Sample Size and Power Analysis: Investigators will have 120 participants in both the intervention and the control group by the six-month follow-up assessment. Considering that the possible sample attrition at the six-month follow-up is expected to be no higher than 20% (n=24), a total of 144 participants will be recruited for the baseline assessment; half will be randomly assigned to intervention (n=72) and another half assigned to control condition (n=72). Using a one-sided nonparametric Wilcoxon rank-sum test and the mixed effect ANOVA at the 0.05 level and assuming continuous scores, this sample size (60 per arm) would provide 80 percent power to reject the null hypothesis of the equality of score-changes for both groups. The key objective of the study is to acquire preliminary estimates of breast cancer screening rates after the intervention to assess if further study is warranted. As such, Investigators stress on the precision with which Investigators can estimate the breast cancer screening rate post-intervention rather than our ability to reject a specific null hypothesis. A sample size of 60 participants for each arm reflects a compromise between keeping the scope of the project within the objectives of a pilot study while ensuring a large enough sample size to estimate the breast cancer screening rate post-intervention.
Quantitative Measures (Baseline, One-Week, and Six-Month Follow-Up Assessment): Measures will be selected based on the frequency of use in cancer and health literature, psychometric properties, previous applications with AI populations, and inputs from the Community Advisory Board (CAB) and focus groups. The primary outcome criterion for efficacy is mammography receipt after the intervention, which will be measured by self-report (yes or no) at six-month follow-up. The measure has been widely used and found reliable in cancer screening research. The secondary outcome criteria for efficacy include breast cancer knowledge, health beliefs, cultural attitudes, and intent to undergo screening. These measures will be administered at multiple points: baseline and one-week post-intervention. In order to assess feasibility, investigators will measure participant satisfaction and intervention effectiveness. These measures will be administered at one week after the intervention. Confounding covariates (e.g., background, sociodemographic, and health-related information) will be collected only at the baseline assessment and used for assessing the influence of such contextual factors.
Quantitative Data Analysis: Prior to our hypotheses tests, group equivalence in terms of baseline characteristics will be examined using t-tests and chi-square tests. For Hypothesis 1, Investigators will compare the percentage of women from each condition who receive mammograms or have scheduled a mammography appointment using a chi-square test. Investigators will supplement this with logistic regression analyses to adjust for confounding covariates. For Hypothesis 2, the averages of score change (pre- to post-test) from the two conditions will be compared using the two-sample t-test, and/or the Wilcoxon rank-sum tests after assessing normality of the scores. The group difference in terms of changes in the given constructs will be tested by a mixed-effect analysis of variance (ANOVA). The mixed-effect ANOVA includes both within-subject (i.e., time: repeated measures) and between-subject factors (i.e., group: intervention versus control) and aims to examine whether there is an interaction between these two factors on the dependent variable. Bonferroni correction will be used to reduce the probability of Type 1 error for multiple comparisons. Investigators will supplement this with a regression analysis of score change in order to adjust for confounding covariates. For Hypothesis 3, averages of general satisfaction and effectiveness scores from each group will be compared using the two-sample t-test. Also, the percentage of participants from each group who endorse "yes" for the intention and recommendation items will be compared using the chi-square test. To minimize a potential non-participation bias, Investigators will closely monitor and compare the first and fourth quartiles of responses for differences in background variables and key constructs. Investigators will also carefully document the response rate over the course of this project. IBM SPSS version 25 will be used for data analyses.
