Specificity is critical to the assessment of perceived benefits. For example, development of a scale to measure perceived benefits of sun protection must take into account the specific action being considered (e.g., use of sunscreen vs. wearing a hat), and the specific benefits being considered (e.g., decreasing likelihood of skin cancer vs. delaying the appearance of age spots and wrinkles). Thus, developing appropriate operational definitions of benefits will continue to challenge researchers as the construct is used with new behaviors. This work will undoubtedly build on the development of valid and reliable scales during the past decade to assess perceptions of the benefits of screening for breast cancer and colorectal cancer.
Benefits of breast self-examination and mammography
Assessments of benefits of breast cancer screening have included
both the behaviors of breast self-examination (Lauver
& Angerame, 1988) and mammography (Champion,
Foster, & Menon, 1997; Champion,
1999). In general, these
scales have good predictive validity. For example, Skinner,
Champion, Gonin, Hanna et al., (1997)
found that perceived benefits for mammography differentiated
between women considering a mammogram and those who were currently
adherent for mammography. Specific items that significantly
differentiated between these groups included finding lumps
early, decreasing chances of dying from cancer, and helping
find lumps before they can be felt. In a sample of low-income
African American women, perceptions of these benefits were
lower for those who had not considered having a mammogram
than for those who had considered the test (Champion
& Springston, 1999).
A measurement study assessed benefits for mammography
screening scale for validity and reliability (Champion,
1999). Items included
not worrying about breast cancer, helping to find breast lumps
early, and treatment won't be as bad (see
Appendix A). Internal consistency reliability of .75 was
calculated for the scale. Confirmatory factor analysis identified
all items as having a Lambda of .40 or greater. Construct
validity was also found through exploratory factor analysis
and by determining that differences in benefits did exist
for persons in different stages of mammography behavior.
Colorectal cancer (CRC)
The development of benefits scales for colorectal
cancer screening has been guided by the same measurement
principles as those for mammography and breast self-examination,
and they have also demonstrated good validity and reliability
et al., 2001). A good example of measurement
specificity can be found in a scale developed by Rawl
et al. (2001), which included questions about
the benefits of finding cancer early and decreasing the chances
of dying from colorectal cancer if one had FOBT, sigmoidoscopy,
or colonoscopy . Reliability was measured using Cronbach's
alpha and was .65 for FOBT, .67 for flexible sigmoidoscopy,
and .70 for colonoscopy. Exploratory factor analysis identified
dimensions for benefits of FOBT, sigmoidoscopy, and colonoscopy
with respective items loading at .54 to .78 for FOBT, .35
to .58 for flexible sigmoidoscopy, and .62 to .72 for colonoscopy.
Theoretically consistent differences were found in all benefits
scales and screening participation. Wardle developed a benefits
scale specific to sigmoidoscopy using a 7-item scale with
5-point Likert-like response scales (Wardle
et al., 2003). Items demonstrated construct
validity by loading at .4 or above on their respective scale.
Internal consistency reliability was .83.
The perceived benefits construct is defined as an individual's belief that specific positive outcomes will result from a specific behavior. Research conducted over the last three decades has demonstrated the use of this construct in predicting behavior, but several measurement issues continue to warrant attention when employing a perceived benefits scale. First, perception of benefit is specific to a behavior and the more specifically the behavior is defined, the higher the predictive validity of the scale. For example, a scale to measure benefits of cancer screening would predictive mammography behavior more poorly than a scale designed specifically to identify benefits of mammography screening per se. Second, because the construct of benefits is most useful when developed as behavior-specific, any attempt to use this construct with a new behavior will necessitate development of items specific to that behavior. Thus, the validity and reliability of measures of the construct will continue to be an important issue as scales are developed to assess the benefits on new health behaviors and health threats. When a new scale is developed, it is important to carefully assess its validity and reliability.