Tip:
Highlight text to annotate it
X
Erik: How do you balance statistical modeling and interviewing to better understand and
deconstruct relationships?
Ben: Well, I try to do both because each has their strengths. Interviews give you rich
insight. I mean it's the use of the network. It's stuff that you maybe couldn't have discovered
in any way. It's not public information. You know, as you and me have had conversations
about you building your network, I've learned so much that often didn't expect. You'll talk
about oh I went to this conference and I did this, or here's how I got to know this person,
and here's how I built that relationship over time, and so there's a real richness to interviews,
and there's the ability to find out the unexpected. At the same time, what statistical analysis
offers, quantitative empirical analysis, is the ability to look at large patterns and
draw pretty rigorous conclusions, if you've structured the data in the right way, you
know, effectively either an experiment or a quasi experiment between inputs or behaviors,
if you will, and outcomes, so you can have a higher degree of confidence in the appropriateness.
So interviews are great for unexpected insight. You know, maybe we know there's a problem
out there. We're not really quite sure, but we know there's some difference between how
successful entrepreneurs raise venture capital and how unsuccessful entrepreneurs are trying
to do it and not succeeding. You know, we know there's probably behaviors, but we're
unclear what those behaviors are -- that's where interviews are great. Alternatively,
once we have some insights, you know, if we can get the data, it's often times then great
you go to a larger sample and make sure that that's really not only there in the statistics,
but the other thing statistics lets you do is it lets you see relative effects, so you
can see how important is having a prior connection versus using a behavior, so you can contrast
the two, or you can look at the interaction, something that's very difficult to do. Well,
if we do these three behaviors together, what happens?