1. You would like to estimate bacterial levels (specifically Escherichia coli) for a large shallow pond. You collect 1-L water samples from 5 random locations in the pond. You measure E. coli levels...

Extracted text: 1. You would like to estimate bacterial levels (specifically Escherichia coli) for a large shallow pond. You collect 1-L water samples from 5 random locations in the pond. You measure E. coli levels using 10-mL subsamples (three per 1-L water sample). (a) This sampling design would be an example of what type of sampling strategy? (b) Table 1 shows the data that you collected. Using equations appropriate to your sampling strategy, calculate the mean E. coli concentration for the pond and estimate the standard error. The estimated volume of the pond is 10,000 m³ or 1×10' L, such that the total number of possible sample units is N= pond volume / sample volume = 1×107 L /1 L = 1×107. (Note, that fi = n/N is very small!) Table 1. Measured E. coli levels (in CFU/100 mL) in sample subsamples. Also indicated are the sample means (y;) and sample sum-of-squares (SSy; i.e. E(yy – y.)³,. Recalculate the (c) mean, treating each subsample as a sample (i.e. Eyi /15, where 2yi is the sum of all measured values). How does this mean compare Subsamples Sy 2,400 Sample 1 2 3 Yi 1 20 80 80 60.0 50 270 50 123.3 32,300 3 120 170 96.7 15,300 4 70 23.3 3,270 5 110 40 30 60.0 3,800 to the mean you calculated in (b)? (d) Assume one of the subsamples was lost (so unequal number subsamples across the samples) and you recalculated the mean as in (b) and (c). How might you expect the two mean calculations to compare? Which version should be a better estimate of the true population mean? Why? [hint: try removing subsample 1 of sample 4]