Dylan Wiliam – Author, Researcher, Trainer and Assessment for Learning Expert  26171


  • Speed dating questions
    • 00:05:20 Favourite number: [Stanley] Skewe’s First Number. 10 to power 10 to power 10 to power 34. Largest number with a practical usage. Discovered as keen teen reading Hardy’s A Mathematician’s Apology.
    • Favourite topic: Number theory. Primes, etc. Beautiful, unuseful (or so they thought at the time). E.g. public key cryptosystems based on factorisation that didn’t seem to have a use originally.
    • Didn’t enjoy applied maths. But was amazed how easy it was to teach.
    • Job if not in education – musician. Started teaching for money to be PA. Worked semi-pro for a few years. Discovered having more fun teaching maths than being in band. Not had time to play, but plans it for retirement.
    • Steps to where is now: Started in tough school with no training at all due to severe shortage of maths and science teachers in London. Had taught for a year in private college in small groups, so learned to teach before he learned crowd control. Had idea of problems in Maths. Used SMILE resources. Had to do 3-4 hours of graft each day in setting work. 3 years later Michael Marland was setting up a school nearby and needed someone with SMILE expertise. Into research and lecturing.
    • Secondment for 2 years for tests for 14 years. Kenneth Clarke: “Elaborate nonsense”. Wiliam and group had wanted projects for assessment at 14, not tests. Then back to lecturing; decided it would be quicker to write up what he’d been doing, including technical issues in psychometrics, rather than finishing PhD. Head of Department at KCL very early. Mecca for assessment specialists is Educational Testing Service in USA. Asked to give a seminar; actually turned out to be what’s called a “job talk” in the USA. Then IoE after that. Wasn’t actually interested in running that. Better number 2 than number 1 and less of annoying jobs than as Vice-Chancellor. 2010 freelance on formative assessment.
    • Big break was being asked to coordinate NC test development in 1989. Had to develop tests in English and Welsh, so big break was because of that.
  • 00:21:00 What makes a lesson successful in Dylan’s opinion?
    • Would have been certain 20 years ago.
    • Now doesn’t know. Doesn’t think anyone does. Paul Kirschner: Learning is a change in long-term memory. Good lesson is one in which things that happened are remembered in 6 weeks’ time. Daniel Willingham: level of cognitive engagement in worthwhile material in lesson.
    • For OfSTED, engagement is “being busy”; cutting up, colouring in. Willingham: memory is the residue of thought – students remember what they have been thinking about. They don’t have to be physically active; important not to confuse cognitive and physical activity.
    • J. M. Barrie Admirable Crichton: I’m not young enough to know everything.
    • Make learning hard, not easy. If too easy, don’t have to work hard enough to lay down memories. Too hard: long-term memory not created. Parts of working memory used in solving problems are also parts of working memory used to create long-term memories.
    • Different types of activity useful for novices and experts. Novices: worked examples. Experts: problems/struggles.
    • 00:24:15 What would he have said 20 years ago? DW: Lots of student conversations, lots of student activity, clear explanations from the teacher. Not a wrong answer, but not as clear about big maths problem of students doing it today and tomorrow, but not in 2 weeks’ time.
    • 00:25:00 Individualised maths scheme. Students would do a worksheet on topic like fractions, then a puzzle, then shape and space, then more on fractions in two weeks. Intuitively, the designer of the scheme had hit on idea of distributed practice. A day on fractions, a day on something else, distributed practice interleaved with other things.
  • Why is Dylan a big fan of the classic SMILE resources, and what did a lesson look like involving these materials?
    • 00:25:50 CB: Individualised cards with a little task on, or 5 questions for student to try.
    • DW: SMILE is:
      1. Way of keeping track of what kids are doing when they’re not all doing the same thing in a class
      2. Set of resources: worksheets, activities, sometimes pages from a textbook
      3. Community of teachers working together to develop practice
    • Started with 1000 cards. 2000 by the time they finished working together. No curricular sequence. Teachers chose the ones for the child. Thought about child, about task, and about best activities to get child to acquire skill. Innovative. Labour intensive. Ultimately abandoned because “teachers were just handing out worksheets”. That’s what happens if you don’t prepare. If you do, then almost all teaching was 1-1 discussions with pupils. They were all working on different topic. Answers freely available in class. After 10 sets of materials they do a test; if OK they go on, if not they can repeat or even go down a level.
    • 00:28:30 Could it work with children today? In Art or D&T kids work on a project – come in, get their work out, may do it for a whole half-term. But would do more collaborative work if he was doing it today. They worked individually and could ask for help, but collaborative work was not integral.
  • Why is it a mistake to plan a lesson on the assumption that students will understand a topic, and what should teachers do instead?
    • 00:29:50 Biggest problem in classrooms: teachers making decisions about learning needs of 30 pupils based on the responses of a handful of confident volunteers.
    • 00:31:00 Most of time we’re teaching something we planned, so you could use a multiple choice with questions A-D (using fingers to show answer) at points through the lessons.
    • We work from the assumption that students will learn effectively from well-designed teaching (kids not getting it is unusual). Wrong! Kid’s won’t necessarily get it. They don’t get it most of the time. Teaching is difficult + learning is complex = check before moving on (and not just the confident ones).
    • You could ask the three kids that got it to work out a way of explaining it for the next lesson, and then ask the rest of the class to vote on who was clearest. Constantly try to use the resources of the class for the whole class’s benefits. Create a community of learners keep everyone together, rather than trying to spread them out (which is the result of so-called differentiation).
    • People worry about higher-attainers. Robert Slavin and others show very clearly that often when pupils teach each other, it’s the ones who give help who have greatest impact on their own learning.
  • Dylan describes a lesson he taught that went badly, and what he learned from it, which leads to a discussion about the dangers of so-called “real life maths”
    • 00:34:40 Was doing p/t Masters at South Bank Polytechnic (he values the Poly standard, so doesn’t rephrase it as Uni on his CV). A module was on problem-solving. A Y9/10 class. Wanted to work on a real problem. Worked in 8-storey cuboid off the Westway. Gave pupils task of designing fire-escape routes for every room in the school. Thought they might come up with “odd-numbered floors walk to the north, even-numbered to the south” because there were stairwells at each end. A group of girls was designing sprinkler systems – not good for his Masters! But they had ‘deposed’ his problem. What they had done is said “what’s the real problem here – it’s fire in schools … let’s go to the root of the problem.” You can’t student-centre the problem-solving unless you student-centre the assessment. Too often we pretend we want the student’s own work, but then it’s assessed through our norms and prejudices.
    • 00:37:40 CB: emphasis on real world learning. As soon as you do this, they will bring in other real world elements that don’t fit into the design. They can’t be predicted, and therefore can’t be directed as easily. DW: Cf his article “Relevance as Macguffin in Mathematics Education”. Hitchcock believed you shouldn’t spend much time on the Macguffin.
    • Example from maths textbook in 1980s: Alan drinking pint of beer in a pub. Drank 3/8. How much left? => “Yeah, that’s what he’s thinking about as he’s drinking his beer.” “Bogus maths”, “waiting for somewhere to happen”.
    • 00:40:00 A kernel of sense is exemplified in Hans Freudenthals approach to ‘realistic maths education’. All maths teachers use metaphors: e.g. positive/negative bank balances, heights above and below sea level, temperatures to teach a concept like +/- numbers. Using physical situations to capitalise on cognitive equipment students have is fine, but too often students get distracted because they know too much about situation. Jo Boaler: “When do girls prefer football to fashion?” … girls did better on football questions than fashions ones because they knew too much about fashion. Distraction from the mathematics of the LI. Realism/utility is a way of engagement, but must be careful of “mathematics looking for somewhere to happen.”
  • We talk through Dylan’s selection of things he’d wished he knew when he started teaching, including ensuring students know you care, and the importance of forgetting for learning.
    00:41:20 Nine things every teacher should know (or, according to DW ‘Things I wish I’d known’):

    1. “Nobody cares how much you know until they know how much you care”. Often attributed to Theodore Roosevelt. Relationships are at the heart of effective teaching. Started a PGDE course with “I know you’re worried about whether students will respect you. I have a different worry: will you respect your students?” Humans with rights. Small, but human. CB: Ref to Beth Lilley NQT podcast. More than just a number in a markbook.
    2. Learning is a change in long-term memory. Paul Kirschner. If learn it today, but forgotten tomorrow, nothing  is learned. Must be sure to teach for the long-term. Marathon, not a sprint. John Mason: “teaching takes place in time, but learning takes place over time”. Lesson observation: “I want to see progress in a lesson” – nonsense; you might only see the seeds of progress.
      • CB: Can you judge the quality of teaching in a snapshot in a lesson? DW: Heather Hills suggests that to get a reliable (not accurate, but reliable in the sense that you’d get same result tomorrow), you’d need to see each teacher teaching six different classes with each one rated by five independent observers. i.e. 30 observations.
      • Those predictions don’t predict how much children are learning, so some ditch observations for Value Added testing. But that is impossible for two reasons:
        1. Statistics gets very complicated. Dan Goldhaber et al looked at value added. Best 20%, next 20% to bottom 20%. Used random effects model and fixed effects model. 9% of best teachers were rated as very worst in the other model. So value-added unreliable.
        2. Most of what gets measured is not what’s most important for long-term development anyway. Kids may not get good results this year, but may be well-prepared for next year. E.g. problem solving, independence, autonomy: may not cover all skills, but will be prepared. Their results may be excellent in 4 years’ time.
      • We can spot extremes. 95% confident that a teacher who looks very very good is not very very bad. And vice versa. But margin of error for teacher evaluation is so large as to be unjustifiable. We should forget teacher evaluation and say “I don’t care how good you are. You can get better. What would you like to get better at? How can I, as your line manager, support you in getting better at the things you think will have the biggest impact for your students.”
      • CB: Easy to spot off-task behaviour, or active cut-and-stick behaviour, especially for a non-subject-specialist observing. But difficult to judge a child sitting thinking at a desk. And teachers’ confidence can be destroyed after a couple of bad observations. Reduce/remove single lesson observations would massively improve stress.
      • DW: I have one question for every teacher: “do you need to get better?” If they say yes, let’s work together. If not, they should be fired. No place in state schools for teachers who don’t think they need to get better.
      • Every teacher is ridiculously optimistic: we’ve taught a difficult topic 10 times, and still hope that this will be the time they get it first time. We always fail (teaching is hard) Good teachers say “What else can I do to get better?” Bad teachers say “What can you expect from these kids?”
      • CB: A lot comes from observer / line manager. Easy to write down “need to improve pace, need to make sure kids make progress” but if person being observed turns round and says “what can you help me with?” that’s more difficult for observer. Not being able/not being willing/not having time to help the teacher.
      • 00:52:00 DW: What is pace? It’s not speed. Pace is making sure as much of the lesson is spent “minds on”, and that often means slowing things down.
      • Timms research on Comparative Studies of Classrooms: In Japan teachers take boardwork very seriously. They practice it [word for this]. 80% of what teacher writes on board is there at the end of the lesson – it’s a summary of the lesson, and is there to be referred back to. But in England we tend to whip images in front of children on IWB, to show the thing they should be focussing on. But sometimes the thing you should be focussing on is what you did 20 minutes ago. “In my view, directing kids’ attention and engagement is actually antithetical to good teaching.”
    3. 00:53:20 Memory is the residue of thought. The harder you work in the learning task (the more you struggle), the better the long-term learning. So OfSTED’s current focus on clarity and smooth lessons and things looking straightforward isn’t just unhelpful, it’s wrong. Robert Bjork (world leader on memory): the quality of performance in the learning task is inversely proportional to the quality of long-term memory. The more struggle, the better the long-term memory.
    4. Learning requires forgetting. 00:54:00 The best time to restudy something is just before you’ve completely forgotten it.
  • Dylan explains why he thinks lesson observations are a waste of time, and describes a model that he believes would support teacher development much better.
  • Why is a good idea to test our students more, and how can we get around the problem that students often hate being tested?
  • What makes good group work, and how can you make the whole group accountable?
  • We discuss feedback in great depth, specifically when it is effective and when it is simply a waste of time. Some of the findings might surprise you.
  • Dylan describes the most important piece of research he has been involved in, and what other piece of research has surprised him the most
  • How would Dylan up-skill teachers who simply do not have time to go on Twitter, read blogs, study research, listen to podcasts, and so on?
  • Why does Dylan like multiple choice questions, but is not overly keen on using them with mini whiteboards or electronic voting systems?
  • What books would Dylan suggest teachers read?
  • Finally, Dylan offers up advice for Heads of Department, and describes how he would change teacher training courses


    [_edit_lock] => Array
            [0] => 1550492103:1

    [_edit_last] => Array
            [0] => 1

    [from] => Array
            [0] => "26170"|2016-11-03

    [www] => Array
            [0] => http://www.mrbartonmaths.com/blog/dylan-wiliam-author-researcher-trainer-and-assessment-for-learning-expert/|Episode page at Mr Barton Maths

    [cast] => Array
            [0] => Craig Barton*presenter|Dylan Wiliam*guest