27 Apr 2017

Bruce's Modelling Gripes, No. 5: None or not many results

The point of a model lies mostly in its results. Indeed, I often only bother to read how a model has been constructed if the results look interesting. One has to remember that how a model is constructed - its basis, the choices you have made, the programming challenges are far, FAR more interesting to you, the programmer, than anyone else. Besides, if you have gone to all the trouble of making a model, the least you can do is extract some results and analyse them.

Ideally one should include the following:
  1. Some indicative results to give readers an idea of the typical or important behaviour of the model. This really helps understand a model's nature and also hints at its "purpose". This can be at a variety of levels - whatever helps to make the results meaningful and vivid. It could include visualisations of example runs as well as the normal graphs -- even following the events happening to a single agent, if that helps.
  2. A sensitivity analysis - checking how varying parameters affects the key results. This involves some judgement as it is usually impossible to do a comprehensive survey. What kind of sensitivity analysis, over what dimensions depends on the nature of the model, but not including ANY sensitivity analysis generally means that you are not (yet?) serious about the model (and if you are not taking it seriously others probably will not either).
  3. In the better papers, some hypotheses about the key mechanisms that seem to determine the significant results are explicitly stated and then tested with some focussed simulation experiments -- trying to falsify the explanations offered. These results with maybe some statistics should be exhibited [note 1].
  4. If the simulation is being validated or otherwise compared against data, this should be (a) shown then (b) measured. [note 2] 
  5. If the simulation is claimed to be predictive, its success at repeatedly predicting data (unknown to the modeller at the time) should be tabulated. It is especially useful in this context to give an idea of when the model predicts and when it does not.
What you show does depend on your model purpose. If the model is merely to illustrate an idea, then some indicative results may be sufficient for your goal, but more may still be helpful to the reader. If you are aiming to support an explanation of some data then a lot more is required. A theoretical exploration of some abstract mechanisms probably requires a very comprehensive display of results.

If you have no, or very few results, you should ask yourself if there is any point in publishing. In most occasions it might be better to wait until you have.



Note 1: p-values are probably not relevant here, since by doing enough runs one can pretty much get any p-value one desires. However checking you have the right power is probably important.  See

Note 2: that only some aspects of the results will be considered significant and other aspects considered model artefacts - it is good practice to be explicit about this. See

20 Apr 2017

Bruce's Modelling Gripes, No. 4: Not being open about code etc.

There is no excuse, these days (at least for academic purposes), not to be totally transparent about the details of the simulation. This is simply good scientific practice, so others can check, probe, play around with and inspect what you have done. The collective good that comes of this far outweighs and personal embarrassment at any errors or shortcomings discovered.

This should include:
  • The simulation code itself, publicly archived somewhere, e.g. openabm.org
  • Instructions as to how to get the code running (any libraries, special instructions, data needed etc.)
  • A full description of the simulation, its structures and algorithms (e.g. using the ODD structure or similar)
  • Links or references to any papers or related models
Other things that are useful are:
  • A set of indicative/example results
  • A sensitivity analysis
  • An account of how you developed the simulation, what you tried (even if it did not work)
For most academics, somehow or other, public money has been spent on funding you do the work, the public have a right to see the results and that you uphold the highest academic standards of openness and transparency.  You should do this even if the code is not perfectly documented and rather dodgy!  There is no excuse.

For more on this see:
Edmonds, B. & Polhill, G. (2015) Open Modelling for Simulators. In Terán, O. & Aguilar, J. (Eds.) Societal Benefits of Freely Accessible Technologies and Knowledge Resources. IGI Global, 237-254. DOI: 10.4018/978-1-4666-8336-5. (Previous version at http://cfpm.org/discussionpapers/172)

12 Apr 2017

Bruce's Modelling Gripes, No. 3: 'thin' or non-existent validation

Agent-based models (ABM) are the ultimate in being able to flexibly represent systems we observe. They are also very suggestive in their output - usually more suggestive than their validity would support. For these reasons it is very easy to be taken-in by an ABM, either one's own or another's. Furthermore, good practice in the form of visualisations and co-developing the model with stakeholders increase the danger that all those that might critique the model are cognitively committed to it.

It is to avoid such (self) deception that validation is undertaken - to make sure our reliance on a model is well-founded and not illusory. Unfortunately, effective validation is resource- and data- intensive and is far from easy. The flexibility in making a model needs to be matched by the strength of the validation checks. To be blunt - it is easy to (unintentionally or otherwise) 'fit' a complex ABM to a single line of data, so it looks convincing. This kind of validation where one aspect of the simulation outcomes is compared against one line of data is called 'thin validation'.

In physics or other sciences they do do this kind of validation, but they have well-validated micro-foundations so the effective flexibility of their modelling is far less than for a social simulation, where assumptions about the behaviour of the units (typically humans) are not well known. That is why such thin validation is inadequate for social phenomena.

Of course the kind and relevance of validation depends upon your purpose for modelling. If you are not really talking about the observed world but only exploring an entirely abstract world of mechanisms it might not be relevant. If you are only illustrating an idea or possible series of events and interactions then 'thin' validation might be sufficient.

However if you are attempting to predict unknown data or establish an explanation for what is observed then you probably need multi-dimensional or multi-aspect validation - checking the simulation in many different ways at once. Thus one might check if the right kind of social network emerges, and statistics about micro-level behaviours are correct and that emergent aggregate time-series are correct and that in snapshots of the simulation there is the right kind of distribution of key attributes. This is 'fat' validation, and then starts to be adequate to pin down our complex constructions.

Papers that have thin or non-existent validation (e.g. they rely on plausibility alone as a check) might be interesting but should not be interpreted as saying anything reliable about the world we observe.

10 Apr 2017

Bruce's Modelling Gripes, No. 2: Making excuses for pragmatic limitations on modelling

We are limited beings, with limited time as well as computational and mental capacity. Any modelling of very complex phenomena (social, ecological, biological etc.) will thus be limited in terms of: (a) how much time we can spend on it (b) how much detail we can cope with (checking, validating, understanding etc.) (c) what assumptions we can check

This is find, but instead of simply being honest about these limits there is a tendency to excuse them, to pretend that these limitations are more fundamentally justified. Three examples are as follows:
  • "for the sake of simplicity" (e.g. these articles in JASSS), this implies that simpler models are somehow better in ways beyond that of straightforward pragmatic convenience (e.g. easier to build, check, understand, communicate etc.)
  • That more complicated models are less complex (e.g. Sun et al.2016) which shows a graph where "complexity may decrease after a certain threshold of model complicatedness". This is sheer wishful thinking, what is more likely to be true is that it is harder to notice complexity in more complicated models, but this is due to our cognitive limitations in terms of pattern recognition, not anything to do with emergent complexity.
  • Changing English to make our achievements sound more impressive than they are, e.g. to call any calculation using a model a "prediction", when everybody else uses this word to really mean prediction (i.e. anticipating unknown data/observations sufficiently accurately using a model).
These weasel words would not matter so much if they were (a) purely internal to the field and everyone understood their meaning and (b) they were not used in public/policy consultations or grant applications where they might be taken seriously. Newcomers to the field often take these excuses too literally and so change what they attempt to do as if these excuses were genuine! This can be an excuse for following the easy option when they should be pushing the boundaries of what is possible. When policy makers/grant funders misunderstand these claims, an inevitable disappointment/disillusionment may follow, damaging the reputation of the field.

References

Sun, Z., Lorscheid, I., Millington, J. D., Lauf, S., Magliocca, N. R., Groeneveld, J., ... & Buchmann, C. M. (2016). Simple or complicated agent-based models? A complicated issue. Environmental Modelling & Software, 86, 56-67. 

27 Mar 2017

Bruce's Modelling Gripes, No.1: Unclear or confused modelling purpose

OK, maybe I am just becoming a grumpy ol man, but I thought I would start a series about tendencies in my own field that I think are bad :-), so here goes...

Modelling Gripe No.1: Unclear or confused modelling purpose

A model is a tool. Tools are useful for a particular purpose. To justify a new model one has to show it is good for its intended purpose.  Some modelling purposes include: prediction, explanation, analogy, theory exploration, illustration... others are listed by Epstein [1]. Even if a model is good for more than one purpose, it needs to be justified separately for each purpose claimed.

So here are 3 common confusions of purpose:


1.    Understanding Theory or Analogy -> Explanation. Once one has immersed oneself in a model, there is a danger that the world looks like this model to its author. Here the temptation is to immediately jump to an explanation of something in the world. A model can provide a way of looking at some phenomena, but just because one can view some phenomena in a particular way does not make it a good explanation.

2.    Explanation -> Prediction. A model that establishes an explanation traces a (complex) set of causal steps from the model set-up to outcomes that compare well with observed data. It is thus tempting to suggest that one can use this model to predict this observed data. However, establishing that a model is good for prediction requires its testing against unknown data many times – this goes way beyond what is needed to establish a candidate explanation for some phenomena.

3.    Illustration -> Understanding Theory. A neat illustration of an idea, suggests a mechanism. Thus the temptation is to use a model designed as an illustration or playful exploration as being sufficient for the purpose of a Understanding Theory. Understanding Theory involves the extensive testing of code to check the behaviour and any assumptions. An illustration, however suggestive, is not that rigorous. For example, it maybe that an illustrated process only appears under very particular circumstances, or it may be that the outcomes were due to aspects of the model that were thought to be unimportant. The work to rule out these kinds of possibility is what differentiates using a model as an illustration from modelling for Understanding Theory.

Unfortunately many authors are not clear in their papers about specifying exactly for which purpose they are justifying their model.  Maybe they have not thought about it, maybe they are just confused and maybe they are just being sloppy (e.g. assuming because its good for one purpose it is good for another).


6 Mar 2017

The Post-Truth Drift -- why it is partly the fault of Science (a short essay)

In a time which talks about being in a "post-truth" era of public discourse and where the reputation of "experts" as a group is questioned, it is easy to blame others for the predicament that Scientists find themselves (e.g. politicians, journalists, big business interests etc.). However, I argue that substantial part of the blame must fall on ourselves, the scientists  -- that we have (collectively), more than anyone, knocked away the pedestal on which they stood.

Firstly, scientists have increasingly allowed their work to be prematurely publicised - "announcing" breakthroughs with the first indicative results (or even before). It is, of course, understandable that scientists should believe in their own research, but it should be part of the discipline that we do not claim more than we have proved. Partly this is due to funding and institutional pressure, to quickly claim impact and progress (I remember an EU funding call that asked for "fundamental theoretical breakthroughs" and "policy impact" in the same project), but again it is part of the job to resist these pressures. More fundamentally, the basis for academic reputation has changed from cautious work to being first with new theories -- from collective to individual achievement.

The result of this over-hyping of results is that science loses its reputation for caution and reliability. This has been particularly stark in some of the "softer" sciences like nutrition or economics. All the clever mathematics in the world did not stop economists missing the last economic collapse - their lack of empirical foundations coming back to bite them. In the case of nutrition, a series of discoveries have been announced before their full complexity is understood.

However this goes a lot further than the softer sciences. The recent crisis in reproducibility in many fields indicates that publication has overtaken caution even for results that are not publicised outside their own field. This indicates that there is an imbalance in these fields with not enough people replicating and checking work and too many racing to discover things first. This is evident in some fields where there are a lot of researchers proposing or talking about abstract theories and not many doing the more concrete work, such as empirical measurement. Reputation should follow when an idea or model empirically checks out and not before.

Measuring academic reputation on citation-based indices reinforces this deleterious trend - one can get many citations for proposing an attractive or controversial idea, but it is independent of whether one was right or not. If we reward academics by their academic popularity with their peers rather than whether they were right, then that will affect the kind of academics we attract into the profession. Many fields are dominated by cliques who cite each other and (consciously or unconsciously) determine the methodological norms.

All fields need some methodological norms, however these norms can come about in ways that are independent of their success or reliability. Papers that grab a lot of attention can be more influential in these terms than those that turned out to be right. All fields seem to adjust their standards of success to ensure that the field, as a whole, can demonstrate progress and hence justify itself. When faced with highly complex phenomena, this can lead to a dilution of the criteria of success so this is achievable. In my field, abstract simulations without strong empirical foundations that provide a way of thinking about issues, gain more attention than they should and, more worryingly, are then advertised as able to perform "what if" analyses on policy inventions (implying their results will somehow correspond to reality). In economics, prediction rather than structural realism was declared as the aim of its modelling, but this weakened to predicting known out-of-sample data.

If all this weakening of criteria were internal and scientists were ultra-careful about not deceiving others into thinking their results were reliable, this would not be so bad. However, whether deliberately or otherwise, far too often the funders/policy makers/public are left with an impression that declared findings are more solidly based than they are. This is exacerbated by the grant funding process, where people who promise great results and impact are funded and more realistic proposals rejected. If one gets a grant, based upon such promises there is then pressure to justify outcomes that fall short of these, and to use language to obscure this.

Finally, when scientific advice and the policy world meet there is often fundamental misunderstanding, and this is partly the fault of the academics. In the wish for relevance and "impact" academics can be pressured into not being completely honest, and providing the policy makers with what they want regardless of whether this is justified by the science. One trouble with this interface is that there is not a clear line of responsibility -- if the advice from the scientists conflicts with those of the policy makers, what are they to do?  If they trust some complex process that they do not understand they are effectively delegating some of their responsibility, if they only trust it if agrees with their intuitions then this selection bias ensures that support rather than critique of decisions gets diffused.

The blurring of political and scientific debate that results from a non-cautious entry into policy debate has resulted in a conflation between debate over method and reliability of results to a confrontation of alternative results. Classically science has not debated results, but rather critiqued each other's method. If there are conflicting results this will not be resolved by debate but by further research. Alternative ideas should be tolerated until there is enough evidence to adjudicate between them. The competition should be in terms of sounder method, not in terms of which theory is better on any other grounds. Thus scientific debate has become conflated with political debate where, rightly, different ideas are contrasted and argued about.

There are some positive sides for science to the "post-truth" tendencies. The disconnected "ivory tower" school of research is rightly criticised. Whilst what academics do should not be constrained, what they use public money for has to be. The automatic deference that academics used to have from the public has also largely disappeared, meaning that results from scientists will be more readily questioned for its unconscious biases and meaning of its claims. To some extent the profession has become more porous with a wider range of people participating in the process of science - it is not just professors or boffins anymore.

Ultimately maintaining academic and research standards is the job of the academics themselves. The institutions they work in, the funders of research and the current governmental priorities mean that other involved actors will have other priorities. Universities compete in terms of the frameworks that government sets (REF, TEF etc.) or league tables constructed on simplistic indices. Funders of research are under pressure to claim research that has immediate and significant impact and publicity. It is only the academics themselves that can resist these pressures and so maintain their own, long-term reputations for independence and reliability. If we do not have these things, why should the public carry on financing us? What will people think of our science in 50 or 100 years time? Let the longer view prevail.

17 Feb 2017

Discussion paper: "Co-developing beliefs and social influence networks – towards understanding Brexit"

Centre for Policy Modelling Discussion paper: CPM-17-235

Abstract.

A relatively simple model is presented where the beliefs of agents and their social network co-develop. Agents can either hold or not each of a fixed menu of candidate beliefs. Depending on their type, agents have different coherency functions between beliefs, so that they are more likely to adopt a belief from a neighbour or drop a belief where this increases the total coherency of their belief set. With given probabilities links are randomly dropped or added but, if possible, links are made to a “friend of a friend”. The outcomes when both belief and link change processes occur are qualitatively different from either alone, showing the necessity of representing both cognitive and social processes together. Some example results are shown which moves a little towards modelling the processes behind divisive collective decisions, such as the Brexit vote.

Downloads

15 Feb 2017

Slides from talk on "Simulating Superdiversity"

#cfpm_org #abm #ethnosim
An invited talk to the Institute for research into superdiversity (IRIS), University of Birmingham, 31st Jan 2017.

Abstract: A simulation to illustrate how the complex patterns of cultural and genetic signals might combine to define what we mean by "groups" of people is presented. In this model both (a) how each individual might define their "in group" and (b) how each individual behaves to others in 'in' or 'out' groups can evolve over time.  Thus groups are not something that is precisely defined but is something that emerges in the simulation. The point is to illustrate the power of simulation techniques to explore such processes in a non-prescriptive way that takes the micro-macro distinction seriously and represents them within complex simulations. In the particular simulation presented, groups defined by culture strongly emerge as dominant and ethnically defined groups only occur when they are also culturally defined.

Slides available at: http://www.slideshare.net/BruceEdmonds/simulating-superdiversity

Linked to the upcoming workshop on simulating ethonocentrism and diversity, Manchester 7/8 July 2017: 

3 Jan 2017

Sad to announce the death of my mother, Anne Gillian Edmonds ("Gill") 1935-2016

My mother died on the 24th of December, 2016. 

She was a loving and effective mother to us four (Bruce, Juliet, Nicola and Malcolm), encouraging us all to be creative, independent and care about social concerns. This little 'wave' of people continues with her grandchildren: Thomas, Ruth, Duncan, Orlanda, Patsy, Sophie, Ben, Iona, Kaila, and Joshua and great-grandchild, William.

As well as raising us mob, she pursued an active career as a social worker for Oxfordshire district council (and in the last couple of years of her work, the RAF). She was an active campaigner on Green and development issues, bombarding David Cameron (her local MP) with letters on these issues before he became PM.  She contributed actively to local life in Burford, volunteering her effort in many ways.

She was a relentless autodidact, achieving an open university degree whilst bringing us up, and continuing on a variety of courses until the very last years of her life.  She spent several years effectively looking after my father as his dementia progressed.

A funeral service will be held at St John the Baptist Church, Burford, on 14th January at 11.00 a.m. All welcome. (Family Flowers only)