No Graham, you have missed what I intended. The data you have provided proves nothing and I gave some hints as to what is required. Strangely getting this data is very difficult, and may not exist. Are you suggesting that your data PROVE that AFDDs make a significant difference to fire statistics as implied by the manufacturers? If so I cannot really see it in your data. The problem with this data is that there is so much variation on a relatively small number that extracting anything much is impossible using statistical techniques. Analysis of variance (ANOVA) is unlikely to give significance, and that is why we need much more, much more specific data. Then a multiple regression analysis will tend to sort out the variation sources and give a useful statistical significance. This is always the problem with statistics, almost no one collects enough data, and yet they expect useful results.

See the document referenced by Graham. Page 9.

Lets look at the number of electrical fires in America. Overall the number is falling very slightly on the trend line but is so small that this is probably insignificant. It may be due to many things, but a slight improvement in the appliance stock is likely, as appliances now tend to take less power than earlier. Even if all of this were due to AFDDs, the cost/benefit is well over to the "not worthwhile" side, because at least 500 million AFDDs were sold over the period at $50 each, and this exceeds the total cost of all electrical fires by 2 times! ($2.5 billion /  $1.3billion). I suppose there might be a small gain per year, but the down trend is only about $10 million per year which is nothing.

I think that any numerate person can see that, even with this simplified statistical data, that AFDDs are not cost effective in the USA.

The present devices need an arc for a period of several cycles at a current of 3A or so, sparks do not cause trips and only a small number of fires are caused by arcs, most are due to overheating from other causes. That is my case.

No Graham, you have missed what I intended. The data you have provided proves nothing and I gave some hints as to what is required. Strangely getting this data is very difficult, and may not exist. Are you suggesting that your data PROVE that AFDDs make a significant difference to fire statistics as implied by the manufacturers? If so I cannot really see it in your data. The problem with this data is that there is so much variation on a relatively small number that extracting anything much is impossible using statistical techniques. Analysis of variance (ANOVA) is unlikely to give significance, and that is why we need much more, much more specific data. Then a multiple regression analysis will tend to sort out the variation sources and give a useful statistical significance. This is always the problem with statistics, almost no one collects enough data, and yet they expect useful results.

See the document referenced by Graham. Page 9.

Lets look at the number of electrical fires in America. Overall the number is falling very slightly on the trend line but is so small that this is probably insignificant. It may be due to many things, but a slight improvement in the appliance stock is likely, as appliances now tend to take less power than earlier. Even if all of this were due to AFDDs, the cost/benefit is well over to the "not worthwhile" side, because at least 500 million AFDDs were sold over the period at $50 each, and this exceeds the total cost of all electrical fires by 2 times! ($2.5 billion /  $1.3billion). I suppose there might be a small gain per year, but the down trend is only about $10 million per year which is nothing.

I think that any numerate person can see that, even with this simplified statistical data, that AFDDs are not cost effective in the USA.

The present devices need an arc for a period of several cycles at a current of 3A or so, sparks do not cause trips and only a small number of fires are caused by arcs, most are due to overheating from other causes. That is my case.