ROAD TO KENYA ELECTIONS 2017: CAN THIS SOFTWARE PREDICT THE WINNER?

By Stan Oyunga (stanoyunga@yahoo.com)

INTRODUCTION
In exactly two years from now (August 2017), Kenyans would have re-elected Uhuru Kenyatta for a second term as President or would have shown him the door and made history as the first country in East Africa to have replaced an incumbent President; or the leading Presidential candidate would have failed to garner the 50% +1 figure required to be declared elected and Kenyans would be preparing for a run-off Presidential Election for the first time.
Whether it will be Uhuru Kenyatta or Raila Odinga or another person altogether who will emerge the winner is hard to predict and opinion polls have proved inaccurate in both the 2007 and 2013 Presidential Elections.
For the first time in history of Presidential elections in Kenya, a software has been developed to do just that: predict the winner of the 2017 Presidential Elections. By using past voting patterns, voter turn-out, projected voter registration and ethnic block voting data, the software has attempted to predict the winner using a baseline data of 80% voter registration and 80% voter turn-out.
Since not all counties will achieve 80% or even 90% voter turn-out, the ultimate winner will depend which candidate can attain the highest voter turn-out in his stronghold. It will also depend if the two main coalitions can retain their core supporters and ethnic block voters.
The Kenya Election Database Version 2.0, is an Election Data Analysis and Strategic Election Planning software that holds full results of Kenya General Elections since 1969, population census for 1999 and 2009 and population projection for 2015 and 2017. It is still the only single source of Kenya General Election Results Data in the market.

COALTION POLITICS

The winner of any free and fair elections in most African countries depends on certain parameters such as Ethnic group block voting, political parties/coalitions and voter turn-out.
The ruling party/coalition usually emerges the winner due to the influence of the incumbency but recent elections in Zambia, Malawi and Nigeria has shown that this is not always the case, when the election is deemed to be free and fair.
Since 2002, the two party/coalition system has been evident in Kenya and has worked well except for the disastrous 2007 Presidential Elections. There is a clear pattern in Ethnic block voting and coalition politics that has created coalition strongholds since no single ethnic group has the numbers to produce a winning Presidential candidate. Refer to my blog “Road to Kenya Election 2017: Who will have the numbers in 2017?” found at www.kenyaelectiondatabase.co.ke .
The two main coalitions in Kenya, the Jubilee Coalition and the Coalition for Reforms and Democracy (CORD), control 18 and 19 counties respectively which can be considered their strongholds. 10 other counties are considered “Fifty/Fifty” (or “battle grounds”) where both coalitions are more or else equal:-

[table id=44 /]

[table id=45 /]

[table id=46 /]

There is no guarantee that both coalitions will hold together for the next two years but it is likely they will and the 2017 General Elections will be a repeat of the 2013 Elections and remain a two horse race. Only one or two dramatic watershed events can alter the makeup of the coalitions and lead to a paradigm shift (e.g. death of a Presidential candidate or Coalition principal/party leader or conviction of Deputy President William Ruto by the International Criminal Court in The Hague).

2017 REGISTERED VOTER PROJECTION

The first step in attempting to predict the results of the 2017 Presidential Elections is to project the number of eligible voters by county as shown in the tables above. To achieve this the software first projects each county’s population for 2017 (based on average annual growth rate of 2.4%) then calculates the number of voters at 40% of the projected population, since the average percentage of Kenyans who register to vote is 40% (equal to 80% of adult population).
The software has calculated Kenya’s population in 2017 to be 48.3 million and estimated the eligible voters at 19.3 million if 80% of adult voter registration is achieved. Some counties may exceed 80% and some may be below, hence the 80% is considered average for all 47 counties.

2017 PROJECTED VOTES

The next step is to project the number of votes each coalition may receive in the 2017 Presidential Elections by county. To do this the data from the software was transferred to a spreadsheet and the projected votes was calculated as a percentage of what the coalition received in the 2013 Presidential Elections for each county. This assumes that the voting pattern in 2017 will follow that of 2013, unless there is a paradigm shift in coalition politics as explained above.
80% voter turn-out was taken as the baseline for each county considering the turn-out for the 2013 General Elections was 86%, the highest in Kenya’s history. The constituency turn-out for the 2013 Presidential Elections ranged from a high of 96% (Rangwe) to a low of 58% (Kilifi North).
For example Nairobi is expected to have a projected population of 3.9 million in 2017 and estimated voter registration of 1.5 million (40%) and voter turn-out of 1.2 million (80%). CORD’s valid vote share in 2013 was 51% while Jubilee’s was 47%, hence CORD’s projected vote in Nairobi will be 640,739 and Jubilee’s will be 590,485. This figure is on the lower side since some constituencies in Nairobi record voter registration of above 100% of adult population due to non-residence of these constituencies coming to register there (voter importation). They include Starehe, Makadara, Embakasi East and Roysambu.
The spreadsheet cell entries are as follows:-
COL H: 2017 Population Projection
COL I: 2017 Registered Voter Estimate (40% of H)
COL J: 2017 Projected Votes (80% of I)
COL K: CORD Valid Vote % in 2013
COL L: Jubilee Valid Vote % in 2013
COL M: Total Valid Vote % in 2013 (K + L)
COL N: CORD Projected Votes (K% of J)
COL O: Jubilee Projected Votes (L% of J)
COL P: Total Votes (N + O)
To obtain the projected votes for the 2017 Presidential Elections I have prepared 4 spreadsheets using the above formulae:-
1) Report 3A-COUNTIES PROJECTED VOTES IF TURN-OUT-80%-ALL
2) Report 3B-COUNTIES PROJECTED VOTES IF TURN-OUT-80%-CORD
3) Report 3C-COUNTIES PROJECTED VOTES IF TURN-OUT-80%-JUBILEE
4) Report 3D-COUNTIES PROJECTED VOTES IF TURN-OUT-80%-FIFTY-FIFTY

2017 PROJECTED RESULTS

According to the spreadsheets which are attached to this blog, if the 2017 Presidential Elections has a 80% turn-out, then the CORD candidate is projected to receive 8.7 million against 7.1 million for the Jubilee candidate, if all 47 counties record a 80% voter registration. If any coalition fails to achieve 80% voter registration and/or 80% turn-out then the projected votes will reduce accordingly or even increase if 90% or higher is achieved for both parameters.
Total projected votes for 2017 is 16 million compared to 12.3 million in 2013. Total projected registered voters will be 19.3 million compared to 14.3 million in 2013 against IEBC voter registration target of 18 million. The IEBC wants to register 8 million new voters by 2017 in order to reach its target of 23 million voters. But the software has projected that IEBC will register only 6 million new voters by the 2017 General Elections

CONCLUSION

The purpose of this software and the spreadsheets is to assist Election Managers in planning the 2017 Presidential campaigns and setting targets that can assure their respective candidates victory. Cutting deals to lure ethnic block votes to their coalitions will be the norm.
It is also to assist Kenyans to see the importance of registering as voters and turning out to vote. CORD strongholds had more potential eligible voters then Jubilee’s but due to failure to register as voters and turn-out in large number to vote in the 2013 General Elections, CORD lost, leading to the infamous “Tyranny of Numbers”.
Overall a total of 5.6 million Kenyans in the strongholds of the 2 top candidates, did not take part in the 2013 Elections, out of which 2 million were registered to vote but did not for various reasons (no one has done any research as to why registered voters do not vote on the actual voting day), 3.6 million did not register at all for various reasons including lack of identity cards but some choose not to do so (boycott?) or lack of interest (millennials or generation y?).
Out of the 2 million registered voters who did not vote in the 2013 General Elections, 1.2 million were from CORD strongholds and over 700,000 were from Jubilee strongholds.
Out of the 3.6 million eligible voters who did not register, 2.4 million were from CORD strongholds and 1.2 million were from Jubilee strongholds.

21-AUG-2015

© 2015 STAN OYUNGA
REPORT-3A-COUNTIES-ELECTION-PROJECTED-VOTES-IF-TURN-OUT-80%-ALL1.xls (257 downloads)
REPORT-3B-COUNTIES-ELECTION-PROJECTED-VOTES-IF-TURN-OUT-80%-CORD1.xls (241 downloads)
REPORT-3C-COUNTIES-ELECTION-PROJECTED-VOTES-IF-TURN-OUT-80%-JUBILEE1.xls (248 downloads)
REPORT-3D-COUNTIES-ELECTION-PROJECTED-VOTES-IF-TURN-OUT-80%-FIFTY-FIFTY.xls (242 downloads)

Leave a Reply

Your email address will not be published.