Hi Martine,

if I understand your setup properly, the correct approach is undoubtedly the two-way repeated measures ANOVA, not so much because the same material was used, but because each procedure is repeated three times. So I would build a model with the concentration of collagen as a dependent variable, the different procedure as a between-subjects independent variable, and the repetition in triplicate as a within-subjects independent variable, possibly including the interaction between these two. In this way, you can independently analyze whether the 8 procedures give different results on average, if repeating the measurement three times returns very different results, and if this occurs especially for a specific group (with the interaction ).

More information here: https://statistics.laerd.com/spss-tutorials/two-way-repeated-measures-anova-using-spss-statistics.php

By the way, if Mauchly's test of sphericity is not significant, it means that you have not violated the assumption of sphericity and that you do not have to adjust for the bias. You can read the following results under "Sphericity Assumed".

Good luck with your project.

As I understand you have a within x within design, with 8 within procedures and 3 within triplicate. If the 8 procedures are a between factor, this woud imply you have 8 times more different subjects.

Certainly, the Two-way Repeated measures ANOVA should work.

That would depend on your data and on the questions you are trying to answer. Data analysis is science, not plug-in. If you don't know how to use ANOVA, please consult a statistician. Be prepared to discuss your data and present graphs to show the patterns you are interested in.

see last chapter of

https://www.researchgate.net/publication/260772180_Developing_multilevel_models_for_analysing_contextuality_heterogeneity_and_change_using_MLwiN_Volume_2

Just to expand on Emanuele's answer where it states:

"By the way, if Mauchly's test of sphericity is not significant, it means that you have not violated the assumption of sphericity and that you do not have to adjust for the bias. You can read the following results under "Sphericity Assumed"."

This is a standard text book interpretation, but isn't always good advice. The Mauchley test lacks statistical power and is not very robust (and arguably addresses the wrong question). Rather than know if there is a violation we want to decide how severe any violation is. This can be done looking at the epsilon statistics. These are 1 if there is no violation and equal to the lower-bound (0.5 for 3 levels, I think) at worst. Generally good advice is to apply one of the corrections if the epsilon estimates fall much below 1 (certainly if lower than say 0.95). The GG correction is conservative and the HF correction liberal, so the GG correction is a safe choice, but HF is often recommended when epsilon values are close to 1 and the degree of violation is known to be small.

İf you have taken the measurements for each procedure in three different times, repeated measures ANOVA for mixed measurements can be considered. Procesures are between subject factors and measures are within subject factor. However, since ANOVA evaluates over the averages, you must take at least two values for each measurement. SPSS will give each measurement error to take the case of a single value. If you have taken 3 measurements at the same time (at the same time) for each procedure, you can use the F test (One-way analysis of variance).

Hi, you have an 8 × 3 design. 8 different procedures and repeated fo three times. for this design you need run two way ANOVA with repeated measures.