Suppose I have access to a population data set. I then perform Simple Random Sampling on this data set to pull a sample. I perform some tests to see if the population data set and the sample data set come from the same population. This results in verification that they did come from the same population.

Since Simple Random Sampling is performed and is successful, can I infer that there is no dependency between each action based on their order?

Thanks in advance.

  • $\begingroup$ What do you mean by "each action"? Your question appears to describe only one sampling operation. $\endgroup$ – whuber May 6 at 16:03

The "success" comes in when you are able to perform SRS without losing any observations because of non-participation, attrition, or other kinds of issues where you were not able to get the observations u wished to obtain.

If you actually succeeded in performing SRS, there is no sense in testing the difference between the population and sample. This test will always be rejected at the Type I error rate or alpha level, which is often 5%.

| cite | improve this answer | |
  • $\begingroup$ This is true. However, my question was "can I infer that there is no dependency between each observation based on their order? " $\endgroup$ – user2253546 Nov 25 '15 at 20:33
  • $\begingroup$ If you did SRS, then yes. $\endgroup$ – Hotaka Nov 25 '15 at 21:40
  • $\begingroup$ Is there a better way of testing if there is a dependency between subsequent observations rather than my hacked together method? $\endgroup$ – user2253546 Nov 27 '15 at 6:31
  • $\begingroup$ You mention no "hacked together method". If you are saying that you tried to do simple random sampling, but are not sure if you did, you'll have to be explicit about you did. But then you must change the title and description of your question. $\endgroup$ – Steve Samuels Nov 27 '15 at 21:10

Not the answer you're looking for? Browse other questions tagged or ask your own question.