When it came to electoral integrity, there was strength in numbers
Stats SA suddenly found itself pulled into the 2019 general election, racing to disprove a damning claim. At stake was no less than the credibility of the result
● As the national statistics office, we had a date with destiny this month. I was very excited to do my democratic duty as a citizen and got up early to go and vote. No sooner had the ink dried on my thumb than reports of double voting emerged and set social media abuzz. Many people were worried.
The media latched onto this and the country and the Independent Electoral Commission (IEC) found themselves in uncharted territory. A potential electoral crisis was on our hands.
For the first time in the democratic era, the credibility of elections could be in doubt and the country could be plunged into a deep constitutional crisis, a very unwelcome one.
A call came through from the IEC and it sent shivers down my spine. The commission requested that we draw a representative sample of voting districts from all nine provinces and establish if there was any validity to the complaints.
I knew this was going to be a very difficult task, but it was also one I could not ignore.
In national statistical organisations, methodology is everything, and Stats SA certainly places great stock in this fundamental truth. We draw up samples every time we embark on large sample surveys. This is what we do regularly — and our methodology produces sound and accurate statistical releases. This area is our forte; we have built our reputation and gained trust based on this.
However, what the IEC presented to us was somewhat uncharted territory. Stats SA had never confronted such an eventuality.
A number of political parties were incensed and very vocally threatened to take the electoral commission to court. It was a monumental race against time as the credibility of the elections was being doubted, even questioned. Calls for a rerun began to surface. This would certainly have been an extremely disagreeable option. I am sure no-one wanted it and it was for this reason that I had to muster the collective experience of Stats SA to help deal with this problem.
When the IEC made the call and later announced that it would approach us, Stats SA was in the spotlight. I consulted my management team. We accepted that this was not our core work, but we felt that we must answer the call when the country needed us. We considered the following in drawing the sample of voting stations:
In the absence of a priori information on statistics related to these double voting incidents, we assumed that the incident being measured happens in 50% of the population.
We used a 3% margin of error to determine the overall sample size. The 3% margin of error tells how much one can expect the survey results to reflect the incident being measured. It measures how effective the survey results are. The smaller the margin of error, the more confidence one may have in the results.
The frame was implicitly stratified by province, local municipality, and ward. This ensured that the sample was spread across these geographic levels.
These specifications yielded a sample of 1,020 voting districts. We had to show our independence and deny political parties’ requests that we add “these two voting districts because we know that there were irregularities there”. Sampling has to be based on sound statistical methodologies, not on anecdotes.
If an election is the cornerstone of multi-party democracy, as a national statistics office we believe that statistics is the factor that illuminates that constitutional order.
Having selected the sample, the second stage was to identify the districts that had potential irregularities.
The voting district’s total number of section 24A votes [in which a voter, typically one away from home on the day, is granted the right to vote at a station at which he or she is not registered] was compared to the total number of votes cast in the same voting district.
Only 13 of the 1,020 sampled voting districts (1.27%) showed the potential for significant irregularities. Based on this, we were confident that, if double voting did occur, it was not significant enough to impact on the results, and informed the IEC that it could release the election results.
Stats SA has a history of working with the IEC.
These past few years we have worked closely on a number of projects. We share our population estimates with the commission.
With regard to the work done for the IEC, population estimates from the 2018 local municipality estimate series were used and projected forwards to the January date when registration was held, as well as to the election day of May 8 this year.
Working closely with the IEC, we have done a lot of behind-the-scenes work on the scanning and digitisation of voter registration forms. We helped the IEC with the scanning and provided it with electronic data sets.
Stats SA has also made inputs to the IEC since the Constitutional Court ruled that the IEC voters roll was invalid as it had less than 35% of actual voters with verifiable physical addresses.
We remain committed to playing a role to complete the project that the IEC has already started. This will be an invaluable collaboration as it will be the first time in our history that we have a complete and coherent address system — and having an address is a human right and restores dignity to our people.
As statisticians and demographers, the value of election registration data cannot be overstated. It is a major asset as a quality check in guiding the work of estimating the population more accurately.
This registration data also makes a major contribution in establishing migration streams. In collaboration with the Council for Scientific and Industrial Research, work around these migration streams is progressing steadily.
SA will soon see the new government assume office and it has a mammoth task in turning around the fortunes of all South Africans. We wish them well during their term of office.
For us at Stats SA, we are confident that the collaborative work we did with the IEC proved that our national statistical office enjoys the trust and respect of South Africans.
Stats SA will continue to guard its independence jealously as it serves the nation by providing numbers the country can trust — and informing the nation about the health and wellbeing of the people of SA.