Robust polychoric correlation

AbstractObjectiveTo demonstrate test–retest reliability (reproducibility) of a new self-administered lifetime diet questionnaire, with a focus on foods relevant to cognitive health in older age.DesignThe reproducibility of dietary recall over four or five life periods was assessed by administering the questionnaire at two time points to an older cohort. The period between questionnaire administrations was 7 weeks. Polychoric correlations measured the association between recall at time 1 and time 2 and the weightedκstatistic measured the level of recall agreement for food groups across the two administrations of the questionnaire.SettingAdelaide, South Australia.SubjectsFifty-two cognitively healthy, older-age, community-dwelling adults completed the Lifetime Diet Questionnaire; mean age 81·8 (sd4·4) years, range 70–90 years.ResultsThe questionnaire showed very good reproducibility in this sample with a mean polychoric correlation coefficient of 0·81 between administration at time 1 and time 2, and an average weightedκof 0·49 for the level of recall agreement between food groups.ConclusionsThe demonstrated reliability of this lifetime diet questionnaire makes it a useful tool to assess potential relationships between long-term dietary intake and later-age cognitive outcomes.

Download Full-text

Asymptotic Robustness Study of the Polychoric Correlation Estimation

Psychometrika ◽

10.1007/s11336-016-9512-2 ◽

2016 ◽

Vol 82 (1) ◽

pp. 67-85 ◽

Cited By ~ 11

Author(s):

Shaobo Jin ◽

Fan Yang-Wallentin

Keyword(s):

Polychoric Correlation ◽

Asymptotic Robustness ◽

Correlation Estimation

Download Full-text

Muscle Dissatisfaction and Muscle-Enhancing Substance Use: A Population-Based Twin Study in Young Adult Men

Twin Research and Human Genetics ◽

10.1375/twin.9.3.431 ◽

2006 ◽

Vol 9 (3) ◽

pp. 431-437 ◽

Cited By ~ 3

Author(s):

Anu Raevuori ◽

Anna Keski-Rahkonen ◽

Richard J. Rose ◽

Aila Rissanen ◽

Jaakko Kaprio

Keyword(s):

Substance Use ◽

Environmental Factors ◽

Twin Study ◽

Genetic Factors ◽

Population Based ◽

Tetrachoric Correlation ◽

Polychoric Correlation ◽

Adult Men ◽

Genetic And Environmental Factors ◽

Best Fit

AbstractIn the population-based FinnTwin16 study, proportions of genetic and environmental factors contributing to muscle dissatisfaction and muscle-enhancing substance use were assessed in 319 pairs of twin brothers: 141 monozygotic (MZ) and 178 dizygotic (DZ) pairs. In addition there were 86 twin individuals from pairs in which only one co-twin responded. Of all respondents, 30% experienced high muscle dissatisfaction. The corresponding proportion of muscle-enhancing substance use was 10%. The subjects were similar in age (23.8 years, 95% confidence interval [CI] 23.76–23.84), body mass index (23.7, 95% CI 23.5–23.9), and waist circumference (84.5 cm, 95% CI 83.7–85.2), independent of their muscle dissatisfaction or muscle-enhancing substance use status and independent of their zygosity. The MZ polychoric correlation for muscle dissatisfaction was .39 (95% CI .17–.58) and .27 for DZ pairs (95% CI .07–.46). The MZ tetrachoric correlation for muscle-enhancing substance use was .65 (95% CI .28–.87) and .56 for DZ pairs (95% CI .26–.78). The AE model, where additive genetic factors (A) accounted for 42% (95% CI .23–.59) and unique environmental factors (E) 58% (95% CI .41–.77) of the liability, provided the best fit for muscle dissatisfaction. The CE model, where common environmental factors (C) accounted for 60% (95% CI .37–.77) and unique environmental factors (E) 40% (95% CI .23–.63) of the liability, provided the best fit for muscle-enhancing substance use. Both genetic and unique (nonfamilial) environmental factors are involved in muscle dissatisfaction in the population. Nongenetic factors (both familial and non-familial) appear to best explain the use of muscle-enhancing substances.

Download Full-text

Bivariate Distributions Underlying Responses to Ordinal Variables

Psych ◽

10.3390/psych3040037 ◽

2021 ◽

Vol 3 (4) ◽

pp. 562-578

Author(s):

Laura Kolbe ◽

Frans Oort ◽

Suzanne Jak

Keyword(s):

Correlation Coefficient ◽

Latent Variables ◽

Ordinal Data ◽

Bivariate Normal Distribution ◽

Polychoric Correlation ◽

Bivariate Distributions ◽

Normal Distributions ◽

Ordinal Variables ◽

Bivariate Normal ◽

Bivariate Normality

The association between two ordinal variables can be expressed with a polychoric correlation coefficient. This coefficient is conventionally based on the assumption that responses to ordinal variables are generated by two underlying continuous latent variables with a bivariate normal distribution. When the underlying bivariate normality assumption is violated, the estimated polychoric correlation coefficient may be biased. In such a case, we may consider other distributions. In this paper, we aimed to provide an illustration of fitting various bivariate distributions to empirical ordinal data and examining how estimates of the polychoric correlation may vary under different distributional assumptions. Results suggested that the bivariate normal and skew-normal distributions rarely hold in the empirical datasets. In contrast, mixtures of bivariate normal distributions were often not rejected.

Download Full-text