topic Sampling groups where variances are similar in SAS Programming

Sampling groups where variances are similar

dustychair — Thu, 03 Sep 2020 20:37:30 GMT

Hi all,

I have a question. I have a data set like below. This is just an example what I have but I have a large sample. The first group has 150,000 people and the second group has 15,000 people. I tested the variance of score of two groups with Levene test and found out that the variance of scores of two groups are different. I would like to select two samples of groups where variances of scores are similar. Also, with that sample I would like to keep the sample ratios similar. What I mean by that for example the first group will have 12000 and the other one have 1200, the sample size is different than the original but the ratio of samples are similar. The point is that I have similar sample size ratio. I have no idea how can I do that. Any help is appreciated.

student_id group score

110 1 430

111 1 622

112 2 530

113 2 532

114 1 410

Re: Sampling groups where variances are similar

PaigeMiller — Thu, 03 Sep 2020 20:59:36 GMT

I have a question and I have a comment.

The question is: what statistic(s) do you want to compute from this sample?

My comment is with regards to your desired to have a sample of 12000 from one group and 1200 from the other group. I don't know why you insist on this, but depending on your answer to my question above, samples of 12000 are rarely worth the effort from a statistical point of view, the extra information you get is not linear with sample size, there is a point of diminishing returns where you get little benefit of (for example) going beyond a sample size of 2000.

Re: Sampling groups where variances are similar

dustychair — Thu, 03 Sep 2020 21:04:42 GMT

Hi @PaigeMiller,

I am going to do DIF (Differential Item Function) analysis after I have two groups who have similar score variances. Sample size is not big deal, It can be 2000 but it shouldn't be less than 1000 since some DIF methods require large sample size. My point is the ratio of two samples should be similar to original one.

Thank you

Re: Sampling groups where variances are similar

PaigeMiller — Thu, 03 Sep 2020 21:12:17 GMT

Okay, thank you. I am not familiar with DIF so I can't comment further.

Re: Sampling groups where variances are similar

Ksharp — Fri, 04 Sep 2020 12:03:24 GMT

Check this .@Rick_SAS wrote a blog about it . But you need SAS/QC .

https://blogs.sas.com/content/iml/2017/05/01/split-data-groups-mean-variance.html