, we could no longer include them
in the study group due to the design of the study.

 Unfortunately, for the second study, children who did not enrol in school
by the age of 12 or so were even more likely to move away from the village.
However, we had a less tight time window for testing, and so we were able to
visit homes on repeated occasions.

Our main method of followup when sending messages via other children failed
for both studies was sending a member of the team - either our dedicated
coordination staff, who also planned testing timetables and gave children
pre-testing snacks, or an interviewer who was planning to interview a parent
- to the child's home.  This was often a sensitive matter as parents had
fears about the study, taking blood samples etc.  This was obviously
time-consuming but we did find that direct contact with the parents helped a
lot, as well as holding school and village meetings before each followup
period.

On a slightly more down-to-earth note - I know that the TEDS study asks
parents for the address and phone number of someone (e.g. a grandparent or
friend) who is not likely to move during the period of the study, as an
additional contact avenue.  Another group employs an ex-social-worker as
their followup liason person.

Katie Alcock

Katie Alcock, DPhil
Lecturer
Department of Psychology
City University
Northampton Square
London EC1V 0HB
Phone (+44) (0)20 7040 0167
Fax (+44) (0)20 7040 8581
Web http://www.staff.city.ac.uk/k.j.alcock

> -----Original Message-----
> From: Shelley Velleman [mailto:velleman at comdis.umass.edu]
> Sent: 12 June 2003 01:15
> To: info-childes at mail.talkbank.org
> Subject: attrition in longitudinal studies
>
>
> Assistance urgently needed!
>
> I need to know what kinds of attrition rates people have had
> in longitudinal studies, especially of very young kids (I hope to be
> following mine from 12 - 42 months).  Especially if you've
> had experience
> with low SES groups, I'd really appreciate you sharing your
> experience,
> including any strategies that helped to reduce attrition.
>
> Please include:
>
> Age range:
> SES:
> Strategies that did or did not work to reduce attrition (Please state
> which!):
> Location:
>
> Many thanks!!
>
> Shelley Velleman
> UMass - Amherst
>
>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/info-childes/attachments/20030612/9d11dd8a/attachment.htm>