Can you guys help a mathlet out.

Statistics question regarding datasets. In studies a larger data set is better for obvious reasons.

But say we have a dataset which is recorded with a known level of inaccuracy, say 40% flat across the board. 40% of our data is inaccurate.
Is there any benefit to increasing the size of the data set given this inaccuracy to reduce the level of uncertainty in the data or is the uncertainty because of this inaccuracy flat regardless of the number of data points.

I think its flat, 40% of your data is untrustworthy or invalid no matter if its 5000 data points or 5 million. But I've been told I'm wrong.

Thanks.