Just Stuff

Just another WordPress.com weblog

Unfitness Flows Between 1991 EHCS and the 1996 EHCS

leave a comment »

Background, the EHCS

In the mid 1990 I worked on the English House Condition Survey at the Building Research Establishment for the UK Department of the Environment and its successors. This was mainly on the physical disrepair but also the analysis of errors in the fitness assessment element of the surveys. The reports on these surveys can be found on the National Archives [1] . Additional material can be found in two conference papers that I published on this work [2][3].

The English House Condition Survey is a periodic survey of the condition of the English housing stock, in the 1990’s this comprised a large scale survey conducted every five years. The survey covered a number of aspects of the housing stock but those that I was particularly concerned with were disrepair and unfitness. The Physical Survey involved sending surveyors to examine the condition of a large-ish sample (ca 20,000) of dwellings selected from the English stock. From their observations a number of estimated repair costs were computed. In addition the surveyours reported on the fitness for human habitation of the dwelling for which they were specially trained to assess against the then current government standards.

Flows From Fitness to Unfitness and From Unfitness to Fitness

The thing that gets my ire is illustrated by figure 1 reproduces from the 1996 EHCS report. This shows that about 30% of the dwellings unfit in 1991 had been rendered fit for 1996 and about the same number of dwellings fit in 1991 had become unfit by 1996. Nothing too outrageous in that we might think. Until that is we realise that these flows between fitness state are derived from those dwellings common to both surveys and we know something about the reliability of the fitness assessments from the call back surveys. I don’t have the relevant data for 1996 to hand and if I did its analysis would be difficult because the over sampling of properties in poor condition in the 1996 callback survey. The data for 1991 is not complicated in this manner and is available in [2] and the relevant table is reproduced here as table 1. What we see in table 1 is that when a dwelling is found unfit on a survey there is something like a 30% chance that it will be found fit if resurveyed even if no work has been done on the dwelling. We may assume that a similar number of dwellings found fit will be found fit on resurvey. Such apparent flows which are just the signature of the fitness assessment uncertainties account for all the flows between unfit and fit states between the surveys.

Figure 1: Change in unfitness 1991-96 from [1]

Now I could extract the apparent flows in figure 1 (or from the tables in the 1996 EHCS report) and use the data from table 1 to estimate the underlying cross flows. Given that the apparent cross flows are comparable to what we would expect just from surveyour variability the results would be very noisy if not meaningless.

Table 1: Numbers of fit and unfit dwellings on initial survey and call-back for the 1991 EHCS from [2]

Why do I take exception to figure 1? It is because the problem of flows between fitness states were known at the time that figure was produced and its originator so informed. The author in question ignored the warnings about this because presumably what was presented told a better story. But a story pretending to be scientific and based on good statistics. It suggests to the present author that all derived statistics emanating from government departments are to be distrusted.

Other Difficulties with Customer

This is in some way typical with the problems I had while working on the EHCS. It always seemed a struggle to get acceptance of what the data was saying against the preconceptions of the DoE side of the EHCS team.

I first encountered this when the initial results for the 1991 EHCS indicated that the disrepair was down on the 1986 EHCS disrepair. It proved necessary to marshal other building statistics to show that the amount of work being done on housing repairs and maintenance had risen by a considerable margin over the period from the building industry activity statistics. I also sent an age with survey forms from 86 and 91 spread over my office floor while I counted the reported disrepair quantities from raw survey forms from the two surveys.

Another problem that I had was the censorship of [2]. Our DoE clients were not happy about my prediction that the 1996 EHCS would see a further small reduction in disrepair over the 1991 survey. The reason for this, if I recall correctly, was that they had already told the Treasury that the public sector disrepair was up and did not want the Treasury to see something that they might interpret as contradicting that. This was despite my prediction being for the stock as a whole, the majority of the English housing stock is under owner occupation and the overall disrepair in the stock is dominated by this sector. It is very difficult to tell from the 1996 EHCS report what the relative movements in disrepair withing particular tenure types was because the disrepair model had changed. The report states that after calibration the disrepair in 96 appeared to be between 0 and 5% down on 1991. This kind of comparison withing tenure types does not seem to have been possible. Attempting this now from the tables in the report seems to indicated that the disrepair in the local authority stock was probably up, but this may have been due to flows out of local authority ownership into other tenures of the best LA stock rather than a deterioration in the condition of specific dwellings (at least for dwellings not in high-rise blocks).

In general I always felt the the DoE part of the EHCS team were not very happy with a lot of the work that I did on the error characteristics of the surveys, nor for that matter with any stock modelling that involved differential equations. If I gave then or now the impression that I had or have a poor opinion of the EHCS this is wrong. I always have thought highly of it, but you can use the data more effectively if you understand its idiosyncrasies.

More to Do on This Post:

Some more needs to be said about the fitness flows including the actual numbers for the flows in figure 1 and something about the peculiarities of the data in table 1. Also something needs to be said about the problems in [2] (which are not in the copy on the RICS website but are in my own PDF which can be found here in red near the end)


[1] 1991 and 1996 EHCS reports http://www.ndad.nationalarchives.gov.uk/CRDA/51/DD/1/quickref.html

[2] R. Larham, “Error Characteristics of the English House Condition Survey”, CORBA95 conference of the RIBS, 1996.

[3] R. Larham, “Modelling the Evolution of the condition of a housing stock”, CORBA96 conference of the RIBS, 1997


Written by CaptainBlack

January 18, 2010 at 15:04

Posted in Maths and Stuff

Tagged with ,

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: