Priority-setting approaches in practice
Penelope Mullen, Peter Spurgeon in Priority Setting and The Public, 2018
A few projects used some form of paired comparison, including some examples of constant sum paired comparison, a two-option version of budget pie. In the survey reported by Whitty (1992) respondents were asked to divide £1 million between: (a) ten people with severe problems and 100 people with less severe problems; (b) services for the elderly and services for children; (c) caring for old people in hospital or in their home; (d) hospital services or immunisation; (e) to have children treated in Colchester hospitals or the same treatment in London hospitals. However, apart from (e), the mean responses were fairly close to an even split. Further, between 107 (30%) and 171 (48%) respondents were unable to express a preference and gave equal amounts. It is not clear whether this resulted from resistance to the question, difficulty in responding to a question received aurally, the amount of money involved or whether it was a true reflection of respondents’ values. Bowling (1993) reports the same question as (a) above being posed, but restricting the responses to giving 100%, 75%, 50%, 25% or 0% to the ten people (the remaining percentage going to the 100 people).
Hypotheses Testing versus Confidence Interval
Shein-Chung Chow in Innovative Statistics in Regulatory Science, 2019
Zheng et al. (2019) conducted several simulations for evaluation of the performance of the proposed simultaneous confidence intervals approach as compared to the method of pairwise comparisons. Simulation results indicated that current pairwise comparison methods lacks the accuracy and reliability of each pairwise comparison since each comparison does not fully utilize all data collected from the three groups, and suffers from the inconsistent use of different equivalence criteria in the three comparisons. The simulation results also showed that the methods using the original version and integrated version of simultaneous confidence interval have significantly larger power compared to the pairwise comparisons method and meanwhile can well control the type I error rate. While the method using the least favorable version of simultaneous confidence interval demonstrated the smallest power among the four methods, it was better able to control type I error rate—thus it is a conservative approach which is preferred for avoiding false positive conclusions.
Sensory Analysis Applied to Cosmetic Products
Heather A.E. Benson, Michael S. Roberts, Vânia Rodrigues Leite-Silva, Kenneth A. Walters in Cosmetic Formulation, 2019
The paired comparison test (or directional difference test; ASTM E2164-08) (ASTM, 2008) is a forced choice test (the assessor must choose one answer) that tests for a specific sensory characteristic between two samples (product gloss, intensity of fragrance, etc.). The samples should be in rank order of presentation (order AB and BA repeated the same number of times) and the probability of random success is 50%. Figure 24.3 shows a score sheet for a paired comparison test of soap fragrance intensity. A significant difference between the samples is deemed to occur at 5% significance.
Binaural model-based dynamic-range compression
Published in International Journal of Audiology, 2018
Stephan M. A. Ernst, Steffen Kortlang, Giso Grimm, Thomas Bisitz, Birger Kollmeier, Stephan D. Ewert
One reason might be found in the different methodology used in both the field test and the paired comparison. The paired comparison, as conducted in this study, has the advantage to allow the subject a direct and instantaneous comparison, which gives the possibility to take even small differences into account. At the same time, the subject is forced to decide for a preferred alternative, and a judgement of equal preference is not possible. In contrast, the sequential character of the field test with much longer familiarisation time for each compression approach might yield a better approximation of an unbiased long-term satisfaction. Thus, both tests represent different and complementary viewpoints on quality perception and should both be considered in a comprehensive evaluation of algorithm performance.
Analysis of the importance of on-field covariates in the German Bundesliga
Published in Journal of Applied Statistics, 2018
Gunther Schauberger, Andreas Groll, Gerhard Tutz
In general, in paired comparison data one has to distinguish between objects and subjects. The objects in paired comparisons are the entities that are compared with respect to a certain underlying latent (or non-observable) trait. In football matches, the objects are the teams that are compared with respect to their playing abilities. The subjects are the entities that perform the respective comparison. For example, in marketing studies one often tries to determine the attractiveness of several products by presenting pairs of the products to participants. Then, the participant (who is the subject of the paired comparison) has to decide which product is more attractive to her or him. In football matches, a single match itself or a match-day, respectively, can be seen as the subject that performs the comparison. The distinction between objects and subjects is particularly important when it comes to the inclusion of covariates. Covariates in paired comparisons can vary
Effects of work schedule and period of exposure on changes in urinary chromium and nickel excretion among rotating shift workers in a stainless-steel plant
Published in Chronobiology International, 2019
Hiroaki Itoh, Rui-Sheng Wang, Syou Maki, Qiao Niu, Huizhen Shang, Yougong Su, Zuquan Weng, Hiroyuki Saito, Nobuhiko Miura, Masaya Takahashi
Our study has several strengths. First, the current study followed up on the same individuals to obtain paired data and perform a paired comparison. Second, a relatively large number of subjects participated in the present study and all participants completed the questionnaire, ensuring statistical power. Although they were not randomly selected, they would not have been biologically special. Three and two shifts are popular work schedules. In that sense, the sample should have represented the stainless-steel plant workers in the country to some extent. Third, we used an inductively coupled plasma mass spectrometer, allowing the accurate detection and measurement of urinary chromium and nickel. As ACGIH employed end-of-shift urine levels for presenting biological exposure indices values, we also used spot urine to ensure comparability. Urinary creatinine levels were also precisely measured using an enzymatic method instead of Jaffe reaction, which is known to overestimate (Kume et al. 2018).
Related Knowledge Centers
- Psychometrics
- Psychology
- Law of Comparative Judgment
- Thurstone Scale
- Rasch Model
- Just-Noticeable Difference
- Potentially All Pairwise Rankings of All Possible Alternatives