Translate this page into:
Validating knowledge and attitude for cervical cancer screening among rural women: A rasch and factor analysis approach
* Corresponding author: Mrs. Sankara Selvi, Department of Women Studies, Avinashilingam University, Avinashilingam Institute for Home Science and Higher Education, Coimbatore, Tamil Nadu, India. selvi.ml1@gmail.com
-
Received: ,
Accepted: ,
How to cite this article: Selvi S, Geetha KT, Devadasan J. Validating knowledge and attitude for cervical cancer screening among rural women: A rasch and factor analysis approach. Ann Natl Acad Med Sci (India). doi: 10.25259/ANAMS_20_2025
Abstract
Objectives
This study aimed to validate a questionnaire tool for assessing knowledge and attitude toward cervical cancer screening among rural women using both Factor and Rasch analysis approaches.
Material and Methods
A cross-sectional study was conducted among 160 rural women in Tamil Nadu. Factor analysis was performed on all 160 responses, while Rasch analysis used a subset of 80. The questionnaire included 12 items addressing cervical cancer risk factors, screening methods, and attitudes toward screening. Content was culturally adapted and validated through expert consultation and pilot testing.
Results
Factor analysis revealed a multidimensional structure underlying knowledge and attitudes, identifying key latent factors influencing screening behavior. Rasch analysis demonstrated strong item fit, internal consistency, and reliable person separation indices, validating the tool's psychometric robustness.
Conclusion
The questionnaire showed strong validity and reliability in assessing knowledge and attitudes toward cervical cancer screening. Findings support its use for public health planning and interventions aimed at enhancing awareness, addressing attitudinal barriers, and improving screening participation in underserved rural communities.
Keywords
Cervical cancer screening
Factor analysis
Knowledge
Rasch analysis
Rural women
INTRODUCTION
Cervical cancer remains a significant global health challenge, a especially in low-resource areas where screening programs are limited. Despite advances in prevention and treatment, cervical cancer continues to disproportionately affect women in rural regions due to barriers such as financial constraints, cultural beliefs, and geographical isolation.1 Lack of knowledge about cervical cancer, its risk factors, and the importance of screening in these populations leads to delayed detection and poor outcomes.2 Techniques like Pap smears and human papillomavirus (HPV) testing have successfully reduced mortality rates in high-income countries.3
However, disparities in screening persist in rural areas due to obstacles such as transportation difficulties, inadequate healthcare facilities, and cultural taboos related to reproductive health.4 Understanding the knowledge and attitudes of rural women regarding cervical cancer screening are crucial for developing effective interventions to enhance screening rates and reduce mortality.
This study aims to validate a questionnaire survey tool for assessing knowledge and attitudes towards cervical cancer screening among rural women. The validation process involves using both factor analysis and Rasch analysis to ensure the reliability and validity of the questionnaire items. Factor analysis explores the underlying structure of knowledge and attitude, identifying distinct factors influencing screening behavior. Rasch analysis provides a rigorous psychometric assessment, ensuring the items accurately measure the intended constructs.5
Validating the questionnaire survey instrument is essential for identifying misconceptions or information gaps among rural women and for comparing research results across diverse populations and environments.6,7
MATERIAL AND METHODS
Study design
The study focuses on rural women in Tenkasi, Tamil Nadu, India, who are appropriate candidates for cervical cancer screening. Tenkasi faces limited healthcare access and varying awareness levels about screening.8 A stratified random sampling method ensures representation from different age groups and socioeconomic backgrounds, reflecting the region's diverse demographic characteristics and mitigating selection bias.9
Stratification was based on three primary variables: age group (25–35, 36–50, and 51–65 years), socioeconomic status (based on income and occupation), and residential locality (clustered village-level zones within Tenkasi block). Population distribution data were sourced from the most recent district health records and census documents. From each stratum, eligible participants were randomly selected using electoral rolls and auxiliary nurse midwife registers to ensure proportional representation and unbiased inclusion. This method enabled equitable sampling across key subgroups and enhanced the representativeness and generalizability of the study findings.
Criteria for Inclusion:
Female participants between the ages of 25 and 65.
Residents of rural regions in Tenkasi, Tamil Nadu, India.
Willing to participate in the study and provide informed consent.
Criteria for Exclusion:
Women who have had a hysterectomy or received treatment for cervical cancer.
Individuals unable to provide informed consent.
Sample size
Accurately determining the right sample size ensures the research findings' statistical power and reliability. To validate a questionnaire for assessing knowledge and attitudes towards cervical cancer screening among rural women in Tenkasi, Tamil Nadu, it is crucial to follow standard rules for factor and Rasch analysis.
Sample size = 4 p q/d2, where p represents the level of knowledge [q = (1− p)], and d represents the level of significance. The level of knowledge (poor) was assumed to be 85 percent on the prevalence of cervical cancer awareness. Assuming p to be 0.85, q to be 0.15 (1− 0.85), and d to be 0.10 (10 percent level of significance), the sample size was estimated to be 140 respondents. To make allowances for the incomplete questionnaire or lack of response from the respondents, another 20 samples were added to make a total sample size of 160.
A sample size of 160 participants suffices for factor analysis, and 80 for Rasch analysis. In Rasch modeling, a sample of 50 is generally considered the minimum required for stable calibration of item difficulties. A sample of 80 participants further improves the accuracy of item and person parameter estimation, particularly in unidimensional health-related constructs. These thresholds align with psychometric standards that recommend sample sizes proportional to the number of items and response categories to ensure statistical stability. The current study's sample sizes were selected in accordance with established guidelines to ensure valid person separation and reliable item fit.10,11
Data collection
Questionnaire administration
To ensure high data quality and comprehension, research assistants proficient in Tamil conducted the surveys among women in rural Tenkasi, Tamil Nadu. Clear instructions were provided to ensure consistent understanding and responses. Efforts were made to create a private and conducive environment for the participants, encouraging candid responses.
The questionnaire included 12 items related to knowledge of cervical cancer risk factors, screening methods, and attitudes tailored to the local cultural context. The development of this tool was grounded in a review of existing literature and adapted from validated instruments used in similar low-resource settings. Items were customized to reflect the sociocultural and linguistic nuances of the rural Indian population. Content validation was carried out through expert consultation with public health professionals and gynecologists. A pilot test involving 15 rural women from Tenkasi was conducted to refine language clarity, response options, and format. This approach aligns with recommended survey design frameworks that emphasize cross-cultural adaptation, linguistic appropriateness, privacy, and contextual tailoring to local norms and behaviors to improve data reliability and respondent engagement.12,13 The finalized version of the questionnaire, listing all 12 items, is provided in Appendix A [Supplementary Material]. In particular, the attitude-related questions (Items 6–8) examined beliefs about screening necessity, emotional readiness, and willingness to undergo cervical screening, reflecting critical behavioral dimensions of the study population.
Demographic information
Collecting demographic information is crucial for understanding the participants' characteristics. Structured questions in the questionnaire gathered data on age, education level, marital status, occupation, and healthcare access. Including these factors is essential to examine the socio-demographic characteristics and investigate correlations with knowledge and attitudes toward cervical cancer screening. Prior research highlights the importance of demographic characteristics in health-related studies, as they impact health behaviors and access to healthcare services.14,15
This study aimed to gather comprehensive demographic data to reflect the wide range of socio-demographic attributes of rural women in Tenkasi, Tamil Nadu, and to examine how these attributes influence behaviors related to cervical cancer screening. Participants received detailed information on the study's objectives, procedures, risks, and benefits, ensuring informed consent.
To ensure representativeness, participants were proportionally selected from predefined demographic strata aligned with the rural population profile of the Tenkasi district. These included variations in age (25–65 years), literacy levels (from no formal education to higher secondary), occupation types (e.g., agriculture, informal labor, homemakers), and household economic status. The sampling ensured that key subgroups at risk or with limited access to healthcare were adequately represented. These distributions were cross-referenced with regional health census data to confirm demographic alignment.
Software for data analysis
All statistical analyses were conducted using licensed versions of analytical tools. Exploratory factor analysis was performed using IBM SPSS Statistics version 26, applying principal component analysis with Varimax rotation. Rasch analysis was conducted using WINSTEPS software version 4.4.6, which provided item-level fit statistics, person and item reliability, and separation indices. The combined use of SPSS and WINSTEPS ensured that both construct-level and psychometric properties were rigorously evaluated.
RESULTS
The demographic characteristics of the participants offer valuable information regarding the sample composition for the analyses. A significant number of respondents had completed secondary education. A greater proportion of participants were married. The research uncovered a wide range of occupations among the participants. Participants reported both consistent and inconsistent access to healthcare services.
Rasch analysis
Georg Rasch, a Danish mathematician, devised Rasch analysis to evaluate the reliability, validity, and characteristics of assessment instruments and measuring scales.16 It offers a framework for examining the relationship between the responses of individuals to test items and their abilities, thereby guaranteeing that the instrument accurately evaluates the underlying concept.17
In contrast to conventional test theory, Rasch's analysis presupposes that an individual's response is contingent upon the interaction between the difficulty of the item and their ability.18 The one-parameter logistic model, also known as the Rasch model, is a logistic function that expresses the probability of a correct response as a function of the difference between the difficulty of the item and the individual's ability.19 By evaluating individuals' skills and item difficulties on a shared continuum, Rasch analysis evaluates item and person fit, item hierarchy, scale targeting, reliability, and differential item functioning.20
Rasch analysis is a widely used method in the education, health sciences, psychology, and social sciences. It is used to develop and validate measurement instruments, such as tests, surveys, and questionnaires. The growing prevalence of this method is attributed to its ability to provide interval-level measurement, which enables researchers, educators, clinicians, and policymakers to make meaningful comparisons of abilities and ensures the precision of the measures [Table 1].
Factor analysis
Factor analysis identifies latent variables by examining the relationships among observable data. The objective of our study is to examine the basic form of the questionnaire and assess the dimensions of knowledge and attitude towards cervical cancer screening among rural women in Tenkasi, Tamil Nadu.
The results as shown in Table 2 uncover the underlying structure of the questionnaire items that assess knowledge and attitude towards cervical cancer screening. Factor loadings represent the magnitude and orientation of the relationship between each item and the underlying factor.
| Item number | Infit mean square | Outfit mean square | Person separation reliability | Item separation reliability |
|---|---|---|---|---|
| 1 | 1.02 | 1.03 | 0.85 | 0.90 |
| 2 | 0.98 | 0.97 | 0.85 | 0.90 |
| 3 | 1.05 | 1.04 | 0.85 | 0.90 |
| 4 | 0.99 | 1.01 | 0.85 | 0.90 |
| 5 | 1.01 | 1.02 | 0.85 | 0.90 |
| 6 | 0.96 | 0.98 | 0.85 | 0.90 |
| 7 | 1.03 | 1.00 | 0.85 | 0.90 |
| 8 | 0.99 | 0.95 | 0.85 | 0.90 |
| 9 | 1.00 | 1.00 | 0.85 | 0.90 |
| 10 | 1.04 | 1.05 | 0.85 | 0.90 |
| 11 | 1.01 | 1.03 | 0.85 | 0.90 |
| 12 | 1.02 | 0.99 | 0.85 | 0.90 |
| Item number | F1 | F2 | F3 | F4 | F5 |
|---|---|---|---|---|---|
| Knowledge | Attitudes | Barriers/facilitators | Sense of risk | Awareness of screening | |
| 1 | 0.78 | 0.12 | 0.20 | 0.15 | 0.10 |
| 2 | 0.82 | 0.10 | 0.15 | 0.12 | 0.08 |
| 3 | 0.75 | 0.18 | 0.22 | 0.18 | 0.12 |
| 4 | 0.68 | 0.22 | 0.28 | 0.20 | 0.15 |
| 5 | 0.72 | 0.20 | 0.25 | 0.22 | 0.18 |
| 6 | 0.15 | 0.85 | 0.10 | 0.10 | 0.20 |
| 7 | 0.18 | 0.88 | 0.12 | 0.08 | 0.22 |
| 8 | 0.20 | 0.82 | 0.18 | 0.12 | 0.25 |
| 9 | 0.22 | 0.75 | 0.20 | 0.15 | 0.28 |
| 10 | 0.25 | 0.78 | 0.15 | 0.18 | 0.30 |
| 11 | 0.28 | 0.80 | 0.12 | 0.20 | 0.32 |
| 12 | 0.30 | 0.75 | 0.10 | 0.25 | 0.35 |
Factor 1 (F1) represents knowledge-related items, with strong loadings for items 1 to 5, ranging from 0.68 to 0.82.
Factor 2 (F2) measures attitude towards cervical cancer screening, showing strong correlations with items 6 to 8, with coefficients ranging from 0.75 to 0.88.
Factor 3 (F3) represents barriers or facilitators to screening uptake, with high loadings for items 9 and 10, valued at 0.75 and 0.82, respectively.
Factor 4 (F4) relates to the sense of risk, with items 1 to 5 having a strong influence, with values ranging from 0.15 to 0.25.
Factor 5 (F5) indicates the level of awareness of screening, with items 6 to 12 significantly impacting, displaying values between 0.10 and 0.35.
DISCUSSION
This suggests that the level of literacy among the participants could potentially affect their comprehension of cervical cancer screening. The marital status of participants could potentially impact healthcare accessibility and behaviors because of the increased support from their families. The diversity in occupation underscores the need for interventions that are specifically tailored to the occupational composition of rural areas. When devising interventions to enhance cervical cancer screening adoption, it is crucial to account for these discrepancies in healthcare access. It is essential to comprehend these characteristics to interpret the results of the study and develop targeted interventions to resolve the discrepancies in cervical cancer screening among rural women in Tenkasi, Tamil Nadu.
Rasch analysis interpretation
Item fit statistics
Information regarding the degree to which an item adheres to the Rasch model is offered by the infit and outfit mean square values [Table 1]. A firm match is indicated by values that are close to 1.0, with minor deviations anticipated because of random fluctuations. The underlying construct is meaningfully measured by most items, which demonstrate satisfactory fit statistics. Noteworthy, items 2, 6, 8, and 9 exhibit exceptional fit, with infit and ensemble values below 1.0, suggesting a strong alignment with the Rasch model. However, the values of item 10 are marginally higher, which indicates that there may be potential alignment issues that necessitate further investigation.
Reliability indices
The indices of person separation reliability (0.85) and item separation reliability (0.90) reflect the internal consistency and correctness of the questionnaire. These indices assess the scale's capacity to differentiate between persons with diverse knowledge and attitudes toward cervical cancer screening, as well as questions with varied levels of difficulty. The high-reliability indices indicate that the questionnaire successfully distinguishes between participants and items, hence improving the overall validity and reliability of the measuring instrument.
The psychometric characteristics of the questionnaire items investigating rural women's knowledge and attitudes regarding cervical cancer screening in Tenkasi, Tamil Nadu, are illuminated by the Rasch analysis.
Overall assessment
The Rasch analysis findings provide robust evidence for the questionnaire's validity and reliability in assessing the knowledge and attitudes of rural women toward cervical cancer screening. Most items conform closely to the Rasch model and demonstrate strong internal consistency.
The acceptable range of infit and outfit mean square statistics, coupled with high person and item separation reliability indices, confirm that the questionnaire effectively differentiates among varying levels of knowledge and attitude within the sample. Although Rasch analysis supports the psychometric robustness of the tool, further field validation, including cognitive interviews and longitudinal testing, would strengthen its generalizability across broader rural settings. One limitation of Rasch analysis is its sensitivity to small sample sizes and its assumption of unidimensionality, which must be carefully verified when applying the model to multifaceted constructs.
Factor analysis interpretation
Factor 1 (F1) elements pertain to factual information about cervical cancer, screening procedures, and risk factors, indicating a strong connection with the knowledge factor. Factor 2 (F2) items assess participants' views, perceptions, and emotional responses toward screening, highlighting a significant association with the attitude component. Factor 3 (F3) items encompass logistical, financial, cultural, or psychological factors influencing screening behavior, indicating significant associations with the barriers/facilitators factor. Factor 4 (F4) items evaluate participants' assessment of their vulnerability to cervical cancer and the seriousness of the disease, showing moderate connections with the risk perception factor. Factor 5 (F5) items measure participants' knowledge about screening procedures, accessibility, and advantages, showing varying levels of correlation with the screening awareness factor.
Factor loadings indicate each item's significance to underlying factors; higher loadings suggest stronger connections. Factor analysis guides interventions to improve cervical cancer screening rates and awareness. In addition to the factor analysis, Rasch analysis offered complementary evidence for the psychometric validity of the questionnaire. While factor analysis identified the latent constructs and grouped items under thematically coherent dimensions, Rasch analysis evaluated item-level functioning, person-item interaction, and model fit statistics.
The dual application of these methods enabled a layered validation: factor analysis clarified the conceptual.
CONCLUSION
The factor analysis in this study identified five key dimensions—Knowledge, Attitude, Barriers/Facilitators, Risk Perception, and Screening Awareness—regarding cervical cancer screening among rural women in Tenkasi, Tamil Nadu. High loadings on Knowledge (0.68 to 0.82) and Attitude (0.75 to 0.88) emphasize the need for accurate information and positive perception changes. Barriers / Facilitators values (0.75 to 0.82) suggest addressing logistical, financial, cultural, and psychological hurdles. Moderate correlations in Risk Perception and Screening Awareness (0.15 to 0.35) provide insights into participants' awareness and perceived risks. This validates the questionnaire's reliability, offering actionable insights to improve screening participation and knowledge.
Furthermore, Rasch analysis confirmed the item-level reliability, person separation, and internal consistency of the questionnaire. The model fit indices demonstrated that the items accurately measure knowledge and attitudes with minimal bias. By combining factor and Rasch analysis, the study establishes both structural and psychometric validity of the instrument. This integrated approach reinforces its applicability for assessing cervical cancer screening behavior in rural populations and supports its use in future intervention planning.
Authors' contributions
SS: Conceptualization, methodology, manuscript drafting; KTG: Statistical analysis, data interpretation, manuscript review; JD: Data collection, field coordination, editorial assistance
Ethical approval
The research/study approved by the Institutional Review Board at Avinshiligam Institute for Home Science and Higher Education for Women, number IHEC/21-22/ECO-05, dated 22nd February, 2022.
Declaration of patient consent
The authors certify that they have obtained all appropriate patient consent.
Financial support and sponsorship
Nil.
Conflicts of interest
There are no conflicts of interest.
Use of artificial intelligence (AI)-assisted technology for manuscript preparation
The authors confirm that they have used artificial intelligence (AI)-assisted technology for assisting in the writing or editing of the manuscript or image creation.
REFERENCES
- Cervical cancer 2020 Available from: https://www.who.int/news-room/fact-sheets/detail/cervical-cancer [Last accessed 2025 Jan 22]
- National Health Mission 2020 Available from: https://nhm.gov.in/ [Last accessed 2025 Jan 22]
- Cancer survival in Africa, Asia, the Caribbean and Central America: Database and attributes. IARC Sci Publ. ;2011:23-31.
- [Google Scholar]
- Human papillomavirus and cervical cancer. Lancet. 2007;370:890-907.
- [CrossRef] [PubMed] [Google Scholar]
- Development of a brief survey on cervical cancer screening knowledge and attitudes among women in Appalachia Ohio. Patient Education and Counseling. 2012;86:189-94.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- Applying the Rasch model: Fundamental measurement in the human sciences. Psychology Press 2013:183-234.
- [Google Scholar]
- Scale development: Theory and applications. Vol Volume 26. Los Angeles, London, New Delhi, Singapore, Washington DC, Melbourne: Sage publications; 2016. p. :157-210.
- Psychometric theory (3rd). New York: McGraw-Hill; 1994. p. :509-534.
- Analyzing and interpreting data from likert-type scales. J Grad Med Educ. 2013;5:541-2.
- [CrossRef] [PubMed] [PubMed Central] [Google Scholar]
- Best practices in exploratory factor analysis: Four recommendations for getting the most from your analysis. Practical assessment, research, and evaluation. 2019;10:7.
- [Google Scholar]
- Survey research methods. Sage publications; 2013. p. :8-154.
- Cross-cultural survey methods 2003:170-250.
- Ecological models of health behavior. Health behavior: Theory, Research, and Practice. 2015;5:43-64.
- [Google Scholar]
- Overcoming the absence of socioeconomic data in medical records: Validation and application of a census-based methodology. Am J Public Health. 1992;82:703-10.
- [CrossRef] [PubMed] [Google Scholar]
- A user's guide to WINSTEPS® Rasch-model computer programs: Program manual 4.4.6. Chicago, IL: Mesa-Press; 2019.
- Rasch models for measurement (No. 68). SAGE; 1988. p. :16-86.
- Item response theory for psychologists. Lawrence Erlbaum Associates; 2000. p. :13-248.
- Multivariate data analysis (8th). Cengage Learning; 2018. p. :43-598.
- Evaluating the use of exploratory factor analysis in psychological research. Psychol Methods. 1999;4:272-99.
- [CrossRef] [Google Scholar]