20.10.2015 Views

A COMPENDIUM OF SCALES for use in the SCHOLARSHIP OF TEACHING AND LEARNING

compscalesstl

compscalesstl

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.

Psychology, 36 of <strong>the</strong> 47 articles (76.6%) where scales were <strong>use</strong>d to assess SoTL outcomes,<br />

<strong>in</strong>ternal reliability was reported. Thirty-one studies (66.0%) reported Cronbach’s alpha values<br />

that were all acceptable (> .70; Cronbach, 1951; DeVellis, 1991), whereas five (10.6%) reported<br />

at least one alpha value that was below <strong>the</strong> acceptable .70 value. One example of report<strong>in</strong>g<br />

<strong>in</strong>ternal consistency is Maybury’s (2013) study of <strong>the</strong> <strong>in</strong>fluence of a positive psychology course<br />

on several measures of student well-be<strong>in</strong>g us<strong>in</strong>g previously established scales and report<strong>in</strong>g <strong>the</strong><br />

Cronbach’s alpha of each scale and subscale. The alphas <strong>in</strong> this study ranged from .72 to .89, all<br />

above <strong>the</strong> .70 m<strong>in</strong>imum <strong>for</strong> acceptable reliability (see Chapter 11 <strong>in</strong> this e-book by Layous,<br />

Nelson, & Legg <strong>for</strong> more detail of this study). Ano<strong>the</strong>r example of report<strong>in</strong>g <strong>in</strong>ternal<br />

consistency is a study of a team approach to undergraduate research that <strong>use</strong>d items that <strong>the</strong><br />

researchers developed, ra<strong>the</strong>r than a previously established scale (Woodzicka, Ford, Caudill, &<br />

Ohanmamooreni, 2015). In this study, <strong>the</strong> researchers developed a 19-item, 5-po<strong>in</strong>t Likertscale<br />

onl<strong>in</strong>e survey with four subscales. Each subscale had a Cronbach’s alpha above<br />

acceptable, even with a relatively small sample size (Woodzicka et al., 2015). Report<strong>in</strong>g <strong>the</strong><br />

reliability of newly developed items <strong>use</strong>d <strong>for</strong> <strong>the</strong> first time is particularly important beca<strong>use</strong> of<br />

<strong>the</strong> lack of established history.<br />

Although, <strong>the</strong> majority of <strong>in</strong>ternal consistency values of <strong>the</strong> scales published <strong>in</strong> SoTL <strong>in</strong><br />

psychology exam<strong>in</strong>ed were above <strong>the</strong> acceptable m<strong>in</strong>imum (83.3% of <strong>the</strong> studies report<strong>in</strong>g<br />

acceptable Cronbach’s alphas), <strong>the</strong>re were several <strong>in</strong>stances where at least one scale assessed<br />

was not. One particular study addressed this issue directly, by assess<strong>in</strong>g reliability and not<strong>in</strong>g<br />

that it was not acceptable <strong>for</strong> one of <strong>the</strong> subscales and <strong>the</strong>n giv<strong>in</strong>g an appropriate rationale <strong>for</strong><br />

still us<strong>in</strong>g <strong>the</strong> subscale <strong>in</strong> <strong>the</strong> research (Boysen, 2015). Specifically, <strong>the</strong> rationale was that <strong>the</strong>re<br />

was a strong <strong>the</strong>oretical relation of <strong>the</strong> items <strong>in</strong> <strong>the</strong> subscale and that <strong>the</strong> items showed equal<br />

patterns of results; <strong>the</strong>re<strong>for</strong>e, <strong>the</strong> researcher kept <strong>the</strong> subscale <strong>in</strong> <strong>the</strong> model even though it was<br />

found to be unreliable (Boysen, 2015). Ano<strong>the</strong>r example of <strong>the</strong> <strong>use</strong> of a scale with poor<br />

reliability is Boysen, Richmond, and Gurung’s (2015) study of model teach<strong>in</strong>g criteria us<strong>in</strong>g <strong>the</strong><br />

Ten-Item Personality Inventory (TIPI; Gosl<strong>in</strong>g, Rentfrow, & Swann, 2003). Although <strong>the</strong><br />

Cronbach’s alpha reliability coefficients <strong>for</strong> four of <strong>the</strong> five subscales were below .70, <strong>the</strong><br />

authors justified <strong>the</strong>ir <strong>use</strong> beca<strong>use</strong> of poor reliability reported <strong>in</strong> previous studies and <strong>the</strong> fact<br />

that each subscale only had two items, mak<strong>in</strong>g good reliability less likely (Boysen et al., 2015).<br />

These are good examples of how to handle situations when data that have already been<br />

collected are not reliable or valid.<br />

Ano<strong>the</strong>r common occurrence <strong>in</strong> <strong>the</strong> report<strong>in</strong>g of <strong>in</strong>ternal consistency is list<strong>in</strong>g <strong>the</strong> reliability of<br />

an established scale from previously published studies ra<strong>the</strong>r than report<strong>in</strong>g <strong>the</strong> reliability of<br />

that scale <strong>in</strong> <strong>the</strong> present study. Of <strong>the</strong> 47 studies that reported scales, 30 studies reported<br />

Cronbach’s alphas from <strong>the</strong> current study and six reported <strong>in</strong>ternal consistency values from<br />

previously published studies, but not <strong>the</strong> current sample. An example of us<strong>in</strong>g previously<br />

published <strong>in</strong>ternal consistency values is a study look<strong>in</strong>g at chang<strong>in</strong>g stereotypes of older adults<br />

(Wurtele & Maruyama, 2013), as measured by <strong>the</strong> Fraboni Scale of Ageism (FSA; Fraboni,<br />

Saltstone, & Hughes, 1990). The authors reported support <strong>for</strong> <strong>in</strong>ternal consistency, construct<br />

validity, and test-retest reliability from previously published work but did not report any<br />

sources of reliability from <strong>the</strong> current study (Wurtele & Maruyama, 2013). Although us<strong>in</strong>g<br />

45

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

Saved successfully!

Ooh no, something went wrong!