ABSTRACT

After analysis of the data, it can be concluded that the LOCI displays excellent psychometric properties. All results show favorable evidence for the LOCI’s statistical and structural validity, and results are compatible with other large scale LOCI validations studies (Roberts & Irvine, 1999). Furthermore the LOCI not only equals the internal reliability scores of the MEQ on measures of morningness, it actually out performs the MEQ on the dimension of "eveningness" (low-morningness, in the MEQ) and introduces the third dimension of Propensity for Sleep Debt, with high reliability scores.

The most impressive scores in this study are the Cronbach Alpha ratings. As can be seen in Table.1, the reliability scores on the LOCI are very high for measures of morning and eveningness, with scores ranging from 0.85 to 0.94. The MEQ is consistently (though not significantly) lower than these scores (Fig. 1). Having more items, the MEQ would be expected (all things being equal) to have greater psychometric stability (Nunally, 1978), so for the LOCI to consistently get higher alpha scores with less items is a testament to its assessment power. The only dimension where the LOCI shows weaker alpha scores was in the Propensity for Sleep Debt Dimension. Though these scores are still high compared to other tests of multi-dimensional construct (Folkard et al, 1979, Torsval et al, 1980, Smith et al, 1989), they do fail to reach the comparable levels to the other dimensions tested.

Fig. 1

This however does not mean that this dimension is invalid. PSD correlates negatively with the measure of morningness, with values ranging from –0.47 to –0.68 (all values significant at 0.05 using a two-tailed test). This would seems as though the Sleep Propensity dimension is measuring the same properties as Eveningness. Yet there was no significant correlation observed between measures of Eveningness and PSD. This suggests that the PSD dimension is indeed a separate dimension.

Reasons for the discrepancy on reliability scores is unclear, though being a newer dimension, it is hardly surprising. Poorer results would naturally be expected, not having had the benefit of large sample validation, such as the number of trials made to validate the MEQ. Further work could also be undertaken in the item selection to improve reliability. Certain questions for this dimension have an ambiguity in meaning that could be refined. For Example, Item 11, which states "on the weekends, I sleep past noon". It is conceivable that this question would give false positives in cases of shift workers, where their normal routine would be to get up late everyday. Though the student population in this study are unlikely to be in this category, some attention should be paid when applying this battery on a more general population.

One of the immediate criticism on the validation of this instrument is the subject pool. In particular, the age pool is rather limited. The mean age of all participants was 24.65, with results closely bunched around this mean. This completely disregards older sections of society, which are known from previous studies to be prone to morningness (Brown, 1993). This observation is confirmed with the data in this experiment, with a positive correlation between age and morningness (0.443 and .515 for the LOCI and MEQ respectably. Both values are significant at the 0.05 level).

Both the LOCI and the MEQ are standardized with samples from university student. Extra validational evidence from a larger sample of the general population would be suggested. In particular, sub-population of night workers could provide useful data for Evening and PSD types. Folkard et al (1979) standardized their scale on such populations. It is know that for student populations the CTQ test has very poor psychometric properties (alpha 0.6, Smith et al, 1989). This discrepancy poses the question of how well the LOCI would perform given an irregular sub population.

The main difference between the MEQ and the LOCI lies on the third dimension that the LOCI introduces, the PSD. Other such multidimensional models have been proposed (CTQ, CCS, DTS, etc.), but none with the statistical properties of the LOCI. Factor analysis shows that the three dimensions proposed have distinct properties. However, as mentioned before, there is a significant (though moderate) negative correlation with morningness and sleep propensity. This leads to question whether there exist a fourth dimension, which would negatively correlate with eveningness? Perhaps the unstable reliability experienced in PSD alpha scores comes from conflict in the components that it asses. The LOCI proposes a model that separates the "morningness-eveningness" scale of the MEQ into two separate, unique dimensions. From the LOCI data it seems that this is a useful distinction to make. Couldn’t this same idea be applied to the energetic/lethargic components that comprise the PSD dimension? Further research may prove four distinct dimensions beneficial in assessment of chronotype, one with morning, evening, lethargy and energetic loadings. Obviously, this is only speculative. It may also lead to a theoretical trap of adding irrelevant categories ad infinitum.

Clearly, the study of diurnal rhythms has many consequences. There are immediate uses for a statistically sound tool like the LOCI, with abounding practical applications. An obvious choice would be shift worker placement and assessment. Yet chronotypes may have further reaching consequences. Studies by Chelmski et al (1999) have shown correlates between the MEQ scores and depression in college students. Their study, however, showed as many as 63.7% of depressed students who did not fit into a clear category on the MEQ. With a more rigorous tool, circadian rhythms can be more accurately determined, with possible uses as a clinical diagnostic tool. The current study has demonstrated that in the sample tested, the LOCI displayed extremely good reliability and structural coherence, and goes a long way towards validating the LOCI.

Chelmski I, Ferraro R, Petros T, Plaud J (????) An analysis of the "eveningness-morningness" dimension in "depressive" college students. ???????

Nunnaly, J (1978). Psychometric Theory (2^nd Ed.) New York; McGraw Hill.