Print

---

Dear All,

 

As promised, a full report of my meeting with the Guardian.

 

Donald MacLeod was there, together with Ian Morton (Press Office) and Jovan Luzajic (Analyst) from UUK, plus Jonathan Waller (Head of Information Services) from HESA and Matt Hiely_Rayner (Researcher) from Campus Pi.  The meeting started off in quite a heated fashion as I elaborated on what I thought were the main defects of their method.  I'm afraid Matt bore the brunt of months of pent up frustration.  He then rather took the wind out of my sails by saying that unit spend statistics were calculated using HESA cost centre student numbers, rather than numbers obtained from JACS codes.  This was certainly not what I found when recreating a few of the King's scores.  In fact, I thought that their method had a knock-on effect resulting in ridiculous top scores, which in turn showed up severe shortcomings in the banding method.  It turned out that Matt was relatively new to Campus Pi and had not been involved in producing last year's figures.  The representative from HESA suggested that JACS numbers may have been used last year, but that he couldn't be certain, as they weren't actually asked to perform the unit spend calculations.  He suggested that they should do so for the forthcoming tables.  I left the situation regarding last year's tables with Donald as a query.

 

When I asked what were the JACS codes used for, Matt said (to be confirmed) entry scores, value added, destinations, aggregating subject scores to produce institutional scores, and NSS (new this year).  This is not such a serious issue with me - but I can well appreciate that it will be for others.

 

Also they refused to budge on missing JACS codes - although hinting that they would look at this in the future.

 

My reservations about banding stem from what I consider ridiculous top scores for unit spend - especially as they do not include academic salaries.  The consequence is that we have bands of huge width, with some empty, and most institutions clustered around a couple of bands in the middle.  It makes no sense at all that an outlier should score twice as much as the majority of institutions.  If an outlier was found to be suspect (for whatever reason) and then discarded, institutions that previously shared a common band would now find themselves spread across several adjacent ones.  Matt seemed sympathetic to this and said he favoured using standard deviations, much as the Times do.

 

I had gone into the meeting fully expecting to recommend to my Principal that King's withdrew from the tables for a year (whether he would have done so is another matter).  I came away thinking that progress of sorts had been made.  My main objective has always been to get the Guardian and Campus Pi around a table - which has been achieved with the setting up of the review group, albeit far too late.  It's very difficult in a one and a half hour meeting to get all the points across and debated thoroughly; so I can understand if people still have serious issues (particularly over the missing JACS codes).  You can count on my full support if you wish to keep hammering away, although time is obviously limited for this year.  Below are Donald's replies to my queries, which I believe demonstratess that we were right in pursuing our campaign and in refusing to be fobbed off by inadequate responses to our concerns.

 

With best wishes,

 

Brian

 

 

Dear Brian
You were right!
I do apologise for the delay in responding but the answers are as follows:

1.  The tables in which the JACS codes are used.

·       Value Added