This is a series of tables and plots that represent the development of the Eye Tracking Suicide Attention Measure. This is not intended to be a “report” that is self explanatory, but I have included annotations throughout to help with interpretation.

The file is ordered in reverse “chronological order”, meaning the latest analyses are at the top of the file (generally)

The earliest plots (those at the bottom) are all about validation that I am processing the eye tracking raw data correctly.

Then come a series of plots that help confirm that I’ve coded it correctly and to begin to get an image of the distribution of fixations.

The next section begins to look at relationships between the data.

The top of the file includes the most high-level analyses of variable relationship.

The following table is a correlation matrix for the main variables in the analysis to date. Note that the earliest “effect” I computed for the eye tracking data is labled “eye.d” and is Cohen’s d calculated for the difference in time spent looking at “me-death” vs “me-life”. Later, I improved on this measure by controlling for the general bias to look at death stimuli compared to life stimuli, this is reported as “meDeath.z” with the “z” indicating the data are in standardized z scores.

Summary descriptives for main variables
eye.d meDeath.z iat.d phq9 gad7 sbq scs rfl
Min. :-0.92273 Min. :-9.6383 Min. :-0.8419 Min. :10.00 Min. : 7.00 Min. : 4.000 Min. :18.00 Min. : 70.0
1st Qu.:-0.25169 1st Qu.:-1.3023 1st Qu.:-0.5139 1st Qu.:11.25 1st Qu.:10.00 1st Qu.: 5.000 1st Qu.:22.50 1st Qu.:124.0
Median :-0.17637 Median : 0.7904 Median :-0.3124 Median :13.00 Median :13.00 Median : 6.500 Median :32.50 Median :137.0
Mean : 0.02755 Mean : 0.0000 Mean :-0.1920 Mean :17.93 Mean :14.29 Mean : 7.857 Mean :33.64 Mean :136.9
3rd Qu.: 0.51184 3rd Qu.: 1.4290 3rd Qu.: 0.2518 3rd Qu.:25.50 3rd Qu.:18.75 3rd Qu.: 9.750 3rd Qu.:40.00 3rd Qu.:152.5
Max. : 0.72037 Max. : 6.8179 Max. : 0.5727 Max. :33.00 Max. :26.00 Max. :16.000 Max. :69.00 Max. :191.0
NA NA NA’s :2 NA NA NA NA NA





The following table contains a series of effect sizes calculated using slightly different methods. For example there is Cohen’s d based on the time spent on a “me-death” fixation vs a “me-life” fixation and there are odds ratios from the same data. There are effect sizes calculated using the number of fixations, and then again but the time spent in fixation. This table is primarily historical and not of direct interest.





The following series of plots are correlation coefficients computed using Bayesian analysis of the main outcome variables.




The remaining plots are all validation plots