What useful information is actually gained from line tests ? ?

[#1]

I do admit that this is both a mischief question and a rhetorical question  :  BUT.
I have an ongoing problem with fluctuating Noise Margins.: variations from  0 db to 17 db with a stable reading rarely lasting for more than a couple of hours.
Plusnet line tests have reported as OK but a maybe a bit slow. 
BT line tests reported as OK. 
The SFI Engineer at my home found no fault when he tested at the master socket. 
I persuaded the Engineer to open the DP and on inspection the joint was seen to be  Wet, Corroded and Green.  A test from the Exchange side of the joint showed over 3 Mb sync.
Since 2007 previous BT Speed tests have shown a typical IP profile of  2000 kbps with many actual throughputs around 1800 kbps and a low of 900 kbps. 
NB.  As of January this year I have also joined the club known as "the BT Speed Tester no longer works for me either".
The repaired joint has gone some way to improve the fluctuating Noise Margin problem.
However,  with the continuing problem,  I am guessing that other equally poor quality joints on other lines are still affecting my line.  The estate was cabled some 30 years ago with aluminium cable and the ongoing saga is being dealt with in Open Question No. 39315379
When all of the above tests were made not one of them located the rather basic problem of a corroded and impaired joint. 
Is this :- 
a).    A tester limitation problem ?  No direct connection to a local loop pair maybe.
b).    A testing procedure problem ? 
c).    A not all tests were actually done problem ?
Finally: my grateful thanks and acknowledgements to the RouterStats program for the evidence that it has provided.

[#2]

I suspect it was in part down to the intermittent nature of the issue, corroded joints aren't always affecting the connection in a lot of cases and if the line is working fine when tested then the issue won't be picked up. It may also be due to the fact that the line is quite long anyway, so a degree of interference is to be expected.
I'm not saying this was definitely the case, just offering possible reasons for it not being picked up. Either way I'll get your fault referred back for more investigation, and am sorry it's been ongoing for so long.

[#3]

Hello Matt,
I thank you very much for your for assistance and for your observations and, forgive me, but I would like to add one small quibble. 
From my end of the line I don't really see the probable line fault as intermittment.
What I believe to be intermittent is whether my neighbour's phones are in use or  "on hook". 
If there are any adjacent bad joints ( as I believe ) then any interference on my broadband frequencies will depend on the frequencies in use on the adjacent lines, i.e. DTMF, calling party signals, etc.  A poor quality joint is likely to act as a form of  diode that varies the condition of the line.  It is this type of condition that caused me to wonder as to whether line testing would or should identify this type of fault. 
Can the line testers be made to simulate on hook and off hook line conditions in terms of frequency and voltage / current ?  This sort of test could  work if adjacent lines were to be compared on a before and after basis.  The use of higher frequncies in the later forms of DSL appears to me to bring the problem of bad joints in the local loop more into question.  It is this line of thinking that caused me to ask the question.

[#4]

Quote from: MK3Chris

From my end of the line I don't really see the probable line fault as intermittment. What I believe to be intermittent is whether my neighbour's phones are in use or  "on hook".

It very much looks like your line IS unstable, it should not be affected as to whether your neighbour is on the phone or not.
Typically cross-line interaction is caused by "spilt pairs" so if you really believe that your neighbour is the cause of this interference then this is something that could be looked into.
You also mentioned bad joints, sometimes these can be picked up by doing a "quite line test" but certainly the BT engineers will have equipment that can do quite comprehensive electrical/electronic line tests

[#5]

Dodgy joints are sometimes improved by the extra voltage from the ringing of an incoming phone call. The engineer will usually ring you to say he is on his way, so when he arrives there's been a recent incoming call and he can't then find the problem!

[#6]

Quote from: jelv

The engineer will usually ring you to say he is on his way, so when he arrives there's been a recent incoming call and he can't then find the problem!

Make sure you only give a mobile number for the engineer to call 

[#7]

A thank you to all for the feed back and suggestions:  and in reply.
@Knowdice.  Good feedback and observations.  I agree that my line was unstable: bur NOT since the repair.  The cleaning, drying and correction of the bad joint has cured that problem.  RouterStats and the Router ( Netgear DG834 V4 with the Broadcom chipset ) show a steady 15 db noise margin for most of the time.  NB. The Router has previously indicated 8 db since Feb. 2007 when I got Broadband from Brightview (Madasafish); so I see 8 db as normal for my line..
The graphs also show the noise margin dropping down to 6 db, this takes between 1 and 2 minutes.  The NM then sits at 6 db for periods varying from 5 minutes up to an hour.  The NM then returns back to 15 db.: the graph return line being almost vertical.  This has been occurring between 3 and 7 times per day at differing times since the DP joint was repaired. There are other additional drops that only go down to 8, 10 or 12 db. 
Before the repair I had similar drops down to 0 db directly attributable to my own phone, but my phone now has no affect at all on the Router figures.  Today ?.  A steady 15 db all morning up to 14:10 then four "dips" so far this afternoon.  My speculation is still that one or more of my neighbour's phones are interacting with my line due to other poor quality joints in the DP, or an adjacent DP.  Hence my question about test information.
@Jelv.  Full agreement.  There was a dip in the RouterStats graph that exactly matched his call.
@Walker23.    Full agreement.  However:  I had asked for calls to be to my land line on this occasion.  I intended that any call would show as part of my evidence.
The later versions of Broadband now use higher frequencies than the original "Mark 1" and the local loop is now more of a transmission line than ever before.  Hence I am still left with the thought that poor quality joints greatly affect the line speed and the line should be specifically tested for this condition.  The fact that a local loop appears to show continuity is maybe not enough.

[#8]

The tests I believe would report line resistance and attenuation and if these are outside what is expected for the line then Openreach should suspect bad joints.