As you can read in this statistics paper, in South Africa, the share of individuals attending school and the share of low-income women-led households is strongly correlated. Why?

  • 1
    As far as I can tell from skimming the linked pages, the correlation displayed is not between school attendance and family poverty for individuals, but rather between the share of individuals attending school and the share of low-income women-led households, which is a very different thing.
    – duplode
    May 23, 2022 at 22:07
  • Thanks for the clarification but as you can read in this link As of 2019, 41.8 percent of households in South Africa were female-headed, which amounted to a total of approximately 7.2 million. I think this reflects the poverty of South Africa in general. However, I will make the appropriate changes: statista.com/statistics/1114301/…
    – gianni404
    May 24, 2022 at 4:22
  • Welcome to Politics! This post may lack appropriate disclosure of affiliation which is required when linking to your own blog. If the work behind the link is indeed your own, please indicate so in your post. For example, by changing 'this' to 'my' to give users a heads-up.
    – JJJ
    May 28, 2022 at 19:45

1 Answer 1


This answers doesn't try to explain the result, but examines the processes one should go through when trying to interpret seemingly paradoxical correlations. This is because the reason for the association may not be known, so I can only look at how to investigate, not give "the answer".

Correlation isn't causation.

First this isn't a "paper", its more an R worksheet showing some charts. So it doesn't attempt to explain the association.

We need to look at what exactly is being measured. The variables are described as "Percentage listing present school attendance as: Yes" and a bit of digging finds it is the percentage of households and data is collected from each ward.

The other variable is "Percentage of women head households with income under R19.6k out of total number of households" again this is per ward. This is interesting, because it looks specifically at "women head households".

The study then finds a correlation between these variables.

Now, there are some immediate questions to ask:

  • Does this mean that the households going to school are poorer?

No, that isn't what is measured. It says that in wards in which lots of households have children going to school there are a lot of relatively poor women headed households.

It is quite possible that such a correlation could occur because the same type of housing is attractive to both young families with children and poorer women-headed households.

*Are there confounding variables?

Several are possible: The age of people in the household, as most people tend to get richer with age, but have children at school when they are young. This might just be detecting the general tendency of older people to be more wealthy and not have school age children.

  • Is the direction of possible causation clear?

I don't think so. Having school age children is correlated with having younger siblings. Young siblings require childcare and that can mean a mother not working. In a women-headed household, if the mother isn't working the effects on income are clear.

The cultural implication of "woman-headed household" needs to be investigated. What does this imply in a SA context?

  • 4
    I wonder if woman headed houeshold is a euphemism for a single mother?
    – Neil Meyer
    May 24, 2022 at 20:04

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .