Fig 1.
Differential correlations due to different subgroup variances.
(A) Density plot of a skewed variable (symptom A) partitioned into 4 quartile groups, demonstrating differential variances across the subgroups. (B) Correlations between 2 skewed variables (symptoms B and C) that are correlated with symptom A and with each other. Across subgroups the correlations vary as a function of different subgroup variances. Subgroup specific correlation coefficients are shown.
Fig 2.
Distributions of the variables.
Histograms of the variable scores’ distributions with the variable scores on the X-axis and frequencies of measurements on the Y-axis.
Table 1.
Distributional parameters of the variables by severity subgrouping.
Table 2.
Pearson product-moment and Spearman rank correlations of the variables.
Fig 3.
Floor effects of negative affect (NA) and paranoia (PAR) by SCL-total score level.
Proportions of mental state (NA and PAR) scores at the floor of their scales by overall severity of psychopathology (SCL-total score). The red curve represents the proportions of floor scores for NA, whereas the blue line represents the same for PAR (scale on the left). Numbers of measurements by overall severity are displayed in the black line (scale on the right).
Fig 4.
Boxplots of the distributions of the negative affect score across severity subgroups.
The “boxes” indicate the interquartile ranges (IQR). The median observations are indicated by thick lines in the boxes. The “whiskers” extend to the highest (and lowest) observations not further away from the box than 1.5 times the IQR. Outliers are represented by small circles.
Table 3.
Mental states (negative affect, positive affect and paranoia) at moment t predicted by mental states at moment t-1, by SCL-severity.
Method: inverse Gaussian regression analysis. Subgroup-specific regression (B) coefficients with 95% confidence intervals.