about sample size

1

Thanks for answering my previous question!

So “a 1-in-100 random sample of the population means that the sample is approximately 1% of the size of the total population.” By “sample”, do you mean each data point within the dataset, or the data as a whole? My understanding is that, for instance, if I select “ACS” 2015 for my data, then the whole dataset is considered to be “a sample”, right? (that represents 1% of the size of the total population in 2015?) Also, if I then choose “ACS” samples for 2014 and 2015, will the combined dataset overall represent 2% of the population?

Sorry for the inconvenience, and thank you for your help.

This shows how large the sample is relative to the total size of the population. For most recent US census years, a 1-in-100 random sample of the population means that the sample is approximately 1% of the size of the total population. Additionally, although this varies based on demographics, in a 1-in-100 random sample each person - on average - represents 100 people in the total sample. Practically, what this means is that you’ll need to use the sample weights (PERWT or HHWT) if you wish to calculate nationally representative statistics.

2

Yes, a sample is the entire dataset available for a given year. So, in 2015 the ACS is a sample of the entire US population, rather than a full survey of every single person within the US population in 2015.

If you choose both 2014 and 2015 ACS samples then you don’t have a 2% sample of the population. Rather you have a 1% sample in 2014 and a 1% sample in 2015. It is important to keep in mind that these samples are drawn at different points in time and are independent from each other. Each sample, in IPUMS USA, represents a nationally representative cross section of the US population at a particular survey year.

I hope this is helpful.