There are two test groups including formal and informal. In formal, we refer to statistical tests, and in informal, we mean graphs. When the sample size is large, the smallest departure from normality is indicated as p-value
The QQ-plot doesn't look normal. Particularly, two bottom points and two top points seem to be outliers. Fit the model without these data points and recheck the assumption.
Dear Mehdi, In my opinion this graph does not show normality in your data. I suggest using Johnson transformation family which is available in Minitab software.
Mehdi Azizmohammad Looha This data doesn't look normal because of those two points in the end and two on top. Dropping those values will make sure the data is normal.
Your plot shows two values that depart from normality on the low side, and one on the high side. In general the problems with departures from normality arise with *asymmetrical* deviations. So the plot you show, while not normal, does not ring alarm bells.
However, It appears that you looked at the normality of the response variable, which seems logical but it is not correct. The assumptions for the general linear model (including regression, ANOVA, and ANCOVA) are that the errors (residuals) are normal and homogeneous (Seber 1966, Neter et al 1983 pp 31& 49, Quinn and Keogh 2002 pp 110 & 280). Evaluation of assumptions in advanced texts, where it occurs, often entails a residual versus fit plot (for homogeneity) and a normal score plot or Quantile-Quantile plot (for normality of the residuals).
Hopefully your graph is for the residuals, but if not see: