I think there are three main purposes for assessment.
- Teachers need to be able to find out what children know and can do, and what they misunderstand, in order to teach them the next steps in their learning.
- Headteachers and school governors need to be able to make sure that every child in their schools is being taught well, and to show that children are doing as well as they would do in a similar school down the road or across the country.
- Governments (local and national) need to know that the school system is improving, and to know where to target support and resources so that differences between areas can be narrowed.
No single assessment, however well designed, can assess everything we need to know in order to make decisions about a pupil, their teachers, their school and the national education system and policy framework.
What we need is a system of assessments, which meet different purposes but are interlinked. This system needs to be designed strategically to offer information for distinctive purposes to different audiences. It would encompass assessment that gives information to policymakers, and richer school and classroom assessments for teachers and pupils. 
Here’s a suggestion of what this system could include:
- formative assessment, using learning progressions, multiple goals and criteria
- low stakes summative assessment
- sample testing at national level
- for now, some form of when ready test of functional English and maths
The key purpose of assessment should be to provide meaningful feedback to teachers in order for them to improve students' learning. The first aspect of a coherent assessment system must be formative, and there is increasing evidence that formative assessment is the next step for raising pupil attainment and closing the gap between the lowest and highest attaining pupils.
For this to work, we need teachers who really understand about assessment, who can design tests and observations that get to the heart of what children can do, and that are based on a clear understanding of what progression looks like within the curriculum. And not just in English and maths.
Low stakes summative assessment:
We need to be able to report results at key moments in time, or key parts of the curriculum. Low-stakes summative assessments could be developed, using banks of questions which teachers could select and use for different pupils, as and when they are ready. In many instances, even at primary level, these assessments could be done on-line, could be responsive to pupil answers, and could provide real-time marking and information. These would need to be developed to assess key aspects of a subject, key skills, knowledge or concepts as defined within a national curriculum.
This kind of assessment would enable reporting of individual pupils' results to their parents. It would allow aggregation of results for subjects, or classes, or groups of pupils, within a school, that could be used by school leaders and governors to monitor the practices and policies within their school, and perhaps in comparison with other schools locally or across networks. It would enable them to ask questions about the effectiveness of their formative assessment and to identify any possible bias in that assessment in order to provide further training and support.
National sample testing
Sample testing at a national level would enable schools to moderate the low-stakes summative assessment, by comparing school results with the national sample results. It would also enable government to monitor national standards over a reasonable time-frame, rather than on an annual basis which is more prone to fluctuation. It could be developed across all subjects, and be administered as part of a planned cycle.
‘When ready’ tests
Realistically, we probably still need some form of national assessment in English and maths at primary level. This could be developed as a test of 'functional English and maths for secondary learning', administered when pupils were ready.
The aim would be that almost all pupils would sit and pass this test at some point during the last year of primary school. Some pupils would take the test earlier, but strong justification would be needed if pupils take the test more than once, in order to avoid pressure to enter pupils early.
Schools could be required to publish numbers of pupils in the cohort who have passed the test by the end of Y6, with government expected to provide additional resources to secondary schools to support all pupils who have not yet passed the test, as well as support to primary schools which struggle.
We need a coherent system of assessments which will provide meaningful information to different stakeholders and can offer checks and balances.
It needs to work for the profession, and to be informed by those in education, so I ask: what’s missing from this alternative model? What would we need to make it work?
Please comment below.
 See Conley, D.T., & Darling-Hammond, L. (2013) Creating systems of assessment for deeper learning. Standford, CA: Stanford Center for Opportunity Policy in Education for more information