A few days ago I posted a list of features that I take to be essential to an ideal report on placement, seeking comments and suggestions. One of the features I mention there is recency. All departments are likely to place more candidates given more time, but this slope is steeper for certain departments. Moreover, placement varies year to year. Thus, one's choice of time frame can substantially alter data on placement. This is the reason that Brian Leiter's numbers for NYU look better than mine (here and here)--I looked at the years 2012 to 2014 (3 years in the recent past), whereas he looked at the years 2005 to 2010 (6 years in the distant past).* Looking at NYU's placement page, one can easily see that the percentage of graduates placed in tenure-track jobs drops as one reaches the present. As I said, this is likely true for all departments. This means that if you look at data in the distant past, it might not matter what the length of the time frame is, but if you look at data ending in the recent past, the length of time frame makes an impact. That is, for NYU for the years starting in 2005, a 6-year time frame has 87% TT placement, a 5-year time frame has 90% TT placement, a 4-year time frame has 88% TT placement, and a 3-year time frame has 90% TT placement. But for the years ending in 2013, a 6-year time frame has 69% TT placement, a 5-year time frame has 65% TT placement, a 4-year time frame has 56% TT placement, and a 3-year time frame has 56% TT placement. Note that even the 6-year window ending in 2013 is associated with much lower placement than any of the windows starting in 2005. It seems obvious to me that we should favor more recent data, since they reveal which departments place students more quickly than others and since they are more relevant to students looking at graduate programs. Beyond that, it is not obvious just what length of time we should choose (3, 4, 5, or 6 years) or just which year we should use as the endpoint.
Yet, one's choice of time frame has a large impact on comparative placement data. Let's compare NYU's placement page to the placement pages of those departments that I found with these methods to have the highest tenure-track placement rates: Berkeley, Princeton, Pittsburgh HPS, and UCLA. If we look at NYU's worst time frame it comes out behind all the others (2010-2013: NYU 56%, UCLA 59%, Berkeley 63%, Princeton 65%, and Pittsburgh HPS 88%). If we look at NYU's best time frame it comes out ahead of all the others (2006-2009: NYU 94%, UCLA 67%, Berkeley 78%, Princeton 86%, and Pittsburgh HPS 93%). If, on the other hand, we look at multiple time frames then a new type of comparison is possible. We can determine, for example, which department has the least low value for tenure-track placement, given any time frame in the period from 2005 to 2013 (with a 3-year minimum time frame and a 6-year maximum time frame). In that case, Pittsburgh HPS comes out on top. It's lowest value is 85%. In comparison, the lowest value for Princeton is 65% (2010-2013), the lowest value for Berkeley is 59% (2009-2012), the lowest value for UCLA is 52% (2009-2012), and the lowest value for NYU is 56% (2010-2013). So if we look at the least low placement for all of these time frames, NYU comes out second to last. Finally, if we look at the full range, from 2005 to 2013, NYU comes out in the middle (Pittsburgh HPS 93%, Princeton 76%, NYU 74%, Berkeley 70%, UCLA 65%).
Suffice it to say, these decisions make a substantial impact on one's results. For that reason, one should attend carefully to justifications on recency and time frame. I will remove the links to Brian Leiter's two posts on placement data here, since I am concerned that they will mislead students. If I had written those posts, I would certainly take them down knowing what I have made clear in this post (i.e. that the numbers for NYU are inflated for the very time frame that Brian Leiter chose to look at, relative to other departments). I have emailed Brian a link to this post.
As for my data, I use the years 2012 to 2014 because those are the most recent years and the years for which I have large data sets. (ProPhilosophy was kind enough to email departments directly in 2012 and 2013, which substantially increased the number of reported hires for those two years.) To go prior to 2012 I would have to either look at individual placement pages for all 118 departments, many of which do not have data of the sort I need, or use what I know to be a skewed sample from the Leiter Reports blog. I have made clear that any rankings I produce are a work in progress and should not be taken as authoritative. (That is one reason I post them to blogs, and not an independent website.) But as time goes on and this process is improved I will have to start making decisions about which time frames matter. I may well follow the lead of David Marshall Miller in reporting multiple time frames, since this might be helpful for students. Suggestions on this point are welcome. (The data that I used for this post are after the break. Feel free to suggest corrections where needed.)
*I hope that this does not need saying, but I am not picking on NYU here. One of my dissertation advisors was at NYU and one of my best friends is currently a student there. I am looking at NYU because it appears to be a focal point in Brian Leiter's criticism of my work. If one were to look at other measures beyond just tenure-track placement, NYU may well fare better than it does here.
Update (7/14/14): In order to satisfy the worry that NYU is particularly burdened by graduates of the JD/PhD program in this measure (2 graduates from NYU left academia for law in this time period, compared to 1 from Princeton, 3 from Berkeley, and perhaps 2 from UCLA), I compared NYU to these other programs while leaving out all those graduates who left academia. In that case, as I point out in the comment below, it is still clear that time frame matters and, in particular, that the time frame of 2005-2010 overly inflates NYU's record (2008-2013 puts NYU in the middle of the group, at 80%, whereas 2005-2010 puts it at 95%, square with Berkeley and Pittsburgh HPS, ahead of UCLA and Princeton. It might be worth noting that with the same methods Fordham University placed 69% of its graduates into tenure-track jobs between 2008 and 2013). See my comment below for details.