Newsletter

December 1996 - Number 3

User Questions

This occasional section of the FLS Newsletter incorporates questions of general interest raised by data users. The section deals largely with questions that clarify existing variables and relationships between variables, respondents, and subfiles in the publicly released databases. Topics covered relate to questions not addressed in public-release documentation.

What does it mean if I find implausible height/weight/age combinations in the IFLS-1 anthropometry data?

The IFLS-1 User's Guide contains a short discussion on data quality for the height and weight variables in Book CA. We suspect that some transcription errors still remain in the anthropometry data. Prior to public release, we had found cases where height and weight were recorded in the reverse order and cases where weight had been recorded in pounds and not kilograms. Those cases were fixed before public release. However, we may not have found all such cases. In addition, we suspect there are cases where the anthropometrist wrote the height/weight values for person N on the line for person N+1 or N­1. Such an interviewer error would lead to oddities such as 80-kilogram five-year olds. Such errors are only a few percent of the total number of people with anthropometry data. If users find such oddities, we suggest they print out all the anthropometry records for the household to see if these are cases where the anthropometrist may have written information on the wrong lines.

We would appreciate receiving a copy of any suggested revisions to height and weight variables that users may develop. We would like to add such suggested revisions to our BUKCCA3.FIX file so other users may use the revisions.

In the IFLS-1 morbidity data (Section MA of Book 3), there are people with "how long ago latest episode started" equal to 95 (MA02B=95 ) with values of 1,2, or 3 in the time-unit variable (MA02C)--what does this mean? Also there are a few people who give a value for "when last felt the symptom" (MA03B) that says they last felt the symptom more than 31 days ago, even though they are only supposed to report on symptoms felt in the last four weeks. What might that mean? Finally, there are a few cases where the value for "when latest episode began" is zero (i.e., MA02B=0 ) and the unit (MA02C )is months or weeks ago. What might that mean?

In regards to MA02B=95 values (and time unit is months/days/weeks), those refer to cases where the symptom has been ongoing for a long time. The 95 code means "95+" in 2-digit fields, so if something started 10 years ago and has continued, that is 120 months or "95+". When checking the specific symptoms where these "95 months ago" responses occur, one finds they are largely for illnesses that have long durations--rheumatism, asthma, skin disorders, etc. In addition, for virtually all of these cases, MA03B is "3", meaning the symptom is still being felt.

We suspect that some of the cases where MA03B ("how long ago did you last feel it") is more than four weeks ago and is larger than MA02B ("when did latest bout start") are those where the respondent thought the question was about the last attack before the one in MA02. Some respondents may have been confused about the incident at which MA03B was directed. It's also possible that some respondents thought they had had a bout in the past four weeks, but when pinned down by the later question, they remembered that it was actually longer than four weeks ago and gave a response accordingly.

As to those 0 time amounts, if someone said the last bout started less than a week ago, the interviewers might have written down "0" for time amount and circled the weeks-ago unit. If the response was "less than one month ago," they could have written down "0" time and circled the months-ago unit. Since the MA02 question did not allow for such answers as "less than," it's not surprising that one might see those 0 time and month/week unit responses. In such cases, the interviewers may not have probed for how many days if less than a week, or for how many weeks if less than one month.