Improving early education outcomes (1)

I have been looking into the first data I received from our Early Grade Reading Assessment, done in grades 1, 2 and 3 across the country. This assessment, developed by USAID for a quick (and somewhat unscientific) snapshot of the reading status of students in (developing) countries, tracks letter-, word- and story reading skills, and it has a tiny comprehension part. Because it is so easily available (and free!), we chose it for a first exploration of our early primary school students here in Liberia.

My most pressing concern presently are the first-graders. With 29 schools to develop, we chose to start with them. So let’s have a brief look at the preliminary data (n = 333, about 80 more data sets coming).

Letter reading

Students had to “read” letters arranged in a 10×10 grid, consisting of mixed (upper- and lowercase) letters, in random order. The task was timed to calculate a “letters/min”-score.

What is an appropriate letters/min-score for students in the early grades?

The literature dealing with child development that I know talks mostly about words/min – actual reading. In the USA, a score of 40 or below in Kindergarten is considered “at risk”. It remains dubious whether this task has anything to say about actual reading skills – in particular in English because of its complicated grapheme-phoneme correspondence.

Figur 1: Letters/min across all 29 schools

The mean scores across all 29 schools for boys and girls are indicated in fig 1. The first thing is – there actually is some learning taking place, no matter how small. The results for grade 3 appear a bit surprising, though.

Luckily, we can have a closer look into these data and separate the scores across regions (our schools are classified into 3 environmental regions which are “urban”, “semi-urban” and “rural”). The distribution was made in a focus group with school administrators, and is thus relatively subjective. “Urban” schools are schools in Monrovia or Paynesville, “semi-urban” schools can be found in major towns and county capitals, whereas the “rural” schools are in villages < 2000 inhabitants.

Fig. 2: Mean score of letters/min, differentiated by region

The spreading explains the poor performance in grade 3, and also highlights one of the central problems of education here in Liberia. Students in rural schools regularly underperform due to cultural constraints, poor teacher motivation, lack of oversight, low interest of parents and guardians, and a multitude of other reasons. Tackling these reasons should keep us busy for the next years!

Positively though the constant upwards trajectory in urban schools. Urbanisation has its good sides, too, apparently.

Word reading

Word reading is measured by reading unconnected words off a 10×10 grid consisting of mainly simple words which occur among the 200 most frequent English words.

Fig. 3: Words/min across all schools

Word reading is much easier to compare across the world, because it is regularly tracked by governments and school authorities (and universities, scholars, and lots of others). It does depend on the word length, though – (Who would have guessed that!?)

For English, anyways, there are fluency standards which apply to native speakers in a fluent and supportive environment. Note the two constraints I mention! These requires a young reader to achieve target rates of 30-80 (grade 2) and 50-110 (grade 3) in autumn. Looking at fig. 3, Liberian students appear to be far behind.

Fig. 4: Mean wpm scores across regions and levels

For a better picture, we did not only look at the data comparing regions, but also decided to understand how the “upper” half of students would perform against the “lower” half of students (lower < median, upper >= median for wpm). This looks quite different now: even some rural students improve, and some semi-urban students possibly went to good preschools. Urban “upper half” students outperform all by far, but it is surprising to me that the “lower half” is left behind in a similar fashion as in the other regions.

What can we conclude?

Remembering the date of the assessment (Nov 2018), fig. 4 illustrates well that good preschools give you the cutting edge to be on the sunny side of the learning outcome distribution. In addition, not all hope is lost for rural schools – even though outcomes are lower than in urban schools, there is measurable progress, which is good news.

All together, the goal to enhance education outcomes remains, but is expanded by a facet of inclusion. Can we assist our teachers in a way that they do not leave 50% of the class behind? I hope time will tell and we will find the right ways.

In addition, a stronger focus on preschool education is recommended. Wherever students start with some extra knowledge and skills, it is self-evident that they will (most likely) perform better than without that. We will focus on this as well.


If you are not skilled in reading statistic, please consider these limitations: The presented graphs are based on a one-time snapshot of students in grades 1, 2 and 3 of the 29 Lutheran schools in Liberia. It is not a longitudinal study tracking the same students over three years – so outcome in grade 3 may or may not be correlated with outcome in grade 1. I am assuming some correlation here for simplicity, but everything needs to be taken with caution.

Leave a Reply

Your e-mail address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.