06.11.2014 Views

learning-styles

learning-styles

learning-styles

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Myers and McCaulley (1985) report a test–retest<br />

reliability meta-analysis on a sample of 102,174<br />

respondents (Table 13) which appears to be robust.<br />

Boyle’s review (1995) notes that the best results<br />

(for Form F) are reported stability coefficients of<br />

between 0.69 (T-F) and 0.78 (E-I), which, though lower<br />

than those in Table 13, are still acceptable. Advocates<br />

who have interpreted MBTI retest scores positively<br />

(eg Carlson 1980, De Vito 1985, Murray 1990) have,<br />

according to Pittenger (1993), used trait judgement<br />

criteria, implying a continuum, rather than type<br />

criteria, reflecting the (allegedly) dichotomous nature<br />

of the scales. This criticism is repeated in reviews<br />

of Form M where it is accepted that MBTI scales<br />

show ‘very high levels of internal consistency<br />

(mostly >0.90) and acceptable [actually very high]<br />

levels of test–retest reliability (0.83–0.97 for a 4-week<br />

interval). However, the authors clearly state that<br />

the MBTI is meant to identify a person’s whole type<br />

(eg ENTP)’ (Fleenor 2001; see also Mastrangelo 2001).<br />

The evidence of whole-type stability from the manual<br />

(Myers and McCaulley 1985) appears to be a little<br />

less impressive, with 65% of respondents maintaining<br />

their type and most of the remaining 35% showing<br />

consistency in three out of four scales (n=424).<br />

The stability of the MBTI type allocations are open<br />

to question in part because the middle scores are<br />

prone to misinterpretation, since they are forced one<br />

way or the other, despite small numerical differences.<br />

For example, Howes and Carskadon (1979) found<br />

that for scores within 15 points of neutral, between<br />

25% and 32% of respondents had changed on the<br />

second test. A meta-analysis of reliability across<br />

210 recent studies (Capraro and Capraro 2002) notes<br />

that most authors of studies using the MBTI do not<br />

engage with issues of reliability at all; however, when<br />

reliability data was available, ‘the MBTI tended to yield<br />

acceptable score reliabilities’ (2002, 596) of around<br />

0.81 (standard deviation 0.08). In addition, Capraro<br />

and Capraro (2002, 599) emphasise that the reliability<br />

of an instrument is context-specific: ‘dependent<br />

on sample characteristics and testing conditions.’<br />

Indeed, while Salter, Evans and Forney (1997, 595)<br />

report ‘some stability (ranging from 0.69 to 0.77)’<br />

over 20 months, they warn that the impact of<br />

environmental factors on changes to individuals’<br />

MBTI scores is under-researched.<br />

A lot of work has been done comparing the MBTI<br />

to other scales, which can be summarised as follows.<br />

McCrae and Costa’s (1989) study indicates that<br />

there are correlations between the NEO-PI scales and<br />

the MBTI, despite the omission of neuroticism from<br />

the MBTI; while Furnham (1996a, 306) detects ‘clear<br />

overlap’, despite promoting the psychometric superiority<br />

of the NEO-PI.<br />

Drummond and Stoddard (1992, 103) note connections<br />

between the MBTI and the Gregorc Style Delineator,<br />

concluding that ‘the Gregorc measures some of the<br />

same dimensions as the Myers-Briggs but uses<br />

different labels’.<br />

Spirrison and Gordy (1994) find the Constructive<br />

Thinking Indicator predictive of scores on the MBTI.<br />

Lim (1994) found moderate relationships between<br />

introversion on the MBTI and abstract and reflective<br />

tendencies on Kolb’s LSI.<br />

Higgs (2001) was able to find only partial correlations<br />

between MBTI type and emotional intelligence.<br />

While there are many attempts to link and correlate<br />

the MBTI with other measures of <strong>learning</strong> style, some<br />

of these (eg Nordvik 1996; or see Di Tiberio 1996<br />

for an overview) seem to be predicated on the belief<br />

that if there are some modest correlations between,<br />

say, three disparate measures, they all somehow<br />

validate one another. Indeed, it could be argued that<br />

the theoretical descriptions of dimensions in the<br />

MBTI differ substantially from dimensions with similar<br />

names in other typologies, since the MBTI is the<br />

only one of these that remains firmly connected to<br />

Jung’s theoretical constructs. This suggests that the<br />

connections with other tests are not of themselves<br />

a good measure of the MBTI’s validity or relevance<br />

to the field of <strong>learning</strong> <strong>styles</strong>, since the field of <strong>learning</strong><br />

<strong>styles</strong> is beset with problems in terms of establishing<br />

shared definitions of key terms.<br />

The huge body of work which exists on the MBTI<br />

must be examined with the critical awareness that<br />

a considerable proportion (estimated to be between<br />

a third and a half of the published material) has<br />

been produced for conferences organised by the<br />

Center for the Application of Psychological Type<br />

or as papers for the Journal of Psychological Type,<br />

both of which are organised and edited by Myers-Briggs<br />

advocates. Pittenger (1993, 478) asserts that ‘the<br />

research on the MBTI was designed to confirm not<br />

refute the MBTI theory’. A good example of this is the<br />

study by Saggino, Cooper and Kline (2001), which<br />

starts from a position which assumes the validity<br />

of the MBTI and tests new versions of it against<br />

itself. As Mastrangelo (2001) argues, the ‘research<br />

[on the MBTI] need[s] to be presented in journals<br />

besides the Journal of Psychological Type … The most<br />

widely used psychological measure should demand<br />

scientific scrutiny to improve service to the public.’ 10<br />

10<br />

Page numbers are not available for online Buros reports from the<br />

Mental Measurements Yearbooks.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!