Table 1.

Composition and scaling of the Health Behavior Framework constructs

Construct and Questionnaire Items	Scaling
Knowledge about breast cancer risk factors: “In your opinion, does ___ increase a woman’s risk of getting breast cancer?” 1Getting older 2High fat diet 3Giving birth to the first child after age 30 4Receiving hits or bruises to the breast 5Excessive fondling of the breast 6Having a blood relative who has had breast cancer 7Having breast implants 8Using chemicals or preservatives in food 9Starting menstruation before age 12 10Having multiple sexual partners	Percent correct derived from 4 response categories (strongly disagree, somewhat disagree, somewhat agree, strongly agree)
Belief in control over getting breast cancer: 11“If you were to get breast cancer, how much do you think you could have done anything to prevent getting the disease?”	High/low belief derived from 3 response categories (a lot, somewhat, very little): high = a lot, low = somewhat or very little
Belief in medical personnel’s control over recovery: 12“How much control do you think medical doctors and nurses have over recovery or survival after breast cancer?”	High/low belief derived from 3 response categories (a lot, somewhat, very little): high = a lot, low = somewhat or very little
Perceived susceptibility: 13“How likely is it that you will develop breast cancer during your lifetime?” 14“Compared to other women your age, how likely do you think it is that you could get breast cancer?”	High/low perception derived from 3 response categories (very, somewhat, not at all): high = one response “very” and other “very” or “somewhat,” low = otherwise
Perceived severity: 15“If a woman is diagnosed with breast cancer, how likely do you think it is that she will be alive 5 years later?”	High/low perception derived from 3 response categories (very, somewhat, not at all): high = not at all, low = very or somewhat
Need for social support to obtain a mammogram: 16“Do you need someone to help you get information on what to do, where to go, or how to get a mammogram?”	High/low need derived from 3 response categories (yes, maybe, no): high = yes or maybe, low = no
Screening concerns: “Are you concerned 17about the radiation (x-ray exposure) from a mammogram?” 18that a mammogram may be painful?” 19that a mammogram may find breast cancer?”	High/low concerns derived from 3 response categories (yes, maybe and no); high = yes or maybe for 2 or more items, low = otherwise
Self-efficacy: “How certain are you that you can 20understand what the doctor is doing when he/she examines you?” 21understand the doctor’s explanation about your mammogram?” 22explain the results to another person?” 23educate another woman about a mammogram?”	High/low self-efficacy derived from mean score of response categories (1 = uncertain, 2 = somewhat uncertain, 3 = neither uncertain/certain, 4 = somewhat certain, 5 = certain): high if mean ≥ 4, low if mean < 4. Cronbach’s α = 0.88 pretest, 0.91 posttest*
Traditional Samoan ethnic identity: 24“God is the unseen head of the family.” 25“It is important to follow Samoan traditions such as faalavelaves.” 26“It is important to practice the FaaSamoa.”	High/low ethnic identity derived from mean score of response categories (1 = strongly disagree, 2 = somewhat disagree, 3 = neither disagree/agree, 4 = somewhat agree, 5 = strongly agree): high if mean ≥ 4, low if mean < 4. Cronbach’s α = 0.66 pretest, 0.68 posttest8
Endorsement of culture-specific beliefs: 27“If I get breast cancer, I will first seek treatment from a fofo or a taulasea.” 28“Breast cancer is spread through the air.”	High/low endorsement derived from 5 response categories (strongly disagree, somewhat disagree, neither disagree/agree, somewhat agree, strongly agree): high = agreement with one or both items, low = otherwise
Agreement with divine intervention: 29“My faith in God will help me not get breast cancer.”	High/low agreement derived from 4 response categories: high = strongly agree or somewhat agree, low = strongly disagree or somewhat disagree

^*Cronbach’s α was computed only for multi-item constructs with raw scores derived as means. Other constructs used single items (belief in control over prevention, belief in control over recovery, perceived severity, social support, divine intervention) or were scored using a counting approach (perceived susceptibility, screening concerns, culture-specific beliefs) or as percent correct (knowledge).