China
Africa
Explore chapters and articles related to this topic
Model Estimation and Evaluation
Published in Douglas D. Gunzler, Adam T. Perzynski, Adam C. Carle, Structural Equation Modeling for Health and Medicine, 2021
Douglas D. Gunzler, Adam T. Perzynski, Adam C. Carle
Ordered-categorical variables in research samples may show a floor effect or ceiling effect. A floor effect occurs when a high proportion of individuals endorse the minimum score on the observed variable. In contrast, a ceiling effect occurs when a high proportion of individuals endorse the maximum score on the observed variable. For an example, the frequency counts in the first category (endorsing the category for “not at all”) were high relative to the other categories in each of the nine items for the PHQ-9 in a sample of 644 hemodialysis patients, thus exhibiting a floor effect. As a consequence, the presence of floor and ceiling effects reduce variability in the reported data with most of the values at an extreme value. Robust WLS or ML with numerical integration can be used to analyze ordered-categorical data with a floor or ceiling effect.
Experiments
Published in Louis Cohen, Lawrence Manion, Keith Morrison, Research Methods in Education, 2017
Louis Cohen, Lawrence Manion, Keith Morrison
Similarly Torgerson and Torgerson (2008) alert researchers to ‘ceiling and floor effects’ (pp. 147–8). A ‘ceiling effect’ is where a test is too easy for the participants, whilst a ‘floor effect’ is where it is too difficult. This rehearses the need not only to pilot the test but to ensure that item discriminability and appropriate scaling have been addressed (see Chapter 27 of the present volume). The authors note that if there is a ceiling or floor effect then it may lead to the false conclusion that an intervention has not worked.
A matter of convergence: Classical and modern approaches to scale development
Published in Francis Guillemin, Alain Leplège, Serge Briançon, Elisabeth Spitz, Joël Coste, Perceived Health and Adaptation in Chronic Disease, 2017
The presence of minimum (floor) and maximum (ceiling) scores has always been understood to be problematic and is often reported (Alonso et al., 1998). It can constrain the ability of a scale to both show the full range of the construct relevant to a particular sample, as well as to show change (for example, it is unknown how far above the ceiling an individual may be, and thus despite showing a decline in the construct being measured, they may nevertheless still be above or at the ceiling). In educational testing this would be equivalent to giving students a test that was too easy (ceiling effect) or too hard (floor effect). It would be seen as a poor test, but in health this may not be so uncommon. For example, screening a population for depression may give a left-skewed effect (low scores) with a sizeable floor effect. In the simulated 20-item scale, only one person out of 300 was at the extreme. The Rasch analysis has the ability to graphically show the relationship between the distribution of persons and items on the same interval scaled metric (Figure 8.4). Thus, the average person ability (−0.030 logits, upper level of graph) matches almost exactly the average difficulty of the item set (centered on zero logits), and consequently the scale is deemed to be perfectly targeted at the people (it was simulated to be so). Logits, or log odds ratio is the unit of measurement derived from modern test theory, but the invariance requirement is only available within the Rasch model, so giving interval scaling (Fischer and Molenaar, 1995).
The Dutch language version of the Pain Disability Index (PDI-DLV): psychometric properties in breast cancer patients
Published in Physiotherapy Theory and Practice, 2023
Elien Van der Gucht, Lore Dams, Koen Bernar, Tessa De Vrieze, Vincent Haenen, An De Groef, Lode Godderis, Bart Morlion, Mira Meeus, Nele Devoogdt
The second limitation relates to the fact that psychometric properties are population- and situation-specific. In contrast to the use of the PDI-DLV as an outcome measure in research, it can be argued that its clinical application will target patients with pain and at different time points after surgery. As only 22% of the subjects in the reliability group rated their average pain as 30 mm or more on the Visual Analog Scale for pain (0–100) (which is considered mild) and we only looked at the psychometric properties at 1 year after surgery, our results warrant caution for clinical use in breast cancer patients with pain at different time points. In addition, pain and disability scores differed slightly between B2 and the other subgroups, adding to the caution with which these findings should be generalized. Furthermore, the overall low pain- and pain-related disability scores in our sample could have possibly resulted in a floor effect. Future research should therefore assess the psychometric properties of the PDI-DLV in a sample of at least 50 participants with higher pain and pain-related disability scores and assessed at different timepoints after surgery. Additionally, it would be valuable to investigate the PDI-DLV’s ability to detect clinically relevant changes in pain-related disability by looking at its responsiveness and minimal clinically important change. A floor effect, on the other hand, may provide a challenge when addressing responsiveness because improvements cannot be evaluated in those patients already presenting with low pain-related disability.
The validity of the Danish version of the Fibromyalgia Impact Questionnaire – Revised applied in a clinical setting: a Rasch analysis
Published in Scandinavian Journal of Rheumatology, 2023
PH Duhn, K Amris, H Bliddal, EE Wæhrens
Examination of the overall separation and reliability indices was performed to evaluate the precision and reproducibility of the item difficulty estimates and estimates of personal disease severity. The person separation index scrutinizes whether the items can discriminate between different levels of participant disease severity (17). The separation indices should be > 2.0 to achieve a desired reliability coefficient of ≥ 0.8 (range 0.0–1.0) for replicability of participant disease severity measures (5). The item separation index is used to confirm the item hierarchy (i.e. construct validity) (17). The item reliability should be > 0.9 (17). Reliability coefficients have a ceiling of 1.0, whereas separation coefficients and indices have no ceiling. An item–person distribution map will be used to demonstrate any floor or ceiling effects. A floor effect occurs when items are lacking at the bottom of the scale to match the worst performing participants, and a ceiling effect occurs when items are missing at the top of the scale.
The Efficacy of Hypnotic Analgesic Suggestions in Chronic Nociplastic Pain: A Randomized Controlled Trial
Published in International Journal of Clinical and Experimental Hypnosis, 2023
Anna Dominika Kaczmarska, Michał Mielimąka, Krzysztof Rutkowski
Although our study focused on comparing the efficacy of the two interventions, some additional findings are worth noting. First, the patients were very eager to participate in the study, and then they were so involved that all completed procedures fully, which is a real rarity. Therefore, the hypnosis treatment had to be attractive and not burdensome for the patients who showed an unprecedented degree of compliance. This is a good premise for both interventions to be effective in the long run. Moreover, we also found that the interventions led to greater pain reductions in the participants who were not taking analgesics and antispasmodics during the study. This finding can be explained by the floor effect, as baseline scores for pain intensity and pain quality were low in the patients who were taking medications. Nevertheless, it is worth mentioning that hypnosis itself reduced the intensity and quality of pain to a level similar to that achieved with medications. Determining the relative efficacy of hypnosis is particularly important in nociplastic pain conditions where nonpharmacological therapies are prioritized, as most medications provide only a modest benefit and are often associated with side effects (Fitzcharles et al., 2021; Lumley & Schubiner, 2019). Our observation requires confirmation, but it is already useful in designing future research.