Looking at Internet searches in the last 3 months to the U.S general election, steady dominance by the overall winner Barack Obama compared to few for Mitt Romney pointed to the relevance of searches in indicating political superiority. With this in mind, my focus shifts to Kenya just 3 months before the general elections.
I focus on the most preferred candidates according to opinion polls carried out by various research firms. Although I am aware that only about 30 percent (14 million) of the entire population are Internet users, we can test the limits of how accurate the Web can be used for elections predictions.
But before I continue to analyze the statistics, I proceed to survey Internet domination ahead of the 2007 general elections when only less than 20 percent (some 8 million) of the population were Internet users. In all but one occasion, presidential aspirant Raila Odinga led the polls, as he did in opinion polls carried out by the Steadman group, although by a larger margin.
According to the Steadman polls, Raila would beat Kibaki in the December 2007 general election by a margin of just 0.5 percent and 4 percent of all the voters participated. In the Internet statistics however, Raila was leading by double digits with average daily numbers showing he was getting almost double the searches. As we all know Kibaki was declared winner amid protest by Raila winning by less than 4 percent.
Five years down the line and the numbers on the Internet are much higher, almost the number of voters in Kenya. The numbers are however much lower given that among the users of the Web are below the age of 18, others are not Kenyan nationals and others might end up not voting.
In my first search using Google trends with the search terms Uhuru, Raila, Ruto, Karua, Kalonzo and Mudavadi, Uhuru leads in the last five years with Raila only leading during the 2008 period, when he had almost double Uhuru’s searches.
My search further reveals that William Ruto has continued to interest many in the net competing with Raila Odinga for the second position.
On a more detailed search using both names, the three candidates again lead with Uhuru Kenyatta and Raila Odinga seeming like the people to beat. In my research over both the past 12 months and 90 days, Odinga has a narrow margin closely followed by Uhuru Kenyatta. The other candidates have far too few searches to be effectively rated.
The Internet searches are in coherence with opinion polls that have put Mr. Odinga ahead of Mr. Uhuru but can they be trusted and do they reflect the facts on the ground?
One thing is for sure, the survey includes larger greater numbers than most opinion polls carried manually and are less likely to be manipulated. However, these statistics are also limited to various geographical zones with access to the Internet or through mobile telephony mostly the urban and peri-urban areas and eliminate votes from sections of the society that would otherwise matter such as the elderly, illiterate and the Kenyans in the rural areas.
The statistics are further limited to single individuals as compared to the facts on the ground where the running mates and people inside the aspirants’ inner circle will play a big role in the influence of the electorate.
As to how effective the Net can be used to predict the outcome of polls in Kenya, attention shifts to March 2013 as to when the true figures will be released by the independent electoral and boundaries commission.