"Mapping Migration" Data Online.

Between January and August 2012 data was collected from 4,015 households from 24 villages in the Buhavu chiefdom of Congo's South Kivu province. The data, with information about 8,199 individuals, includes: complete individual-level migration histories (from where, when, why, to where, etc.), village-level NGO and project histories, household-level information about contributions to public goods projects, and much more.

The data is available at: http://petervanderwindt.com/research/mapping/

The dataset comes at three levels and can be downloaded in .zip format:

1. Household level: 4,015 units. Size and format: 69 MB and .dta.
2. Individual level: 8,199 units. Size and format: 97 MB and .dta.
3. Movement level: 18,275 units. Size and format: 128 MB and .dta.

The website also includes all the instruments (surveys, protocols, etc.) used to collect the data.

This data is being used, among others, for my job market paper. However, to make my contribution to research on Congo and migration as large as possible, I make this data available publicly for everybody to use.

For any comments, questions, or errors, do please send me an email at:

pv2160[at]Columbia.edu
 

Migration in Congo's South Kivu province.

To obtain a deep understanding of migration patterns in the DRC I conducted a survey in 24 randomly selected villages of the Buhavu Chiefdom in Congo's South Kivu province -- an area known for its large migrant populations in part due to conflict.[1] The research villages are located close to Lake Kivu and have in recent years been a safe haven for people (mainly Tembo and Rwandan (Hutu and Tutsi)) fleeing attacks by a.o. the FDLR in the mountains more to the west. This is also the area where spent most of the 18 months that I've been in the Congo, including three months living inside these villages to conduct ethnographic research on native-migrant interactions.

For the heads of a total of 4,015 households -- and their spouse(s) -- I collected information about all villages the individuals have lived in during their life (for at least a month), the year for all migrations between these villages and the reason for these movements, including return migrations. The data is rich and includes information for 8,199 adults and a combined total of 18,282 movements.

Migration features prominently in the Buhavu chiefdom. Of the individuals in our sample 85% have moved at least once in their life, and more than two-thirds currently live outside of their village of birth.[2] Migration patterns in the Buhavu chiefdom are best characterized by rural-rural migration. There are two major cities located near the Buhavu chiefdom: Bukavu and Goma, respectively the capitals of the South Kivu and North Kivu provinces. Of the 18,282 movement destinations mentioned, Bukavu and Goma were mentioned only 511 (3%) and 775 (4%) times. The main reason to visit these cities was for work or family, and these visits were often brief -- the majority moved back to the rural area that same year, and almost 90% moved back within two years.[3]
 

Fig 1: Migration in Congo's South Kivu province
Click to enlarge!

The survey also collected detailed information about the reasons for migration asking for each of the 18,282 movements why the movement took place. We find that while people migrate for a diverse set of reasons, two stand out: marriage and security. The figure above plots the number of movements by reason for migration per year, divided by gender. Among the ethnic groups in this area of the Democratic Republic of Congo men marry women from outside their village. The women then moves to the village of her husband. This dynamic is well illustrated in the figure by the absence and importance of marriage as a reason for migration for women and men, respectively.

Unsurprisingly, migration due to insecurity -- indicated in red -- is an important driver for migration in this region. In 2004 in particular we see a large spike in insecurity-induced migration. The reason for this spike was a rebellion by the rebel general Laurent Nkunda who moved in 2004 from Goma to Bukavu right through the Buhavu Chiefdom, which culminated with the struggle for control of Bukavu. The lower panels in the figure plot the share of individuals that moved per year. We find that almost 30% of the population moved in 2004. While migration levels are no longer at its 2004 level, around 10% still moves every year.

Disclaimer:

This is the first set of result from work in progress. Given the rich data more will be done with it.

Footnotes:

[1] The dataframe used was a dataset created together with the International Rescue Committee and CARE International in 2010 that includes all villages in the Buhavu Chiefdom. The selection of the villages was conditioned on the presence of migrants (more than 25 migrant households) and basic safety conditions for the surveyors.

[2] A large survey conducted in 2007 in over 600 randomly-selected villages throughout Eastern Congo confirms these numbers and finds that a full 71% of individuals in the South Kivu sample were reported as having fled at least once at some point during the period from 1996 to 2007 due to armed activities by organized armed groups or militias. See: here.

[3] To be confident that this result is not driven by individuals that did not return to our survey villages I compared per village the household list created from the census in 2012, with household lists that was created in 2010 for another project. The presence of a household on the latter but not on the first would evidence for death or outmigration. For those households that migrated out of the village I obtained their ethnic membership, the reason for outmigration and their destination. Few moved to Bukavu or Goma.

After many months...

a blog post.

It has been quiet for one major reason: I haven't been in the field yet this year. Yes, I miss Africa. I've kept myself busy though over the last months. Four activities stand out:

1. Spring semester 2013 I TA-ed at Columbia for the course "Advanced Topics in Quantitative Research: Limited and Qualitative Dependent Variables" (more here). With the students we worked through the models in detail, and then coded it up in R. I learned a lot. 
2. I've been presenting my work at different conferences: I'm at ten already this year and counting. The most recent two were EPSA (Barcelona, June 20-22) and the Peace Science Conference (Milan, June 24-26). The next one is APSA (Chicago, end-August).
3. With Neelan (Columbia), Ty (Chicago) and Maarten (Wageningen) we collected data in Sierra Leone. We wrote the design (we pre-registered the study at EGAP), and set up the computer code for analysis. The data is in since a month, so a paper will be written soon.
4. Very important, I have been working on my jobmarket paper in recent months. At the end of the year I'll be on the infamous academic "jobmarket", and need something to show of my skills. Writing this is very enjoyable, and while doing so I am analyzing the data collected in the Congo in recent years and come across interesting research articles written by colleagues -- it is largely because of this that I'll be adding posts to this blog again in the weeks to come.

Merry Christmas.

Merry Christmas!

 

For the occassion a lovely song (ht my little brother):

 



Not Random Links.

1. A very interesting read: "Broaden Your Horizons by Randomizing Your Reading Material" (h/t Ali). Since reading "The Diceman" by George Cockcroft (on the right below "BOOKS THAT MADE AN IMPACT ON ME") I have a random number generator on my phone that is normally used to select what to eat from a menu. And, now I'm here in the Netherlands, whether I should go to University by bike or foot (the latter is more healthy, but takes 15 minutes longer).
2. Pics from the WGAPE meeting last May at UC Berkeley.
3. The movie "Stealing Africa" on Dutch TV: info here. Why Zambia doesn't profit from higher prices of copper. Answer: the big foreign companies earn the money and avoid paying tax in Zambia. (h/t Maarten)
4. This is brilliant. A randomly generated math paper was accepted to the "Advances in Pure Mathematics". Read also the reviews by the referees: a.o. "We can’t catch the main thought from this abstract." Duh!
   
5. Interesting read about the administor-faculty ratio at universities. Mmm. Maybe I should become an administrator. They also seem to earn more. (h/t Jenn)
6. Interesting post on iRevolution on the use of email data to estimate international migration rates. See also this for very interesting work done by Blumenstock in Rwanda by making use of cellphones.  

BBC's "Masters of Money": Keynes, Hayek, Marx.

Presented by Stephanie Flanders (BBC's economics editor), "Masters of Money" is a 2012 three-episode BBC documentary about Keynes, Hayek, and Marx (each their own episode). It's quite good (h/t Carlos):

The first episode about John Maynard Keynes:

As events are unfolding in and around Goma.

A blog about the Congo, and mainly about my experiences in Eastern Congo, and I still haven't written about the current events taking place in and around Goma. Well, there are people reporting about this in a much more informed and better way than I could ever do. Please follow among others @texasinafrica; @jasonkstearns; @kvlassen; and @schlindweinsim on Twitter about how events are unfolding.

Today yet another expat friend safely arrived in Bujumbura. I'm happy about that. Unfortunately, my Congolese enumerators, many of them I now consider friends, are still in the area. Yesterday one of them uploaded the picture and text below on Facebook:

 

 

"Merci de prier pour la fin de la guerre en Rd Congo. Pensez à tous ces enfants abandonnées, sans parents ... Clique J'aime et écris ta prière en faveur de ce pays au coeur de l'Afrique. Que Dieu vienne en aide mon pays la RD Congo !!!"

With some of them (both Goma and Bukavu) I'm in almost daily contact and they're doing ok. I don't belief much in praying as the quote wants us to do, but I do hope that: 1) President Kagama has told and keeps on telling the M23 to "behave", and that 2) the FARDC and others do not go on pillage (and worse) sprees.

And how long is your… survey?

When waiting for my plane at JFK last Monday, I was asked to fill out a survey. The US Department of Transport (DOT) wanted to know about the behavior of travelers in order to improve JFK airport. Good initiative, but how useful is a survey like this? It's not random at all! Most people said 'no' when the surveyor asked people to fill out the questionnaire. I said 'yes', but only because I'm doing survey work myself and felt for the surveyor. Let's say that the population consists out of 1% friendly people, and that only friendly people (and the rare confused grad-student) say 'yes' to a survey. Let's say that in this world only friendly people like candy-machines, and that that is what they suggest in the survey as an improvement to the airport. The DOT will thus conclude that "100% of people surveyed want candy-machines" and will spend a lot of money to build candy-machines at JFK. But they’ll build 90% too many of them! It’s a bit stupid example, but you see the point. One needs a proper random sample, and just asking people on the fly is really not doing that.

 

But there is another problem. When I was about 2 minutes into the survey I was no longer interested and, to be very honest, didn't really read more than half of the question before answering. My answers were probably off several times. Now I'm probably particularly quick to lose interest in things, but what does this say about the quality of answers to, for example, lengthy surveys we conduct in Africa for research?

It's our enumerator sitting comfy in that chair. :)

The DOT survey (at least that’s what was written on the front page) takes around 15 minutes. The survey we conducted in the Congo between 2010-2011 with around 10,000 households, took on average two hours! The design can be found here. Now, of course, I'm a particularly impatient person (much more than any African villager I have met). Moreover, our enumerators sit down with the respondents and ask the questions to them instead of the respondent filling out the survey alone. But still, two hours on average per survey cannot do the quality of replies much good.

Of course this was not the first time we thought about this. Two years ago, when creating the design documents, we were very well aware of this potential problem. As a result, we implemented the survey with two variations to learn about this:

1. Random ordering of questions: The first variation was the random ordering of a set of questions. That is, half of the surveys (randomly selected) had sections ordered as X Y Z, and another half of the surveys had the order: X Z Y. 
2. Mandatory break: Second, in half of the surveys (randomly selected) we instituted a mandatory break. We were able to do this because we did not make use of hardcopy surveys, but PDAs. We had programmed the PDA in such a way that at a certain point it would say “Now take a 30 minutes break”, and the PDA would be blocked for 30 minutes. The idea here is that after taking a break the person being interviewed (and the surveyor) would start fresh again improving the quality of the responses. 

Needless to say we added several trick questions to the surveys in order to measure whether people answered carefully. By doing so we can now learn whether the length of the survey is important.

Wageningen University: Here I come.

After a cancelled flight last week due to Hurricane Sandy, I'm currently at Heathrow waiting for my flight to Amsterdam. Destination: the Development Economics Group (DEG) at Wageningen University and Research, where I will be a pre-doctoral researcher during November and December.

I look much forward for two reasons. First, I'll by away from all the districtations in New York. While I have several papers up and running, the most important one (the jobmarket paper) still has to be written. I have many ideas but have not yet had the time to sit down and write it. Second, I expect the DEG a really good place to be. There are a set of researchers doing very interesting work on the political economy of development (impact evaluations, lab-in-the-field-experiments), and I look forward working together. As a start: I'm presenting tomorrow at their seminar.

:)

Blogging has been slow. I leave this Wednesday for two months to the Netherlands (to Wageningen University) and have been working hard to get things finished before, and have a bit of a social life. As an "I'm-sorry-that-I-haven't-blogged-in-a-while" hereby a picture that I'm sure will make you happy for the rest of the day: 

 

(ht to my little brotter)

Vampire Squid From Hell.

Was reading the magazine NewScientist yesterday and there was a short article about the Vampyroteuthis Infernalis, literally "Vampire Squid from Hell". They look like this:

The Vampire Squids from Hell use their thin, rectractable filaments like a fishing line, letting them drift and... collect bits of waste. These guys "munch on any dead plankton, crustacean remains and faecal matter that happens to pass by, making them the only cephalopod not to hunt living prey." Well. That's a bit disappointing given their impressive name and appearance. [1]

In the same issue NewScientist reports research by Kyung Jin Min and colleagues (recently published in Current Biology) that finds how "Eunuchs had an average lifespan of 70-years -- 17 years longer than their non-castrated contemporaries". One theory is that "... testosterone-fuelled reproduction comes at a price. When testosterone is lacking, the body shifts amino acids useful for cell proliferation to cellular defence." Mmm. Given I'm a social scientist, what about this one: Women make men do self-destructive things in order to have sex.

[1] From recently published research in Proceedings of the Royal Society by Henk-Jan Hoving.

What to write in your acknowledgements or preface?

In the last few days I've read several articles by Oded Stark who does very interesting work on (among others) migration. Some of his most interesting articles were written in a time when not only the research's identification strategy was important, but also the idea (these days the balance seems to shift sometimes too much to the first). And interesting research was very much put forward in the 80s and 90s by some of my favorite authors: Oded Stark (economics), Mark Rosenzweig (economics), James Scott (polisci and anthropology), and the like. A bit like some of my current day favorite authors: Herbert Gintis, Samual Bowles, and the like.

In the acknowledgements of his 1985 paper "Motivations to Remit: Evidence from Botswana" with Robert Lucas in the Journal of Political Economy, Oded Stark writes:

“Since this is a joint product, the authors would like to blame each other for all remaining errors.”

Nice!

And yesterday I started (re-)reading James Scott's 1976 "The Moral Economy of the Peasant: Rebellion and Subsistence in Southeast Asia". A must-read (together with all his other books) for those working in the developing world. In the preface he writes:

"At this point in the standard preface it is customary for the author to claim total responsibility for error and wrongheadedness and to absolve others of blame. I am not so sure I want to do that. While I am happy to stand or fall with what I have written, it is also clear that I have learned so much from so many scholars that a great many of us are implicated in this enterprise. If it should turn out that I am on the wrong track, I suspect that many of them are on the same errant train with me!  

I wish also to report that my wife and children, who have their own scholarly and other concerns, had virtually nothing to do with this volume. They were not particularly understanding or helpful when it came to research and writing but called me away as often as possible to the many pleasures of life in common. May it always remain so."

:)

Peter the pessimistic 'Expert'.

The presentation last week (see previous post) went well and the days have been busy since. The main goal for upcoming weeks is to get words on paper. That is, write my dissertation. But, of course, more than that happened last week: among others me being an 'expert' for a SIPA course. This was quite fun.

Students of Columbia's Master "MPA in Development Practices" have to do an assignment in which they have to "apply the concepts and tools from the readings, as well as the class discussions and practical exercises, in order to produce a coherent, well justified design for a country-specific development program." This they then have to present to a panel of four experts who give comments. In addition to three real experts (a professor from SIPA, a professor from the National War College, and the chair of the Millennium Promise), I sat in as well.

A total of four projects were presented, each by a group of 6 SIPA students. Each group received a problem and several (hypothetical) millions of dollar. And each group then had to design a development program (+- 40 pages each) to solve one of the following problems, in brief:

1. Youth unemployment in Tunisia;
2. Low levels of agricultural productivity in Haiti's northern corridor;
3. Slow recovery of Sri Lanka's eastern province;
4. Provincial malnutrion in Mozambique.

Needless to say I know little about Tunisia, Haiti and Sri Lanka, and only a bit more about Mozambique because of my dissertation's Africa focus, and found myself using what I learned from old development economics courses. Of course I also made some remarks regarding evaluations, cellphones and surveys, which is much closer to my current work. Of the four people in the panel I was probably most pessimistic, though. There were three major points that I emphasized:

1. Incentives: All four presentation expected complete government buy-in and participation. I emphasized that they should not take this for granted. While outsiders might have the best of the project in mind, bureacrats or local government officials might not. The latter's incentives might be different: they might see a development project as a way to get a new car or some extra money to buy a television. They might also not been paid for months and very unwilling to take on extra work.
2. Local capacity: All four projects involved new, additional management structures and the need for local capacity. The to-be-created management structures were to be embedded in ministries. Some others structures would be cross-ministry with bureaucrats from different departments sitting in. All projects, of course, were directly reporting to the prime-minister or president. This held even for the 'small' projects; say $10 million. Important is to keep in mind that each of these projects will join another several hundred (if not thousand) projects that already take place in the country: there are so many World Bank projects, each developed country has a large set of development agencies and NGOs, etc. All of these projects want to put management layers like this in place -- often burdening developing countries' governments that are underfunded and do not have enough staff in the first place.
3. Tradition: Each of the four projects would spend a lot of time and money on trainings, workshops, etc. For example, to decrease corruption in Tunisia it was suggested to have a set of information meetings to discuss ways to increase transparency. Or to increase productivity in Haiti it was suggested to hold meetings with farmers and tell them about new farming techniques and seeds. I pushed back on this a lot for two reasons. First, these issues (corruption, gender, farming, etc.) do not change from one day to the next. Farming, to give one example, is based on centuries of tradition. They are often the result of structural issues and do not change due to a workshop. Using one type of seed might not depend on output but risk (i.e. decreasing variability in output). Second, who are we to come in and say "you do X, that's wrong, do Y". I have been working in Congo for the last 4-5 years and still do not know the place. We do not know a country after reading 2 or 3 books about it. I really think that we -- us young-dog but well-meaning Westerners -- have to be more humble.

Was I too pessimistic. Maybe. Maybe not. It is important to mention them I think, though. SIPA students go on to be the world's leaders and while these were hypothetical cases, several of these students will definitely be spending 10s if not 100s of millions of dollars for developing projects in the very near future.

"Characterizing Migrant Integration" at Columbia.

This Wednesday (October 3) I present the paper "Characterizing Migrant Immigration: An Analysis of Pro-Social Behaviors in the Congo" at the Comparative Politics Seminar -- jointly sponsored by Columbia's Department of Political Science and the Institute for Social and Economic Research and Policy (ISERP). This is joint work with Neelanjan Sircar and is work in progress. Chris Blattman agreed to be the discussant. So it is a good Seminar and a good Discussant, now let's hope the paper is up to standard! A preliminary abstract: 

In this paper we seek to learn how migrant populations integrate into new communities. We do so through an analysis of pro-social behaviors -- benevolence and reciprocity -- between natives and migrants in the Democratic Republic of Congo. We design a novel set of dictator games that takes into account the dyadic and network aspects of pro-social behavior, which allows us to retrieve empirically robust estimates of migrant integration.

Feel free to join! Lunch and refreshments will be served at 12:15 and the seminar will run from 12:30 until 2:00. It takes place in the Lindsay Rogers room (Room 707), International Affairs Building at Columbia University.

Fun.

Crowdseeding also in Sweden.

Am happy to see that "crowdseeding" is picked up as a strategy for data collection in addition to "crowdsourcing", and "bounded crowdsourcing". At the bottom of the slide:

Crowdseeding:

• Representative samples
• Small costs

From the presentation "ICT and Conflict Prevention: The Case of a Quantitative Early Warning System" by Gerd Hagmeyer-Gave from the Stockholm International Peace Research Institute (SIPRI).

More on crowdseeding here.

Crowdseeding in Oslo.

Today I presented at the two-day conference "ICTs and the Global Governance of Peace and Security" at the University of Oslo in Norway. It is in interesting get-together with participants from Cambridge U, Oxford U, PRIO, SIPRI, UNDP, EUI, etc. More information here.

I (again) discussed the benefits and limitations of "Crowdseeding" based upon our experiences with Voix des Kivus in the Democratic Republic of Congo. If you're interested, my slides are here.

Random:

• Other than stop-overs by plane, the last time in Norway was about 15 years ago with mom, dad and brothers for holiday (read: the typical Dutch family with car + caravan). Except for the long winters that people keep on talking about here, it's a gorgeous country!
• Next time I do take a jacket with me. It's cold.
• And... it's expensive. To give an idea, a 25cl bottle of wine (+- a big glas of wine) from the minibar is almost $25. One can by a farm in France from that!

Expensive (1 $ is about 5.5 NOK)...

 

 

But beautiful!

Survey Response to Moral Issues.

By making use of a survey 'magic trick' researchers show how people can be tricked into reversing their opinions on moral issues, even to the point of constructing good arguments to support the opposite of their original positions. More in the video below and here.

 

The article continues: The researcher have previously reported this effect, called 'choice blindness', in other areas, including taste and smell and aesthetic choice. “I don't feel we have exposed people or fooled them,” says Hall, “Rather this shows something otherwise very difficult to show, [which is] how open and flexible people can actually be.”

In our household survey in Congo we did something similar (design document: here), although we did not have a magic trick and we were interested in something different. In particular we were interested in what we called the respondents' "social desirability bias". An important subject for academics and policymakers that conduct surveys is: In how far does the reporter actually tell the truth? Is the response given really his or her opinion, or did he or she just reply in such a way to give (what the reporter thinks is) "the correct answer" -- the socially desirable reply.

The final evaluation report was published several months ago (here) but we are currently in the process of writing the secondary analysis report, and this social desirability bias is one of the many things we look at. How? We have two statements:

• A “Many NGOs in the region believe that elections are not appropriate to choose community representatives when it comes to a position of technical responsibility. “
• B “Many NGOs in the region believe that elections are always the best way to choose community representatives, even for positions of technical responsibility.”

The statements are followed by the question: “Do you agree that elections are the best way to select representatives?” Now 50% of the respondents received first statement A, and then sufficiently later in the survey statement B. For the other 50% of respondents A and B are swapped. As a result, for each individual we receive two observations: their response to “Do you agree that elections are the best way to select representatives?” after statement A, and once after statement B. The social desirability effect is the effect of the prime on the likelihood to agree with that statement. Respondents that give socially desirable answers will say "yes" to the question after statement B, and "no" to exactly the same question after statement A. In addition to this within-person analysis, we also conduct a between-person comparison where we use only whichever statement was asked first for each person (which removes consistency biases), and regress the answer on the prime given.

More soon!

Can cellphones be used to learn about conflict events?

Between 2009 and 2011 Columbia University implemented Voix des Kivus in Eastern Congo; an SMS-based pilot project to obtain high quality data about conflict events in real-time from hard-to-access areas. To obtain high-quality data Columbia invented "crowdseeding" -- in contrast to data-collection projects based on "crowdsourcing". Last week the academic study came out discussing data quality and using the data for a downstream experiment to assess the conflict effects of international aid. The abstract:

Poor quality data about conflict events weakens humanitarian responses and hinders academic research on the dynamics of violence. To address this problem we piloted a data gathering system in Eastern Congo in which reporters in randomly selected villages reported on events in real-time. We describe the data generated through this system and use it to implement a downstream experiment that illustrates how the data can be used. We take advantage of exogenous variation in the allocation of development aid to assess whether aid is associated with increased or reduced violence. Our data suggests that aid had a negative effect on conflict. Critically, by exploiting the continuous nature of our data, we also highlight the sensitivity of estimates of effects to the timing of measurement.

The complete study can be found HERE.

I'm on Twitter.

Since a few minutes I'm on Twitter. Let's see if I can get into this:

https://twitter.com/vanderwindt