Become a Creator today!Start creating today - Share your story with the world!
Start for free
00:00:00
00:00:01
Episode #60: Evan Sinar
129 Plays8 years ago
On this week’s episode of The PolicyViz Podcast, I’m pleased to welcome Evan Sinar to the show. Evan is the Chief Scientist and Vice President at Development Dimensions International (DDI) in Pittsburgh. Evan is a Industrial and Organizational Psychologist (I/O Psych)...
The post Episode #60: Evan Sinar appeared first on PolicyViz.
Recommended
Transcript
Introduction to JUMP's Software & Guest
00:00:00
Speaker
This episode of the PolicyViz podcast is brought to you by JUMP's Statistical Discovery Software from SAS. JUMP's powerful, easy-to-use visualization capabilities allow you to both explore your data for hidden insights and create interactive graphics that tell a compelling story. Enhance your presentations with dynamic graphics powered by world-class analytics in JUMP.
00:00:23
Speaker
Visit www.jmp.com to download a 30-day free trial to see for yourself how with JUMP, data visualization and exploratory analysis go hand in hand.
00:00:46
Speaker
Welcome back to the Policy Viz Podcast. I'm your host, John Schwabisch. I'm joined today on the show by Evan Sinar, a chief scientist and vice president at Development Dimensions International outside of Pittsburgh. Evan, welcome to the show. Great. Thanks so much, John, for inviting me.
00:01:02
Speaker
Well, thanks for coming on. You have quite an interesting background, a field that I'm not that familiar with. So I'm excited to learn more about what you do, both the DDI and some of the other things that you work on with big data and data visualization.
Evan Sinar's Role & Background
00:01:15
Speaker
But why don't we start, maybe you just give folks a little bit information about yourself and about the field that you work on.
00:01:22
Speaker
Absolutely. So, yeah, I'm in a role right now where I'm in a role as our chief scientist and vice president of a group called the Center for Analytics and Behavioral Research here at DDI. And the short version of that is really the hub for all things research at DDI. And that could range from research we're conducting with individual clients
00:01:42
Speaker
about the effectiveness, the program evaluation of the systems they have in place. We're a leadership consulting firm, so we're often asked to conduct analyses and help clients conduct their own analyses on how well those programs are working and how they can work better.
00:01:57
Speaker
So it's one facet of my group's work. We also get involved in some trend research studies where we gather data ourselves, either by compiling data from the simulations and assessments that we run when evaluating individuals for leadership roles, or we also conduct some large-scale survey projects. And so my group is involved with compiling those and ultimately analyzing and then representing and presenting those data back out.
00:02:23
Speaker
through some market facing research and then again paired with the client specific outputs that we're providing. And in terms of my background, and you alluded to this, my background's in industrial organizational psychology, and we can talk a bit more about that in a minute.
00:02:39
Speaker
In part, as part of my role, we're often asked to produce and conduct this research and then present it back out. So I've had the chance to do some authoring on this topic. So within the field of IO psychology, we recently, myself and some colleagues, we're editing a book on big data at work.
00:02:56
Speaker
and they had asked me about writing the data visualization chapter, which at the time, I hadn't really developed much depth in the area other than just some general interest. And after getting into it, I spent probably the better part of a year really digging into the topic and then turning that back around into this chapter. So the ability to really take a deeper dive and author some content has been an increasingly important part of my role for what I do here at DDI.
00:03:22
Speaker
Can you talk a little bit about what it means to do IO Psych? First off, it's an awesome abbreviation, but can you talk a little bit about what that is? I mean, you talked a little bit about what you do at DDI, but what is your sort of field in general? What does that mean to work in that field?
IO Psychology & Workplace Success
00:03:39
Speaker
Sure. So it's an applied facet of psychology, of course. So it's got all the psychology background that you would expect, but it's applied specifically towards the workplace. So I think the way I tend to think of it is it's science of people at work.
00:03:54
Speaker
And what we try to do is we help organizations build engaging safe and productive workplaces. And we try to help employees succeed. We work very closely with the leadership at organizations to give them the skills they need and the tools they need to build and sustain a positive work environment for employees because there's a long history of research showing that the more satisfied employees are, the more engaged they are, it leads to healthier and more successful organizations overall.
00:04:24
Speaker
A major focus is also on matching employees to companies. So what are the skills that employees are going to need for particular jobs and how can we help make sure that they have them as well as help them grow and develop those skills once they're on the job.
00:04:39
Speaker
So there's a real deep foundation of psychometrics, statistics, analysis as part of the field because there's a high degree of rigor that goes into these processes because if you're making decisions about people, this is a critical life decision for many people where they work, how they work, how they progress through these organizations. We as a field want to make sure that those decisions are made rigorously
00:05:04
Speaker
fairly with high degree of validity. So there's a strong research heritage there in the field.
00:05:10
Speaker
And I would assume that the field has been collecting data on characteristics of people and characteristics of employees and using statistical models and that sort of thing. How has the use of data changed in the field over the time that you've been working in the field? Have you seen data change to big data? Have the types of data that people are collecting and using and how they collected, has that changed? How is that field evolving with this newfound sort of appreciation, as it were, appreciation or value of data?
Impact of Big Data on Decision-Making
00:05:41
Speaker
Absolutely, yeah. You're absolutely right. So there's a long history of data and the gathering of data. I think what's really changed is on the audience side from the perspective of business decision makers. So for a while there, and I think it's still largely continuing, it seemed like just about every major business publication was noting the big data trend, Harvard Business Review and other publications that senior business leaders were reading and are reading. And so that really
00:06:09
Speaker
really drove an increased level of attention, awareness, and really I would say respect for data to drive business decisions. And so that gave us as a field an opportunity to connect the work that we've historically been doing to the other side of the equation, which of course is the demand and interest coming from the business leaders.
00:06:32
Speaker
I do think that the amount of data has been increasing and and this of course is a relative thing every every field I think has different thresholds for what they would consider big data and so we're working we as a field work with multinational corporations which of course have in some cases hundreds of thousands of employees and so that.
00:06:48
Speaker
That adds up pretty quickly to some large data sets, especially when you consider we're often looking across many different variables for those individuals. So the use of surveys to evaluate employee engagement, for example, or some of the new tools that are coming online, such as wearable devices that will help
00:07:08
Speaker
organizations understand where individuals are going, who they're interacting with. So there's other new data sources coming online that are continuing to contribute towards those these of big data, the volume, the variety, the velocity. I would also say that there's been a distinct trend towards data-driven decision-making, which also means that it's away from more of a gut instinct form of decision-making.
00:07:34
Speaker
And increasingly with our clients and I would say with the business community more broadly, that's rarely enough anymore. It's rarely enough just to have your experience drive a decision. There's typically some expectation and really demand at this point for evidence to back up the decisions. There's often high degree of financial or operational risk associated with these decisions and organizations are viewing data as a key input to that to help them make those decisions accurately and effectively.
00:08:03
Speaker
And have people in the field sort of change how they approach their analysis of data when the data are getting bigger? Or are the tools, I mean, what are the sort of standard tools that people in the IO psych field are using? Can they sort of handle the new wave of data? Or is it sort of a sea change in the tools that people are using the analysis they're conducting?
00:08:23
Speaker
Absolutely. Yeah. So I think a little of both, certainly. I think there are some foundational types of analyses, like there are forms of regression, for example, linear regression that have been in place for decades, of course. But now I think there's a push towards some of the more advanced regression models and more advanced statistical models that deal with
00:08:42
Speaker
a wider variety of data sets, data sets with a wider range of distributional characteristics, for example. So I think that that's still, I think the foundations are there with some of the standard analysis tools still being appropriate, but the recognition is increasing quite quickly towards some of these more advanced measures. And that's where I think the
00:09:06
Speaker
The bridge occurs between a field like IEL psychology and the broader data science community and really what we can learn from them, but also what they can learn from us. I think something that IEL psychologists bring that other fields are a bit newer in is some of the way of thinking about data coming from people and the fact that
00:09:26
Speaker
People need to know that data are being gathered and used appropriately. Sometimes there are legal issues associated with how decisions are made based on data. There are certainly ethical issues involved. And I see IO psychology is having a rich background there, whereas on the data science side, some of those processes are set up based on different types of data that may not
00:09:48
Speaker
necessarily be drawn directly from people. And so there's a nice opportunity intersection point there. And that's been a strong trend we've been seeing within our field. We even have a conference coming up later this year focused specifically on that topic. And that's where I think the overlap occurs and where we can pull in some of the tools from the data science community. But at the same time, we want to be able to share some of our knowledge about people needs to be handled appropriately.
Role of Data Visualization in Business Intelligence
00:10:14
Speaker
Let's shift gears a little bit. You've talked about sort of the new forms of data, but you also mentioned a couple times how important it is to communicate those findings and that analysis to different types of audiences. So I'm curious how data visualization comes into play in your field and how important it is and how people are embracing data visualization or maybe they're not embracing data visualization.
00:10:37
Speaker
Absolutely, yeah. We've certainly seen the thirst for an increase, and I think it relates to a lot of the earlier trends we talked about with just the sheer amount of data that organizations are looking to process and process quickly. So the pace of business and the pressure towards rapid decision making really hasn't changed, but at the same time, the amount of data has increased. And I think that's something that likely occurs across many fields. When that pressure point occurs, when those two trends collide, that's where the visuals
00:11:05
Speaker
opportunity comes in and why visualization is so valuable in our field as well as others with the, of course, the perceptual skills we can bring to how data are explored and ultimately explained to that broader audience. I think for IO psychologists, we often are in a very direct liaison or translator role between really science and business. So translating the complexity of the data into what business is going to find as
00:11:33
Speaker
intelligence that's going to drive their actions forward and so you've got the data perspective but then you've also got the business attention on making decisions as a result of that. What you also have in many cases is a very mixed audience in terms of technical depth. So certainly the rate of analysts and technically skilled individuals is growing within organizations but likely not at the same rate that data are being factored into decision making so
00:12:02
Speaker
The range of individuals who are making decisions based on data has expanded much more quickly than the pure quantitative analyst role.
00:12:11
Speaker
And so that's an opportunity, of course, to use visualization as that explanatory tool, translation tool to convey the complexity of data in a way that you can draw in a broader audience. And of course, that audience has a tremendously valuable perspective with what they know about business. They may not be quantitative analysts in the same way, but visualization is a way to connect data to them and them to the data and use that to guide decisions as a result. Right.
00:12:39
Speaker
I want to ask about the challenges that people in your field have with data visualization. I mean, it sounds like you have a similar challenge that lots of people face, which is a diverse audience, some with technical skills who sort of understand the concepts and other people who don't. But are there specific challenges that come with, in some ways, trying to match the employer and the employee together when it comes to visualizing data that may be sort of distinct or unique from other fields?
00:13:07
Speaker
Yeah, that's a great question. I think you're right. There's certainly quite a bit of overlap. I do think that particularly when you get into applications of data visualization that are designed to extend across an entire organization, and that's a wide range of jobs, a wide range of roles, a wide range of experiences, I do think that that audience tends to vary tremendously on their level of really visual literacy.
00:13:32
Speaker
And so the ability to provide data in a format visualize it in a way that connects with that broad audiences is certainly as critical in our field as others.
00:13:42
Speaker
You know, I think that the, you know, some of the challenges that again, it's, you know, it's hard to say if they're entirely unique, but I certainly feel them pretty, pretty profoundly in the work that I do. I think one example is of the, what I sometimes think of as the five second rule for visualization. So if someone doesn't get it in five seconds, then they think it's not doing the job. They really have to move on. And I think that, that can be a very, that can be a challenge when you're dealing with.
00:14:07
Speaker
complex data. So not every data set can be simplified so that someone can instantly understand it within just a few seconds. In some cases, the goal actually is within a limited amount of time to actually engage someone with the data, not just have them fully understand it. So that's the factor.
00:14:26
Speaker
I think the bias for action is there, certainly driven by business pressures. Again, I may not be entirely unique to business settings, but the action ability update, it can't just be interesting. It has to have the ability of driving some decision.
00:14:42
Speaker
One that I think might fall a little bit in the unique area potentially is, and this is I think something that even IO psychologists struggle with at times, is that we do have a deep research heritage. A lot of that is based on statistical significance testing for how we present and analyze data.
00:15:01
Speaker
and visualization in some ways pushes the statistical analyses into the background a bit and it can lead to something that I sometimes think of as optical significance. So a trend in the data that someone might see and want to take action on, but that doesn't have the corresponding statistical rigor behind it.
00:15:22
Speaker
And that's something that can be challenging for an IO psychologist to guide decision makers through and help them navigate through that topic. Just one other I'd mention is that many senior leaders and organizations have built their career on the strength of their gut instinct, their experience. And in many ways, the
00:15:42
Speaker
orientation towards data-driven decision making, whether it be visual or other types of analytics, that can be threatening for senior decision makers. Visualization, while it can be a communication tool, it also puts increased attention on using data to drive decisions, which isn't always compatible with the organizational culture and the skills of the current leadership in that organization. Right. That's really interesting, especially this sort of
00:16:07
Speaker
I like this optical significance because it's true. You show someone a bar chart and they think it all really matters, but maybe it's just statistically that they're not that different. So that's really interesting. When it comes to creating visualizations, and you work with a lot of people in obviously a DDI and in the field more generally with helping them do a better job with their visualizations, what are some of the biggest mistakes you think that people make, either in IO psych or more broadly?
Common Pitfalls in Data Visualization
00:16:33
Speaker
Sure. So I think the first one that comes to mind is that people are using tools to create visualization, whether it be a tool like Excel, for example, that they're using to create visualizations. They often are generating a visualization directly from the tool, and that becomes the final visualization. So they're using all the defaults of colors and formats and borders and all the other
00:16:58
Speaker
aspects that go into the data or even just through how they build the visualization. If they're not closely attentive to how those visual properties are actually affecting the message and the clarity of the message, I think the overuse of those defaults can lead to a very cluttered and ultimately much more confusing visual than it needs to be.
00:17:22
Speaker
I think what that leads to is you're using visual cues and again colors are great example but but shading whether you're are whether you're receiving some of the data into the background by using gray versus some selective color highlighting there's a tendency to really overuse visual cues for what really are trivial or not important properties for the data.
00:17:45
Speaker
Another one I would call out is the neglect for what I think of as the annotation layer that sits on top of the visualization. So it's rare that the visualization itself will tell the full story on its own. The analyst has done so much work to get to that point and they need to take some additional steps to add some annotations or in some cases it's as simple as the title or something that
00:18:09
Speaker
really makes it clear what someone should see in the data, what the analysis has brought to that point in adding those annotations to the data. So it's rarely the case that a visualization will be fully ready to share, just coming straight out of really any package someone is using to visualize.
00:18:26
Speaker
There's a couple other I would add. I think there's sometimes a neglect of the opportunity to apply visualization techniques to open-ended data. And in our field, we deal with organizations who are pulling in a large number of customer comments or employee comments, for example. So that open-ended form of data, I think, sometimes gets missed as an opportunity to apply visualization to how they make sense of and take action on the unstructured forms of data.
00:18:54
Speaker
And then I'd finally call out the fact that I think in many cases individuals will revert back to the standard forms of visualization, like pie chart for example, which of course has its own challenges, or even bar charts and column charts, which can be very appropriate in many cases. Bar charts and column charts in particular, they can present a very clean view of the data.
00:19:15
Speaker
But I think there's an opportunity for individuals to go a bit beyond the standard tools. And I've been real excited by some of the research coming out that shows that some of the more unique visualization types can ultimately be more memorable. And I think people don't always remember the mundane. They remember things that are a bit more unique. And they give individuals and organizations aren't taking that into account to try to learn about some of these new techniques and build them in and try them out, experiment with them. They may be missing an opportunity to draw on
00:19:44
Speaker
some of the more powerful and unique visualization types out there. Yeah, that's right. You hit on one of my big pet peeves, which is using the defaults in any tool. For some reason, defaults in most tools are just plain terrible. There are a few things in the field, in the data visualization field, that get people sort of riled up. Like there's the pie chart debate, there's the color blindness issue, there's dual access debates. And people tend to take strong lines on those. Are there any of those sorts of issues where you're like,
00:20:14
Speaker
You're militant or have really strong feelings like absolutely no pie charts ever. Are there any of those sorts of things that really get you heated up that really get you upset when people do that? Oh, absolutely.
00:20:28
Speaker
I do think pie charts, I feel like once you get into the data visualization field, it's kind of almost a rite of passage to really take a hard look at yourself and what you're really thinking about pie charts. Certainly, I always put that, when I talk about them, I easily put a parenthetical around it. If you must, there are some ways that you can use pie charts effectively, but generally, it's going to be the wrong choice much more often than the right choice.
00:20:55
Speaker
I think your point about color is a fabulous one because color, first of all, it's a powerful visual property that is dramatically overused. I think these rainbow color scales or even the red, yellow, green scales, which we see so often in dashboards and other business applications, those have many challenges to them. There's challenges with how you use a rainbow scale with continuous data, for example, versus categorical data.
00:21:22
Speaker
Or the other aspect that we do run into, because we work with multinational companies in many cases, is that color doesn't mean the same thing globally. So red is a great example where red is seen as the low category, the warning signs, the issues, the problems in a data set or in a dashboard. But in some cultures and countries, red is
00:21:42
Speaker
Is it is a sign of is a color of celebration of positive news i think it's easy for people to rely on some of those standard but not always accurate conventions for color.
00:21:55
Speaker
I think 3D is another one that I tried to get out of the visualization as quickly as I can and I think that along with pie charts, I think it's a feature of the data. It adds nothing to the clarity of the data and in fact, it often obscures the data because you're adding in
00:22:13
Speaker
a dimension of the data that doesn't contribute towards a higher accuracy. I talked earlier about one of the phrases that really will get me up on my soapbox pretty quickly is that visualization is about simplifying the complex.
00:22:30
Speaker
And that can be true, but we live in a complex world and not everything can be simplified. And I think of it as conveying the complex. And I think simplifying the complex as a phrase actually minimizes the potential of visualization to navigate an audience through the complexity. But you can't always simplify it because there is true depth to the data that you often don't want to strip away with something that's overly simplistic.
00:22:58
Speaker
I mentioned the annotation and the use or not use of the annotation. One other technique that we think about best practices is around the small multiples approach to visualization and that's often an antidote to someone trying to layer too much on just one graph in one visual versus
00:23:20
Speaker
a small multiples approach which will take more of a panel view to split that data out across geographic regions for example or across different variables. So rather than trying to pack them all onto the same graph to the point where it's really impossible to see the trends, to pull them apart really can add quite a bit of clarity to the message.
00:23:43
Speaker
Those are a few of my...
Tools for Beginners in Data Visualization
00:23:46
Speaker
I want to close up just by talking about tools for a moment. You've written a few things, a few posts on tools and resources that people should use or can use for their data and their data visualization needs. Are there particular tools or platforms that you think are especially useful, especially for people who might just be getting their feet wet with data visualization?
00:24:13
Speaker
Absolutely, and I think your last point is such an important one because the audience is so broad. What I target is something I think of as really guerilla data visualization. So what are some tools that you can pick up and get up and running quickly? So a very quick learning curve. Generally, I do try to orient people towards tools that are free or available already on their computers. And obviously, there's as many other layers of commercial tools that
00:24:40
Speaker
that sit on top of that. But frankly, I often find that I can do whatever I want with the tools that I either have or that are freely available. And I think that's a huge strength of the data visualization interest over the past few years, as many more of these coming online. So to get specific, one of the tools that I find absolutely fantastic and use it almost daily at times is a tool called Raw. So it's raw.densitydesign.org.
00:25:07
Speaker
And it's an extremely easy to use tool for generating a very wide range of visualization types. So when you're looking to experiment with some new visualization types that really bring clarity to complex data, I see that tool as really well suited to that. You can actually drop data directly in from
00:25:28
Speaker
From a table form into the interface and with a few clicks build some very engaging forms of visualizations you can actually export that export those files into a graphics editor and do additional annotation or color changing to it.
00:25:44
Speaker
I do think there's quite a bit of their tremendous benefits to excel excel i think can be one of the worst offenders i think we can stick with the defaults but it can be cleaned up in a way that that can be very powerful and actually can with the right considerations be be pretty valuable to folks and of course that's something that many individuals already have access to i think there's a there's a tool called color brewer.
00:26:07
Speaker
When we think about color and how the appropriate selection of color is so critical, Color Brewer is a phenomenal site for helping you pick color palettes that, first of all, allow you to represent data in an accurate way when you're looking across the color scale, but also to avoid issues with color blindness or making something print friendly.
00:26:27
Speaker
A couple that I would also call out when looking at text analysis, there's a tool called Voyant, which I find extremely useful for exploring text-based data. So if you're dealing with comments or any type of unstructured word-oriented data, that could be valuable. There's also another site called WordTree, which also allows you to look for key patterns in data. What types of words are often used together?
00:26:55
Speaker
So those are, I guess, my short list of sorts of tools that are my go-to sources, what I tend to be using. There's many others beyond that, but I do think that someone getting started in visualization doesn't have to come up with a huge list, nor do they have to spend an immense amount of time to at least experiment with tools like those.
00:27:14
Speaker
Yeah, I think that's right. I think you have that exactly right for the people who are the beginners or even waist deep or even neck deep. It all sort of depends on who you're trying to communicate with and maybe Excel or raw or some of these other tools you've mentioned are all that's needed and maybe you don't need to learn JavaScript, but people get excited about that. Everybody wants to create what the New York Times is doing these days.
00:27:38
Speaker
Well, Evan, this has been really interesting. I'm really interested in learning about this field IO psych. So thanks so much for coming on the show and sharing all this great information. Absolutely, John. I really appreciate it. This was fun to talk through these topics and certainly a great admirer of your work as well. So thanks so much for having me on the podcast. Ah, thanks so much. And thanks to everyone else for tuning in and listening to this week's show.
00:28:03
Speaker
please feel free to let me know if you have any comments or thoughts or suggestions about changes to show or other guests that you'd like to hear from. So please feel free to drop me a line on the website or on Twitter or via email. So until next time, this has been the Policy Viz Podcast. Thanks so much for listening.
00:28:32
Speaker
This episode of the PolicyViz podcast is brought to you by Jumps Statistical Discovery Software from SAS. Jumps powerful, easy-to-use visualization capabilities allow you to both explore your data for hidden insights and create interactive graphics that tell a compelling story.
00:28:49
Speaker
Enhance your presentations with dynamic graphics powered by world-class analytics in JMP. Visit www.jmp.com to download a 30-day free trial to see for yourself how with JMP data visualization and exploratory analysis go hand in hand.