Anne Schulz May 4, 2018:

@Susan If what you now quoted is the "Methods" to the "Results" part presented in your query text, "Gesamtmittelwert" correspond to "Mittelwert über alle Dimensionen" (which does not sound like "mean of means" to me). For translation, anything should work (overall/global mean, grand mean, sample mean), as based on Mechthild's quote from the English Wikipedia, "grand mean" seems to be used not only in the sense of a mean of sub-means, but also in the sense of the overall mean. <br />NB: Calculating a Gesamtmittelwert in the sense of mean of means would make sense if one was interested in the variation between the individual dimensions' means. (For example, with a phobia assessment: "Does a group showing large differences between mean claustrophobia index and mean social phobia index have another user rating than a group with similar indices on all sub-scales?") This does not seem to be the case in your study.

Susan Welsh (asker) May 3, 2018:

correlations This passage from my text seems to address what Anne asked about: "Zur Bewertung wurde sowohl ein Mittelwert für jede Dimension als auch ein Mittelwert über alle Dimensionen hinweg berechnet. Darüber hinaus wurde die Korrelation zwischen dem Nutzerrating, den Mittelwerten der Subdimensionen und dem Mittelwert über alle Dimensionen berechnet."

MechthildOpp May 3, 2018:

In this set of 8 samples {1,1, 1, 1, 2, 2, 3, 5} the sample mean is 2.
Suppose you decompose this into disjoint subsamples, e.g. {1, 1, 1, 1, 3} with mean 1.4, and {2, 2, 5} with mean 3. You may now form the mean of means, which is the mean of 1.4 and 3, namely 2.2. Here you see that the mean of means (2.2) is different from the sample mean (2). If all subsamples had had the same size, one can prove that the mean of means and the sample mean have the same value. Since the mean of means is very dependent on the sample sizes, the sample mean is the more meaningful term. That is the "Gesamtmittelwert". You don't grasp this difference with the term "grand mean", "overall mean". But if you make it clear what the "sample" is that you are taking the mean over, e.g. all observations, you may say "the mean over all observations". Or, if you termed this large set as "sample, you may say "sample mean". Everybody knows what that is.

Susan Welsh (asker) May 3, 2018:

@Anne and Mechthild Yes, there are sub-dimensions. I'm going to have to read back through the article, now that you've helped me understand the concept.
Wikipedia writes: "The grand mean is the mean of the means of several subsamples, as long as the subsamples have the same number of data points.[1] For example, consider several lots, each containing several items. The items from each lot are sampled for a measure of some variable and the means of the measurements from each lot are computed. The mean of the measures from each lot constitutes the subsample mean. The mean of these subsample means is then the grand mean."

Anne Schulz May 3, 2018:

As I understand it, the 'grand mean' is a mean of means used for estimating variance between sub-groups. Clinical studies tend to be less sophisticated than that, and based on the quoted phrase only, 'Gesamtmittelwert' could just be the overall or global mean of all items as opposed to the means of sub-groups of items (sub-scales, dimensions). Is there any more context to show whether or not the variance between sub-scales is important for this study?

Elif Baykara Narbay May 2, 2018:

cumulative mean/average The first term that came to my mind was the "cumulative mean" or "cumulative average". I don't know much about statistics yet it may be worth to add these terms to your search.
I didn't hear of a grand mean, neither.