OPINION | The struggle is real to reach con(census) on the homeless

Imagine the scenario. A major parastatal acknowledges that it has failed to deliver in the past. It seeks assistance from members of civil society who have the expertise. And a perfect partnership of public and private actors brings about a result which is good for everyone.  

A fantasy in SA? Sadly yes. But what makes it sadder is that it could easily have come true. Stats SA is the parastatal in question, mandated to deliver the decennial census that is a touchstone for government policy. As I heard its deputy director-general of operations say at the start of this year’s census, speaking in Durban: “The census is not just about the present, it’s about the future. It tells us how many school places we will need in five years’ time, how many university places in 18 years’ time and how many graves in 60+ years’ time.”

The census is a huge task and it is right that a substantial budget is assigned: a figure of R3.2bn is quoted. But that’s because it matters and — as with everything in government — it matters most to those who are poor and vulnerable and with limited resources to opt out. So the homeless, a repeatedly neglected part of our society, also need to be counted in the census.

And that was the admission that Stats SA made to the National Homeless Network. That they had probably undercounted the homeless in the 2011 census. Their figure was 6,000. For which city? we asked. Six thousand for the whole country was their reply — and we fell about laughing since each of our major cities has at least that number and more.  

They invited members of the National Homeless Network to their plush modern offices in Tshwane (ironically built on a site where homeless people used to stay) so that we could propose ways to help Stats SA to improve the count of the homeless in the next census. They told us that they wanted to live up to their watchword that “everybody counts”. That was more than three years ago, in November 2018. We made multiple suggestions. And Stats SA went very quiet. Fair enough. These are bureaucrats and they have their own ways of doing things.

To our delight three years later, last November, they contacted the National Homeless Network again, now keen to work in a concrete way with organisations on the ground to prepare for the February 2022 census. Three months was not a lot of time but certainly something could be done.

Except that nothing was done for three months. Despite multiple e-mails and phone calls following up from us, all we were told was that a policy was being discussed to ensure a national rollout plan and we would hear in due course. A glimmer of activity in Gauteng in late December turned out to be a false dawn.

And then in late January we were told: there is no national policy, there is no national rollout plan. But if your members would like to contact their local Stats SA branches and offer their help they might respond. There was only one week to go to census day.

I am not the statistician-general. But one of the few things I know about censuses is the importance of consistency. There is little point in measuring the rainfall in Cape Town if you then use a different way of measuring it in Durban because then you cannot compare. Hence the value of a national policy and a national rollout for counting the homeless.

The challenge Stats SA had set itself was a big one. The reference night for the 2022 census was February 2/3 and the intention was to get a snapshot of everyone who was inside the borders on that night. For you and I, we can answer those questions (online, by phone or face to face) later in February, as long as we refer back to that night. But Stats SA decided that transient populations needed to be physically counted and interviewed on that very night. People passing through airports and hotels. But also people who are homeless.

Assuming that the likely population of homeless people is closer to 100,000 than 6,000 that is not a small task. It is an achievable one with a bit of planning. But as the old saying goes, those who fail to plan, plan to fail.  

Stats SA are not expected to know the whereabouts of homeless people, or how to approach them, or how to build trust with them, or how many are likely to be in shelters vs on the streets. But the members of the National Homeless Network do know that and more besides.  

That is the expertise that we persistently offered to Stats SA and which they ignored for the last three months (even though they had originally asked for it). And then, with literally days to go before the census, individual Stats SA offices started speaking to homeless NGOs: Can you help us with a map? Do you have a list of shelters (the ones on the list we have are all closed)? Can you help us train the counters? What would you advise in terms of security? When would be the best time to try to interview people?

All great questions. But not the same questions from each office. And no attempt by Stats SA to harmonise the response.

Different members of the National Homeless Network answered the questions they were asked, and arranged the assistance that was requested — and often did more than requested — all of this at our own expense despite being under-resourced NGOs with very few staff. Seemingly, the census 2022 budget of R3.2bn does not have space to remunerate organisations for their invaluable expert advice.

In the end, on the night of February 2, the much-heralded night of the census, a range of interventions happened. But because they were inconsistent, the results that come out will not be comparable.  

In Durban, the 220 counters (who had already been trained in how to speak to homeless people) went out with a clearly articulated map plus homeless guides to conduct the census. This was reasonably successful, except that we probably only counted 50% of the community.  (A “dry run” would have shown that twice the level of resources were needed to count everyone on the night.)  

In central Cape Town, the homeless did prepare a map but then, despite making themselves available, they were not allowed to help guide the counters because the Stats SA people did not want to be watched by anyone.

In central Johannesburg, the group started so late that they only attempted to count the homeless in Park Station and ignored the hundreds sleeping within a few hundred metres outside.

In Bloemfontein, the start was even later – after food and speeches from Stats SA officials and the Free State premier — so in the end they did not count anyone at all, despite a team of homeless experts having been assembled to help.

In Pretoria, they were keen to get a map of hotspots but then went out with almost no homeless guides and only focused on a small number of hotspots. In Bellville and Fish Hoek, homeless guides and maps were provided as requested and then ignored. In Randburg, the counters also refused the advice they were offered from local homeless experts; but in any case their ability to function was impeded as they had been issued with torches but no batteries. 

In Pietermaritzburg, there were homeless experts on hand to assist; but no matter since the Stats SA tablets had not been charged before load-shedding and so no-one could do any counting anyway. (And that is in addition to the huge numbers of counters across the country who could not use their tablets because they were not working, or not configured, or would not connect to the internet).

In most other parts of the country from which we received reports — Kempton Park, Alberton, Boksburg, Benoni, Mogale City — Stats SA was generally described as “useless”. In some parts of the country where we were asked to provide links to local homeless organisations, no-one was ever contacted. And, with very few exceptions, Stats SA counters were sent out into the dark streets of our cities with none of the police back-up that had been recommended and promised.

An inconsistent process is bound to lead to inconsistent results. Judging from what we have heard, it looks as if Stats SA will end up with a rather bizarre count of homeless people from the night of February 2/3: about 3,000 for Durban, 300 each for Pretoria and Johannesburg, who knows what for Cape Town, and close to zero for the rest of the country. And that is what they will proudly present to the government as a basis for plotting future policy.

I don’t want to suggest that there are not people at Stats SA — established officials and census enumerators — who are skilled and dedicated to the task (though some among those I have met are not up to it). But any individual’s ability to complete a task is limited by the expertise of the organisation, and crucially in this case the willingness of the organisation to seek outside expertise when they need it.  

The homeless count was the first test of Census 2022 and, if the test is thoroughness and consistency, Stats SA has failed spectacularly.  

A journalist who asked the statistician general about these problems was told that these were “teething problems that will be ironed out in a day or two”. Given that they conducted an expensive pilot of the census a few months ago, it seems rather late to be having teething problems. And, in any case, the only opportunity to count the homeless has now passed, so presumably the statistician general does not actually believe that “everybody counts”. 

Maybe they will be ironed out in a day or two: these issues and also the problems of tablets that don’t work, temporary staff without contracts, and field workers who have not been trained. We will soon see because Stats SA will have the chance to demonstrate that the failure to count thousands of homeless people is not going to be followed by a failure to count millions of residents of informal settlements.  

And what of the rest of us?  We have the chance to enter our own data online: all we have to do is register on their system. If you have not done so, don’t worry: only 0.5% of households had registered by February 3. But Stats SA has it under control — they extended the deadline by an extra two days to capture the other 99.5%!  Let’s hope the system can cope with processing 8,000 registrations per minute.

It would be funny if it was not so important.  The results of this census will be quoted endlessly by politicians and government officials over the next 10 years as a justification for all kinds of policies. The cost of a bad census will, as the DDG pointed out, be felt not only in the present but in the future. And that cost will again be borne mostly by the poor. 

Only a really hardened cynic would suggest that a government that does not care about the poor has a vested interest in conducting a census that undercounts the poor. Because, after all, if they are not on a spreadsheet, then we don’t have to provide any services for them, do we?

Raymond Perrier is director of the Denis Hurley Centre Durban and co-founder of the National Homeless Network.