Thursday 31 March 2011

On why 15+ interviews and no match

According to "charting outcomes for the match" which is an NRMP publication. there is a "optimum number" of interviews that yields the most chance of matching. This is 8 to 12 institutions on the rank order list.

The percentage of applicants who match (IMGs - are called as independent applicants in the publication) increases with every single addition to the rank order list (ROL). The match rate is lowest for the ones with lower ROLs, however at the high end of the spectrum you will see that with 16-18 interviews +, there is a drop in percentage of applicants matching. 

They do not go on to address as to why this happens, but nonetheless it exists.

--------------------------

addendum:
After I wrote this article, there was some discussion on forums if this is for real. So I decided to do an in depth article which explores the same. The appended article is as follows


In order to further scrutinize the phenomenon of increasing ROL and diminishing chances of match. I decided to show subjectively the results.


Source of Data

Data was complied from charting outcomes 2011 publication from NRMP. Seven subjects that are most relevant to IMGs were chosen. The number of independent applicants (IMGs, USIMGs) to the number of contiguous ranks (rank order list) were obtained for Internal medicine, family medicine, psychiatry, pediatrics, neurology, anesthesia and emergency medicine.

Normalization of Data

Because the number of applicants in each specialty is disparate and comparison would have been difficult (e.g 275 unmatched psych applicants with 1 program on ROL as opposed to 1390 internal medicine applicants with 1 program on ROL) all the numbers were normalized using a min-max method. thus putting each speciality with in a 0 to 1 range. where zero is the minimum and 1 being the maximum. Although a word of caution here, i am going by the numbers provided by the NRMP, in their graphs they do not specify the units on the ordinate (y-axis), the abscissa (x-axis) is number of programs on the ROL.


Fig 1. Unmatched applicants and length of ROL

Groups 1

As seen in figure 1 where i have grouped anesthesia, emergency medicine and neurology, both EM and anesthesia have a terminal bump when the ROL approaches sixteen. Since the data does not extend beyond 16 it is difficult to speculate whether the trend will continue to increase or stay the same. Surprisingly, only neurology did not have the terminal bump.


Fig 2. Unmatched applicants and ROL length (group 2)
Group 2
Then next I compared the results from Internal medicine, family medicine, psychiatry and pediatrics (the four pillars of IMG residency -- read about it here) and I see the same trend. There is a terminal increase in the unmatched applicants beyond 15+ interviews. Again, since the data does not extend beyond 16 programs on ROL, it is difficult to speculate if it progressively increases or stays the same.

The degree of 'unmatchedness' is low but still significant to show up as a pattern in the specialties studied.


as for the reason why this is, it is still not clear. it could be hypothesized that either the applicants with long ROLs do not perform well on the interview, or they may disclose that they have 15+ interviews due to which they do not get ranked, perhaps there is an element of arrogance during interview process. Since it is not known, perhaps the readers can leave a few words in the comments section?


Addendum 2
-----------------
One of the readers of the blog (Dr. Vishakha D.) made some excellent points. I am going to add them here to highlight them. Some possibilities as to why this may be are:

How NRMP defines the term "Match Rate". It's the rate of matching into the first specialty ranked on the RoL.
 

1. One possibility is that people with longer RoL's have multiple specialties on them - possibly highly competitive ones like Derm on top and easier ones like IntMed at the bottom, making it likely that they rank into IntMed and thus count as a "Match failure" according to NRMP.
 

2. Second, even if the calculated RoL length was based on the number of contiguous ranks given to the specialty of first preference, we must remember that NRMP does allow us to rank programs which we haven't interviewed at! In fact I think 15-30% of candidates actually ranked programs that they hadn't interviewed at (NRMP Applicants' survey). The longer RoLs may represent people who did this.

3. Third, some of the ranks in these XLRoLs (extremely long RoLs :P, >16 ranks) may represent Preliminary year or Transitional year positions which automatically get grouped under Internal Medicine - likely having an unknown effect on the list.

4. Fourth,Unknown if  NRMP calculated the length of RoLs as the total of programs on both the primary and supplemental RoLs or just the primary. 


The entire comment can be read in the comments section below

    6 comments:

    1. Even though the article mentions "as for the reason why this is, it is still not clear. it could be hypothesized that either the applicants with long ROLs do not perform well on the interview, or they may disclose that they have 15+ interviews due to which they do not get ranked, perhaps there is an element of arrogance during interview process." I am very glad this is being discussed out here with multiple team members using collective intelligence to explain this phenomenon. A few posts had some exceedingly valid points again reiterating that that scores and accomplishments on paper may not translate to an impressive effective interview performance.

      Though it is less likely that there is a wrong way of placing the ROL (dictum: how you rank the programs decides WHERE you match - how the programs rank you decides IF you match at all), it is more likely that something with the applicant was amiss. Again one has to consider statistical errors like very small sample size (there are a miniscule number of applicants with 15+ interviews compared to most who are in the 5 to 12 range) which can amplify variations at the extreme high end of the spectrum.

      ReplyDelete
    2. Thanks for your informative post, now suppose someone was to get 15+ interview invites and keeping this information in mind decides to only interview at 10, does disreagrding the extra interviews negate the negative effects produced by the increased number of interviews?

      ReplyDelete
      Replies
      1. there is a big element of performing well/poorly in the interview, that may affect the outcome for those with many interviews. this is a trend and not a definite. IF you get more interviews my recommendation is to go for as many as you can and perform in an outstanding manner. Good luck!

        Delete
    3. I think this has more to do with how NRMP defines the term "Match Rate". It's the rate of matching into the first specialty ranked on the RoL.

      One possibility is that people with longer RoL's have multiple specialties on them - possibly highly competitive ones like Derm on top and easier ones like IntMed at the bottom, making it likely that they rank into IntMed and thus count as a "Match failure" according to NRMP.

      Second, even if the calculated RoL length was based on the number of contiguous ranks given to the specialty of first preference, we must remember that NRMP does allow us to rank programs which we haven't interviewed at! In fact I think 15-30% of candidates actually ranked programs that they hadn't interviewed at (NRMP Applicants' survey). The longer RoLs may represent people who did this. On a hilarious aside, this may also explain why that post-16 slump is not seen in Neurology - we brain nerds are good at brainwork and are very rational, we don't do silly or wishful things like that - at least not those of us who have 16 interviews! ;D

      Third, some of the ranks in these XLRoLs (extremely long RoLs :P, >16 ranks) may represent Preliminary year or Transitional year positions which automatically get grouped under Internal Medicine - I don't know what effect this may have.

      Fourth, I'd like to know if NRMP calculated the length of RoLs as the total of programs on both the primary and supplemental RoLs or just the primary.

      Those are a few possibilities.

      ReplyDelete
      Replies
      1. great pointers Dr. Vishakha! I am going to include these in the main body of the article above as an addendum with attribution to you. Are you applying this year? or are you in residency already?

        Delete
      2. Sure, no problem. I'm applying this year. Racking my brains over my RoL right now!

        Delete

    HOT Topics ! Popular by Demand