Prevalence, Symptom Profiles, and Correlates of Mixed Anxiety–Depression in Male and Female Autistic Youth
Round 1
Reviewer 1 Report
Comments and Suggestions for Authorsthank you for revising your manuscript.
one minor issue:
"was undertaken by JASP" - maybe better to write: Bayesian statistics was done in the statistical software JASP (JASP Team, 2024).
and in reference list: JASP Team (2024). JASP (Version 0.19.0)[Computer software].
I congratulate for using Bayesian statistics (and likely the default values in JASP; as the prior is not reported). But in your case, it does not - as wished - clarify the "no sex difference" issue. The data is either too noisy or N too small to state that there is no sex difference.
That is, a BF10 = .429 means that BF01 = 2.331 and hence below 3 (and BF10 larger than 1/3) and this is commonly interpreted as anecdotal.
regarding the other three BF10 values: only factor 3 with BF10 = .209 lends support for the null hypothesis by common interpretation of the Bayes factor.
this should be addressed in the discussion. It does not reduce the impact of your work. Quite the opposite, it may spur replication attempts.
Author Response
Comment 1:
"was undertaken by JASP" - maybe better to write: Bayesian statistics was done in the statistical software JASP (JASP Team, 2024).
Author response: We have now done that.
Comment 2: and in reference list: JASP Team (2024). JASP (Version 0.19.0)[Computer software].
Author response: Done.
Comment 3: I congratulate for using Bayesian statistics (and likely the default values in JASP; as the prior is not reported). But in your case, it does not - as wished - clarify the "no sex difference" issue. The data is either too noisy or N too small to state that there is no sex difference.
That is, a BF10 = .429 means that BF01 = 2.331 and hence below 3 (and BF10 larger than 1/3) and this is commonly interpreted as anecdotal.
regarding the other three BF10 values: only factor 3 with BF10 = .209 lends support for the null hypothesis by common interpretation of the Bayes factor.
this should be addressed in the discussion. It does not reduce the impact of your work. Quite the opposite, it may spur replication attempts.
Author response: We have now added the following comment to the Results:, end of section 3.2.2:
That is, whilst independent samples t-tests indicated no gender effect for any factor (p = 0.301 / 0.114 / 1.000 respectively), Bayesian independent samples t-tests were also performed to estimate the likelihood of gender effects for each of the three main factors via JASP (JASP Team, 2024). Assuming the prior is a Cauchy distribution centred at zero with a width of 0.65 (corresponding to an 80% probability of an effect size between -2 and 2), anecdotal to moderate evidence for no gender effect (i.e., evidence for H0) was found; BF10 = 0.361 for Factor 1, 0.683 for Factor 2, and 0.225 for Factor 3..
We have also added the following comment to the Discussion, end of section 4.1:
Finally, whilst gendered effects can never be entirely ruled out, it does not appear to be a noteworthy factor in these results. The Bayes Factor Robustness check (JASP Team, 2024) indicated that the only evidence for a gendered effect would require a Cauchy distribution width less than 0.27, and that would only apply to a single factor (factor 2). All other assumptions pointed towards stronger evidence for no gender effect, however it should be acknowledged that Cauchy prior widths between 0.4 (factor 3) and 1.6 (factor 2) would be required to reach at least moderate evidence in favour of the null hypothesis (i.e. BF10 scores below 1/3).
Reviewer 2 Report
Comments and Suggestions for Authors I read the last version, the manuscript has been sufficiently improved and now warrants publication in NeuroSci.
Author Response
Thank you.
Reviewer 3 Report
Comments and Suggestions for AuthorsThank you for an opportunity to review the manuscript in its revised form. In my view, it is ready for publication
Author Response
Thank you.