The Four Parliamentary by-elections on 29 August went well for Azimio and poorly for Kenya Kwanza.

As expected UDA won Rongai in Nakuru, ending the Moi "dynasty" with the defeat of the last family member Raymond, but lost both Pokot seats to KUP candidates (KUP is a local Pokot party but Azimio-affiliated so far) and as expected lost Kitui Rural to Wiper.

​With one other typo corrected (Kilifi North I had wrongly recorded as ODM not UDA) this is the current stage of the parties. I've assigned all independents (who have declared a side now) to their relevant alliance and moved UDM, NAPK, one UPIA MP and MDG from Azimio to Kenya Kwanza to match their recent defections.

This gives KK a narrow majority of 17. Enough to win set pieces like the Speakership but not enough to govern reliably.

Five more updates including final senator and governor for Kiambu

Basically, an extremely good match. No presidential result exceeds 3% more than the maximum of the other results, except Mombasa (incomplete, governor cancelled), Laikipia (incomplete) and Bungoma (incomplete). There is no indication of material presidential vote inflation compared to the other elections.

Given its role as one of a dozen grounds for Azimio's contesting the 2022 presidential results, the four way match analysis proved quite popular, creating a second peak of interest last week and driving unique visits for the month over 10,000. Which was nice.

Some small updates once more, the main addition being Tana River senator and women rep results (complete) and some more partials.

Incremental changes only, with some partial women rep results and the complete Samburu senator results for the first time

ps this has hit diminishing returns - harder and harder to find anything new. Unless someone can uncover a reliable source, there is not much more I can do.

A few people may also have seen this little gem being peddled by the Azimio folks.

Its a silly lie of course - I'm not even on Twitter - but it does reveal a contempt for the truth which should be a warning to Kenyans. These guys just lie.

 The latest news is that Charles Kibiru's results (Jubilee, Kirinyaga Governor) were skipped by the RO when announcing the governor result (probably because he had conceded) so were not transcribed by any of the media. I had a blank entry for him but no number - twitter has since provided a number which looks plausible and which would pretty much exactly fill the gap. Women Rep and Senator are still missing several smaller candidates so we don't have a proper four way match but the two way Pres to Gov is now good if the Kibiru number was good (which I think it is).

UPDATED 23 AUGUST with 6 amendments, including correcting the misattribution of the registered voters to the Homa Bay governor result (apologies). Updates are marked

Warning: This is still incomplete. Many of the results have never been published in full and this is a Frankenstein's monster, reconstructed from fragmentary media reports in the absence of any portal with the combined results.

Key: Red=too incomplete to publish. Yellow-won by KK. Blue won by Azimio. Grey-independent or deferred.

Where you have much lower senate or women representative results, its because I am still missing a few candidates' votes. To assess the presidential poll for integrity, you should compare against the maximum of the other three. If you think you have the complete results for any of the reds, please let me know 

Generally the match is very good and much better than previous elections. The only case I have found which is clearly anomalous and relevant is Kirinyaga, where there genuinely seems to be 20,000 more presidential votes than in any of the other polls.  At the moment I cannot think of a reasonable explanation for this.

Meru is because Tigania East presidential results are still unpublished anywhere I can trust

As this issue forms a small part of the Azimio petition, I think its important to get whatever facts we have out there.

Correction: one ODM Tana River seat was wrongly attributed to UDA in the heat of the moment.
Below is the final state of the house after the 12 nominated MP slots were allocated proportional to party strength. There is one change as I had an error in the women rep numbers earlier. this is before the poaching started

Comparing my eve of poll prediction with the final (with one presentation change to visually separate the county Women Representatives from the constituency MPs as it was difficult to keep track otherwise):

Most predictions were pretty spot on. Jubilee indeed collapsed across much of ex-Central Province (which I had predicted but not fully taken into account) but ODM did slightly better to compensate. This left the two alliances balanced (with Azimio four ahead and four elections delayed) and the four true independent constituency MPs and one true independent woman representative nominally holding the balance of power (for a day or so, before KK started to "acquire" them).
​
Since that time UDM has defected en masse, and KK has also brought on board most independents and a couple of individual  Azimio MPs, giving them a narrow - too narrow - majority of circa 19.

The final 5-way level 1 match (i.e. comparing who won in each contest).

Samburu and Isiolo are the only outliers presidentially, one in each direction

I'm now working on a four-way match report at the more detailed level (minus the MPs for now), but it will be very incomplete as the results for the down-ballot races released by the media are often missing smaller candidates and there are no published official results on a portal.

Details for the record:
​1) Official 34C results including - presumably - reconciliations of every constituency 34A set vs 34B by IEBC staff and including all 291 results:

2) My independent count based first on scanned 34Bs and then amended when IEBC announced corrected (verified) versions. I captured the delta between the two each time for posterity. That count ended around 5pm when I concluded that Ruto's margin exceeded the remaining votes. At the time I had 38 unverified results, including 2 where the 34B was corrupted and one where it was entirely missing (Fafi). Fafi has since arrived in the portal and I have incorporated (unverified) but the 34Bs for Tigania East and Kilifi North remain unusable, so I still have no results for them:

As you can see, the differences are tiny. 

So, in summary, the 34C totals are consistent with the 34Bs as available on the portal and then as announced as final following checking and tidying up during the IEBC Staff and agent verification. So, if there is a problem with the results being somehow "fake", it's not a problem with the IT subtotal not matching the forms, its a problem with the uploaded form 34As and 34Bs and with the entire process of results verification of the physical forms by the IEBC staff and agents. 

As requested, a list of all the winners and their parties in the 47 * 3 county posts, coloured by alliance.
(updated - one error found and fixed in Kilifi)