Iraq Maps

Maps Illustrating Main Street Bias and other Sampling Issues

In The Lancet paper the authors write:"The third stage consisted of random selection of a main street within the administrative unit from a list of all main streets. A residential street was then randomly selected from a list of residential streets crossing the main street. On the residential street, houses were numbered and a start household was randomly selected. From this start household, the team proceeded to the adjacent residence until 40 households were surveyed." They have clarified that main streets are "major commercial streets or avenues".

What does this mean in practice? What kinds of areas can and cannot be sampled? What are the chances of selecting one kind of household compared to another kind of household? The above language is too vague to answer these questions definitively. For starters, we cannot be sure of what is a major commercial street or avenue. But it is possible to explore some scenarios and we do this below. We encourage everyone to do their own explorations based using (the miraculous) Google Earth.

Note that the researchers have not yet released information on which towns they have visited so the discussion below necessarily is about what might have happened if these places were sampled.

---

This map covers an area of Kirkuk. We highlight the roads that we consider major commercial streets or avenues in red. We highlight the crossroads to the major roads in yellow. The blue shaded area contains residential building that could not be sampled under the authors' stated scheme and accepting our classification of major commercial streets or avenues. If these were indeed the major commercial streets chosen in The Lancet paper, then only a tiny sliver of Kirkuk could have been sampled. A more liberal rendering of major commercial streets or avenues would enable deeper penetration.

Note that the sampling scheme used in The Lancet study can also reach households located off of crossroads to main roads,  because, although the starting point is always a household located on a crossroad to a  main road,  some of the 40 households nearest to the first household may be located on side roads to the selected crossroad. Yet they will still mostly be close to the crossroad which served as starting point for the sample.

---

This map of Tikrit really pushes the envelope in our view on what qualifies as a major commercial street or avenue. Indeed, Tikrit would have to be quite a bustling commercial zone for all the red streets in the picture to meet the criterion.

Again, red shading indicates a main road and we have allowed many such roads. The blue roads are the cross roads to the main roads and the pink shaded areas cannot be reached by the sampling scheme even under the liberal rendering of main roads (The picture is not quite complete. There should be a bit more pink shading.) There are still big stretches of territory outside the sampling scheme.

The researchers in The Lancet study have stated that the interview teams arrived in a town and then drove down the major commercial road writing down the names of the cross streets on pieces of paper, and still carried out all the 40 interviews, in a single day. This statement might indicate that there were only relatively few streets considered as main streets. Unfortunately the list of streets considered as main streets in The Lancet study is unavailable.

---

And what about rural areas? The researchers also sampled rural areas. It is not clear to us how the major commercial street sampling would work when there is no major commercial street available. Consider the map of Akre district. It is possible that if Akre district was selected in the early stages of the randomization procedure then the interview team would have been sucked right to the main road in the town of Akre.

---

More generally, think about your home town. What are the cross streets to the main roads? For most towns we know of there will be great ambiguity in this question but it will still be clear that most households will not meet the sampling criterion.

And again, please go to Google Earth and explore Iraq.