Let's take a look at some techniques that are popular for getting into the idea of what are the root causes. Here's a list that will be going through in this session: brainstorming; affinity diagram; Pareto analysis; fishbone, what is also called the Ishikawa diagram; 5-Why Analysis or what I like to think of, my five-year-old's asking five whys and many more whys than five because they keep asking why for every response that you give them for the previous why. You say, "Don't put your hands in the socket." "Why shouldn't I do that?" "Because you might get a shock." "Why will I get a shock?" Those kinds of things are what we are talking about, more from a process perspective when we're talking about a 5-Why analysis. Just going through the rest of the list. The things that you'll be seeing in this session, cause effect matrix and scatter diagram. Scatter diagram is where we start bleeding into from the idea of exploring root causes to starting to test them. Scatter diagram is already starting to test the relationships based on doing an x-y analysis on a graph and saying, "Does it look like there's a relationship?" There we're on the border of formulating hypotheses and starting to test them when you're thinking of a scatter diagram. Let's take a look at the first technique here, and this you may be familiar with. You may have heard this term being thrown around quite a bit, brainstorming. This is what we do in teams in terms of, "Let's brainstorm to come up with a way of doing something." That's exactly the idea when you're using it for a process improvement project. What are you doing when you're brainstorming? You're getting people who may be from different levels of expertise, who may be from different areas of expertise into a room, and you are trying to get their ideas together. Now, the important thing when you're doing this in a commercial setting, is that when you do have people who have diverse views, there can be problems. There can be problems of somebody might be more dominant in terms of not letting others speak or there might be some people who don't speak up. In terms of trying to find a way of having an effective brainstorming session, you can structure the brainstorming session in many ways. You can have rules or you can have ways in which people can give their anonymous opinions or anonymous ideas before you start discussing them. You can have rules as to every person has to speak once and you take that down and then you open it up for everybody else. Or you can say there can be a way of giving your opinion, giving your assessment of the situation, giving your best judgment in terms of anonymous pieces of paper or on software in a survey and then using that in some way to collect the information. The idea being that you want to be able to get all the diverse points of view. You want to be able to get that idea that's way out there and somebody should not be afraid to give that idea that's way out there. When you're thinking of innovative ideas, sometimes we think about them as being laughable ideas and laughable ideas are to some extent good when you're talking about a brainstorming session. The idea of somebody simply coming up and saying, "What if we could do this?" This brings to example, a particular instance of brainstorming in health care where somebody said, "How about if we have a drive-through flu clinic?" It was laughable idea at that time but it ultimately became a reality for hospitals to do drive-through clinics. Again, the idea of brainstorming is to get a free flow from ideas without there being any judgment of those ideas at that point in time but trying to get everybody to discuss those ideas. In the same way as brainstorming, there's the idea of affinity diagram. This is a little more systematic than an affinity diagram. Brainstorming can be made systematic based on an affinity diagram. It can also be based on another technique. It can be also be made systematic based on another technique called a nominal group technique. That's another technique that gets used in order to get people's ideas and get them together so that you can do something about them. But what is the affinity diagram? Affinity diagram is basically an idea of categorizing different reasons for a problem, so not only collecting the data but also categorizing it. Having people give their opinions, having people give their ideas, and then categorizing them into different categories, putting them in different clusters, putting them in different groups, so that they can be more manageable in terms of doing something about them. Now these could be ideas that are coming from people, or these could be a list of customer complaints. You could be taking those lists, that list of customer complaints and saying, these are really about our drivers who are delivering things to our customers. These are really about our food quality and what we're making in terms of our take-out service. These are really about the service level in terms of our order takers and how they are taking the order. You could take customer complaints of some number and start to categorize those and have people categorize those. Have people who know about these things categorize those. Now, for any of these kinds of people involved, categorization or data collection kinds of exercises, when you have people from the process, you are getting an additional benefit in terms of involving the people that are working in the process. You are also serving that purpose of getting buy-in from the very people who are going to be taking the changes that you might get from this project, from this initiative, and using them. That's also a bonus if you can call it that on top of you being able to speculate on root causes. But let's move further into the affinity diagram and see how you would create one, get into the specifics of it. There are four steps that we talk about when we say come up with an affinity diagram. For these, you need materials like sticky notes. If I may use a brand we talk about Post-its and Post-its are what we use in order to collect people's ideas. So sticky notes, marking pens, you need a large surface like a table. If it's going to be a flat table or it can be a wall, or where you can put different Post-its and start to group them together. What I have done for these kinds of exercises usually is I use an easel board and the big chart paper that you get from the easel board. Those are the ones that you can actually paste on the wall. The advantage that you get from that is that you can take those with you and then codify them, put them into an Excel spreadsheet in terms of codifying them as data for future use. That's something that you can do. But what are the steps for an affinity diagram? You get the ideas from people or you collect complaints from people. You display the ideas so that people can see them. Then you start sorting them into groups and then you create group headings. How does this actually work in terms of a real example? Here we have an example of delayed food delivery that I made up and here are different reasons for why that food delivery might be delayed. Your Y is, the outcome is the delayed food delivery. The X's maybe all of these different things or some kind of grouping of these different things. So this would be step 2 in terms of displaying the ideas. Step 1 would have been collecting the ideas and you didn't see that happening here. But step 2 is displaying the ideas on a wall or on a table. Now next, what I'd like you to do is take these very ideas that you see on this slide, and use them to categorize it. What I'd like you to do is think about categorization of these different reasons for delayed food delivery. Something that you should be able to relate to. Try to put them in categories. Think of 3,4,5 categories that you would put them in and see what you find and then I'll show you my categorization and we'll see how that matches up. You had a chance to think about this and put those different reasons into different categories. What I did was I put them in four different teams. One team was order taking, one was planning the resources, one was factors affecting drive, the environmental conditions affecting the drive, and one was driver capabilities. Those are the four areas in which I clustered the different reasons that you saw on the previous slide. Let's take a look at how I clustered them. Here are the four categorizations, the groups that I have formed. Then I gave them those four names that you saw earlier. You might recognize those four names simply by looking at this but here are the four names. The order taking was made up of the six reasons. Then you had planning made up of five reasons. You had environment made up of three and drivers made up of- A couple of points of caution when you're trying to do something like this. You are obviously going to or you are probably going to have some disagreements on what goes where. You may have some disagreements of what to call what category, and there may also be this temptation of creating a miscellaneous category where you put things that don't fit anywhere else. Similarly, there might also be the temptation of taking something and putting it in two categories. The last one is not that bad, I guess if you had to take something and put it in two categories, it's saying that there might be two ways of thinking about that particular cause for the effect that you're getting, so that might not entirely be a bad thing. These would be the four group headings that I would have used if I was doing this. Like I said, you may not have gotten the same, so there's no right answer here, but this is how you would start using the affinity diagram. Let's take a look at some other techniques that can be used for something similar, and these are quantitative techniques. The quantitative techniques that I point out here are cluster analysis; which is a way of taking data and trying to put it in groups based on similarities, based on different variables that might be assessed for that particular product or process that you're looking at. That's cluster analysis. Factor analysis is another quantitative technique, another statistics based technique. Then multidimensional scaling, which is used very commonly in marketing circles to figure out customer segments, and what are the different aspects of a particular product that customers like? If we can segment customers based on similarities or things that they prefer from a particular product. All the winners and all the qualifiers or the Kano characteristics that customers might prefer from a particular product. Those you can cluster them together on the basis of this technique of multidimensional scaling. Moving on to a different technique here, this is called the Pareto analysis. The Pareto analysis is something that get used in many different contexts, it's what is popularly known as the 80-20 rule. You may have heard of this as the 80-20 rule. The origins of this interestingly came from an Italian philosopher who simply made the observation that the 80 percent of the wealth was owned by 20 percent of the people. That 20 percent of the wealthiest people owned 80 percent of the wealth. That's where the 80-20 rule came from, and nowadays it's used in many different contexts. We think about it in terms of inventory and how we should be thinking about important inventory that we should be focusing on. Here you see an application of it that was popularized by Juran, the idea that 80 percent of problems can be traced to 20 percent causes. That's why we should be focusing on those 20 percent causes, on those trivial, on those vital few that are causing most of the effects. You shouldn't be worrying about the trivial ones, there may be the many trivial that you can actually ignore or not pay that much attention to in terms of causing the defect. What does the Pareto analysis look like? How is it done? You basically take the frequencies of different things, and you go with the frequency going from left to right, highest frequency to lowest frequency. What you have here in this chart is an example of defects that were found on a car dashboard. Again, data that I made up but it's simply giving you an example of saying that there is a high number of defects that came from loose panels, which is the first area, and those account for a high percentage. If you combine the first and the second, that gets you to that 77 percent number. If you take care of those two types of defects, what you're getting is 77 percent of the problems are being taken care of if you take care of those kinds of defects. Then you can move on and you see that there are many trivial ones. You can see the hundreds getting all clustered together up there because it's not even making a tiny difference to the cumulative percentage. Just to describe what you see in this Pareto chart, you have the bars that are describing the actual number of defects in each category, and then you have this line that's going from left to right which is the cumulative percentage. You're starting with 44 percent, you're getting to 77, you're going on and on until you get to 99 and then it stays at 100 percent, so it's basically 99.98. 99.99, so it's staying at 100 percent as you're going on and on from there. That's the idea of a pareto chart. How do you use this? Well, it's telling you where you should be focusing your attention first, and then maybe you can prioritize based on these what you should be focusing on. Next is the idea of a fishbone diagram. In some ways, this is very closely related to the idea of an affinity diagram. How is it related? Well, in an affinity diagram, what you're doing is you're taking many ideas and you are trying to group them into small groups. In the case of the fishbone diagram, what you're doing is you're taking a big idea and you're trying to break it down into smaller ones in some way. You start with an effect and you start looking at the causes. The reason it's called a fishbone diagram is it's shaped like a fishbone. You start on one end with the main effect, and you start looking at the main causes, and then you divide it up among the different sub-causes, the sub-bones, so to speak of that. Again, it can be done in both directions, I said you can start from the bigger ones and break it down into smaller causes, or you can go the opposite direction. You can say, we started off with the smaller ones, we started grouping them into these particular areas. Similar to an affinity diagram, this can be done using sticky notes, getting ideas from people, and starting to group them in some way. What you can use in terms of spurring some conversation, or if people are not talking in that fishbone diagram exercise when they're all together in a room, you can use some generic categories. The problem with generic categories is then people try to force things into those categories. If you can use generic categories without forcing the team to have certain things in those categories, these categories can be helpful in terms of thinking about where the problems might be coming from or thinking about areas that we should be considering. Here are some generic categories. In marketing, we popularly talk about the four Ps of marketing. Here we have the six Ms if we can call them that of process improvement of the fishbone diagram. You have machines, methods, materials, measurements, and because we wanted to force this into Ms, we're calling environment, mother nature, and because we wanted to force it into Ms, we are being politically incorrect and calling people as manpower. These are the six Ms that could be generic categories that you could be thinking about when you're trying to get people to think about fishbone diagram categories. The other one that you can have is policies, procedures, people, plant, and technology. If you're thinking more from a service perspective, these might be more applicable. In the service industry when we do a fishbone diagram, these are more readily applicable if you're thinking of generic categories. Finally, what you can do is you can think about each of the steps in the process as categories. Here's a defect here's an effect, and I want to think about the causes, you can have each of the steps in the process as being the main bones and then you can start thinking about the different sub-bones in there. Here's a made-up example of a technique, so a fishbone diagram that I put together for delay inpatient discharges. Here I'm using an example from health care. There's a delay in inpatient discharges that they're thinking about, why does this happen? There are four different big reasons. It could be the patient not being ready in some way, so you have three sub reasons there. Procedures, you could have missing information, insurance could be causing some delays, there might be some changes in billing toward the end. On the bottom half of this, you have doctors and registered nurses. It's something that they did with the procedure or did not do with the procedure for that to take place, that's why there was a delay. That's why the patient had to wait in order to get discharged. This would be how you would be thinking about a fishbone diagram. Next, let's take a look at a related technique that you can take from the fishbone diagram and go even deeper, and that's the idea of the 5-Why Analysis. The 5-Why Analysis, the one that I referred to as being that toddler, that four or five-year-old who keeps on asking the question, why? It's the same idea when you're thinking about it for process improvement. You can start off with the causes that you find from the fishbone diagram. You can use this as a follow-up to the fishbone diagram and start asking the question why. The important thing about using the 5-Why Analysis, and this is very popular with Toyota. They use this all the time when they're thinking about root cause analysis, when they're thinking about systematic going after root cause analysis. The point that is emphasized when you think about Toyota using the 5-Why Analysis, and these points are from Jeff Liker's book, The Toyota Way. The point that Toyota emphasizes is that we should follow the chain. We shouldn't be jumping to conclusions, that we should be slow in going towards the root cause and not jumping to conclusions. It's not only that you want to get to the root cause, but you want to make sure that you don't miss any cause that you might not have thought about. What Jeff Liker also talks about Toyota as doing with the 5-Why Analysis is branching off from within the 5-Why Analysis. Toyota thinks about this as there may be a causal chain of reasons that may branch off and go into different branches off multiple 5-Whys that you might be thinking about. Those are the things that you might want to think about when you use a 5-Why Analysis to systematically get to the root causes, to think about root causes that might be giving you the problems or maybe giving you opportunities for improvement. Here's an example of a 5-Why Analysis. Now, this is not made up. This is from Jeff Liker's book. It's an example of the below target production from a particular assembly line, from a particular task. If it is falling below target all the time, it means that they're not able to make enough. Why are they not able to make enough? Because they were losing production opportunities. Why were they losing production opportunities? Because there was a deficit within the cycle time. Why was the cycle time not enough? Why was there a deficit? Because the cycle time was used for other tasks. Why was the cycle time used for other tasks? Because loading machine was taking too long. Why was loading machines taking too long? Because the operator was walking a long distance. What does this get to? This gets to an actionable item, something that can be done in order to take care of this problem of below target units per hour. Like I said earlier, this could branch off into multiple reasonings. This might be one way of taking care of this particular problem, but there might be other ways that are branched off from any of the previous whys that you may have looked at and you may be thinking about going more in-depth into why those things are happening. Next, let's take a look at another technique called the cause effect matrix. Now you looked at the fishbone diagram as a way of looking at one effect and the many causes. The cause effect matrix gives you a matrix of many different effects and then looking at the causes simultaneously and thinking about it in that way. It's used to get some sense of what may be the different things that customers care about, and then go back to the idea of how do things that we do in the process affect what customers care about. You start off with getting critical to quality characteristics from customers, and you get weights for those critical to quality characteristics. Those weights add up to 1. This may sound familiar to you. We use something like this for project selection. It's something similar to that. The idea here is you get weights from customers for our critical to quality characteristics, and then you try to co-relate things that you're doing in the process. You try to find a relationship or you try to get information from people who know about the process, about the relationship between different things that are being done in the process with the different critical to quality characteristics. I think this can be best seen with an example so let's take a look at an example here. Here's a cause effect matrix for ceramic vases. What are our customers caring about in ceramic vases? They are caring about glaze, weight, whether it's sturdy or not, and whether it's precisely made. Those are the things that customers care about. You see the weights, those would have come from some kind of customer analysis, and those weights add up to 1, so we have 0.5, 0.21, and 0.2. Those weights adding up to one tell us what is relatively more important for the customer, and here you can clearly see that for the customer, the glaze is much more important for the customer. That's what they care about when they're looking at ceramic vases. They are the different features of the process that are on the top row and they're talking about what mix to be used in order to make those ceramic vases? What are the different kinds of materials that we've put in? How much water has been put in? What is the temperature at which the kernel was when these were fired? Then how was the cooling done? Whether it was rapid cooling, whether it was slow cooling, whether it was natural cooling, how was it done? Those are based on people who know about this process, experts on this process, we have a rating going from 0-9. Here's the key for that rating. Zero, meaning no relationship, to nine meaning a strong relationship between what we're doing. That can be a positive or a negative relationship when you're thinking about what is being done in the process, what are the different inputs to the process in terms of the mix of things and things like that and how is that affecting each of those critical to quality characteristics. Whether it's affecting it at a nine, which is a strong relationship, or whether it's a zero with no relationship, or it's a three, with some relationship. What you can do with this is, you can come up with a total weighted average for each of those features or each of those things that you're doing in the process and what this is telling us is that cooling has the highest effect on customer satisfaction. A cooling has an effect on glaze, cooling has an effect on sturdiness. Glaze happens to be very important, so 9 times 0.5 gives you 4.5 and then 9 times 0.1 gives you 0.9. So 4.5 plus 0.9 plus 3 times 0.2 gives you 0.6. That adds up to six. That becomes the most important thing to focus on when you want to get more customer satisfaction. Cooling would be the first thing that you should be paying attention to based on this cause effect matrix. Let's take a look at this technique, which is the last technique that we'll talk about in terms of speculating about what might be the relationship. As I said earlier, the scatter diagram is bordering on the idea of you're formulating hypothesis, but you're also starting to test the hypothesis. What is a scatter diagram? It's a basic X-Y kind of a chart. It has one aspect on the X and one aspect on the y. We usually like to think about Y as being the outcome and X as being the input for that outcome. Y as being the dependent variable and X as being the independent variable. But when you're doing a bivariate analysis, it doesn't really matter where you put it, what you're trying to look for or what you're looking for from a scatter diagram is, is there any relationship when we put this on an X-Y chart, doesn't reveal any pattern when we put something on an X-Y chart. Now, you can get the same information from simply looking at a bivariate correlation, or you can get at least some of the similar information. From a scatter diagram, you'll be looking at all different patterns from a bivariate correlation. You can look at the relationship between two variables from a correlation perspective, going from minus one to positive one, whether it's when one increases the other goes down, or whether it's when one increases, the other also increases. Let's take a look an example here to see what a scatter diagram looks like. Here you see the number of changeovers and then you see the overhead costs and what do you see from here? You can see some pattern here. As the number of changeovers goes higher, the overhead costs are getting higher and higher. Now it's not going to be perfect, but there's some pattern here in terms of the relationship between them. The advantage of the scatter diagram is it can show you a linear relationship, it can show a nonlinear relationship, it can show you an inverse U relationship, U relationship or any other kind of an S curve relationship simply by looking at the picture of the X-Y kind of a graph, simple graph over here. Finally, you can also talk about this from a correlation perspective that there's a 52 percent correlation between number of changeovers and overhead costs. These are all the techniques that you can use in order to hypothesize about cause effect relationship.