Any maths / statistics whizzes here?

[DimPrawn]

Is this a gut feel or your expert opinion?

My expert opinion is:

what is statistically significant number?
can they do the thing more than once?
is the 0.4 constant every week?
Does the 100% change as a sample?
When you need the answer?
Do I have to show my working out?
Can I change my answer later?
Where do I send the invoice to?

[Uncle Albert]

How many of the sample are contractors? I for one would have staff to do "the thing" on my behalf and on behalf of my company secretary and fellow shareholder.

[DimPrawn]

You're a bit thick huh? That's why I like you, easy to entertain.

Six Sigma - Wikipedia

[ladymuck]

My expert opinion is:

what is statistically significant number?
can they do the thing more than once?
is the 0.4 constant every week?
Does the 100% change as a sample?
When you need the answer?
Do I have to show my working out?
Can I change my answer later?
Where do I send the invoice to?

Rookie mistake putting invoicing last

[WTFH]

I am looking to see if I can extrapolate some numbers based on the information set I have.

If a sample is 100% and each week 0.4% "do a thing", how long would it take on average for all (or at least a statistically significant number) the people to do "the thing", I assume it's not as simple as 100 / 0.4 = X number of weeks

The answer, my friend, is blowing in the wind.

[NotAllThere]

I've subcontracted the question, to two of my sprogs and my dad. I don't do stats, because I have a degree in maths.

Not heard from my dad yet. And sprog one has declared that he studied physics to avoid this type of problem.

The other sprog (doing a degree in pharmaceutical engineering, where you really need to be good at stats - though I'm worried that she's enjoying working on this problem on a Friday night) has determined that the minimum number of weeks is 1/0.004 = 250 weeks. Trivial. Now we need to put in the figures for where we have duplication, and that involves a hypergeometric distribution.

So far it looks like it'll be considerably more than 250 weeks to get a 90% chance of the entire population being covered.

[DealorNoDeal]

I tweaked a program I'm working on, for random walks in asset prices, to test this out.

1000 people
Each week 0.4% (4 people), selected at random, do the thing.

Running 1000 trials...

After 250 weeks, 633 people have done the thing at least once
After 500 weeks, 865 people
After 1000 weeks, 982 people

To reach 1000 people, with a reasonable confidence level, I had to extend the number of weeks to 4000.

[ladymuck]

I tweaked a program I'm working on, for random walks in asset prices, to test this out.

1000 people
Each week 0.4% (4 people), selected at random, do the thing.

Running 1000 trials...

After 250 weeks, 633 people have done the thing at least once
After 500 weeks, 865 people
After 1000 weeks, 982 people

To reach 1000 people, with a reasonable confidence level, I had to extend the number of weeks to 4000.

In your model could some people do the thing more than once and others not at all?

Or does it assume that once done, the pool of people who need to do the thing is reduced?

[jamesbrown]

I am looking to see if I can extrapolate some numbers based on the information set I have.

If a sample is 100% and each week 0.4% "do a thing", how long would it take on average for all (or at least a statistically significant number) the people to do "the thing", I assume it's not as simple as 100 / 0.4 = X number of weeks

Well, it would take [1-0.004]^n = 0.5 for 50% to have done that thing or n = log(0.5) / log(0.996) or n = 172.9399 weeks. Not sure what "average" or "statistically significant" means in this context, but choose something other than 0.5 and solve again.

All this says is that the 0.4% of the people that did the thing last week and are no longer in the pool, so you're fishing from a smaller pond each week. If you mean something different, it will be different.

[DealorNoDeal]

In your model could some people do the thing more than once and others not at all?

Or does it assume that once done, the pool of people who need to do the thing is reduced?

I simulated a 1000-sided dice being rolled 4 times per week. So, any 4 numbers from 1-1000, could come up each week.

After 250 weeks, on average, 663 numbers will have come up. After 500 weeks, 865 numbers etc.

Of course, that may not have been what the OP meant.