As far as I am aware, the FFMQ should only be used to measure facets, not a single composite score. I think the 'observe' subscale in particular does not correlate as well as other subscales.

I would agree with Karen on this one, particularly vis-a-vis the fact that observe subscale scores tend to predict outcomes quite differently from some of the others. A copy of a recent paper in which the predictive utility of aggregates versus the individual components in physical health is attached. Be interested to hear a bit more about your research and/or what you find.

I think there is nothing wrong with using the overall FFMQ score as a mindfulness score. It's true that the observe sub-scale tends to have weaker correlations with some other constructs, but why is this a problem? The FFMQ seems to have good intercultural validity and performs very well in all studies I know it was used. That it has 5 sub-scales is an advantage in my opinion, because you get more information. Why should an overall score which is composed of more than one factor be less valid or usable than the score on a one-factor mindfulness instrument?

I think mindfulness is a very complex, not yet finally operationalized concept and the FFMQ gives consideration to that by measuring more than just one facet. Additionally the 5 factor structure seems to be very stable and reproducable.

Hi Jonte,

I agree with what you're saying--it makes sense to me. I don't quite understand why the weaker correlation for the Observe subscale is a problem for using the total FFMQ score. Perhaps others can weigh in?

Thanks,

Jon

The problem is that the observe subscale has been shown to have a weaker relationship across several studies therefore it might actually tap a different concept. i did a review of studies using the FFMQ not that long ago to see if any used a composite score and I found none. There is nothing to stop you using a composite score ( I would check the internal consistency first though to make sure the observe scale is measuring the same thing as the other items); I doubt if you will be able to publish using a composite score. Is there any specific reason why you want to use a composite score when you can get more specific information from the sub-scales?

I'd have to agree with Karen on this one. Global FFMQ scores will almost certainly have a tendency to obscure the "active" components of what is a very complex psychological process/state. Psychometrically, reliability may be OK but almost any 39 item measure will produce decent reliability estimates in part because of method bias. In the absence of very good validity data, aggregating is not really capturing "more" of this complex state/process I guess I'm just not really sure why a global score would be preferred.

I think it would indeed not be advisable to use ONLY the composite score. However when you report all subscale scores, in my opinion the composite score gives a valuable overview. Concerning the observe-scale, I would not say that it taps a different concept just because of weaker inter-subscale-correlations. If the observe-facet is a component of mindfulness or not, I think, is not a question of statistical considerations, but of philosophical ones.

Thank you all for your input. My interest is in using overall level of dispositional mindfulness to predict other psychological components. My concern with using each of the five facets individually is the increase in likelihood of Type I error. In other words, if I put each of the 5 facets on the FFMQ into a regression as the predictor, and my criterion has, let's say, 4 facets, that is messier than being able to use just an overall mindfulness score as the predictor. Does this make sense? Can anyone offer guidance? Thanks!!

Hey Jon. It's an interesting (and common) problem when dealing with multi-componential psychological metrics. One solution is to try to use theory to guide prediction via the components (i.e., make predictions about differential association and/or directionality). Alternately, you could use a forward entry type approach in the regression and let the algorithm "choose" the facets that are best predictive. Three, you could see if the outcomes you're interested in are actually different (i.e., what are their associations) and evaluate whether you need all of them. I tend to favor the first approach but think the alternates also have their place. Good luck.

Hi Jon, probably you have probably solved the problem and moved on but FWIW here are my thoughts.  I had a pretty close look at this last year, including Baer's development work on the FFMQ and on the earlier KIMS (where much of the observe subscale items originated).  The observe subscale is problematic in a number of ways.  First of all it only behaves as expected with trained and experienced meditators and therefore is not that good at measuring dispositional mindfulness.  With untrained people, the problem goes beyond low correlations to correlations in the wrong direction.  While positively correlated with openness, emotional intelligence and self-compassion (as expected), it was also positively correlated with dissociation, absent-mindedness, psychological symptoms, and thought suppression (contrary to predictions).  I think the observe items that originated in the KIMS are not well written and seem to describe a narrow intense focus instead of a broad open receptive awareness. Those that came from the FMI are better and worked better in the confirmatory factor analysis.  Interestingly I think the other four facets in the FFMQ have much better psychometrics, which the observe subscale would drag down if all five were put together to form a composite measure.  Great question Jon and I hope it worked out!  BTW if anyone has experience with the MAAS (or another mindfulness scale) in relation emotion processing and regulation I would love to hear about it.

Hello! Hereen et al. (2011) refers to « recent studies » using the global score of the FFMQ, without citing any sources. She gives a global score and suggests that the french version might be used as a global score in relationship to other psychological variables.

http://www.sciencedirect.com/science/article/pii/S1162908811000211

If the question is still valid (and someone looking for an answer):

Following is the paper from journal Psychological Assessment (Page 2, 2nd para) that suggest that five facets in FFMQ can be clubbed to get a total score ("global measure of mindfulness"):

https://self-compassion.org/wp-content/uploads/publications/FFMQ.SCS.FactprStructure.pdf