 Let's get started, it's on the hour as Dickel would say at our至 on events. Lovely to see you all. How nice to see a few of my friends from Faren as well. I guess they're only here because they want to hear what my voice sounds like.. Dyna'r ysgol i ffynolfi'r cyflawniol. Felly, dywedaeth bod wedi yw'r llfa yma ar y cyflodi fod yw eu hyd sefydliad. Mae winnod, dydy ben fydd yn fwy fyddio'r dweud â'r dweud i chi, ac mae'r fydd gweithio'r dweud. Fy yn jem Imag, mae'n ddweud i'r profesi yng nghymryd y byddwch yn y UK, ac yn cyfieithio ar y Cymru yn llesyn y byeunol yno, ychydig yn ymddi'r ysgol. Bydd y cyfnodd, dywed am y cyfnodd cyfnodd o'r cyfnodd cyfnodd ymddi'r cyfnodd. Felly, dywed am ymddi'r cyfnodd ymddi'r cyfnodd o'r cyfnodd ymddir yma. Yn ymddi'r bod yn ymddi'r cyfnodd, mae'r bach gŵr yn ymddi'r gynhyrch. Felly, ydw'n credu cwestiynau ymddi o'r gweith o'r chart. I'll keep off an eye on the chat log and yeah, by the way, if you have a piece of paper and a pencil beside you, there's a little exercise about halfway through. So let's start by running through some basic assumptions we have when we're doing science, our usual theory of knowledge creation. The basic unremarkable approach we follow when we try and create knowledge. We work with a formally expressed general statement, which has got the potential to explain things. That's our theory. We make a deduction that if a theory is true, then we'd expect to find a relationship between at least two variables A and B. There's our hypothesis. We define exactly what we need to measure in order to observe A and B, the operational definition. And we make the measurements by making observations according to a careful research design. Now, that bit's important, very important. It's probably the heart of what we're doing because a good design allows us to draw conclusions above the hypothesis in a way that tests our belief, tests the fit of data to hypothesis, rather than just saying, ooh, I can see a fit without considering alternative possibilities. And finally, there's a verification. We draw implications back to the theory. Well, that's all straightforward enough. When it's developed sufficiently, we use the theory to generalize, to clarify principles, and eventually to recognize laws out there. Now, notice the assumptions that we're making. We're assuming, first of all, that we can analyze phenomena in terms of variables. We're assuming that we're engaged in an act of discovery, that there's principles out there to be discovered by a dispassionate objective observer. A bit more subtly, we're also assuming that given the evidence, we're always able to distinguish between what's true and what isn't true, so that we can identify the real reason for phenomena. And our purpose is to build theories, general statements which have validly explained what's going on. And then there's Occam's razor. In other words, we should always value the simpler explanation over the more complicated explanation. And finally, once we've developed those theories sufficiently, we should seek to apply them for productive purposes. There's Occam's practicality theories aren't there to hang about in the abstract they're there to be put to use. Now, here's where I'm coming from. That works very well for the physical and the biological sciences, and for their applications and subjects like engineering, for example. And it certainly acts as a model for classical psychology, the sort of psychology I was trained in, probably 50 years ago. But recently people in the social sciences, including psychology, as well as sociology, and social systems researchers have been expressing, increasingly expressing reservations about that approach, as do people in the organizations that they study. Let's look at some of those reservations. And as we do so, please, I'd like you to think of your own organization, the own firm in which you work, the university in which you work, your own friends and neighbours, if you're in the blessed situation of retirement. When you look at the phenomena on the ground, you notice people tend not to ask the question whether the area was here. They tend to ask, look, what's the issue? What are the issues? They're not neutral about it. They're stakeholders with a position to maintain. Sure, as supposedly neutral researchers, we select useful hypotheses to test, but our choice depends on the effort we've put into how and what we've verified in the past. As practitioners in our different roles, we care for what we're doing. We've got a position to maintain and defend. We want to do good by our organization drawing on our particular functional area expertise, and it matters to us that we get it right. Reality is different for the different people involved. Reality is different for the academic and the university administrator. I'm sure I can hear all the academics groaning when I say that. It's different for the full professor and the tenure track lecturer. Different things matter to each. They notice different things. They seek to explain, understand different things. Reality is different for the dean of arts faculty and the dean of engineering. We care for what we're doing from our own stake holding. Truth, truth isn't there to be discovered in an absolute way. Yes, we stick to the evidence and we use evidence to work towards a consensus about what matters, but that's often a process of negotiation towards an agreement. And we might sometimes agree to differ. We satisfy, in other words, we invent a just sufficient outcome that fits the evidence. People don't talk about issues being solved. They talk about them being resolved for the time being. And the purpose of the inquiry isn't so much to explain us to understand sufficient to be able to act. Ogham's Fraser, I hear you throw the term around the euro and fill. It certainly needs throwing around. I've got reservations about it. The phenomenon on the ground out there are quite complex. As Herbert Simon said when he defined complexity, quotes given the properties of the parts and the laws of their interaction, it isn't a trivial matter to incur the properties of the whole. And then we have the cybernetician, Ross Ashby, and the notion of requisite variety. The complexity of our explanations, he says, must match the complexity of the phenomena. And the simpler explanation may miss something that's obvious to people with different stake holdings. Application. Well, you know, our theories are never complete and never finished. Our actions take place alongside the development of our understanding of what's going on. I wonder what your reactions are so far. Just thinking about all of that. Notice how the language has become vaguer. We've abandoned, for a moment, the search for truth out there. Absolute explanations. Have we thrown the baby out with the bathwater? Have we abandoned all rigor? Yeah, tagline. Absolutely. No hypotheses would look at that in a moment. Vague language, though. What can we do about it? How can we handle the situation? Well, there is a way of being rigorous. The constructivist approach. It flops out of work by a clinical psychologist in the USA seeking to develop a new theory of personality, a chap called George Kelly. And he offered, he ended up offering to us not just a theory, but a complete epistemology, a theory of knowledge, an alternative theory of knowledge. And I think the best way I can get that across is by giving you a story. Let me tell you a story. One day back in the 1940s, Kelly spent his morning with one of his doctoral students in a tutorial. In the tutorial, he encouraged the student to state a research question, set up a hypothesis, no and alternative tagline, deduce what kind of data would test the hypothesis, draw a suitable sample. He helped to decide on how to gather and analyze the data, draw conclusions that arrived at informative answers to the research question and generalize the outcomes in a lawful manner. Then he went off to lunch. And after lunch, he spent an hour doing therapy with one of his clinical clients. In that therapy session, he asked the client to state what he thought was distressing him and causing the symptoms, asked him what he believed would make life easier, asked how he might check the belief through his behavior in the following week and come back with his thoughts on what had worked and what hadn't worked and asked what sort of a difference it might make if he behaved in the way it worked. And you see what we're saying. What's the difference between what was happening in the morning and what was happening in the afternoon? Each session was seeking to make sense, seeking to construe experience. And what Kelly is famous for in my field is offering us the model of the person as scientist. But he says something much more profound than, you know, we're all scientists trying to make sense of things. He's saying there's no difference between what the laboratory scientist does and what the man or woman in the street do. The epistemology is the same. The theory of knowledge is the same. What constitutes proof is the same in each case. We need to talk a little about his theory. But first of all, I've got an exercise. Have you all got a piece of paper and a pencil beside you? I'd like you to do something. There's an invitation to an exercise. The back of an old envelope will do. Everyone got one? Just say yes in a nearby chat perhaps. Sumo, Dali, Phil, Tagline, Extender. OK, right. I'd like you to think about your experience in second life. How do you construe the people you interact with? And I'd like you to write down the names of six avatars known to you in second life across the top of the paper like so. I've written down Jenny, Fred, Sally and three other avatars that I know. Please write down your own six. If you write down the names of anybody here, no looking over each other's shoulders. And again, let me know when you've got six names written down there. Let's keep it confidential to yourself for the time being. Keep it in RL, not SL. But Sizzigie, I'm going to be doing one of the team building exercises in a couple of weeks time based exactly on this. And it'll be about sharing. So in the meantime then, has everybody got six names written down? Again, could you say yes, yes, yes? Lovely. And now what I'd like you to do please is to address the following question. Please look at your avatars number one, number two and number three and answer the following question. Which two are alike in some way and thereby different from the third? What do two of them have in common in contrast to the third? There may be several similarities and differences that occur to you. Just focus on one for the time being. And write down what they have in common. Oops, where's it gone to? Write down whatever it is two of them have in common on the left hand side. And the contrast that characterises the odd one out on the right hand side. And again, I'll pause for everyone to do that. And this time the pause might be a little longer because I want you to write something down that's as explicit as possible. You know, if you're writing down nice versus nasty, in what way nice, in what way nasty. I'm sure you're a kind person, you will be doing, you won't be saying that. Or maybe you will be. OK, so again a phrase on the right, on the left that characterises the two that go together. And the contrast to that, not necessarily just the logical opposite, but you know the in what way different on the right hand side. You might not realise it, but what you're doing is creating a five point rating scale. The phrase on the left anchors the one end of the five point scale. And the phrase on the right anchors the five end of the scale. There's what I wrote down. And you know, please don't write that down yourselves. Really it's got to be your own construct, not mine. But I wrote down that the two what they have in common is that they're relaxed and laid back. Whereas the chat on the right is a bit of a warrior. OK, has everybody got a phrase? And the opposite will come to that in a moment. I'm just waiting for everyone to have something written down on the left on the right. I want you to give each of your avatars a rating on that scale from one to five. Depending on whether the phrase on the left applies to them. Or the phrase on the right applies to them. Just write their rating down below the name. So here's what I did. Jenny and Sally are relaxed and laid back. They both get a rating of one. Remember, relaxed and laid back is the on the left is the one end of the scale. In contrast, Fred is a bit of a warrior. I gave him a three. Notice once I thought about it, Flora and Bert are even greater warriors than Fred. And so they got ratings of four and five. So please, could you write down your own ratings on whatever phrase you wrote down, giving each a rating of one to five? Yes, that's right, tagline. Opposite, but contrasting rather than opposite. Nice, not nice doesn't tell us very much. Nice, as opposed to totally nasty, begins to provide us with a particular contrast. The rating scale you wrote down wasn't mine. And that is so important. It was yours. I didn't lay down on you a way of thinking about six avatars. You used one of your own constructs, one of the ways you have to construe your social environment in second life. It's called a construct. We've identified a construct, a way in which you have a making sense of your social world. The construct you wrote down will be almost certainly different to mine. And by the way, we're not looking for right ways or wrong ways of seeing the social world your way and your defining the terms of discourse. And so on, we would continue. I would offer you another three, verse number two, verse number five, verse number six, for example, in what new way are two of them the same and one different. Don't repeat yourself a new way in which two are the same and one different. I'll show you a typical outcome of the process in a moment. But now we've defined what a construct is. Right? It is a tagline, has it? Yeah, contrasting attributes of an individual. And we'll see, Rockhound, how we can get or multivariate about these polar opposites. Now that we've defined what a construct is, let's dodge back to Kelly's theory and say a little bit about how he construed people's construing. Kelly did something that's quite unusual in psychology, actually. He offered an explicit theory stated formally, stated in terms of a fundamental postulate and 11 corollaries. His fundamental postulate says the following, the world out there is real, as real as any traditional null hypothesis testing account of existence. But he also said the world in here is equally real, in here in the head, and that people operate by building internal representations to match what they see of the external world in order to anticipate their future experience. And they do that whether they're scientists or not. His corollaries deal with how people alter their constructs in the light of experience, developing their internal model to match that experience. And it describes in various ways how they interact with other people in order to share their models. And that's how people are scientists, says Kelly. That's how we as researchers can develop our own models of human behaviour without necessarily having to accept the positivist assumptions I outlined earlier, without having to accept a real world that's true out there, but rather living with a world that we construe, a world about which we invent constructions, obviously still in accordance with evidence. But let's not be salipsist about this. There isn't a single true world out there. Knowledge is a social construction. That, after all, is why scientists hang about in groups, build a conferences, share one another's knowledge. It's open to rigorous analysis nevertheless. For example, and in conclusion, here's what my own construing looks like after I completed the exercise by eliciting seven more of my own constructs. What you see in front of you, by the way, is called a repertory grid. That's one of the things I've specialised in down the years. There's eight of my constructs, I think in terms of how relaxed and laid-back they are, but also the extent to which they're dominant, whether they're into building things in second life, how careful they are with their money here, different ways in which I feel close to them. Notice the constructs are bipolar. That's so important. You have to ascertain the contrast to capture a person's meaning. Look, think about it. The word good is meaningless. Good means one thing if we're talking about a student's essay, good as opposed to poor. Good means something completely different if we're talking about good versus evil. Just feel the weight of that word evil reflecting back on our use of the word good. Good by itself doesn't tell us anything. Meaning is carried by contrast. Explicit or tacit. One of the things about doing a repertory grid is that it helps people to surface their tacit constructs as well as their more obvious and evident ones. The other thing to notice is that constructs should be as specific as possible if they're going to be useful. Construct number four, for example, initially I was thinking two of them are mean and one of them is generous, but then I got a bit more specific. I operationalised it in terms of these two are careful with their Linden Dollars whereas this guy spends a lot of money in Second Life Marketplace and is prepared to travel to in-world markets and get terribly upset when the market in-world no longer exists because the persons sold the site in which he was doing sales. Okay, so here we are in a constructivist world. Isn't all of it a bit vague and woolly? Well, no, because notice there are numbers there as well as words. In fact, meaning is being expressed by in terms of the number locating the avatar with respect to the words. You need both. We all remember the old saying, if a thing exists, it exists to some degree and if it exists to some degree, it can be measured. I'm just pausing to read what you're saying, Tagline. Yeah, yeah, yeah. I once did a study in which I looked at what bankers, how bankers construed an effective lending officer and it was quite funny to see how they construed someone who was good at making commercial lending decisions to small businesses. And then I asked the banks, look, I'm going to use a double-blind technique. I want you to identify the people who you know have succeeded in their predictions about business success and which lending officers have succeeded in their predictions about business success and which lending officers have not been as effective. And would you believe the banker would not take part in that part of the exercise? It was just close to the bone, I think, to be making statements about who was more effective and who was less effective for the banker, in ways in which we can do that for ourselves in our repertory grids. OK. So, yes, relationships between these characteristics. We've got some numbers, folks. So we can do some quite advanced analyses. For instance, we can do a cluster analysis on that grid. We reorder the constructs and the avatars. So the ones with the most similar ratings lie side by side. And then we group the individuals according to their degree of similarity. So have a look at, oops, I've skipped one. Bear with me, I've got to go back. One. How do I go back to number 12? Yep. Have a look at that. The tree structure, bottom right-hand corner in red. That tells us something about how the ratings group the individual avatars. So Sally and Jenny received ratings that are almost the same. In fact, on that underground, their ratings are much at about 96%. Bert and Flora are rather different to the others. At best, Flora matches with Sally as low as the average. Matches with Sally as low as about 69%. By the way, I mean, that's a graphic representation. I'm saying about, you know, we can print out matrices of the exact similarity values. That grouping is information about how I think. Now have a look at the tree, the dendrogram in blue on the right-hand side, halfway up. There's evidence there about how I'm thinking. It shows the structure of my thinking and the implications I might draw. For instance, look at the ratings along the relaxed construct and I could meet construct. They're very highly matched. If you follow the lines across until they meet the common apex on the right-hand side, they're matched at about 93%. There's a sense in which whenever I think of an avatar as relaxed and laid back, there's somebody I wouldn't mind meeting in real life. Whereas wherever I think of an avatar as a bit of a warrior, well, to that extent, our relationship is more likely to be confined to second life. I wouldn't especially be excited about meeting them in real life. There's the other high match, the second one down and the third one down, matched at 93%. I feel I can trust someone that I care a lot for and sometimes a bit cautious with respect to somebody that I feel neutral about. So, all that we've done is worked out the sum of differences vertically downwards and expressed as a percentage score in the red dendrogram. Then we've worked out the sum of differences across between every construct and every other construct and used them to reorder the constructs, so that those with the most similar ratings, right side by side. Yes, there will be. I can send the whole thing to you. So, that's one way of looking at the data we have in the repertoire grid. There's another way. Another form of analysis, equally rigorous principle component analysis. It's kind of like a light factor analysis. It's the same repertoire grid. This time the constructs are organised according to the patterns of variance in the ratings. Plotted along two orthogonal axes, the first two orthogonal axes that come up in the analysis. You can say something about the avatars in terms of the different, the distance between each avatar. And you can say something about the similarity between constructs in terms of the angles between them. Oh, Rockhamd, you can make all sorts of decisions with this. Next time you decide to buy a car, right? List six possible cars, six to eight possible cars. Do the exercise which two are like in one different in some particular way. Get out as many distinct constructs that you have. Then go to the motoring press and have a look at what the reviewers are saying and pick up any of their professional constructs and then use those for ratings. Then here's where the decision comes. Add an additional car. It's called my ideal car. Rate that car on all of the constructs, yours and the professional ones. And then buy the car whose ratings lie closest to the ideal. Okay, I tend not to use principle components analysis very much. It can be quite complicated to explain principle components et cetera to a client, whereas the cluster analysis, things lying together, close together in a tree structure is easier to explain. However, there's one last thing I want to draw your attention to and that is the bit at the very bottom. There's a plot of the proportion of variance explained by the various components. And as a very useful rule of thumb, the total variance explained by the first two principle components, the ones that are graphed up there, tells you something about the cognitive complexity of a person's thinking about that topic. Think about it. The more components it takes to account for all the variance in the ratings, the more complex the thinking, the fewer components. Well, I'm only thinking in one or two ways, black and white terms. Here, in this grid, my thinking is very simple. If you add up the percentage of variance accounted for along the horizontal axis, there it is on the right-hand side. And along the vertical axis, there it is, labelled at the bottom of the vertical axis, you can see that 90% of the variance in all those ratings is accounted for by only two components. I've only got a few distinct ways of thinking about all these people in this grid anyway. And so to conclude, there we are, we've covered a lot of ground. Thanks ever so much for your patience. There's really just three things that matter to me that you can... Oh, sorry, what are they? Sorry, okay. Just to go back to Sisiqi's question. I've lost it. Hang on. I've just moved the camera too far back with first two components. It's a very wise question. There's a lesson to be learned in that question. That was, hold on, number 13. What are the components? Well, you tell me. And the important lesson is that the analysis doesn't tell us the name of the component. We have to make a judgement reflecting the constructs which lie closest to that horizontal line, to give the horizontal line a name. And similarly, vertically, to give that vertical component a name, we've got to think in terms of the constructs which are angled most close to it. This is the problem for any multivariate analysis. You can be as numerate as you like. At the end of the day, you make a personal judgement as to what you're going to label as the underlying variable. Hopefully you'll do that with a well-trained mind, following some careful principles, and argue about it with your colleagues who have also conducted the same study. And look back at the literature and see what terms other people are using for those kinds of data. But the actual name doesn't come out from the analysis. You have to decide. Okay. So, if you're going to take everything away from this, I think three things matter. I wonder if you agree. Number one, people behave as scientists as they seek to make sense of their experience by mapping their, their construing onto the events they encounter and by caring about whether other people see those experiences in the same way or not. Scientists may have a few more precise ways of making sense of their experience of the data, but they're subject to the same needs to negotiate shared understandings with other scientists, with other people as they seek to map their data onto their hypotheses. Science depends on interaction between scientists. If you've done a study and you've got your results and you keep them in your own head, you haven't done science. You've got to share it, write an article, have it reviewed, put it out into the community of scientists, and enjoy the critique. It's a way of making sure that your measurements are reliable. And remember, if you're a hypothetical deductive positivist believing in discovering truths out there, you can't identify valid data if it's not reliable, if your measurements aren't reliable. So, there you go, both the scientists and the person in the street are doing ultimately the same thing. That's Kelly's epistemology. It's called constructivism, and it's a different way of looking at the world than positivism. I'm going to argue that it's suited to some of the subject matter that psychologists have to do these with. There you go. That's it. Got any questions?