Classical Test Theory Methods Research Articles

Reliability of a probabilistic knowledge structure

Indexes for estimating the overall reliability of a test in the framework of knowledge space theory (KST) are proposed and analyzed. First, the possibility of applying in KST the existing classical test theory (CTT) methods, based on the ratio between the true score variance and the total variance of the measure, has been explored. However, these methods are not suitable because in KST error and true score are not independent. Therefore, two new indexes based on the concepts of entropy and conditional entropy are developed. One index is used to estimate the reliability of the response pattern given the knowledge state, while the second one refers to the reliability of the estimated knowledge state of a person. Some theoretical considerations as well as simulations and an empirical example on real data are provided within a study of the behavior of these indexes under a certain number of different conditions.

The Patient Activation Measure-13 (PAM-13) in an oncology patient population: psychometric properties and dimensionality evaluation

BackgroundAccurate assessment and enhancement of health-related skills among oncology patients are pivotal for optimizing cancer care. The Patient Activation Measure (PAM-13), a questionnaire designed to reflect an individual’s knowledge, skills, and confidence in self-healthcare management, has been validated across diverse countries and settings. Concerns have been raised regarding the cross-situational applicability, as patients with specific diseases and cultural backgrounds interpret questionnaire items differently. This study aimed to examine the structural validity and psychometric properties of the PAM-13 in an oncological patient cohort.MethodsBaseline data from a longitudinal non-randomized controlled study involving cancer out-patients (n = 1,125) from Comprehensive Cancer Centres in Southern Germany were analysed. The German version of the PAM-13 was employed. With classical test and item response theory methods data quality, reliability, convergent and structural validity, as well as psychometric properties were assessed. Exploratory (EFA) and confirmatory factor analyses (CFA) were employed to investigate the postulated unidimensionality of the underlying construct. With a partial credit model (PCM) we examined item fit, targeting, local independence and differential item functioning.ResultsParticipants were predominantly female (73.0%) with a breast cancer diagnosis (41.3%). While items were generally well-accepted, ceiling effects were observed and a high mean PAM-13 score (69.7, SD = 14.2) was noted, potentially compromising responsiveness to interventions. Reliability was adequate (Cronbach’s α = 0.81), person and item separation reliability were good to excellent (0.81 and 0.99, respectively). Explorations of the unidimensionality of the construct (EFA, CFA, PCM) yielded inconclusive results, hinting towards a two-factor solution. Item difficulty rankings deviated from the original. No differential item functioning was identified, and local independence was confirmed.ConclusionsWhile the PAM-13 serves as a valuable instrument for comprehending and promoting health-related skills in cancer patients, the identification of ceiling effects, disordered item-difficulty rankings, and inconclusive findings regarding unidimensionality contribute to the expanding body of evidence, emphasizing the dependency of PAM-13’s validity and reliability on distinctive characteristics within the population under investigation. Future research should prioritize refining or adding PAM-13 items to better capture the specific health-related challenges within diverse populations, paving the way for more effective patient engagement strategies in oncology.Trial registration numberDRKS00021779

Construct validity of the Braden scale in acute- and long-term care settings in Austria: A structural equation modeling analysis

Objective: The Braden scale is frequently used to assess pressure ulcer risk in health care settings. Selected psychometric properties have been tested using various methods of classical test theory in international studies. However, limited information on construct validity is available. Aim was to determine if the Braden subscale items correlate with the construct pressure ulcer risk and whether the construct validity concerning the factor structure of the Braden scale is adequate in acute and long-term settings.Methods: A quantitative design with secondary analysis of data from one acute (n = 328) and eight long-term care facilities (n = 311) in Austria was used to test construct validity. Data analysis included principal axis factor analysis with Promax rotation and assessment of internal consistency, followed by structural equation modeling.Results: For the acute care setting, a structure equation model with two latent factors and for the long-term care setting with one latent factor was tested according to principal axis factoring results. The Braden subscale items correlated with the construct pressure ulcer risk. Almost all examined model fit indices were within recommended reference values. Thus, the construct validity of the Braden scale was adequate in both settings.Conclusions: The factor structure in the acute care setting did not match that in the investigated long-term care setting. Further research regarding the construct validity of the Braden scale is therefore necessary.

Nonparametric Kernel Smoothing Item Response Theory Analysis of Likert Items

Likert scales are the most common psychometric response scales in the social and behavioral sciences. Likert items are typically used to measure individuals’ attitudes, perceptions, knowledge, and behavioral changes. To analyze the psychometric properties of individual Likert-type items and overall Likert scales, mostly methods based on classical test theory (CTT) are used, including corrected item–total correlations and reliability indices. CTT methods heavily rely on the total scale scores, making it challenging to directly examine the performance of items and response options across varying levels of the trait. In this study, Kernel Smoothing Item Response Theory (KS-IRT) is introduced as a graphical nonparametric IRT approach for the evaluation of Likert items. Unlike parametric IRT models, nonparametric IRT models do not involve strong assumptions regarding the form of item response functions (IRFs). KS-IRT provides graphics for detecting peculiar patterns in items across different levels of a latent trait. Differential item functioning (DIF) can also be examined by applying KS-IRT. Using empirical data, we illustrate the application of KS-IRT to the examination of Likert items on a psychological scale.

Defining a universal measurement unit and scale for gross motor development.

The ability of children to accomplish progressively more difficult gross motor tasks follows a predictable sequence that has been well documented as part of development. Current existing instruments were developed independently using classical test theory methods which led to the lack of a universal measurement scale and unit. The purpose of this study was to test a specification equation, anchored to commonly accepted and reproducible tasks in gross motor development, to generate a universal measurement scale and unit of measurement, called the Gross Motor (GM) unit. We rated component measures for each of the gross motor development tasks on the Gross Motor Function Measure-66 (GMFM). The GMFM is a gross motor development measure created with Rasch measurement theory to quantify observed difficulty levels measured on an interval scale. Component measures for body position, movement, and support were based on hypothesized contributions to gross motor development based on theory. Forward stepwise linear regression was used to test a specification equation. The specification equation was anchored to reference points to fix a unit size. Our specification equation explained 87% of the variance in observed gross motor task difficulty. Predicted difficulty for gross motor tasks was strongly associated with observed task difficulty (r = 0.94, p < 0.0001). Our specification equation was anchored to 1) lying supine (0 GM units) and 2) walking unsupported (100 GM units) setting the size of the GM unit to 1/100 of the distance between lying supine and unsupported walking. Our specification equation allows for experimental testing of gross motor development theories. This approach provides a framework for refining our understanding and measurement of gross motor development and creates a universal scale and unit. We expect that this will facilitate placing many, if not all, current gross motor development instruments on the same measurement scale.

Evaluation of the psychometric performance of the Spanish and Catalan versions of the patient reported experiences and Outcomes of Safety in Primary Care (PREOS-PC)-Compact questionnaire

Background Patients provide a unique, irreplaceable, and essential perspective in evaluating patient safety. The suite of Patient Reported Experiences and Outcomes of Safety in Primary Care (PREOS-PC) tools are a notable exception to the scarcity of patient-reported patient safety measures. Full evaluation of their performance has only been attempted for the English version, thereby limiting its international applicability. Objectives To assess the psychometric performance of the Spanish and Catalan versions of the PREOS-PC-Compact. Methods Cross-sectional validation study. We used Classical Test Theory methods to examine scale score distribution, internal consistency, and construct validity; and Item Response Theory (IRT) methods to further explore construct validity. Results 3287 patients completed the Spanish version, and 1007 the Catalan version. Similar results were obtained for both versions. Confirmatory Factor Analysis supported a single construct for each scale. The correlations between PREOS-PC-Compact scales and known group analysis suggested adequate construct validity (inconclusive for known groups at the provider level). All four multi-item scales demonstrated adequate internal consistency reliability (α > 0.7), which was only confirmed for test-retest reliability for ‘Practice activation.’ A sample between 60–90 patients per practice was estimated sufficient to produce scores with reliability > 0.7 for all scales except for harm scales. IRT models showed disordered thresholds for ‘Practice activation’ and ‘Harm burden’ but showed excellent fit after reducing the response categories. Conclusion The Spanish and Catalan versions of the PREOS-PC-Compact are broadly valid and reliable tools to measure patient safety in Spanish primary care centres; confirmation of lower-than-expected test-rest reliability merits further examination .

The HAIR-QoL measure Part 2: Validation of an instrument to measure of the severity and impact of chemotherapy-induced alopecia (CIA)

ObjectiveChemotherapy-induced alopecia (CIA) is one of the most common and distressing side effects related to cancer treatment. Few patient-reported outcome measures (PROMs) have been developed specifically to measure the patient-experience of hair loss during chemotherapy treatment. In part I of this research, a set of provisional questionnaire items were developed for a new tool. This article focusses on the validation and psychometric testing of the CIA-specific scoring items in the provisional HAIR-QoL measure. MethodsThe provisional HAIR-QoL questionnaire, comprised of 43 CIA-specific items, was administered to a large sample of patients (n = 228) from 8 international centres, who were undergoing chemotherapy. Psychometric testing used a combination of Classical Test Theory (CTT) and Item Response Theory methods; Principal Components Analysis (PCA) and Rasch Analysis, respectively. ResultsFrom the initial item pool of 43 items, 28 items satisfied the criteria for inclusion into four components. The four subscales identified were as follows, 1) Concern about change in appearance, 2) Image of self, 3) Preparedness for hair loss and 4) Impact of hair loss on you and others. The results of the quantitative analyses, PCA and Rasch analysis, suggested modest amendment to the number of items to the provisional HAIR-QoL measure. Psychometric properties were reported to be very good/excellent for all four subscales and were found to be sensitive to detecting group differences. ConclusionThe refined items which comprise the final HAIR-QoL measure were found to tap into several important psycho-social domains of concern for patients who experienced CIA. The subscale scores have the potential to inform future decision-making on patients’ unmet information and support needs and pave the way for personalised psychological care and more robust research on efficacy of scalp cooling.

A Comparison of Covariates, Equating Designs, and Methods in Equating TIMSS 2019 Science Tests

This research aimed to compare the equated scores by the methods based on classical test theory (CTT) and kernel equating, using covariates design (NEC) and anchor test design (NEAT). TIMSS 2019 science test scores equated by both Tucker, Levine true score, Levine observed score, equipercentile equating (pre-smoothing and post-smoothing) methods in CTT, and linear and equipercentile methods in kernel equating. Additionally, the covariates in NEC design were “home resources for learning,” “student confidence in science and mathematics,” “like learning science,” “instructional clarity in science lessons,” “math achievement,” “sex,” and “speaking the language of the test at home”. The equating results in NEC were compared with those in NEAT and EG. The participants comprised 1699 4th-grade students who attended the e-TIMSS 2019 in Canada, Singapore, and Chile. Results were analyzed according to equating errors and differences between equated scores. The research concluded that math achievement and home resources for learning could be used as covariates in NEC to equate the science test in case equating could not be done in the NEAT. However, when the other variables were used as covariates in NEC, the equated scores were very similar to the EG. Also, Tucker (CTT) and post-stratification (kernel) yielded similar equated scores in linear equating, and these methods were similarly different from kernel linear equating in EG. In equipercentile equating, the equated scores obtained from the post-smoothing (CTT) and EG were close to each other but slightly differed from post-stratification.

Development and validation of the patient-reported outcome scale for chronic kidney disease

PurposeThe patient-reported outcomes (PROs) measuring patient’s experience and perception of disease are important components of approach to care. However, no tools are available to assess the PROs of chronic kidney disease (CKD). This study aims to develop and verify a PROs scale to evaluate clinical outcomes in CKD patients.MethodsThe theoretical structure model and original item pool were formed through a literature review, patient interviews and references to relevant scales. The Delphi method, classical test theory methods and item response theory method were used to select items and adjust dimensions to form the final scale. Altogether 360 CKD patients were recruited through convenience sampling. CKD-PROs could be evaluated from four aspects, namely reliability, content validity, construct validity, responsibility, and feasibility.ResultsThe CKD-PROs scale covers 4 domains, including the physiological, psychological, social, and therapeutic domain, and 12 dimensions, 54 items. The Cronbach’s α is 0.939, the split reliability coefficient is 0.945, and the correlation of the scores each item and domain’s coefficients range from 0.413 to 0.669. The results of structure validity, content validity and reactivity showed that the multidimensional measurement of the scale met professional expectations. The recovery rate and effective rate of the scale were over 99%.ConclusionThe CKD-PROs scale has great reliability, validity, reactivity, acceptability and is capable of being used as one of the evaluation tools for the clinical outcomes of CKD patients.

Using item response theory in the assessment of the financial well-being scale: an application in Brazil and India

PurposeThis study aims to evaluate Netemeyer and colleagues' much cited financial well-being scale in Brazil and India and compare responses from different demographics. It also compares the results using two analysis techniques, item response theory (IRT) and confirmatory factor analysis (CFA).Design/methodology/approachA total of 994 survey responses from Brazil and 1,081 from India were collected. IRT and CFA models were used to analyse the data.FindingsThe results demonstrate the two-dimensional structure of the financial well-being scale and show that different items are differentially useful in measuring the construct across different groups. These findings may support the scale's future refinement and use in applied studies that will target specific groups (e.g. males, females, younger respondents and older respondents).Research limitations/implicationsThis study serves as an example to others who can explore the advantages of IRT over classical test theory methods to assess the psychometric properties of scales aimed at measuring latent constructs of interest in the field of marketing.Practical implicationsThe correct diagnosis of financial well-being is important to guide interventions by governments and non-governmental entities, as well as by financial institutions interested in better understanding individuals.Originality/valueThe authors show how the identification of the characteristics of scale items provided by the IRT technique allows for a better understanding of its properties and how it can be improved.

ANALYZING THE QUALITY OF MEASUREMENT INSTRUMENTS OF MULTIPLE CHOICE QUESTIONS ON CLASS XI ECONOMICS MATERIAL IN PUBLIC HIGH SCHOOL 3 GORONTALO THROUGH CLASSICAL TEST THEORY AND RASCH MODELS

One form of evaluation of student learning outcomes is the Final Semester Examination. This exam is designed to measure the extent of achievement of educational objectives. A good evaluation must meet several criteria, including good item validity and reliability, a variety of difficulty levels, and the power of differentiation. This study aims to describe the results of a comparative analysis of the quality of measurement instruments in the form of multiple-choice questions using the classical test theory approach and the Rasch model in terms of validity, reliability, difficulty level, and question differentiation. Data were obtained through a website that presents multiple choice exam results of grade XI students at SMA Negeri 3 Gorontalo, consisting of 26 female students and 10 male students. The results showed that in the instrument validity analysis, the Rasch model showed more valid items with a determination category of 0.4 &lt; pt measure corr &lt; 0.8. This means that the Rasch model provides a better analysis compared to the classical test theory analysis. In the reliability analysis, the reliability value of items in the Rasch model is higher but in almost the same category. In analyzing the difficulty level of the instrument, the classical test theory approach shows that the items are in the easy, medium, and difficult categories, so they are still considered capable of measuring students' abilities. However, in the Rasch model, items are only in the very easy, difficult, and extremely difficult categories. In analyzing the power of differentiation, the classical test theory method and the Rasch model have not provided good enough results to identify respondents in several groups based on their level of understanding

Linguistic and Cultural Validation of the Diabetic Retinopathy Knowledge and Attitudes Scale (DRKA) in a Chinese Population.

To linguistically and culturally adapt the 31-item Singaporean Diabetic Retinopathy Knowledge and Attitudes (DRKA) questionnaire for a Chinese population and assess its reliability and validity using classical and modern psychometric theory. A total of 230 patients with diabetic retinopathy (DR) were recruited, and of these, 202 valid responses were analyzed. Rasch analysis and classical test theory (CTT) methods were used to analyze the fit statistics of the Knowledge (n = 22 items) and Attitudes (n = 9 items) scales, including the functionality of the response categories, fit statistics, person and item reliability and separation, unidimensionality, targeting, differential item functioning (DIF), internal consistency, convergent validity, and known-group validity. After revision, both the Knowledge and Attitudes scales were unidimensional and had good measurement precision (Person Separation Index = 2.18 and 1.72) and internal consistency (Cronbach's α = 0.83 and 0.82). While the items in the Knowledge scale aptly targeted participants' ability level, targeting of the Attitudes scale was slightly suboptimal, with items too easy on average for participants' ability level. There were no issues with DIF and item fit, and the scales showed good known-group validity (scores increased as education level increased) and convergent validity (high correlation with the DRKA Practice questionnaire). After a thorough language and cultural verification process, the Chinese version of the DRKA is culturally appropriate and has good psychometric performance. The DRKA questionnaire may be useful to assess patients' DR-related knowledge and attitude level, as well as inform specific education interventions and optimize patients' ability to manage their condition.

Adoption and standardization of attitude scale for polio framed on oversees population from underdeveloped area of West Bengal

Background: Study revealed fear from polio vaccination programme yet not completely has eradicated from the minds of immigrated communities resides in backward areas of Bengal. Due that, pre-established attitude tool was adopted for the formation of new tool for targeted population. Methods: Interestingly, attitude tool found almost similar result after 5 years of its first administration (on minorities). Foundation of tool was set on 3-point Likert type scaling style. Classical test theory (CTT) was followed in which popularity index and discrimination index were taken for item-analysis for pilot study. Results: CTT rejects 6 items and test-retest reliability with high r value 0.94 suggests the tool has excellent reliability. Standardized tool finally comes in 31 items with 9 different dimensions. Conclusions: CTT method as handy method for small sample item-analysis where popularity index and discrimination index used to refine values on which item-total correlation was administered on Pearson’s correlation at 99% confidence interval. Application of the same tool after some years in CTT method in constructing attitude tool have successfully applied.

Detecting anomalous experiences in the community: The Transpersonal Experiences Questionnaire (TEQ).

There is growing recognition of the value of researching anomalous experiences in the general population to aid our understanding of the psychosis continuum. There are key differences in aims, foci and epistemologies of existing measures, with varying utility for specific research designs. This study addresses gaps in the literature by developing a measure of anomalous experiences with utility for longitudinal (time-sensitive) research, and with particular reliability for people towards the upper (high scoring) end of the continuum. An online sample was recruited from the general population to provide questionnaire data for two study parts: (A) item selection and (B) psychometric evaluation. For Part A, both classical test theory and item response theory methods were used to select which items to be included from an initial pool of 57, generated from individuals with persistent anomalous experiences. For Part B, psychometric properties of the resulting measure were evaluated using exploratory and confirmatory factor analysis and tests of reliability and validity. Scores were provided by 532 participants, from which a 19-item scale, the Transpersonal Experiences Questionnaire (TEQ), was developed. The TEQ was found to be a unidimensional scale, with satisfactory internal consistency (0.85), good test-retest reliability and convergent validity. The TEQ can be used as a unidimensional scale to detect anomalous experiences in the general population, with particular reliability for people with higher incidence of these experiences.

The EvMed Assessment: A test for measuring student understanding of core concepts in evolutionary medicine.

Universities throughout the USA increasingly offer undergraduate courses in evolutionary medicine (EvMed), which creates a need for pedagogical resources. Several resources offer course content (e.g. textbooks) and a previous study identified EvMed core principles to help instructors set learning goals. However, assessment tools are not yet available. In this study, we address this need by developing an assessment that measures students' ability to apply EvMed core principles to various health-related scenarios. The EvMed Assessment (EMA) consists of questions containing a short description of a health-related scenario followed by several likely/unlikely items. We evaluated the assessment's validity and reliability using a variety of qualitative (expert reviews and student interviews) and quantitative (Cronbach's α and classical test theory) methods. We iteratively revised the assessment through several rounds of validation. We then administered the assessment to undergraduates in EvMed and Evolution courses at multiple institutions. We used results from the pilot to create the EMA final draft. After conducting quantitative validation, we deleted items that failed to meet performance criteria and revised items that exhibited borderline performance. The final version of the EMA consists of six core questions containing 25 items, and five supplemental questions containing 20 items. The EMA is a pedagogical tool supported by a wide range of validation evidence. Instructors can use it as a pre/post measure of student learning in an EvMed course to inform curriculum revision, or as a test bank to draw upon when developing in-class assessments, quizzes or exams.

Rasch validation and refinement of the Lymphedema Life Impact Scale version 2 in an Italian cohort with secondary lymphedema

Purpose To produce and validate an Italian version of the Lymphedema Life Impact Scale version 2 (LLISv2-It), a tool measuring the impact of lymphedema on health-related quality of life, and investigate its main psychometric characteristics. Methods After translation and cross-cultural adaptation of the LLISv2, we administered it to 156 subjects with secondary lymphedema (upper or lower limb), together with (depending on the limb involved) either the Disabilities of the Arm, Shoulder and Hand questionnaire (DASH) or the Lower Extremity Functional Scale (LEFS). We analyzed the performance of LLISv2-It using Classical Test Theory and Rasch methods. Results Cronbach alpha was 0.89. Item fit statistics indicated that item #12 was underfitting (Infit MnSQ = 1.56; Outfit MnSq = 1.75). Test-retest reliability of the 17-item version (without item #18, not scored, according to the original authors) and of a 16-item version (without #12 and #18—a unidimensional item set) were both excellent (ICC2.1 = 0.93; 95%). The minimum detectable change was 8.9 points. The correlation of LLISv2-It with DASH and LEFS was r = 0.81 and −0.57, respectively. Conclusions The Italian version of the LLISv2 is reliable and valid. To make the tool more psychometrically sound, we recommend to not calculate items #12 and #18 in the total score. IMPLICATIONS FOR REHABILITATION The Lymphedema Life Impact Scale (LLISv2) is particularly useful as a validated condition-specific health-related quality of life measure for patients with chronic edema/lymphedema. We have validated its Italian version (LLISv2-It). We also examined the psychometric properties of the LLISv2 in individuals with secondary lymphedema of either upper or lower limb, using both classical test theory and Rasch techniques, to provide insights for further refinement of the questionnaire. Based on our results, we recommend not to count item #12 in the total score (besides #18, as already suggested by the original authors). Moreover, we determined that the minimum detectable change of the LLISv2 was 9 points. This information is helpful in clinical practice to understand if the change in score is not due to chance (measurement error). Although further research is warranted, the present validation study of the LLISv2 increases the confidence in the metric quality and clinical utility of the scale to assess lymphedema-specific health-related quality of life in individuals with secondary lymphedema of either upper or lower limb.

Parallel exploratory and confirmatory factor analysis of the Hungarian Fear of COVID-19 Scale in a large general population sample: a psychometric and dimensionality evaluation

BackgroundThis study aims to confirm validity and reliability of the Hungarian version of Fear of COVID-19 Scale (FCV-19S) and evaluate its dimensional structure.MethodsCross-sectional survey was carried out in 2021 among Hungarian general population. In addition to classical test theory methods, construct dimensionality of FCV-19S was assessed using EFA with principal axis factoring method and CFA with diagonally-weighted least squares estimation. Fear score was compared in age, gender, educational level, vaccination and infection subgroups.ResultsSignificant differences in FCV-19S mean scores were observed between three subgroups (age, gender, vaccination). Items showed good internal consistency (Cronbach α = 0.88). EFA identified two latent factors (eig = 4.2 and 1.02), though parallel analysis supports the one-factor model. The two-dimensional structure was confirmed by CFA, items 3,4,6,7 correlated with Factor 1 (physiological fear), items 1,2,5 with Factor 2 (emotional fear).ConclusionThe Hungarian version of FCV-19S seems valid and reliable. The EFA identified two-latent factors (emotional and physiological fear), that was confirmed by CFA. The two-factor structure had better model fit, though its’ acceptance is limited.

Evaluation of the Desire to Avoid Pregnancy Scale in the UK: a psychometric analysis including predictive validity

ObjectivesTo evaluate the psychometric performance, including predictive validity, of a UK version of the Desire to Avoid Pregnancy (DAP) scale.DesignProspective cohort study for psychometric evaluation.SettingUK.ParticipantsWomen in the UK aged 15...

Implementing Item Response Theory (IRT) Method in Quiz Assessment System

A quiz is one of the methods to test students’ abilities. Estimation of ability scores can be calculated by using the Classical Test Theory (CTT) and Item Response Theory (IRT) approaches. IRT is more sensitive to item characteristics that can estimate students’ ability. This study implements IRT in a quiz assessment system, compares the results of the CTT and IRT, then analyses the functionality and usability of the built system. The system was tested by 50 respondents using 30 question items with 5 answer choices. The collected data is automatically assessed by the system using CTT and IRT calculations. The results showed differences in ranking between CTT and IRT methods. Using the IRT method, there were 14 respondents who experienced an increase in rank, 19 respondents experienced a decrease in rank, and the remaining 17 respondents did not experience a change in ranking. The functionality of built system was tested with the Blackbox Testing method and it has passed all the test functions. The usability of the system was also tested by involving 23 respondents using the System Usability Scale (SUS) questionnaire and the SUS score showed that the system can be accepted by users.

Facial Disability Index in Adults With Peripheral Facial Palsy: Rasch Analysis and Suggestions for Refinement

ObjectiveTo investigate the psychometric properties of the validated Italian version of the Facial Disability Index (FDI), a patient-reported outcome measure widely used to assess individuals with peripheral facial palsy. DesignMethodological research on cross-sectional data from a convenience sample. SettingOutpatient university rehabilitation clinic. ParticipantsA total of 168 (N=168) outpatients (66% female; mean age, 44±15 years) with peripheral facial palsy of diverse etiology (48% postsurgical, 31% Bell palsy, 8% posttraumatic, 3% congenital, 11% other medical conditions) and severity at the first visit. InterventionsNot applicable. Main Outcome MeasuresThe 2 FDI subscales, physical function (FDI-PHY) and social/well-being function (FDI-SWB), were separately analyzed using classical test theory methods and Rasch analysis. ResultsCronbach α was 0.79 in FDI-PHY and 0.74 in FDI-SWB, while item-rest correlation ranged from 0.36-0.67 in FDI-PHY and from 0.43-0.68 in FDI-SWB. In the FDI-PHY, we deleted 2 underused response categories, rescoring the remaining ones. In the FDI-SWB, some response categories did not function as expected by the Rasch model: their collapse into a 4-level format solved this problem. In each subscale, all items fitted the Rasch model except item 4 (eye tearing/becoming dry) in FDI-PHY that showed an unexpectedly high response variability. The person separation reliability of both subscales indicated that they are useful only for group-level judgments. In both subscales, principal component analysis of the residuals showed unidimensionality and absence of locally dependent items. No significant differential item functioning concerning sex, age, or time from paralysis emerged. ConclusionsOur study demonstrated overall positive psychometric characteristics of FDI except for the functioning of the response categories. We propose a refined version with 4 response categories only and related conversion graphs that may improve the interpretability, feasibility, and metric performance of this tool.