It would suit the right-wing of politics to ignore the profound impact of poverty on education. Nat blogger DPF ran the line, complete with his usual bombastic ignorance, just yesterday:
Let’s put this one to bed. Even if this was true (it is not), this is an announcement on education, not welfare. Turei seems to say we should do nothing to improve the education system while some families are poorer than others. How depressing. I want to see more families doing better, but there is no magic wand. Getting people out of poverty is often a generational thing as you have to confront parenting skills, welfare dependency, employment, drug and alcohol issues, and oh yeah education.
But let’s deal with the big lie. I call it a lie, because the amount of research on what influences educational outcomes is massive. There have been over 50,000 studies. Over 800 meta-analysis done involving 200 million students. Professor John Hattie has done a meta meta analysis of all these studies and identified 138 factors that influence educational outcomes. Not one factor, but 138. Greens think there is just one.
Now socio-economic status is important. It definitely is an influence. There have been 499 studies that looked at its effect. But is it the biggest influence. No. Is it second? No. Third? No. Top 10? Still no. Top 20? Still a no. It is No 32 and home environment by the way is No 31.
So the next time the Greens say the key reason for educational decline is poverty or income inequality, don’t beat around the bush. Call them a liar.
DPF is referring to research described in his book Visible Learning and its related work. It is (predictable but) unfortunate that Hattie’s book should be interpreted and used in this way, because here’s what Hattie himself says in the introduction:
It is not a book about what cannot be influenced in schools — thus critical discussions about class, poverty, resources in families, health in families, and nutrition are not included — but this is NOT because they are unimportant, indeed they may be more important than many of the influences discussed in this book. It is just that I have not included these topics in my orbit. [p ix]
Hattie says factors he has not considered may be highly important. So DPF is wrong that Hattie’s work proves the limited significance of socioeconomic factors, end of story, you can save yourself some time and stop reading this post right now.
Still reading? OK – here’s the long version. DPF’s fallacy is based on a superficial reading of summaries such as this:
Socioeconomic status (marked with a red dot) is Rank 32 out of 138 in this list. As above, Hattie himself admits that his analysis of socioeconomic effects is limited, he regarded many factors as outside the scope of his study, but he has included some data (the main measure was “parental income” where limited effects at the upper range dilute significant effects at the lower range – a measure of poverty would have been more sensitive).
Hattie’s socioeconomic status ranks at only 32 because he used an insensitive measure, because it has a narrow definition, and because of the way Hattie’s study is structured. Most importantly – Hattie’s data is from studies relating to schools, but in his section on socioeconomic status he notes that the main impact occurs before school years begin:
It is likely that the effects from socioeconomic resources are more influential during the pre-school and early years of schooling. For example, Hart and Risley (1995) showed that when students from lower SES groups start school, they have, on average, spoken about 2.5 million words, whereas those from higher groups have spoken 4.5 million words: this demonstrates a remarkable difference in what students bring to school. The lack of resources, the lower levels of involvement in teaching and schooling, the lesser facilities to realize higher expectations and encouragement, and the lack of knowledge about the language of learning may mean that students from lower SES groups start the schooling process behind others. [p 62]
A major part of Hattie’s analysis focuses on the nature of the student when they start school (their background). It is here that we see the effects of socioeconomic status, though it is not being measured directly. The following quotes are all from Chapter 4 The contributions from the student:
The fundamental argument in this chapter is that students not only bring to school their prior achievement (from preschool, home, and genetics), but also a set of personal dispositions that can have a marked effect on the outcomes of schooling. While there is no doubt that schools can affect both achievement and learning dispositions, the origins of both are often well in place before the child enters the school yard. For achievement, there are influences from genetics and early development, very early home and social experiences, and opportunities for learning from birth to five years (e.g., preschool and other early interventions). [p 40]
It is certainly the case that by the time the child enters school, family, preschool, or genetic factors will have already played a major role in generating subsequent differences in school-based achievement. [p 42]
Children in the bottom quartile at 22 months “are significantly less likely to get any qualifications than those in the top quartile”, suggesting that “before children have even entered school, very substantial signals about educational progress” are evident (Feinstein, 2003, p. 82). The effects of social class (based on parental occupation) were marked at 22 months, and if anything, the variability increased over time. This dual influence of early achievement and socioeconomic resources contribute much to what a child brings to school. [p 42 my emphasis]
Where does this show up in Hattie’s list (figure above)? In Student effects. The two top ranked effects in the list are both student effects, they are Self-reported grades i.e. self assessment/confidence, and Piagetian programs i.e. level of cognitive development. These and other student effects are substantially determined by pre-school factors and experience (outside the scope of Hattie’s study) including socioeconomic status.
In summary, Hattie’s analysis of socioeconomic factors is both too little (by his own admission) and too late (focused on school years not the crucial prior experience) so (because of the way he has structured his study) where socioeconomic impact comes out most strongly is in factors relating to student background, such as his Rank 1 and Rank 2 effects.
Of course poverty matters. Duh.
Here’s one of the conclusions of a meta-study of socioeconomic factors (one of the ones cited by Hattie in fact):
Of all the factors examined in the meta-analytic literature, family SES at the student level is one of the strongest correlates of academic performance. At the school level, the correlations were even stronger.
This review’s overall finding, therefore, suggests that parents’ location in the socioeconomic structure has a strong impact on students’ academic achievement. Family SES sets the stage for students’ academic performance both by directly providing resources at home and by indirectly providing the social capital that is necessary to succeed in school (Coleman, 1988). Family SES also helps to determine the kind of school and classroom environment to which the student has access (Reynolds & Walberg, 1992a).
And here’s an analysis of Hattie’s research, including his treatment of socioeconomic factors: Snook, Ivan; O’Neill, John; Clark, John; O’Neill, Anne-Marie; Openshaw, Roger (2009) Invisible Learnings? A Commentary on John Hattie’s book: Visible Learning: A synthesis of over 800 meta-analyses relating to achievement. New Zealand Journal of Educational Studies 44.1: 93-106.
Personal note – I’m on sabatical from The Standard for a couple of months yet, this is a one-off post.