 |
Is it appropriate to add up scale scores on the EMI-2 to get a total motivation score? |
|
I want to conduct some multivariate analyses using the EMI-2 but with so many subscales my sample size is not big enough. How can I reduce the data? |
|
I've seen a criticism of the EMI that it is not theoretically derived. Is this true? |
|
The EMI-2 only has two fitness-related subscales, strength and endurance, and nimbleness. What about all the other potential fitness-related reasons people have for exercising? |
|
Do you intend making any further modifications to the EMI-2 or is this the final version? |
|
Can the EMI-2 be used with children and adolescents? |
Is it appropriate to add up scale scores on the EMI-2 to get a total motivation score? No! The way I see it, the EMI-2 is designed to tap the direction of motivation rather than the intensity, although there's obviously some element of 'intensity of direction' as it were! Sorry, that's a bit cryptic - I mean that it taps the extent to which a respondent is motivated to exercise for various distinct reasons rather than just how motivated to exercise they are. A total score would be pretty meaningless, in my view.
I want to conduct some multivariate analyses using the EMI-2 but with so many subscales my sample size is not big enough. How can I reduce the data? Yes, this is a problem that is a quite serious practical limitation of the instrument that I have always recognised. It's not appropriate to simply add up subscale scores to reduce the data (see above). One thing I once toyed with to reduce the data to manageable proportions was to test a smaller set of higher order models on the basis that if these models fit the data well enough, one could collapse scores on the first-order factors. If you are familiar with the paper describing the development of the EMI-2 (Markland and Ingledew, 1997), you will recall that we grouped conceptually-related subscales and tested five submodels, because of the sample size problem in testing the model as a whole. More recently, I tested a set of higher order models based on these groupings. Although the fit of these higher order models was generally good, I was not too happy with this approach for conceptual reasons. For example, the Interpersonal Motives grouping comprised social recognition, affiliation and competition. Although conceptually related in the sense that they all tap interpersonal reasons, scores on these subscales are clearly not driven by the same thing at a higher order level - one might exercise for affiliation reasons for example, but not for social recognition and competition. There's no reason to posit some kind of overarching drive to exercise for social reasons that leads people to exercise for affiliation AND social recognition AND competition. So I abandoned that approach. I think that one can still make a case for grouping subscales based on the good fit of the submodels in the EMI-2 development paper, the empirical correlations among the subscales and providing that the internal consistency of the grouped items is acceptable. For example, the correlations among those three interpersonal subscales were actually quite strong. I'd be interested to see how journal reviewers/editors viewed that approach! Actually, in effect this would not be any different to what happens in practice with some instruments designed to tap exercise motives that use a much smaller set of subscales such as Silberstein et al.'s (1988) Reasons for Exercise Inventory - they simply group together rather diverse sets of motives into fewer but broader factors than the EMI-2. Nevertheless, grouping them together results in a loss of information - for example, one would have a picture of the effects (if the subscales are used as independent variables in the analyses) of interpersonal motives overall/in general on whatever the dependent variable is but not what the effects of particular motives are. A much sounder thing to do to reduce the data is to only use selected subscales based on a priori predictions.
I've seen a criticism of the EMI that it is not theoretically derived. Is this true? Partly, yes, I hold my hands up! The initial item pool for the original EMI was developed from an open-ended question asking people why they exercise. However, it was not entirely dust-bowl empiricism. The EMI was designed to be used to test theoretically-derived questions, especially from the perspective of Deci and Ryan's (1985) self-determination theory. Naively, I had at first intended to be able to group subscales into intrinsically-oriented and extrinsically-oriented motives. That probably would have made it more clearly theoretically-driven and satisfied the critics but obviously it's not that simple. As we pointed out in Markland and Ingledew (1997), some motives might be more clearly labelled intrinsically-oriented (such as enjoyment, affiliation) and others more extrinsic (e.g. health pressures). But for other motives it depends on how the individual experiences them. The question I have for those who criticise the EMI-2 for being atheoretical is "Tell me what theory provides a conceptual basis for determining a priori what the full range of exercise motives is?" I don't believe that simply labelling an instrument as the Some Theory or Other Exercise Incentives Questionnaire makes it any more theoretically-driven than the EMI-2.
The EMI-2 only has two fitness-related subscales, strength and endurance, and nimbleness. What about all the other potential fitness-related reasons people have for exercising? Good question. We began with a set of items tapping a wider range of fitness motives but they simply didn't factor out. It seems that, for our development samples at least, individuals just didn't distinguish so finely between different aspects of fitness. Individuals who exercise in specific or specialised contexts would probably be more acutely aware of the different parameters of fitness and distinguish between them accordingly. Loze and Collins (1998, Journal of Sports Sciences, 16, 761-767) developed a muscular development scale which they administered to four groups of exercisers, along with two EMI subscales (weight management and appearance). They found that their new scale was reasonably factorially distinct from the two EMI subscales, although there was a degree of cross-loading of muscular development items on weight management (which is quite an interesting finding in itself). Their new scale discriminated between resistance exercisers and those who engaged in other types of exercise. Researchers examining exercise motives in certain contexts might like to obtain Loze and Collin's scale and add it to the EMI-2. However, as they only factor analysed their new items with weight management and appearance items from the original EMI, you should bear in mind that the muscular development items might load on other subscales of the EMI-2.
Do you intend making any further modifications to the EMI-2 or is this the final version? Measurement refinement never ends and the EMI-2 is by no means perfect. However, I do not intend making the measurement of exercise motives my life's work. So at the moment I don't see myself doing any major revisions to the EMI-2. If there's anyone out there who wants to take on that task, though, I'd be willing to help.
Can the EMI-2 be used with children and adolescents? Some of the items in the EMI-2 are not relevant to younger age groups (e.g. 'I exercise to help me look younger). Ingledew and Sullivan (2002) modified the instrument for use among younger (11-13 year old) and older (17-19 year old) German adolescents. The English translations of the items for this version of the EMI are available on the German EMI page of this site (click on the Other language versions button). Researchers investigating exercise motives among younger age groups might want to use these items rather than the adult version.