I do agree with Sundara opinion because some method can affect the results especially this bird which cosmopolite around the world because of its distribution. So, what did the method you use?
Did you estimate the number of sparrows in a geographic area? If so, that sounds very much like human events where crowd sizes are estimated. Such estimates can vary wildly. I think the idea is to count the number in a given area and multiply by the ratio of a larger size area to the area counted, where you think that the distribution is fairly uniform. The problem is that it will not be approximately uniform over a very large area, so I suppose that you need to take "sample" areas to count that are relatively widespread. This is not really my area of expertise, but perhaps you could search the internet for methods of estimating crowd sizes. - The toughest part, I imagine, would be getting an idea as to how accurate your results might be.
I am interested in birds, and would like to hear anything that you discovered.
It has occurred to me that this sounds like what a survey agency, such as the US Bureau of the Census, would call a cluster sample, where here you do a census in each block. So actually, it could be done simply by just laying out your geographic region into a grid of equal size squares, and just doing a simple random sample of those blocks, where you just count the sparrows in each block of the sample. That way, the sample unit is a square of land, and you count the sparrows in each unit chosen. Then it is just a simple random sample where any survey sampling book and many online sites can give you estimators for totals or means with their accompanying estimates of variance.
So this becomes simple random sampling, where you can get a measure of accuracy for sampling error, but don't forget nonsampling error, such as measurement error while counting. You could research the concept of "total survey error."
Also, you could look into stratified random sampling. For example, in urban applications, you might stratify by buildings and open areas.
One resource would be Cochran, W.G.(1977), Sampling Techniques, 3rd ed., John Wiley & Sons. You may be particularly interested in chapters 2, 4, and 5 on simple random sampling, and stratified random sampling and estimation.
Another very well written, but expensive book, is Lohr, S.L.(2010), Sampling: Design and Analysis, 2nd ed., Brooks/Cole.
Jim
PS - You may want to investigate the term "area sampling."
PSS - My experience in continuous data does not always carry over well to discrete data. There are some differences, so please bear that in mind. (Jan 2015)