The instrument at the center of the study is the Pica, ARFID and Rumination Disorder Interview–ARFID Questionnaire, abbreviated as PARDI-AR-Q. It grew out of the PARDI, a structured clinician interview developed by an international team that includes Rachel Bryant-Waugh of King’s College London and the South London and Maudsley NHS Foundation Trust, one of the co-authors of the new paper. The questionnaire version condenses that interview into a self-report format, asking respondents about sensory-based avoidance of food, lack of interest in eating or appetite, fear of aversive consequences of eating, and the impact of eating problems on everyday functioning. Because ARFID presentations differ so markedly from person to person, capturing these distinct dimensions separately is critical. A person who avoids fruit because of its texture needs a different clinical conversation from someone who eats reluctantly because they fear vomiting, and a good screening tool must be able to distinguish between them.

Translating a psychological questionnaire is far more involved than swapping words between languages. Concepts such as food neophobia, sensory sensitivity, and appetite must resonate with the everyday vocabulary of the target population, and the response scales must behave in the same way statistically. To establish whether the Turkish PARDI-AR-Q met these standards, a research team led by Makbule Esen Öksüzoğlu of Kastamonu University, together with colleagues from several Turkish hospitals and universities, recruited 1,343 university students aged 18 to 40 from three Turkish universities. The sample was 53.4 percent female, with a mean age of 23.6 years. Participants completed the Turkish PARDI-AR-Q alongside a battery of comparison measures: the Nine-Item ARFID Screen, known as NIAS, the Food Neophobia Scale, the short form of the Eating Disorder Examination Questionnaire, and the Depression Anxiety Stress Scale-21. A subset of 138 participants filled out the PARDI-AR-Q a second time roughly four weeks later, allowing the researchers to test whether scores remained stable over time.

The statistical workup was thorough. The team examined internal consistency, the degree to which items within a scale measure the same underlying construct; test–retest reliability, the stability of scores across weeks; exploratory and confirmatory factor analyses, which test whether the questionnaire’s structure matches the theoretical dimensions of ARFID; measurement invariance across sex, age, ARFID-risk status, and weight status; and both convergent and criterion-referenced validity. Each of these analyses answers a different question about the instrument, and together they provide a comprehensive portrait of how the Turkish version behaves in practice. The study received ethics approval from Uşak University and was conducted in accordance with the Declaration of Helsinki, with written informed consent from all participants.

The headline finding is that the Turkish PARDI-AR-Q performed remarkably well. Exploratory factor analysis supported a four-factor structure mirroring the intended dimensions of the instrument, and confirmatory factor analysis showed excellent fit for a correlated four-factor model, with a comparative fit index of 0.990, a Tucker-Lewis index of 0.985, a root mean square error of approximation of 0.049, and a standardized root mean square residual of just 0.017. For readers unfamiliar with these statistics, values close to 1.0 on the first two indices and low values on the latter two indicate that the observed data align almost perfectly with the proposed model. In plain terms, the questionnaire measures four distinct but related facets of ARFID exactly as designed, and it does so in Turkish as cleanly as psychometricians could hope.

Reliability figures were equally impressive. Internal consistency, measured by Cronbach’s alpha, reached 0.95 for the total score, which is excellent, and ranged from 0.86 to 0.92 across the subscales, spanning good to excellent. The four-week test–retest reliability, expressed as intraclass correlation coefficients, fell between 0.921 and 0.947, meaning that a person’s scores changed very little over a month in the absence of any intervention. This stability matters enormously for research: if a screening tool gives wildly different results from one month to the next, it cannot be used to track symptoms or evaluate treatment. The Turkish PARDI-AR-Q passes that test with room to spare.

Validity evidence came from several directions. The PARDI-AR-Q subscales correlated most strongly with the corresponding subscales of the NIAS, with correlation coefficients ranging from 0.663 to 0.760, exactly the pattern expected if both instruments tap the same constructs. Participants who scored above the established NIAS cut-off for ARFID risk scored significantly higher on the PARDI-AR-Q, providing criterion-referenced support. The researchers also demonstrated scalar measurement invariance across sex, age group, ARFID-risk status, and weight status, which means the questionnaire functions comparably for women and men and for younger and older adults, allowing meaningful score comparisons across these groups. On the descriptive side, endorsement of diagnostic and risk-related items was generally low to moderate: 19.8 percent of participants reported perceiving an eating problem themselves, 18.8 percent said a professional had identified a nutritional deficiency, and 17.9 percent reported that others had noticed an eating problem. These figures suggest that eating-related concerns are far from rare in young adult populations, even if few meet full diagnostic criteria.

Why does this matter beyond Turkey? Psychometric validation studies can seem dry, but they are the quiet infrastructure of mental health science. Without validated tools in local languages, entire populations are effectively invisible to epidemiological research, and clinicians must rely on informal judgment or instruments that were never tested for their patients. ARFID in particular has been under-recognized in adults, with much of the literature focused on children and adolescents. The new study is explicitly the first to validate the PARDI-AR-Q outside English-speaking populations, making the Turkish version a template for adaptations elsewhere. The authors note that the questionnaire can now serve as a short, practical screening instrument for Turkish-speaking adults who may be experiencing ARFID symptoms and who might benefit from further assessment or support, opening the door to prevalence studies and clinical trials in a population of more than 80 million people.

The researchers are candid about the limits of their work. The sample consisted entirely of university students, a group that is young, educated, and not representative of the general adult population, let alone of clinical populations. Future studies, they write, should test the questionnaire in people with diagnosed eating disorders and in more diverse groups, including adults outside higher education. A screening tool validated in students may behave differently in older adults, in people with chronic medical conditions, or in psychiatric inpatients. There is also the inherent limitation of self-report: ARFID can involve limited insight, and some individuals may underreport symptoms, which is why the questionnaire is positioned as a screening and research tool rather than a substitute for a full clinical interview such as the original PARDI.

Even with those caveats, the study represents a meaningful step forward for a disorder that has struggled for recognition. ARFID is not picky eating, and it is not a phase that people simply grow out of; it is a condition that can compromise nutrition, growth, and quality of life across the lifespan. Giving clinicians and researchers in Turkey a rigorously validated, quick-to-administer measure of its core symptoms makes it possible to find the people who have been suffering quietly, often mislabeled as merely fussy or anxious around food. As awareness of ARFID grows worldwide, studies like this one quietly ensure that the science of measurement keeps pace, so that no one’s eating difficulties go unseen simply because of the language they speak.

Subject of Research: Validation of the Turkish PARDI-AR-Q questionnaire for assessing ARFID symptoms in adults

Article Title: Validity and reliability of the pica, ARFID and rumination disorder interview- ARFID questionnaire (PARDI-AR-Q) in Turkish adults

Article References: Esen Öksüzoğlu, M., Günal Okumuş, H., Kaşak, M., Çelik, Y. S., Öğütlü, H., & Bryant-Waugh, R. (2026). Validity and reliability of the pica, ARFID and rumination disorder interview- ARFID questionnaire (PARDI-AR-Q) in Turkish adults. Journal of Eating Disorders. https://doi.org/10.1186/s40337-026-01750-3

Image Credits: AI Generated

DOI: 10.1186/s40337-026-01750-3

Keywords: ARFID, PARDI-AR-Q, eating disorders, psychometric validation, questionnaire, Turkish adults, test-retest reliability, factor analysis, screening tool, food neophobia, mental health assessment, Validity

The conceptual foundation of the CEES is deceptively simple but represents a significant departure from earlier measures. Emotional eating, in this framework, is a multidimensional construct with four distinct dimensions: overeating in response to positive emotions, overeating in response to negative emotions, undereating in response to positive emotions, and undereating in response to negative emotions. Most existing questionnaires capture only the negative-overeating quadrant, which means researchers have historically been blind to people who celebrate with food, who suppress their appetite when distressed, or who cannot eat when they are happy. The Turkish validation study is the first to bring this complete four-dimensional framework to a Turkish-speaking population, filling what the authors describe as a genuine gap in the available assessment instruments.

To test the scale, the research team recruited 897 healthy adults between the ages of 18 and 65. Methodological studies of this kind require careful statistical architecture, and the investigators split their sample into two demographically matched subsamples of 449 and 448 participants. The first subsample was used for exploratory factor analysis, a technique that allows the underlying structure of a questionnaire to emerge from the data rather than being imposed in advance. The second subsample was reserved for confirmatory factor analysis, which tests whether the structure discovered in the first half holds up in an independent group of respondents. This two-step design is considered best practice in psychometrics because it guards against the common pitfall of building and validating a model on the same data.

The exploratory analysis delivered a clean verdict. The Turkish CEES reproduced the expected four-factor structure, corresponding to positive overeating, positive undereating, negative overeating, and negative undereating, and these four factors together accounted for 70.0 percent of the total variance in responses. In questionnaire research, explaining seven-tenths of the variance is an unusually strong result, indicating that the items cluster tightly around the intended dimensions. Internal consistency, which reflects how well the items within each subscale measure the same underlying construct, was excellent across the board, with Cronbach’s alpha values ranging from 0.948 to 0.962. Values above 0.90 are typically regarded as excellent, and these figures place the Turkish CEES among the most internally consistent instruments in the emotional eating literature.

Reliability over time was assessed by asking a subset of participants to complete the questionnaire a second time, two weeks after their first administration. The intraclass correlation coefficients for the four subscales ranged from 0.80 to 0.84, and the total score achieved an intraclass correlation of 0.91, indicating that people’s answers remained stable across the two-week interval. This test-retest stability matters because emotional eating is conceptualized as a relatively enduring behavioral tendency rather than a fleeting mood-dependent state. A questionnaire whose results swing wildly from one fortnight to the next would be of limited use for tracking individuals in research studies or, eventually, in clinical practice.

The confirmatory stage of the study involved a sophisticated comparison of competing statistical models. The researchers tested a standard confirmatory factor model, an exploratory structural equation model, and a bifactor exploratory structural equation model, the latter allowing for both a general emotional eating factor and specific subscale factors simultaneously. The bifactor model that incorporated three additional specific factors for semantically overlapping items demonstrated the best fit to the data, with a comparative fit index of 0.996, a Tucker-Lewis index of 0.994, a root mean square error of approximation of 0.089, and a standardized root mean square residual of 0.043. Fit indices close to or above 0.95 for CFI and TLI, and below 0.08 for SRMR, are conventionally interpreted as indicating excellent model fit. Notably, the bifactor solution supported the four subscale factors alongside only a weak general factor, which carries an important implication: emotional eating in this framework is best understood as four related but distinct patterns rather than a single unified tendency, and researchers should analyze the subscales separately rather than relying on a total score alone.

A further question in cross-cultural psychometrics is whether a questionnaire functions equivalently across different groups. If an instrument measures something different in men than in women, or in younger versus older adults, comparisons between those groups become meaningless. The Turkish CEES team therefore tested measurement invariance across gender, age, marital status, and body mass index categories. At the subscale level, changes in fit indices were generally consistent with metric and scalar invariance across sex, marital status, and age, meaning the scale measures the same constructs in the same way across these groups. The authors do, however, urge caution in two respects: the baseline models showed limited absolute fit, and the comparisons across body mass index categories were less straightforward, so BMI-related group comparisons should be interpreted with care until further evidence accumulates.

Convergent validity, the demonstration that a new scale correlates with other measures in theoretically expected ways, was established through correlations with four established instruments: the 13-item Eating Disorder Examination Questionnaire, the eight-item Depression Anxiety and Stress Scale, the 16-item Difficulties in Emotion Regulation Scale, and the 14-item Perceived Stress Scale. The Turkish CEES scores exhibited the theoretically consistent associations with eating disorder psychopathology, emotional dysregulation, stress, and psychological distress. In other words, people who reported more emotional eating of a given type also tended to report more eating-related problems, greater difficulty managing their emotions, and higher levels of stress and psychological distress, exactly the pattern that theories of emotional eating would predict.

The practical significance of this validation extends well beyond the statistics. Türkiye has a growing community of researchers working on obesity, disordered eating, and the psychological drivers of dietary behavior, but until now they have lacked a Turkish-language instrument that captures the full positive and negative, overeating and undereating spectrum. With the CEES now available in Turkish, studies can distinguish, for example, between someone who eats excessively when anxious and someone who stops eating when celebrating, two behaviors with potentially very different metabolic and psychological consequences. The four-dimensional profile could also help clinicians and dietitians understand why some patients gain weight under stress while others lose it, a distinction that single-dimension scales have obscured.

The authors are careful to note the limits of their work. The validation was conducted in healthy adults, and the clinical utility of the Turkish CEES in patient populations, such as individuals with diagnosed eating disorders or obesity undergoing treatment, requires further validation before the scale can be used confidently in clinical settings. The study also received no specific external funding, and the authors declare no competing interests. Even with those caveats, the study represents a substantial methodological contribution: a rigorously tested, four-dimensional, psychometrically strong Turkish measure of emotional eating, validated in nearly nine hundred adults with state-of-the-art factor analytic techniques. For a field that has long relied on instruments capturing only a fraction of the emotional eating phenomenon, the arrival of a comprehensive Turkish-language tool is likely to reshape how researchers in Türkiye study the tangled relationship between feelings and food.

Subject of Research: Validation of the Turkish version of the Comprehensive Emotional Eating Scale, a four-dimensional measure of emotional eating

Article Title: Psychometric properties of the Turkish version of the comprehensive emotional eating scale

Article References: Psychometric properties of the Turkish version of the comprehensive emotional eating scale. (n.d.). https://doi.org/10.1186/s40337-026-01773-w

Image Credits: AI Generated

DOI: 10.1186/s40337-026-01773-w

Keywords: emotional eating, psychometrics, scale validation, overeating, undereating, eating disorders, emotion regulation, factor analysis, Turkish adults, measurement invariance, test-retest reliability, Journal of Eating Disorders