I use qqnorm to plot my data as the photo attached. Does this plot indicate the data is normal distributed?
$\begingroup$
$\endgroup$
6
-
3$\begingroup$ This question may be better suited to the Cross-Validated stack exchange. However, I believe a Shapiro-Wilk test can give you an actual value for determining normality. $\endgroup$Kai– Kai2022-10-02 00:30:30 +00:00Commented Oct 2, 2022 at 0:30
-
5$\begingroup$ In a word, No, that doesn't look normally distributed. However, people often think they need normality when they don't at all. $\endgroup$Gregor Thomas– Gregor Thomas2022-10-02 00:49:15 +00:00Commented Oct 2, 2022 at 0:49
-
1$\begingroup$ It looks like a right-skewed distribution with discrete support $\endgroup$Henry– Henry2022-10-02 02:38:08 +00:00Commented Oct 2, 2022 at 2:38
-
1$\begingroup$ It would be fair to say this distribution is "approximately" Normal, for whatever such a vague statement might be worth. If your concern is whether it has the principal characteristics of any random sample of a Normal distribution, the answer is definitely not: there are too many collections of tied values, as evidenced by the strings of adjacent horizontal points. $\endgroup$whuber– whuber ♦2022-10-02 15:35:37 +00:00Commented Oct 2, 2022 at 15:35
-
$\begingroup$ What's the data and the model? Often tied residuals are the result of fitting a model with categorical predictors only to count data. And in that case approximately Normal is the best that you can expect to get. $\endgroup$dipetkov– dipetkov2022-10-02 16:17:18 +00:00Commented Oct 2, 2022 at 16:17
|
Show 1 more comment
1 Answer
$\begingroup$
$\endgroup$
1
On this plot, the sample doesn't look normally distributed at all. But for the normal distribution, there are formal statistical tests.
-
1$\begingroup$ A few words of explanation would be welcome: exactly what features of this plot cause you to conclude the data do not look normally distributed? $\endgroup$2022-10-02 15:36:08 +00:00Commented Oct 2, 2022 at 15:36