What type of statistical analysis do I need to do for this?

[bluebirdc]

For one of our assignments, there is a hypothetical experiment in which the eggs of clams from different populations are attempted to be fertilized, and , in an attempt to see how many were viable and how many were not. Anyhow, we are supposed to do statistical analyisis and give the results, but I am unsure of what tests to run. Any direction would be appreciated! Thank you!

[Unity+]

For one of our assignments, there is a hypothetical experiment in which the eggs of clams from different populations are attempted to be fertilized, and , in an attempt to see how many were viable and how many were not. Anyhow, we are supposed to do statistical analyisis and give the results, but I am unsure of what tests to run. Any direction would be appreciated! Thank you!

Well, that would depend on what you are trying to look for. If you want to determine what the approximate population mean is for the amount that are fertilized then you can use confidence levels in order to make a prediction with a certain percentage.

 

Find a sample and test for what the experiment is looking for. Use the standard deviation equation and find the sample mean:

 

[math]s=\sqrt{\frac{\sum x^2-\frac{(\sum x)^{2}}{n}}{n-1}}[/math]

 

Once you find the standard deviation, find the t-score(I assume you know what those are). For example, let's say your sample size is 100 and you want a 90% confidence level then the degree of freedom(n-1) is 99 and the t score would be 1.290(calculated with http://www.danielsoper.com/statcalc3/calc.aspx?id=10 if you need to calculate it). Then, apply the values to the following equation:

 

[math]EBM=t_{\frac{\alpha }{2}}\frac{s}{\sqrt{n}}[/math]

 

This will be your error bound.

 

Then, you will want to apply this EBM to the confidence interval.

 

[math](\bar{x}-EBM,\bar{x}+EBM)[/math]

 

Then, once the high and low bound are found then apply that information to give a report. From the information you have given, I would say that is the best analysis you can use.

Edited April 28, 2014 by Unity+

[studiot]

Agreed the 't' test would be appropriate.

 

Unity, since you are doing so well, perhaps you would like to explain about samples and populations since the object of the exercise is to compare populations by comparing samples from different clam populations.

 

A question for bluebird

 

Do you think this should be a one tailed test or a two tailed test?

Edited April 28, 2014 by studiot

[Unity+]

Agreed the 't' test would be appropriate.

 

Unity, since you are doing so well, perhaps you would like to explain about samples and populations since the object of the exercise is to compare populations by comparing samples from different clam populations.

 

A question for bluebird

 

Do you think this should be a one tailed test or a two tailed test?

Well, if I know what you mean about comparing the two, I will explain the importance of using t scores with sample means and samples standard deviations.

 

The whole point is to obtain a value that is as close to the population mean of fertilized clam eggs as possible, t scores are used to be able to adjust for the statistical difference between the sample mean and deviation and the actual population mean and standard deviation. The larger the confidence level being used the wider the confidence interval will be because with only a sample we want to have some leeway for whether the actual population mean is within the confidence interval or not. Also, increasing the sample size will decrease the width of the confidence interval because the more that are sampled the closer the sample mean gets to the actual population mean.

 

I hope this is what you and the OP are looking for.