Table of Contents


Key Terms

Survey: Data collected from the initial survey including topic ranks, frequency ratings, and intensity ratings.

Scanner: Data collected in the fMRI scanner one week after the survey. Includes feeling ratings, intensity ratings, and reaction time.

Event type: A statement can be one of three types: a worry statement, a rumination statement, or a neutral statement.

Frequency rating: Participant rates, on a scale of 0-4, how frequently they think about the current statement.

Feeling rating: Participant rates the valence of the current statement, 1 being negative, 2 being neutral, and 3 being positive.

Intensity rating: Participant rates, on a scale of 0-4 (survey) or 1-4 (scanner), how intensely they feel when thinking about the current statement.

Topic Rank: Participant ranks a set of topics (i.e school, finances, health) from 1 to 12 based on the likelihood that they would worry about the topic, 1 being the topic they are most likely to worry about.

Reaction Time: Time it takes for participant to rate a statement in the scanner, in milliseconds.


Scanner Analyses

Rating Responses

Descriptive Statistics

neutral (N=585) rum (N=780) worry (N=780) Total (N=2145) p value
FeelingRate.RESP < 0.001
- N-Miss 12 23 15 50
- Mean (SD) 2.216 (0.537) 1.297 (0.564) 1.154 (0.415) 1.496 (0.675)
- Range 1.000 - 3.000 1.000 - 3.000 1.000 - 3.000 1.000 - 3.000
IntensityRate.RESP < 0.001
- N-Miss 6 9 9 24
- Mean (SD) 1.259 (0.550) 2.341 (0.908) 2.684 (0.899) 2.170 (1.005)
- Range 1.000 - 4.000 1.000 - 4.000 1.000 - 4.000 1.000 - 4.000



Feeling Ratings by Event Type

Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
EventType 416.975 208.487 2 2054.235 891.852 0 
## 
##  Welch Two Sample t-test
## 
## data:  Response by EventType
## t = 5.6293, df = 1389.8, p-value = 2.185e-08
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  0.09315381 0.19280124
## sample estimates:
##   mean in group rum mean in group worry 
##            1.297226            1.154248

Worry and rumination statements were rated more negative than neutral statements in terms of Feeling rate. Worry statements were rated significantly more negative than rumination statements, on average.


Intensity Ratings by Event Type

Over all 3 feeling ratings: negative, neutral, and positive

Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
EventType 45.141 45.141 1 1502.061 72.479 0

When measured across all worry and rumination statements, worry statements were rated significantly higher in intensity than rumination statements, on average.




Over just the negative feeling ratings

Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
EventType 20.137 20.137 1 1191.575 36.49 0

When measured across only worry and rumination statements that were rated “negative”, worry statements were rated significantly higher in intensity than rumination statements, on average.






Reaction Time (Scanner)

Descriptive Statistics (RT)

neutral (N=585) rum (N=780) worry (N=780) Total (N=2145) p value
FeelingRate.RT 0.370
- Mean (SD) 1026.414 (520.609) 1040.940 (540.083) 1004.187 (490.926) 1023.614 (517.325)
- Range 0.000 - 2987.000 0.000 - 2983.000 0.000 - 2981.000 0.000 - 2987.000
IntensityRate.RT 0.659
- Mean (SD) 1030.660 (518.506) 1034.144 (534.219) 1011.136 (531.376) 1024.827 (528.797)
- Range 0.000 - 2667.000 0.000 - 2916.000 0.000 - 2932.000 0.000 - 2932.000



FeelingRate Reaction Times by Event Type

Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
EventType 1079168 539584.2 2 2054.156 2.548 0.078

Reaction times for Feeling Ratings were not significantly different between Event Types (neutral, worry, and rumination).



IntensityRate Reaction Times by Event Type

Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
EventType 233814 116907 2 2104 0.491 0.612

Reaction times for Intensity Ratings were not significantly different between Event Types (neutral, worry, and rumination).




Reaction Time over duration of the task

  ReactionTime
Predictors Estimates CI p
(Intercept) 1128.57 1065.54 – 1191.59 <0.001
TaskTrialNum -3.73 -4.65 – -2.80 <0.001
Random Effects
σ2 239541.71
τ00 Subject 31373.11
ICC 0.12
N Subject 39
Observations 4290
Marginal R2 / Conditional R2 0.013 / 0.127


Reaction Time became faster over time.




Survey Analyses

Factors affecting Survey Ratings

Ratingtype = frequency or intensity, RNTtype = worry or rumination
Sum Sq Mean Sq NumDF DenDF F value Pr(>F)
Topicrank 36.175 36.175 1 1914.117 107.903 0.000
RNTtype 12.523 12.523 1 1915.219 37.354 0.000
Ratingtype 0.075 0.075 1 1913.776 0.223 0.637
Topicrank:RNTtype 1.208 1.208 1 1914.119 3.602 0.058
Topicrank:Ratingtype 0.054 0.054 1 1913.776 0.160 0.689
RNTtype:Ratingtype 2.504 2.504 1 1913.776 7.470 0.006
Topicrank:RNTtype:Ratingtype 2.784 2.784 1 1913.776 8.303 0.004


Frequency Ratings by Topic Rank

As the topic rank increased, the average Frequency Rating decreased, especially for worry statements.



Intensity Ratings by Topic Rank

As the topic rank increased, the average Intensity Rating decreased.



Cognitive risk and Symptom Measures

Score Distributions



Correlations: WRG Survey and Time 1 Cog Risk and Symptom Measures


Survey v. Scanner Responses

Survey v. Scanner Intensity Ratings

## 
##  Pearson's product-moment correlation
## 
## data:  fr2$Response and fr2$qualInt
## t = 13.004, df = 1539, p-value < 2.2e-16
## alternative hypothesis: true correlation is not equal to 0
## 95 percent confidence interval:
##  0.2689420 0.3589472
## sample estimates:
##       cor 
## 0.3146517

Scanner Intensity Ratings and Survey Intensity Ratings are significantly correlated.

Trial by trial

Df Sum Sq Mean Sq F value Pr(>F)
qualInt 1 129.007 129.007 169.113 0
Residuals 1539 1174.021 0.763 NA NA

Compared by trial, Survey and Scanner Intensity Ratings are significantly correlated.

Multilevel Model

Intraclass correlation for Scanner Intensity Rating is 11.27%, indicating a lot a variance within-person around their mean

Also, when regressing Intensity Ratings on Time/Trial Number, there is enough variability to model the intercept as random, but there is not enough variability over time to model the slope as random.

Multilevel model predicting Scanner Intensity: including a random intercept and random slopes of survey intensity and reaction time.

This model demonstrates significant:
a fixed effect of time
a fixed between-person effect of survey rating
fixed within-person effects of survey rating and reaction time
  Response
Predictors Estimates CI p
(Intercept) 2.51 2.38 – 2.64 <0.001
C.TaskTrialNum -0.00 -0.01 – -0.00 0.015
cwc.surv.int 0.35 0.25 – 0.44 <0.001
cprsmn.surv.int 0.28 0.05 – 0.51 0.015
cwc.rt.scaled -0.11 -0.16 – -0.05 <0.001
cprsmn.rt 0.00 -0.00 – 0.00 0.453
cwc.surv.int *
cprsmn.surv.int 		0.06 -0.13 – 0.25 0.545
Random Effects
σ2 0.56
τ00 Subject 0.15
τ11 Subject.cwc.surv.int 0.04
τ11 Subject.cwc.rt.scaled 0.01
ρ01 0.25
-0.54
ICC 0.24
N Subject 39
Observations 1541
Marginal R2 / Conditional R2 0.101 / 0.320


C.TaskTrialNum: centered trial number (so zero equals trial 27.5)

Cwc.surv.int: centered within cluster/within person survey int ratings (0 equals each SUBJECTS avg survey int rating)

Cwc.rt.scaled: centered within cluster/within person scanner RT for int ratings (0 equals each SUBJECTS avg RT [divided by 500 to make the numbers smaller])

Cprsmn.surv.int: GROUP centered avgs for survey int ratings (0 equals the avg of the person avgs across the group)

Cprsmn.rt: GROUP centered avgs for scanner RT for int ratings (0 equals the avg of the person avgs across the group)


A plot showing both within-person (green) and between-person (turquoise) effects of survey intensity of scanner intensity ratings.