When we wish to do assessment for learning - to ensure we are teaching efficiently - we often find that our data is inaccurate due to learners’ surprisingly sophisticated copying skills. If we decide to use test conditions to assess learners we may have difficulty ensuring test conditions because of lack of space in our classrooms or lack of cooperation from our learners or we may find that our data is inaccurate because our learners are unable to think clearly in test conditions.

solution: collect accurate assessment for learning data 

If we want assessment for learning data to be accurate, we need to take care when collecting it, however this takes time ...

timely practice's well presented, accurate assessment for learning data and targeted resources in turn makes more time for assessment for learning

To collect accurate assessment for learning data we need to work with the learner - but initially the learner will resist our efforts. Most low attaining learners are more motivated by avoiding shame in maths lessons than they are in learning more maths. 

If copying is possible our data won't be accurate. If we have to enforce exam conditions, we won't be able to use that time for teaching - so timely practice takes a third way.

We make assessment for learning accurate, as each learner is randomly assigned an assessment question, so the possibility of copying is dramatically reduced, and since learners soon spot this, they generally don't even try to copy. We make assessment for learning low stakes, concentrating on what learners know, we appear disinterested in what learners don't know until we are ready to teach them. However once we teach some new learning, we do everything we can - teach on firm learning foundations, remind learners of similar skills they already have, give help and give feedback - to ensure that the learning becomes embedded learning. 

Once learners feel that we genuinely want to find out what they know in order to be able to teach them better and they see we value what they know already they are less resistant to assessment. 

Since we track the depth of embedded learning, the teacher only needs to assess a topic once, when they return to teach the topic again they will find really accurate assessment for learning data which makes planning teaching quicker and easier.

The time and effort we spend in collecting the assessment for learning data pays teacher time dividends almost immediately as

• planning is quicker (we know what learners know and so what to teach them next),
• resourcing of lessons is quicker (as practise-learn worksheets are available for each layer),
• assessment is quicker than marking (single code - no other writing),
• we have less behaviour management problems (learners are less stressed and more engaged in learning)

Later we see more benefits

• when we return to teach a topic again we need do no further pre assessment,
• we very rarely teach the same thing again,
• we no longer have pressure to "teach extra lessons",
• we can, on occasions, "activate" low attaining learners as teachers, 
• we are less busy in the classroom - so we may have time to do in class assessment too.