I 'm having difficulties understanding when we can say something about just the sample, or about the population as a whole as well. In almost all previous exams, there's a question in the form of 'according to the output, what can we say about the effect of ... on the sample and/or on the population'.
I'm inclined to think that we can always only say something about the sample, given that our experiment is just a random sample of a much bigger population, but according to the answer models this is not the case.
Could someone help me out in distinguishing these two? E.g. which data in the SPSS output belongs to what, or when will the effect of sample/population not be the same?
Basically, you can ONLY answer questions about the Population based on formal tests (t-tests, F-test etc.). So for example let's assume you have a one-way ANOVA and want to test the effect of 3 treatment groups. Treatment group 1 has a mean of 130, group 2 a mean of 140 and group 3 a mean of 160. The ANOVA then tests if the mean difference that you observe in the SAMPLE, is also the case in the POPULATION!
So whenever a questions asks about the SAMPLE -> Look at the results of your actual data (Descriptives, Plots etc.) in the example above, we would answer that there is clearly a mean difference in the Sample (-10 between G1 & G2, -20 between G2&G3, -30 between G1&G3)
Whenever asked about POPULATION -> Look at significance of tests. Let's say ANOVA yielded the p-value 0.3 - we'd say that there's no mean difference at population level, as there was no significant result
I don't really get the idea of interaction effects.. "the two profiles are roughly parallel so there is no clear interaction bewteen location and yearthese two parameters"... but I find the profiles look pretty parallel??
parallel = no interaction (the graph in the picure is just slightly non-parallel, so you can still say that there is no (significant) interaction)
for non parallel graphs the Field book says:
- Non-parallel lines on an interaction graph show up significant interactions. However, this doesn’t
mean that non-parallel lines always reflect significant interaction effects: it depends on how nonparallel
the lines are.
- If the lines on an interaction graph cross then obviously they are not parallel and this can be a dead
give-away that you have a possible significant interaction. However, if the lines of the interaction
graph cross it isn’t always the case that the interaction is significant.