when you have for exemple 3 varieties of plant and each variety is submitted to 3 levels of NaCl concentration (2,4,6 g/l) and you like to see if a treatement can alleviate the effect of NaCl concentration. So you have 3 variables which are concentration, variety and treatement...you can use GLM and ANOVA test to see if the interaction is significant or no. Until this step it is easy ! But if i need to compare for exemple the effect of the treatement for (plant from variety 1 under 4 g/l of NaCl with) with plant from variety 2 .?? i means , interaction gave you in general the significance.. But how to compares means between the same variety or between two variety ? thanks for all answers .
I don't know in SPSS, but in SAS we can compare the means of the interactions through statistical tests like Tukey and Bonferroni. Depending of the number of comparisons, these kind of test is better than t test because this is not protected to ensure the level of significance to all comparisons but the level of significance is valid to each comparison, increasing the general chance of error.
In SPSS look for terms like lsmeans, tukey, t test, post hoc test, means comparison inside the glm module.
That should be part of the coefficients of your model.
The interaction between variety (categ. 3 levels) and NaCl (cat. 3 levels) is represented by (3-1)x(3-1) = 2 coefficients in the model. The meaning of the coefficients will depend on the coding you have chosen. For the most commonly used dummy-coding you can chose variety 1 as the reference level for variety and 4 g/l NaCl as reference level for NaCl; then the coefficient "Variety=2" would give you the difference vetween Var.1 and 2 at a NaCl conc of 4 g/l. If the reference for NaCl is 2 g/l then you would have to calculate the sum of this coefficient and the coefficients for "NaCl=4" and the interaction coef. "Var=2:NaCl=4".
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA