#31468 | AsPredicted

'Predicting cognitive resilience among low-income children - Part 2'
(AsPredicted #31468)

Created:       11/20/2019 06:52 AM (PT)

This is an anonymized version of the pre-registration.  It was created by the author(s) to use during peer-review.
A non-anonymized version (containing author names) should be made available by the authors when the work it supports is made public.

1) Have any data been collected for this study already?
It's complicated. We have already collected some data but explain in Question 8 why readers may consider this a valid pre-registration nevertheless.

2) What's the main question being asked or hypothesis being tested in this study?
We have completed the brain and behavior analyses specified in our first pre-registration. Here, we seek to use demographic data to follow up on these initial results; this pre-registration, though unconventional, is meant to document our plan ahead of time, given the large amount of data and analysis flexibility possible within the ABCD dataset.

Note that we performed only analysis 1a from our initial pre-registration, given the lack of available T2 cognitive test performance data (preventing 1b) and the lack of available ROI results from the resting state data (preventing 2a and 2b). In contrast to our initial prediction, we found that stronger lateral frontoparietal network (LFPN)-default mode network (DMN) between-network connectivity was marginally associated with lower cognitive test scores for kids below the poverty line, a relation in the opposite direction from what has been observed in prior studies involving children from higher-income families. As a post-hoc analysis, we explored whether the predicted association appears in the higher-income sample, finding that it does, in the expected direction. Further analyses confirmed that there is a significant income (above/below poverty) by LFPN-DMN connectivity interaction in predicting cognitive test performance.

Here we seek to further explore this unexpected association by asking whether certain environmental variables determine whether LFPN-DMN connectivity is positively or negatively associated with cognitive test performance across individuals. Specifically, we ask whether greater LFPN-DMN connectivity may be optimal under certain environmental circumstances.

3) Describe the key dependent variable(s) specifying how they will be measured.
Cognitive scores: Z-score and sum of Flanker, DCCS, and WISC matrix reasoning at T1
Resting state connectivity (average strength of correlation across nodes within/between specific brain networks/nodes) between LFPN and DMN, as identified using the ABCD in-house processing
Environmental variables: 31 of ABCD’s pre-scored variables, as listed below.
Home (18): child race/ethnicity (8-level: White, Latinx, Black, Asian, Native American, Native Hawaiian, Other, Mixed); parents’ highest level of education (max of parent, partner); generational status (whether the parent, child, or other family member was born outside of the United States); parent marital status; financial stress (sum of 7 yes/no questions); number of people in home; hours/week child spends in another household; ethnic identification overall; family environment conflict youth report; parental monitoring; parenting acceptance subscale; self-reported discrimination mean; sum of being negatively affected by reported negative life events; sum of being positively affected by reported positive life events; any siblings (yes, no); parent aggressive behavior (ASR raw score); parent intrusive behavior (ASR raw score); parent withdrawn behavior (ASR raw score)
Neighborhood (9): neighborhood safety parent report mean; percentage of population aged >=25 y with at least a high school diploma; income disparity; percentage of occupied housing units without complete plumbing; percentage of families below poverty; percentage of civilian labor force population aged >=16 y unemployed; grand total of crime reports; estimated lead risk; adult violent crime reports
School (4): school environment subscale sum; school involvement subscale sum; school disengagement subscale sum; school setting (9 possible answers, including public school, private school, etc.)

4) How many and which conditions will participants be assigned to?

5) Specify exactly which analyses you will conduct to examine the main question/hypothesis.
Data from children living in poverty will be split into a training (2/3) and testing (1/3) set. We will perform analyses using 5-fold cross-validation on the training set; findings from the training set will ultimately be used to predict cognitive test performance in the testing set.
1) Perform a confirmatory factor analysis to see whether home, neighborhood, and school variables can be separated into distinct factors. If this achieves adequate fit (significantly better fit than a single factor model and CFI>9), perform a linear mixed effects model predicting cognitive test performance from an interaction between LFPN-DMN connectivity and each factor score, with age as a fixed effect and random intercepts for testing site and family (in the case of multiple children from the same family).
2) Perform a ridge regression predicting cognitive test performance from an interaction between LFPN-DMN connectivity and each environmental variable of interest.

6) Describe exactly how outliers will be defined and handled, and your precise rule(s) for excluding observations.
We will exclude children who do not provide data on all three cognitive tests, as in our prior analyses.

7) How many observations will be collected or what will determine sample size?
No need to justify decision, but be precise about exactly how the number will be determined.

As many as usable at T1 (baseline), according to the criteria outlined above.

8) Anything else you would like to pre-register?
(e.g., secondary analyses, variables collected for exploratory purposes, unusual analyses planned?)

As explained in 2, we have already performed one set of analyses on the data. We have not yet performed any analyses on environmental variables.