SAA Analyses

Main takeaway: IPTS variables are VERY short term and the interaction doesn’t pan out when done prospecitvely within a day using EMA, or at all using DD. This doesn’t mean the theory is bad, but points to how it could be a very proximal predictor – OR a correlate (which isn’t so good). Also, we have issues with power given that we are looking at consecutive responses/days.

Testing IPTS Theory in Online EMA

Basic question? Is the IPTS interaction of loneliness (sub for thwarted belonging) X burdensomness associated with contemporaneous and lagged SI?

Model input

IPTS_EMA_INT<-lmer(SI~Burdensome_cent*Lonely_cent+(1|subject),data=ONLINE_EMA)
IPTS_EMA_INT_LEAD1<-lmer(SI_lead~Burdensome_cent*Lonely_cent+(1|subject),data=ONLINE_EMA[DAY==DAY1,])
IPTS_EMA_INT_LEAD2<-lmer(SI_lead2~Burdensome_cent*Lonely_cent+(1|subject),data=ONLINE_EMA[DAY==DAY2,])

Model results

Note: Lags are limited to the same day
  SI SI lead SI lead 2
Predictors Estimates CI p Estimates CI p Estimates CI p
(Intercept) 2.49 1.90 – 3.08 <0.001 2.55 1.92 – 3.18 <0.001 2.78 2.09 – 3.48 <0.001
Burdensome cent 0.50 0.43 – 0.58 <0.001 0.26 0.16 – 0.36 <0.001 0.12 -0.02 – 0.26 0.100
Lonely cent 0.47 0.40 – 0.54 <0.001 0.16 0.07 – 0.25 <0.001 0.21 0.08 – 0.33 0.001
Burdensome_cent:Lonely_cent 0.14 0.08 – 0.19 <0.001 0.01 -0.06 – 0.08 0.728 -0.03 -0.13 – 0.07 0.595
Random Effects
σ2 3.68 3.81 3.80
τ00 4.57 subject 4.86 subject 5.50 subject
ICC 0.55 subject 0.56 subject 0.59 subject
Observations 2889 1739 952
Marginal R2 / Conditional R2 0.089 / 0.593 0.015 / 0.567 0.008 / 0.595

Plot of significant contemporaneous interaction

Testing IPTS thory in Daily Diary

Same question as above, but on a different time scale

Model input

IPTS_DD_INT<-lmer(SI~BURDEN_cent*BELONG_cent+(1|subject),data=ONLINE_DD)
IPTS_DD_INT_LEAD1<-lmer(SI_lead~BURDEN_cent*BELONG_cent+(1|subject),data=ONLINE_DD[(date+1)==date1,])
IPTS_DD_INT_LEAD2<-lmer(SI_lead2~BURDEN_cent*BELONG_cent+(1|subject),data=ONLINE_DD[(date+2)==date2,])

Model results

Note: Lags are limited to consecutive day (or two days), not the next day of data (e.g., if P does data on 10/20, 10/21, and 10/23, there won’t be any lag for 2 days on 10/20 b/c there are no data on 10/22)
  SI SI lead SI lead 2
Predictors Estimates CI p Estimates CI p Estimates CI p
(Intercept) 5.42 5.00 – 5.85 <0.001 5.41 4.95 – 5.88 <0.001 5.59 5.06 – 6.12 <0.001
BURDEN cent 0.26 0.17 – 0.35 <0.001 0.05 -0.06 – 0.17 0.361 -0.03 -0.16 – 0.09 0.597
BELONG cent 0.42 0.33 – 0.51 <0.001 0.05 -0.07 – 0.17 0.385 0.07 -0.06 – 0.21 0.308
BURDEN_cent:BELONG_cent -0.01 -0.07 – 0.06 0.866 -0.00 -0.09 – 0.09 0.960 0.02 -0.09 – 0.12 0.728
Random Effects
σ2 1.51 1.61 1.57
τ00 2.19 subject 2.39 subject 2.73 subject
ICC 0.59 subject 0.60 subject 0.63 subject
Observations 944 652 510
Marginal R2 / Conditional R2 0.084 / 0.627 0.002 / 0.599 0.001 / 0.635