help
help > RE: second-level analysis in Conn
Sep 25, 2015 02:09 PM | Alfonso Nieto-Castanon - Boston University
RE: second-level analysis in Conn
Hi Diana,
Yes, your interpretation is perfectly correct. Connectivity between right Intracalcarine and right FO cortex appears negatively correlated with symptom scores (i.e. higher connectivity between these regions in children with lower symptom scores). In addition there also appear to be an interaction with age, with stronger associations with symptom scores (and perhaps also higher average connectivity as well, see point below) in older children (age group 3) compared to younger children.
There is only one subtlety here that affects the interpretation of the first three bars. Could you please let me know whether the symptom-score covariates are centered or not? If they are centered, did you center them individually within each age-group or globally across all subjects?; and if they are not centered, is a symptom-score equal to zero meaningul/interpretable? This affects the interpretation of the sign and effect-size of the first three bars in your plot -in other words, whether on average the connectivity between right ICC and right FO is positive or negative/anticorrelated-, which in turn affects the correct interpretation of the associations with symptom scores -e.g. higher connectivity in children with lower symptoms may reflect stronger/more-positive connectivity in these children or weaker/less-negative conectivity in these children-. Generally the effect-sizes in your first three bars represent the average level of connectivity within each age-group at the zero level of your symptom score covariate. If the symptom score covariates are centered globally (subtracted the average across all of your subjects) then that zero-level represents the average symptom-level across all of your subjects (and the same level across the three age-groups), if they are centered separately (subtracted the average within each age-group separately) then that zero-level represents the average symptom-level within each age-group separately, and if they are not centered (they represent "raw" symptom scores) then that zero-level represents an actual zero-value of your original symptom variables (again the same level across the three age-groups).
Thanks
Alfonso
Originally posted by Diana Parvinchi:
Yes, your interpretation is perfectly correct. Connectivity between right Intracalcarine and right FO cortex appears negatively correlated with symptom scores (i.e. higher connectivity between these regions in children with lower symptom scores). In addition there also appear to be an interaction with age, with stronger associations with symptom scores (and perhaps also higher average connectivity as well, see point below) in older children (age group 3) compared to younger children.
There is only one subtlety here that affects the interpretation of the first three bars. Could you please let me know whether the symptom-score covariates are centered or not? If they are centered, did you center them individually within each age-group or globally across all subjects?; and if they are not centered, is a symptom-score equal to zero meaningul/interpretable? This affects the interpretation of the sign and effect-size of the first three bars in your plot -in other words, whether on average the connectivity between right ICC and right FO is positive or negative/anticorrelated-, which in turn affects the correct interpretation of the associations with symptom scores -e.g. higher connectivity in children with lower symptoms may reflect stronger/more-positive connectivity in these children or weaker/less-negative conectivity in these children-. Generally the effect-sizes in your first three bars represent the average level of connectivity within each age-group at the zero level of your symptom score covariate. If the symptom score covariates are centered globally (subtracted the average across all of your subjects) then that zero-level represents the average symptom-level across all of your subjects (and the same level across the three age-groups), if they are centered separately (subtracted the average within each age-group separately) then that zero-level represents the average symptom-level within each age-group separately, and if they are not centered (they represent "raw" symptom scores) then that zero-level represents an actual zero-value of your original symptom variables (again the same level across the three age-groups).
Thanks
Alfonso
Originally posted by Diana Parvinchi:
Hi Alfonso,
Could I please get your input on a graph that I have attached. Just to remind you of our analysis, we are looking at the correlation between functional connectivity and symptoms severity in children with autism spectrum disorder (ASD). We have three age-groups of cohorts with ASD. I have attached a screen shot of a graph showing 6 regressors. The first 3 are the age-groups and the last three are the symptom scores within each group. My interoperation of this graph is that functional connectivity between right Intracalcarine and right Frontal Orbital Cortex is negatively correlated with severity of Repetitive Behaviour. Meaning, as connectivity between these two regions increases, repetitive behaviour decreases. Is this correct? Also, if we'd like to report this finding in a manuscript, how could we provide details of this effect (p-value, beta which are specific to this significant effect)? Many many thanks for your help!
Best,
Diana.
Could I please get your input on a graph that I have attached. Just to remind you of our analysis, we are looking at the correlation between functional connectivity and symptoms severity in children with autism spectrum disorder (ASD). We have three age-groups of cohorts with ASD. I have attached a screen shot of a graph showing 6 regressors. The first 3 are the age-groups and the last three are the symptom scores within each group. My interoperation of this graph is that functional connectivity between right Intracalcarine and right Frontal Orbital Cortex is negatively correlated with severity of Repetitive Behaviour. Meaning, as connectivity between these two regions increases, repetitive behaviour decreases. Is this correct? Also, if we'd like to report this finding in a manuscript, how could we provide details of this effect (p-value, beta which are specific to this significant effect)? Many many thanks for your help!
Best,
Diana.