Gradschool

Seeing networks clearly

/ shariqueorg / 1 Comment

A practical challenge people have when they enter a new firm, school or social organization is figuring out how things “work” there. Sometimes you might ask yourself “how do I get people on board with my new idea” or “who really makes the decisions here” or “who should I talk to in order to get the latest gossip”? Our instincts are to answer these questions by resorting to two types of information we can easily gather:

  • Who these people are (e.g., how long have these people been at the firm, their personalities, where they got their degrees, or how smart they are, etc.).
  • What these people are (e.g., their formal positions in the company’s organization chart).

Actually, our instincts are pretty good. People who have been around longer probably do have more interesting things to say or hold more power; people who have a certain personality type might indeed be more approachable, charismatic, or influential; a person who holds a fancy title should be able to get more things done; people who have higher IQs probably do have more knowledge.

But for every example of the above, we have all seen examples of the exact opposite. In fact, some might argue that we see more contradictory examples than we see confirmatory ones. This raises a puzzle: Why are individual traits only sometimes predictive of the things we care about?

Network analysts suggest that answering the who people are and what people are questions is insufficient for understanding power and influence. Answering these questions will give you only crude signals for what you are trying to learn about: power, influence or knowledge. The underlying causal force is actually people’s social networks: the pattern of social relationships that link people together in an organization or community.

Who’s got the power? 

Consider the following R&D unit in a large technology firm called Micron (name changed). Alexis was recently hired to help turn around this unit as Director of R&D. She is a recent graduate of a top MBA program and was previously a product manager for a breakthrough consumer app. Her references all praise her deep technical knowledge and her superb leadership and organizational skills.

Micron has been facing some big challenges in their R&D unit over the past few years. For instance, the R&D team has lost two Directors over the past three years. The unit has also been unable to produce any new breakthrough product ideas or technologies.

Micron is really counting on Alexis to figure out what the problem is and to fix it.

Today, the unit consists of three teams (Design, led by Waylon; Hardware, led by Justine; and Software, led by Chris.) The design team has three additional members (Case, Brogan and Rodrigo), the hardware team includes Frank, April and Kinley and the Software team consists of Leo, Eva, Sasha and Kim.  The formal organizational chart of Micron is depicted below.

Alexis is an astute student of organizational politics. She knows that fixing things in any organization starts by answering the following question: Who has got the power?

Screen Shot 2017-04-25 at 10.35.09 AM.png

Obviously, I haven’t given you the necessary information to help you answer this question yourselves. But a smart observer might come up with the following answers:

  • The person who has been at Micron the longest (it is Brogan, but he’s not the most powerful)
  • The boss (well, that is Alexis, and no he isn’t the most powerful)
  • The most friendly and outgoing (that is Sasha, no not her either).
  • Someone who has the most technical knowledge (Justine, but it is not her).

To answer this question we must begin by asking: What is power? 

Power is the ability to mobilize resources to get things done.

Breaking that answer down, powerful people are those who can get access to scarce resources faster than others, convince other people to act in a certain way or towards a certain goal, and see the potential political obstacles in the way and have the support but also foresight to avoid these obstacles.

The answer to the power question, according to network theory, is: people who are in central positions in social networks are powerful. They are privy to knowledge (through their social ties) that not everyone holds, they can influence others through their social relationships, and see better than others the political obstacles in their way.

Lets begin by looking at another graph. Below, I’ve plotted a network where the nodes (the ovals with names in them) represent the same people we saw in the organizational chart above. The nodes are connected to each other by edges which in this case represent mutual advice giving and receiving (but could represent other relationships such as trust or friendship). By looking at this graph, who do you now think is the most powerful person in this unit?

Screen Shot 2017-04-25 at 10.50.42 AM.png

The answer appears to be Eva. She is connected to more people and more of the groups than anyone else. Kim, for instance, is only connected to her little team, and Case, Brogan and Rodrigo are also minimally connected. Alexis, though she is the boss, has connections to the top of each team, but the information she gets is highly filtered. Not surprisingly, Alexis also goes to Eva for advice.

And who is Eva? She’s actually just another software engineer, but one that somehow has built a network that gives her insight and influence. The important thing to learn from Eva is not that Eva has some characteristic or trait (or even a combination of these), but rather that the network provides independent information that helps predict who has got power beyond observable traits. 

Lesson #1: The real way that work gets done in a company is represented by the informal network.

Network Positons

The above example is illustrative of how networks provide different information than the org chart or people’s observable characteristics. There is an entire science on interpreting and analyzing these network graphs (sometimes called sociograms).

Screen Shot 2017-04-25 at 2.24.39 PM.pngAbove is a stylized network, called the “Kite Network” developed by Professor David Krackhardt of Carnegie Mellon University.

Like the “real” network above, the kite network has nodes that are more powerful than others. Which node is the most powerful in the kite network?

Screen Shot 2017-04-25 at 2.27.11 PM.png

One possible answer is node D. The reason is that it has the most number of connections. Indeed, D is powerful. It has a type of centrality in the network called Popularity centrality or Degree centrality. If you want to get many people on board with an organizational change, or organize a party, D is your node. You can calculate degree centrality by merely counting the number of connections that a node has.

Screen Shot 2017-04-25 at 2.30.09 PM.png

Another answer is either F or G. The centrality of these nodes is a bit harder to see. They have what is called Farness centrality. If you count up the number of “hops” on the network it takes to get from one node (say, A) to all other nodes (B … to … J) and take the average, you get farness centrality. F and G have the lowest farness (or highest closeness) which means it takes a lot less time for information (or disease) to get from F and G to everyone else. Research has shown that Farness/Closeness is correlated to how fast ideas, knowledge, information spread out from a starting point.

Screen Shot 2017-04-25 at 2.42.06 PM

Finally, H has what we call Betweenness centrality. Betweenness measures the extent to which information must travel over a certain node in order to get somewhere else in the network. In other words, nodes high in betweenness are bridges that connect otherwise disconnected parts of the network.  There is a extremely large body of research showing that individuals who are high in betweenness have access to diverse information in their organizations and are often the source of creative ideas, have greater bargaining power, and experience superior career outcomes.

Lesson #2: Power accrues to those who are central in the informal network.

Now that you know some of these concepts, it might be useful to draw out your own network. Here is a short assignment you can do to understand your network better.

How well do managers perceive social networks?

All of the analysis above is premised on the notion that a manager can actually see the informal social networks of the people they are managing.

Research suggests that most people (including managers) do not accurately perceive the social networks in their companies. Indeed, there is growing evidence that people are even wrong about their own friendships—that they think think someone is friend, but that person does not feel the same way. However, research (e.g., Krackhardt 1990) also shows that there is substantial variation in which managers/employees have more (or less) accurate perceptions about a network. Managers with better perceptions are more powerful—i.e., better able to get things done.

Consider Silicon Systems (from Krackhardt (1992)) a technology company undergoing some organizational turbulence because of the not-so-friendly behavior of one of the managers, Ev. The situation at Silicon Systems had gotten so bad that the technicians working under Ev wanted to form a union to protect themselves from him.

Screen Shot 2017-04-25 at 3.45.52 PM

The story of Silicon Systems is interesting and you can find more details here. The short of it, however, is that a lot of the managerial problems at Silicon Systems were in part due to the poor perceptions of the social network—by the manager in question, but also by the top leadership.

Here is a visualization of the “actual” advice and friendship networks at Silicon Systems. They were constructed by asking people who they sought advice from and who they considered friends at the company.

Advice Network at Silicon Systems

Screen Shot 2017-04-25 at 3.55.46 PM.png

Friendship Network at Silicon Systems

Screen Shot 2017-04-25 at 3.55.53 PM.png

By looking at these two networks you notice two things. The networks are different, but also that in the Advice network, Ev, the trouble manager is much more central than in the friendship network. This is because he is smart and a high performer. However, the workers under him are not particularly happy.

A key player among the technicians working for Ev is Chris. Chris is a nice guy, he’s not particularly charismatic or among the best technically, but people do like him. Chris listens to what people have to say and people trust his judgment and advice. In fact, Chris’s opinion matters to them and the other technicians value his opinion when making their own decisions.

Now that you have seen the advice and friendship networks at Silicon systems. Let’s compare the perceptions of Chris (the informal leader) and Ev (the formal leader).

Chris’ perception of the social network at Silicon Systems

Screen Shot 2017-04-25 at 3.55.59 PM.png

Consider Chris’ network perception. Note how central Chris is in the friendship network. He’s the guy that most technicians trust (you can see that in the friendship and to some extent the advice network above). Though Chris isn’t 100% accurate in his perception, he’s got a decent understanding of the network at Silicon Systems. He is a bit more popular in his mind than he is in reality, though further away from his group of social connections he is a bit more clueless. But he does have some idea about how things work at Silicon Systems.

Now consider Ev, the problem manager. Below is Ev’s perception.

Ev’s Perception of the Networks at Silicon Systems

Screen Shot 2017-04-25 at 3.56.07 PM.png

Ev has no clue how things work at Silicon Systems. He has no idea that people are so tightly knit. This affects his ability to manage well and definitely weakens his power within the firm.

Ev is a problem. He’s smart, productive, and critical to the firm. But he is clueless. If you were Ev’s boss, what would you do?

Lesson #3: Power accrues to people who can accurately perceive the informal social networks in their firms.

Advertisement

PhD Student Stars

/ shariqueorg

After I defended my PhD dissertation in March of 2010, I decided to send my friend Chris (who is a star Informatics professor now) an e-mail summarizing what I had learned during that experience. As I read this email 7 years later, there is little I would change about the advice I would give to a new PhD student. Indeed, I give very similar advice to my own students, some of whom are now professors at great universities themselves.

Here is the text of the original email:

———-

So, now with a PhD (well, enough signatures to get me a PhD) in hand. I thought I should perhaps write some of my thoughts down about what I learned throughout the process. Primarily, I learned that “research” is much like any other job, perhaps even akin to making “widgets” in a factory. There is a process. Although I haven’t figured the entire process all out yet, particularly the publishing part, which is now going to be the primary interface between me and the production of widgets, I think I have come up with an outline for a theory.

Prelims

Before I started graduate school, even before I started my MS (I think), I read the website below. It gave me the best advice in terms of a general framework about how I should think about acting/behaving during my time in graduate school. I believe it was what helped me get admitted, finish, and find a job.

http://www.psychwww.com/careers/suprstar.htm

I would recommend any graduate student read it and take it to heart. When I started graduate school for my master’s degree, I tried to model myself after these suggestions. Though others might argue otherwise, I think, for the most part I worked an average of around 5-6 hours of real work per day, for at most 6 days a week – putting peak times aside. I mostly worked at school. I think most faculty knew my name and I personally asked almost all faculty to come to my presentations.

I expect to work significantly harder during my faculty job. Raising the average real hours worked a day to 7 or at most 8.

Some observations about “poorly” performing students:

  • The students who do the worst in graduate school are not present on campus and in the office on a regular basis. This conforms with the visibility hypothesis. Being on campus is important. First, you work. Second, you can talk with other students to resolve your problems. Talk to faculty and be a part of the intellectual life of the place. That means attending talks, giving talks. Even the “mindless” chitchat often contains important pieces of knowledge, gossip, tips and tricks, linkage into important networks that will provide guidance and encouragement during your PhD and beyond.
  • Students who perform poorly often reinvent the wheel. They do not take good advice from others – both explicit advice and what would I consider “implicit” advice (e.g. modeling yourself after the best of the cohorts above you.)
    • This includes writing papers. The structure of research papers is quite standard. This includes how to write introductions, results sections, etc. However, it also consists of due diligence on statistical procedures, etc. I have learned this through trial and error. But I often look at other good papers that try to do “similar things” (broadly defined) to see what types of other tests, etc. I should do before I wrap up my paper.

    • It also includes presentations. Particularly glaring is the absence of students at other people’s presentations. I am often surprised by this since academic output consists of two tangible products: papers and presentations. Just as writing good papers requires reading good papers, giving good presentations requires going to good presentations. And much like how writing good papers requires the ability to take and give productive criticism, so does presentation.

  • Read. I am often just in AWE of students’ lack of knowledge in their own field of study. I have encountered many students who are totally unaware of the basic – that is core – papers or ideas in their field. Not that I am the most well read person in the world or even the program, but I do work quite hard to keep abreast of recent literature (less so these days), the news, and the classics (putting a lot of time into this right now.) Reading and digesting the literature puts ideas, especially theoretical ideas, in context. Reading is important, as is remembering what you read. We all make mistakes. I might cite Smith’s 1975 paper, while it might be Smithe’s 1975 or 76 paper. But my “hunch” is that even when we make mistakses these are good heuristics for remembering papers, linking names to concepts (Granovetter -> Weak Ties) to era’s (1970’s) and linking these with each other into a “network” of sorts of concepts, authors, and eras. Knowing these basic things, will give individuals a good lay-of-the-land with respect to where the holes are in the research, where the interesting problems are, and where your own research can fit in. It also goes back to “re-creating the wheel.” A good knowledge of the current and past literature will give you, in addition to a better theoretical lens with which to view your research, ideas about data, about survey instruments, about methods, and about framing research as well. – A good quote about the importance of reading can be found here:

    “My first rule was given to me by TH White, author of The Sword in the Stone and other Arthurian fantasies and was: Read. Read everything you can lay hands on. I always advise people who want to write a fantasy or science fiction or romance to stop reading everything in those genres and start reading everything else from Bunyan to Byatt.” – Michael Moorcock

     

  • Do not wait for feedback to do work. I often notice students with a paralysis of sorts when it comes to doing anything that they have not gotten explicit directions from their advisors or approval from them for some reason. Keep playing with your data and your ideas. Feedback is slow, people are busy, and even when you do get feedback – remember that no one knows your data and the methods you used to analyze it better than you. Keep plugging away. I kind of have a heuristic about “regression analysis.” Once I get a “main effect” to be significant – I try (though, I increasingly notice that I often fail on some dimension) to do all I can to make it disappear (in theoretically justified ways of course). If I do get it to stay, then I am more confident. If it disappears, then you have to start searching for theory again (especially if you didn’t include the variable that made the effect disappear for a theoretically justified reason.)
  • Don’t listen to all the advice you get from your advisors. They are busy and they are human. Take all the comments in, make appropriate changes, and argue back when you have you. You will have to do it for the rest of your life with reviewers anyway. “Critiques” are not always correct.
  • Don’t TA too much. I see some students overload with TAs even in their 8th or 9th year (yes!). I think a manageable number of TA’s per semester is three if you got your research organized and you are in your 2nd or third year. If your research is a mess, keep it to 2 TA’s a semester. Here is a simple formula. Assuming that an average student can TA three classes per semester (not all unique) – that is 6 total classes a year earning $28,800 per annum without significantly extending their time in the PhD program. Now assume that any additional TA above this 6 TAs per year will increase the length of time you stay in the PhD program by 3 months (that’s just 1/4 of a year) and that your opportunity cost of staying in the PhD program is 80,000 (an above average salary for a master’s student). That decision to TA just that one extra course will cost you 20,000-4800 = 15,200. That is probably a low end of the estimate. Increasing the number of extra months that you might stay because of an extra TA by another month will increase this to over 20k lost. Bump up the salary… and you see the point. TAing, even if it just adds a “few” months will really hurt your pocket book. The next point is related more to time in the PhD program.
  • Little rules, big rules. Making sure you don’t break the little rules will help you make your deadlines on time. Finish your classes, your first paper, and your second paper ON TIME. That is almost like the first commandment of the PhD at Heinz. The little rules at Heinz are quite simple. These are the major milestones of the PhD here and are almost sacred. Doing this will provide you with enough structure in the formative periods of your PhD that will take you along through your proposal and defense. The more important thing coming out of finishing your FP and SP on time is that this will give you the “meta skills” to get you organized for your proposal and your dissertation. Finally, as a secondary note regarding the FP/SP deadlines is that there is an organizational memory. Everyone knows who didn’t finish their papers on time. Faculty have long memories as well. They are more lenient (with risky topics, etc.) when people make sure they obey the little rules. So, if you follow the little rules, you can break the big ones. Breaking the big rules is where the fun is.
  • Again, time. The academic job market penalizes “long” PhDs. This is a qualitative observation. Though there may be a handful of PhD’s who finished after 9 years who ended up with jobs in academia – the fact of the matter is that it is really looked down upon. Six year might be the peak of the neutral point at which it is OK not to have finished your PhD by this time, after that your prospects of landing a good academic job decline quite dramatically, and it snowballs to almost nil by 8th or 9th year.
  • Be nice to other students. Word spreads about “assholes” (this is a technical term – see Van Maanen 1978). We all have made faux pas’ in our lives. Probably tons of them. But consistent “assholeary” is bad. Be trustworthy and others will trust you and even let others know they trust you too.
  • Everybody here is pretty smart. It is not just you. It is hard work that creates the gradient on which good graduate students vary. Hard work is demonstrated by being on schedule, writing, reading, and working every week day for at least a few hours on your research (on average.) I am often surprised at how easy this is, and how some people just do not get it.
  • Know when to quit. Get real advice. Don’t stick it out longer than you have to because of your ego.