Answer: Choice B
We are 90% confident that the mean weight of all newborn babies is between 6.5 and 8.5 pounds.
Explanatiion:
The idea is that if we generate say 100 confidence intervals, then about 90 of them should capture the true population mean (mu) somewhere in them. Choice A is close to being true, but replace "sample" with "population" to make the statement fully correct.
We don't need a confidence interval to estimate a sample statistic. A confidence interval is used to estimate a population parameter. Choice C is a trick answer and a common misconception about confidence intervals. Choice D is a slight variation of choice A. These facts allow us to rule out choices A, C and D.