In this ‘Ask the expert’, Emily Jones, Research Director at NFER, answers some FAQs on interpreting assessment data.
The assessments I use provide different data outputs. What is the best output to use when analysing pupil performance?
Analyses on pupil performance are most easily carried out when raw scores are converted to more useful score outputs such as standardised or age standardised scores. Although you can compare the raw scores of pupils sitting the same assessment, for example, to rank the pupils in your class, raw scores from different tests cannot be directly compared as even if the tests have the same number of marks, they are extremely unlikely to be of the same difficulty. Standardised scores (and age standardised scores) from different tests on the other hand can be compared. This means that a pupil can complete different assessments in two school years and a comparison of the scores can show you whether or not the pupil has progressed in line with the national cohort.
What analyses should I perform with standardised scores?
It can be helpful to monitor both individual pupils and cohorts over time to check that they are progressing appropriately. By completing assessments at the start and end of the school year or at the same time point in consecutive years, you can track the pupils’ learning within or across school years. A child with standardised scores of 112 and 110 on consecutive assessments (i.e. who is gaining a similar standardised score each time) is making average or expected progress; they have maintained their position relative to their peers. If a pupil’s standardised score goes up by a considerable amount, it means they are making more than average progress (this may happen, for example, if a pupil with a score previously well below average has been having additional support and is catching up with their peers). Conversely, if a pupil’s standardised score falls by a considerable amount, they are making less than average progress from that starting point and may need to be monitored more carefully and extra support provided.
What other useful analyses could I perform on my pupil assessment data?
As well as looking at whole test data, it can also be helpful to look at the data from a question level perspective. For example, you could compare how well your class has completed individual questions in comparison to the national sample on which the tests were standardised. Looking for patterns in the questions that your pupils completed better or less well than the national population, may reveal the strengths and weaknesses of your pupils and enable you to tailor your teaching to meet their needs.
How can I manage my assessment data?
A lot of assessment data is generated by just one pupil over the course of their school career, never mind by an entire cohort. It is important to have a system which allows you to manage the data easily over time. Online systems have several advantages such as automatic generation of useful score outputs from raw score entry, enabling several authorised people to access the same data easily, and allowing data to be manipulated so that the most informative analyses can be carried out. Be selective though, as some online systems generate extraneous reports and more information than is useful.
Written by Emily Jones, Research Director at NFER.
Emily has been with NFER since 2001 and has worked on a wide range of projects, centred on developing test materials in science and mathematics, including for the Key Stage 1 and Key Stage 2 statutory tests in England and the Australian statutory NAP test.
Emily is responsible for the development of NFER tests in mathematics, reading, grammar and punctuation and spelling. She is also a consultant member of the TIMSS 2019 Science and Maths Item Review Committee and in recent years has contributed to the development of their new practical science e-assessment tasks, as well as developing and trialling the standard science questions.
Do you have a question on assessment that you’d like to put to one of our assessment team? Send it through to us at firstname.lastname@example.org.