Timeline for Can I use Z-score with skewed sample but normal distributed population?
Current License: CC BY-SA 4.0
16 events
when toggle format | what | by | license | comment | |
---|---|---|---|---|---|
Jul 12, 2023 at 18:27 | history | became hot network question | |||
Jul 12, 2023 at 14:33 | comment | added | Dave | Welcome to Cross Validated! Could you please say why your samples are skewed when the populations are not? | |
Jul 12, 2023 at 14:30 | history | edited | kjetil b halvorsen♦ | CC BY-SA 4.0 |
added 324 characters in body
|
Jul 12, 2023 at 11:48 | answer | added | Stephan Kolassa | timeline score: 6 | |
Jul 12, 2023 at 11:35 | comment | added | Peter Flom | OK, I get you want to look at differences due to medication. But what about medians rather than means? Also, do you want to account for sex? If so, you would need some kind of regression (but there's a danger of overfitting). | |
Jul 12, 2023 at 11:32 | comment | added | Kimberley | Thanks for your answer @PeterFlom We would like to compare the means of birth weight to see if there are differences due to medication use (thats the idea behind the analyses). And the outlier is correct, its was one big baby :) almost five kg | |
Jul 12, 2023 at 11:30 | comment | added | Frank Harrell | $Z$ scores are only for normally distributed data, and they use standard deviations which are largely for symmetric distributions (of which normal distributions are examples). They needlessly complicate the problem and I seldom recommend $Z$ scores for any purpose. | |
Jul 12, 2023 at 11:28 | comment | added | Peter Flom | I think there is a preliminary question, which a lot of people skip. What, exactly, do you want to compare? Are you sure it's means? Also, for the baby boys, you have one big outlier. Have you checked if his weight is accurately entered? (Data entry errors happen a LOT more than people think they do). | |
Jul 12, 2023 at 11:09 | comment | added | Kimberley | The 6 data points are only the boys. Plot for girls: imgur.com/0ouJeQN I also added the raw data in my main post! And sample isnt even 24, its 6 boys + 8 girls | |
Jul 12, 2023 at 11:08 | history | edited | Kimberley | CC BY-SA 4.0 |
added 300 characters in body
|
Jul 12, 2023 at 11:01 | comment | added | Stephan Kolassa | Ah, so you also have a gender variable, that makes it a bit more problematic.... That plot only shows six data points, are these just the boys or just the girls? | |
Jul 12, 2023 at 10:59 | comment | added | Kimberley | Thanks for your reply! The genders combined and women separately have a normal distribution, for men it does not. See imgur.com/1n0VmxD for the Q-Q plot | |
Jul 12, 2023 at 10:59 | comment | added | Stephan Kolassa | If you only have 24 data points, you could edit your post to include the raw data. We might be able to help you better with that. | |
Jul 12, 2023 at 10:46 | comment | added | utobi | In my experience, weight data pretty nicely fit to the normal distribution. Can you post a quantile-quantile plot o your sample? I think you should be using a t-test and not a z-test, since the population variance is, I guess, unknown. | |
S Jul 12, 2023 at 10:26 | review | First questions | |||
Jul 12, 2023 at 10:47 | |||||
S Jul 12, 2023 at 10:26 | history | asked | Kimberley | CC BY-SA 4.0 |