Oh No-ah you didn't: Demographic behaviour and gender egalitarianism

Recently Noah Smith made a tweet seemingly questioning the relationship between demographic processes and feminism, drawn from cross country comparisons from Pew. I understand that this is a tweet, and so that there was not much room to expound a detailed explanation of a theory, but much of the claim seemed to rest on the idea that because the United States had higher levels of childbearing and earlier marriage rates than some countries which are stereotypically stricter in terms of gender norms, this somehow refuted the standard claim that increasing women's agency was associated with lower fertility and later marriage.

There are a number of issues with this supposed refutation, and I'll discuss these in a little detail in the following three sections.

Are cross national comparisons valid?

The claim that Noah seems to advance here is that because at one point in time the correlation between demographic outcomes and gender equity is not strongly negative, this refutes the idea that there is an effect here. This seems to fundamentally misunderstand the claim that demographers make: this is that changes over time will tend to decrease fertility rates within the same setting. Considering the stylised diagram, Noah is effectively looking along the green line, and using the variability between countries (red) in TFR to claim there is not correlation between demographic outcomes and egalitarianism.

Now it should be noted that the evidence when looking on a between country basis is somewhat mixed. However, using cross sectional comparisons to draw conclusions about trend over time, without accounting for the starting point (the variation in the country TFR in the example at the green line is solely due to variation at the start, since the slopes are parallel) gives a deeply misleading picture. In reality of course, we are unlikely to see such a neat picture as in the toy example- indeed there is no requirement for societies to transition to low fertility at the same rate with increasing cultural and gender changes under second demographic transitions model. As such, neither the existence of between country demographic variation, nor the fact that there is no homogenous pattern, necessarily refutes the point that gender equality would be associated and perhaps causally with decreases in fertility rates.

What is the model of change?

Another problem is that Noah is incredibly unspecific with his terminology, so we don't have a real logic of change model to deal with. One of the major transformations we have seen that would be widely accepted as a measure of gender equality would be the increase in female educational enrolment, as we can see in the UK and France in the figure below.

Source: Ni Bhrolchain and Beaujouan 2012

Now, this establishes that there is an increase in the indicator we are claiming is a measure of gender equity, how does this fit into our logic of change model for demographic outcomes? The figure below presents age specific fertility rates which begin when the woman leaves full time education

Source: Ni Bhrolchain and Beaujouan 2012

Overall, we notice very little change between cohorts, once we take into account the termination of educational enrolment. This leads us to the conclusion that there is a strong association between the two to the extent that as much of 60% of fertility postponement is explained by our indicator of gender equality, namely female education. Therefore, once we actually operationalise gender equity beyond more than stereotypes, some reasonably strong relationships emerge

What is the hypothesised relationship?

The final problem with the claim is that gender equity will necessarily be associated with decreases in behaviours like fertility. This is not unreasonable at relatively early stages of the Second Demographic Transition: the difficulty in role combination between work and motherhood and educational enrolment (as we have seen already) would be expected to decrease the intensity of all demographic behaviours. However, the assumption that this relationship is monotonic has little basis in demographic theory. Indeed, once we get to a certain stage, gender equality is expected to increase fertility level: this has been proffered as an explanation for European fertility variation with more gender equal countries in Scandinavia benefiting from higher fertility rates, with more traditional countries in the South experiencing drag on fertility as conservative social norms clash with educational and labour market requirements for women. The underlying assumption that feminism will reduce fertility universally is assuming that a country is at an early stage of gender development and the second demographic transition: I am not sure that this hold for the dataset that Noah is using. 

The Accuracy of Rapid Diagnostic Tests for Malaria [Reblogged]

I had the privilege of giving an interview to Thomas Locke from Fight Malaria. I'm reblogging the transcript, with the audio available via their iTunes and Spotify channels. Full technical details are available in the working paper of initial research from the Portsmouth Brawijaya Working Paper Series (Number 8): Field interviewers effects on the quality of malaria diagnosis in Malawi published in conjunction with Dr. Ngania Kandala and Prof. Saseendran Pallikadavath.

The original is hosted on the Fight Malaria blog

Hello, I’m Thomas Locke and this is Five Minutes, the podcast that brings you closer to the people fighting malaria.

Today I’m joined by Dr Mark Amos to discuss the accuracy of malaria testing. So, how accurate are Rapid Diagnostic Tests, or RTDs, tools that are becoming increasingly popular? And how do they compare to traditional lab testing?

This is Five Minutes with Dr Mark Amos.

Mark thanks for joining me.

Thank you very much.

Talk me through how you’re assessing the accuracy of malaria testing.

This is essentially a validity study. We’ve been comparing two methods of diagnosis. We’ve been comparing field diagnostic methods with lab-based methods. And we’re assuming that lab-based methods here are the gold standard against which we’re judging the accuracy both in terms of false negatives and false positives for testing administered in the field.

Presumably, field tests are more accessible than lab testing?

They are, there are a number of advantages to field testing. It’s very rapid so you can give a diagnosis to the potential patient straight away. You can also carry it with you so you don’t need people to turn up to a lab or worry about transport. The concern is that it is being tested in an uncontrolled environment, which is why we wanted to look at the accuracy of the testing mechanisms available.

How did you conduct this research and what were the outcomes?

Our research was using Demographic and Health Survey data which is available from the DHS website. We found that there was a reasonable degree of accuracy in most cases, but there was some variation in the degree of accuracy and testing depending on the interview team. Now, for false positives, this was actually around about 15% of variation in the level of false positives was attributable to the interview team. However, we did actually find for false negatives, that around 48% of variation in the rate of false negatives was attributable to the interview team. Now, obviously false positives are reasonably serious, Malawi is resource-poor setting. We prefer not to have false positives if it’s all possible, it’s a bit of a waste of resources. However, probably false negatives are, in a sense, the more serious false diagnosis, you’re potentially saying to someone who does have malaria that they’re okay. That has a number of implications, it may delay them seeking care or treatment for malaria or it may actually put more seeking care treatment for malaria at all.

What sort of data did you collect?

So the data we’re using is actually secondary data. It was collected by the DHS program in Malawi and we access their data remotely. We have two measures within the dataset: the diagnosis from the field test and the diagnosis from a lab-based result, and we simply compared the two.

The DHS, is that state-owned by the Malawian government, is it USAID, who owns the data?

It’s funded by USAID, it’s a survey that has been running for a number of decades, it started off as the world fertility survey. It’s a series of cross-sectional surveys in all areas, where there are reasonably high levels of fertility. It captures a number of dimensions of data that might be useful ranging from contraceptive use, maternal health care utilization, to things like AIDS prevalence or the prevalence of malaria.

You found out that there is some degree of accuracy with both methods of testing. What are your next steps?

There are two major steps. Firstly, although there is some variability in false diagnosis, in a way the fact that we’re attributing this to interviews is actually a reasonably positive step forward, it is actually something we can do about it. It’s not a function of the underlying test, it’s a function of the interview a team administering the test. So there might be potential ways of looking at better training or encouraging interviews to deliver tests in a different way, which might actually increase the accuracy so there’s a kind of a positive policy story type there. The other major thing that we want to do going forward is we want to expand our analysis to look at, hopefully, all of the demographic and health survey datasets across sub-Saharan Africa. So we’ll be able to compare between countries and see sort of whether there is something unusual about Malawi or whether this is a generalizable finding.

Dr Mark Amos, thank you.

Thank you very much.