The Data Liberation Project

A few days after the May 6, 2018 parliamentary, the Ministry of Interior (MOI) published the results. The initial results were in a PDF format, with limited information. They lacked a breakdown of participation or voting per district. As an avid analyst with an interest in data and patterns recognition, that was very frustrating.

PDF from MOI
Initial results in PDF, from MOI

Fortunately, the MOI released the full results a week later. However, the MOI also published these results in a static PDF format, grouped by the preliminary counting committees, severely limiting any possible analysis and cross linkage.

Consequently, I started working on possible methods to convert and reorganize the data to make it easily accessible, and then cross linking it to gender and sect, to allow a deeper and richer analysis. So I took my home district Saida as a case study and went to work.

The process was divided into three phases:

  • First step, converting the static PDFs into machine readable Excel
  • Reorganizing the data by neighborhood
  • Adding the Sect and gender labels of each ballot bureau, by using the MOI data base.
reorganization by neighboorhood
Reorganization by Neighborhood

It took me two weeks to finalize the Saida district (one of the smallest in Lebanon), allowing me to use the newly organized data to write a deep analysis of the district’s results, learning about the turnout and voting patterns by gender and sect. Moreover, the data opened up another level of analysis, such as the linkage between women turnout and voting, and the presence of strong women candidates.

adding the sect and gender column
Adding the Sect and Gender Columns

However, the amount of time and efforts the first case study took, convinced me that this should be a larger team effort. I contacted several institutions to propose possible cooperation. The National Democratic Institute (NDI) expressed interest and they offered to support the project, and this is how the Data Liberation Project (DLP) started. After the initial proposal was accepted, I hired several young and enthusiastic experts, organized the workflow, planned the different phases, and prepared the tasks. The Project was launched and we slowly started the conversion and the reorganization of the elections results.

Two months later, the data was finalized, and after a long and assiduous editing and corroborating effort, we submitted the finished project, to the satisfaction of NDI and team.

Consequently, it is my honor and privilege to freely offer the full results of the Lebanese 2018 parliamentary elections, cross checked with gender and sects, for the general public, students, and experts.

The results are availed upon request, via a google drive link, by contacting me directly or by using this email: dlp.lebanon@gmail.com

Finally, I would like to thank the whole NDI team, for their unwavering support, professionalism, and help in completing such an important project, that aim to provide free and open data, considering how hard it is to get unbiased, objective, and solid data in Lebanon. It goes without saying that the project would not have been possible without the young men and women who worked tirelessly for its success, especially Fouad Saoudi and Lana Skaffi, among many others.

Advertisements

The US Midterm Elections!

Usually US mid term elections are a routine affair, where the people cast a vote supporting or censoring the current president and his policies.

This year the elections are crucial, to the US, and dare I say, to the world. Trump won the presidency riding a wave of authoritarian popularism, that spread to Eastern Europe and reached Brazil. Using xenophobia, latent racism, and the economy changing from an industrial focus to a modern digital and open one, these autocrats have reached power and started dismantling the Democratic Institutions that allowed them in.

This election is the Gettysburg of this age. Does the hightide of authoritarian popularism wane at the shores of Lady liberty, or does it engulf the world in another age of extremisim, similar to the 1930’s

In a few hours we will see, it all rests on turnout, on the willgness of young people and minorities to vote, en mass!

Go vote!

A detailed analysis of SAIDA’s 2018 electoral result!

Rafik Hariri’s legacy can be traced back to this once small city, on the Mediterranean, with its plentiful orange orchards and long history. He was born there, and his siblings and distant family are still living in the city. Rafik Hariri helped rebuild and renovated Saida’s infrastructure after the 1982 Israeli invasion, building roads, school, and various projects.

Thus, the two Sunni MPs of the city have always had a symbolic importance. In 2009, for the first time since the formation of the Future Movement (FM) party,  it was able to win both seats. Unfortunately, in 2018 FM failed to repeat this feat and their old adversary Oussam Saad took the second seat in the place of PM Fouad Sinioura.

It is worthy to note that in 2009 Mrs. Hariri won with 25,500 votes to Oussam Saad’s 13,500. Indeed, MP Hariri received almost 64% of the total vote, with a record setting 68% participation, the highest in Lebanon. In 2018 Future Movement numbers fell in Saida to 15,308 against 10,255 for Oussam’s, with only a 54% participation!

Picture6

So what happened? In this deep analysis of the results, I will try to find patterns in the numbers, and draw some conclusions. I will tackle this in two sections: turnout and voting patterns! This analysis has been made possible with the help of NDI, with whom i have been working on a data liberation project, to transform the Ministry of Interior official detailed PDF results into machine readable excel, and add sect and gender information. To learn more about this project please use this link.

 

First, turnout:

Picture1

It is clear that turnout fell significantly, especially among Sunni! Based on the a preliminary analysis of the numbers (link) this trend among Sunni’s seems widespread to most districts with Sunni majorities (Beirut 2, Tripoli, …) Indeed, during the last 9 years 10,000 new voters registered in Saida, with a majority of Sunni, however less people voted in absolute term in 2018 than in 2009

Even more significant, there are more female than male voters in Lebanon. Almost 53% (according to UNDP). Additionally, in relative numbers, women turnout is higher than men’s, yet in Saida, Sunni women turnout was lower than men, at odds with Shia and Christian women!

Picture4

 

Picture3

Women turnout, as previously mentioned, was lower than men’s. Even though the head of FM list in the district was a women, but that did not raise the turnout.

Although, Sunni women voted less than men did, but they supported MP Bahiaa at a higher percentage than men. (48.37% of women, voted for MP Hariri, as opposed to 44.90% of men)

This are very important data points, and once the rest of the districts are analyzed, it will be possible to draw correlation between several variables. For example is women turnout affected by the presence of  strong women candidates, and do women vote more for women candidates?

 

Second: voting patterns:

I believe that Saida’s voting patterns might hold true for most of the other districts, and i will try to show that in future analysis. Indeed, Shia voters voted in mass and with a very high adherence to party instructions. Meanwhile Sunni voters were lethargic, and their vote was split and dispersed among many lists (see Beirut II as a perfect example), severely lowering its effectiveness. On the other hand, the Christian voters were more fired up, and more disciplined, but divided into different parties (FPM, LF, …) this sometimes played against them, in Jezzine for example.

Thus in Saida, Sunni turnout fell heavily, and from a 64% support to FM in 2009, to just 46% in 2018. On the other hand, shiaa voters in Saida adhered strongly to their party’s instruction and voted, cross confessionally, to their sunni ally at a very high 73% rate. This trend of shia voters, is especially salient in the Southern and Baalbek districts reaching high 80 or even 90% adherence. In Jezzine too this was even more evident, shia voted at a dizzying 92% for their Christian ally Azzar.

Picture6

Picture2

 

Lastly, preparation, training, and information, pay and pay very well. Saida suffered of one of the highest rates of invalidated ballots, at 3.14%. Meanwhile, in Jezzine it was half of that, at 1.55%. With the Ministry of Interior being busy with the overall organization of the election, it fell to the parties and NGOs to train their people and their supporters on how to vote and how to help voters during the process.

Thus, certain parties (Lebanese Forces come to mind) were able to prepare their electoral machine, train their people and observers, inform their electors, and they reaped the benefits, by doubling the number of their MPs. For example, their regions of Bcharreh had the lowest rate of invalid ballots at 1.71%. Others did not, and the invalid ballots rate in their strongholds was fairly high, robbing them of precious seats. For more analysis on the invalid ballots and turnout, check this link.

Finally, it is worthy to note that despite rampant sectarianism and the ugly inflammatory rhetoric that preceded the elections, there was small but significant cross-sectarian voting. 12% of Saida’s shia voted forMrs. Bahiaa Hariri and 34% of Christians, and she received 1,100 Christian votes for Jezzine. I firmly believe that political parties, focusing on issues politics (waste management’s, electricity, and environment) with a clear and organized agenda can make significant headway in the Lebanese political arena.

This leads us to the a very important question on why the civil society wave started in 2016 with Beirut Madinati receiving more than 35% of the vote in Beirut’s municipal elections, did not make a significant breakthrough in the parliamentary elections, winning only one seat. A question that I will deal with in depth in the next several articles.

Details of the mistakes

Unfortunately, as i mentioned in my analysis of the turnout numbers, there are two mistakes that i was able to catch in the official numbers.

the first one:

it concerns Zahleh’s fifth counting committee, where a ballot box usually holding 600 registered persons plus or minus (as inferred by most ballot boxes in this sub-committee) had 315,617 as the number of registered. so to solve this issue i assumed that it was 617.

mistake one

The second is similar to the first, but in Tripoli. Here too, a ballot that was supposed to hold in 600’s was put in for 62,581. again i corrected it as 625.

mistake two

A first look at the numbers!

The detailed results are finally out, so after a thorough analysis of the turnout and other general numbers, I arrived to these conclusions. Unfortunately, before going into the analysis, it is important to note that I discovered two errors in the documents published on the official website (one added 62,000 and the other added 315,000 to number of registered, heavily skewing the turnout). I tried to correct them to the best of my ability. You can find more details about these errors in here. Fortunately, these errors are limited to the registered numbers, and do NOT affect the results. However, the fact that there are two errors might indicate that there are more…

First, the ‘official’ turnout number of 49.2% announced by the Ministry of Interior on Monday May 7, is not exact.*

The turnout number of the 2018 election is 48.02% (1,861,203 voted out of 3,875,981) There is a 5.35 % drop in Turnout between 2018 and the 2009 elections. Most of the districts saw a drop. The only district that showed an improvement in turnout was the Bekaa III (Baalbeck – Hermel), with an impressive 9.44% improvement. Meanwhile, another four direct held almost the same turnout between 2009 and 2018 Mount Lebanon IV (Chouf Aley) with -0.38 , South III with +0.38, Beirut (as the districting of Beirut changed between the elections, I had to make a combined turnout for both Beirut I and II) with +0.61%, and Mount Lebanon I (Jebeil Kesrouan) with -1.16.

turnout 2018 2009

Concerning, the blank and invalid ballots percentages, there was a similar trend for the blank ballots with 0.81% in 2018 compared to 0.68% in 2009. It is worthy to note that in three districts the blank ballots rose above the average reaching more than 1%, in South I (Saida and Jezzine), South II (Tyre – Zahrani), and North II (Tripoli, Minyeh, Dinnieh). In South II, the rate is understandable as there was only two competing lists, and many felt that neither represented them. So is the South I rate, where the field was highly politicized, but with no civil society list presents. However, for North II, it is a point that I will tackle after talking about the invalid votes.

As expected, there was a large increase in invalid ballots, due to the introduction of a new proportional electoral system. The rate of invalid ballots jumped almost 300% from 0.61% in 2009 to 2.09% in 2018!! In a district by district comparison the higher trend was similar in most, with the exception of the North II (Tripoli, Minyeh, Dinnieh) district that had 3.52% of invalid ballots, a full percentage point above all other districts!!

invallid ballot

Both in blank and invalid ballots North II district is above average, in a statistically significant way. It would be interesting to keep an eye on this district, with all the judicial procedures being made. This district, has a history of having slightly above average Blank and invalid ballots, but that alone does not explain these big numbers. Indeed, in 2009 Tripoli was at 1.15% for invalid compared to a national average of 0.61%, and 1.14% in blank ballots for an average of 0.68%. However, it was not the district with the highest number, and was fairly in line the overall curve.

*All conclusions and numbers in this analysis are based on the detailed results published by the Ministry of Interior on this website. If anyone is interested by the excel sheet with all the numbers extracted from the official PDF documents, you can find it here.

Still no official results!

It’s been three days, and no official results has been made available. the Minister of Interior announced on TV the list of the winning candidates, without Akkar, where apparently recounts were being done. The results were read live, and no detailed results were supplied.

Officially that was it.

Meanwhile, unofficial numbers are floating around, such as these results of Beirut-2, and a semi-official list of winning candidates and their preferential votes was circulated on whatsapp.

We are still missing number of voters, invalid and white ballots, lists votes, losing candidates results… Unfortunately, the Interior Ministry official website is still empty and the wait goes on!

Prileminiary analysis!

The official results and detailed numbers are still not out, casting a lengthening shadow on the electoral process.

Nevertheless, there are a few conclusion that can already be made:

1- the monolithic iron control of the Shiite duo (Amal and Hezbollah) over their constituency and regions, is intact, almost absolute, and crushing. They won a full slate in two southern regions, in a PROPORTIONAL law, with the precense of a small but significant minority in one of these regions.

2- the Lebanese Forces have significantly benefitted from the law, as their support is strong but scattered over many regions. Their numbers of MPs jumped by over 100%

3- The Aounist were able to limit their losses and maintain a large (possibly the largest) parliamentary block

4- Future Movement faced a daunting fight and had two challenges to overcome (check this for a more detailed analysis): a- energize their base and encourage the reluctant to come back into the fold. b- to significantly raise the turnout in all their regions, to increase the threshold and consolidate their support, while limiting the small lists ability to win seats. FM was successful in energizing its base but failed at raising the turnout. The result? FM parliamentary block has shrunk.

5- consequently, following the previous points, we now have a large block (that should reach 40+) formed by Hezbollah and Amal, including around 10 Sunni MPs, in addition to all the Shiite MPs except one.

Once the final results and detailed numbers are out, I will write a series of analytical articles.