2017 Polls Analyses and Assumptions Using Math

Much of the analyses found on social media about the outcome of August polls have largely been informed by gut-feel and, depending on the analyst, fuelled by ethnic bigotry.

And if you have been listening to TV panelists during prime time news, you will agree with me that opinions are long on popular public chatter but short on trend analyses and number crunching.

Using the latest IEBC figures, good old Microsoft Excel and factoring 2013 voting patterns, I took advantage of the long Eid ul Fitr weekend to undertake emotion-free analyses of the chances that the two leading candidates for the presidency, Uhuru and RAO, have of occupying State House come August.

In my analyses, I have found that if RAO in August ‘inherited’ 5% of votes that UK got from EACH county in 2013 (but using latest 2017 official figures), the incumbent could still register a marginal win by approximately 128,000 votes.

But, hey, I do not claim to have monopoly of accurate assumptions. Mine are based on math, reasonable probability and, importantly, 2013 voting patterns.

What I did was study the actual stats from the 2013 polls. I looked at number of votes cast per county, voter turnout by county, and the votes that the two leading presidential candidates, RAO and UK, got in each county, all 47 of them. My analyses have been made simpler by the fact that like in 2013, RAO and UK are again the two leading candidates in 2017.

Different from the 2013 polls, IEBC created three more national ‘constituencies’ after recent voter recruitment drives. These are 1. Huduma centres 2. Universities and 3. Diaspora.

For these three constituencies, I have assumed that the two candidates will get votes at percentages similar to those heavily cosmopolitan Nairobi gave them in 2013.

In IEBC’s voter recruitment drives held post-2013, it is clear that zones perceived to be pro-Jubilee registered comparatively higher numbers of new voters than did NASA strongholds.

And did you know that Kajiado, which boosted in just four years its 2013 number of voters by 20%, and Mandera, are among counties that exceeded by a wide shot IEBC new voter registration targets? Interestingly, in 2013 Mandera defied the voting pattern of other NEP counties. Mandera is the only county in its province that overwhelmingly voted UK (93.2%) in the last polls. In the other counties of NEP, RAO beat UK.

It will also be interesting to see how Kajiado votes this time. In 2013, RAO and UK got 44.8% and 52.8% votes from this county respectively. But Kajiado has since registered a massive wave of new voters. I believe majority of these are natives of central Province. Why? Because Kajiado borders populous Kikuyu areas and, secondly, the House of Mumbi is certainly the single largest immigrant community in this county.

Anyone can make assumptions on how the 2017 polls will turn out. In my analyses, I decided to look at a reasonably possible worst case for the incumbent, UK. I assumed that relative to the 2013 polls, his performance per county – including his strongholds, will this time in percentage terms worsen by 5%. Note use of ‘5%’. Not ‘5% points’.

If Uhuruto’s camp haemorrages 5% of the percentages won in each county in 2013, it means someone has to benefit from this loss. So I applied this 5% loss of Uhuru percentage votes per county in favour of RAO.

In short, I assumed that come August, RAO will ‘inherit’ this loss of UK votes. Let’s call this assumption the Uhuruto voter haemorrage scenario.

And I have used latest IEBC figures per county in my assumptions. I looked at what each of the two leading candidates got in 2013, and what the voter turnout was in each of the counties then. I have further assumed the same voter turnout per county in 2017 as it was in 2013.

Then I used the latest 2017 numbers from IEBC, ‘discounting’ 5% from the percentage (not the number) of votes that UK got from each county, including Uhuruto strongholds, in 2013, and applying that ‘discount’ to boost RAO’s tally, based on what the latter got per county.

Let me illustrate what I mean, using one county as an example:

From IEBC stats, 90.8% of all votes cast in Kiambu in 2013 were, needless to say, in favour of UK. So what I have done in my analysis is discount the 90.8% votes by 5% to come to 86.3%, which is the percentage that I have, perhaps harshly, assumed UK will get in August.

And since RAO actually got 7.9% of votes cast in Kiambu in 2013, my assumption is that in 2017, he will harvest an additional premium of 5% of the percentage vote he got in 2013. Thus the 7.9% for RAO in 2013 becomes 8.3% in 2017.

In short, I have assumed that UK’s loss in each of the 47 counties will become RAO’s gain. And when I simulated the Uhuruto voter haemorrage scenario across all counties (plus the three ‘constituencies’) on MS Excel, it turned out that UK would garner 8,018,906 votes against RAO’s 7,890,093.

Moreover, I have not seen the need to add Mudavadi’s votes to RAO’s tally. Mudavadi garnered 484K votes in 2013. Whilst this figure seems significant, the newly registered voters in pro-Jubilee zones will put a check on that.

For example, Nyanza and Western combined registered about 752,000 new voters since the last elections. On the other hand, Central and Rift Valley combined registered close to 1.4 million new voters! And, mark you, I have not even included Upper Eastern’s (Meru) avalanche of new voters.

Also, I have heard some observers argue that Isaac Ruto will inflict significant loss of votes for the Jubilee camp in Rift Valley, and Kipsigis areas in particular. I am not sure Isaac Ruto has enough clout to spoil the Jubilee party in a significant way.

Consider this: At the height of the spat between the two Rutos, William and Isaac, Charles Keter, a Kipsigis from Isaac’s backyard, was appointed to the Cabinet, thereby occasioning a by-election. Contrary to widespread expectation, the vacant seat was not won by Chama Cha Mashinani but by Jubilee.

In any case, voters in 2017 are far more enlightened nowadays to discern when a candidate is a ‘spoiler’.

Overall, if I were Jubilee, I would work overtime to creatively charm, and win, new voters from NASA areas, thereby creating splits. Jubilee seems to be doing just that. On the other hand, NASA would do well to avoid talk of ‘us’ and ‘them’.

Why? Because their greatest chance arguably does not lie in traditional NASA areas but in Jubilee ones. Remember, NASA seems to have more political goodwill now in Jubilee areas than it did in 2013. In fact, Raila’s chances lie in his camp prying significant chunks of votes from Jubilee zones, and Nairobi.

Thus NASA should be busy staging charm offensives in pro-Jubilee zones, not antagonising them by using innuendoes that have a whiff of revenge, or hate. But alas, they dont have much time.

As was the case in 2013, voter turnout in their respective strongholds will be extremely key for the two front runners.


4 thoughts on “2017 Polls Analyses and Assumptions Using Math

  1. Hk

    Nice analysis. However I note your bias. My heading would have been ” how jubilee will win ” then perhaps you can analyse with bias again ” how NASA will win ” overall draw conclusion from the two analysis.


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s