Motherhood in the Making: Key Determinants of Parenthood Motivation in Young Adult Women

Round 1

Reviewer 1 Report

This manuscript investigates an interesting and useful question about parental motivation predictors among young females in Croatia. I find the question interesting and the methods adequate; however, there are a few problems in the way the research is described and especially in the analyses that make it difficult to judge the quality of the analysis or the conclusions drawn from the analysis. Below, I list some specific concerns. Hope this is helpful.

·       

Some of the demographic statistics are unclear. For example, what is “living standards” based on? Was there a question on a demographic survey that tells us about the living standards of the participants? Also, for the religiosity scale, was it made clear to the participants what 1 and 10 stand for? Was there an option for “not religious at all”? Relatedly, how are the 32% and 28% estimates for “not very religious” and “very religious” generated? How are the researchers defining “not very religious” and “very religious”?

·       It would be a good idea to add an appendix or online supplement to the manuscript and add all scales and questionnaires and present individual items that fall under the various factors within scales (for example, what exactly are the survey items that measure the “financial problem” factor under “negative motivation”  in the Childbearing Motivation Scale?) It’s generally good practice to present all research materials.

·       It sounds like data were not collected on a university campus but through social media. Student status, however, was used as an exclusion criteria. I’m not sure I understand the rationale for surveying students only. What is special about students in relation to the research question being investigated here?

·       Results are not explained clearly. For example, on line 198: “significantly lower”. What is this referring to? What is significantly lower than what? Another example is on line 203, where authors seem to be still describing bivariate correlations but they use inaccurate language and describe a “strongest predictor”. This is inaccurate in two ways: first, when describing correlations, nothing is a predictor; second, comparisons that lead to statements such as “strongest” are not analytically made here.

·       Tables 2 and 3 would be more legible and easier to read if abbreviations were replaced with the full titles of constructs.  

·       Tables 4 and 5 do not resemble standard regression models. Please consider including the intercept, measures of variance (SE or CIs), t values, etc. Also, what is F referring to? The authors note that F refers to “change in R2 significance)”. But what is this change? Did the authors run two models and then compare them? It’s a change from what to what?

·       Authors state, on line 217, that attachment dimensions were excluded from the models. But these are, in fact, included in Tables 4 and 5.

·       I’m not sure I understand why there are two sets of regression models. It looks like the models from Table 6 are run step-wise. But there is no discussion of this or any rationale for why this is done. Other decisions also seem arbitrary and are not explained or motivated. For example, why are these specific variables selected to be included in the various models?

·       I’m not familiar with the term “criterion variable”. Is that the same as “outcome variable”?

·       Is “religion” the same as “religiosity” in Table 6? It sounds like it’s a new variable describing religious affiliation. Relatedly, do we know the religious affiliation of participants?

·       Not sure what Figure 1 shows (is it number of people who provided support for each policy?) or what its significance is in relation to other data points.

Discussion

·       How are “moderate levels” defined? Is it the midpoint at the scale?

·       The Discussion does not really get into a “discussion” of the results and implications. The explanations are very brief, unclear, and sometimes repeated word for word (see lines 301- 302 and 313-314.

·       The grounding research on sexual risk and attitudes, mentioned on line 318 is not described in relation to the research. Attachment and emotion regulation are discussed next, but we don’t know at this point how the findings on these constructs relate to research on sexual risk and attitudes.

Author Response

Dear Reviewer 1,

Please see the attachment bellow for detailed answers to your generous feedback.

Thank you on behalf of all authors!

Author Response File: Author Response.pdf

Reviewer 2 Report

- The introduction is very long, perhaps it can be summarized. Introduction would benefit from schemes that relate all the theories and frameworks that are explained.

- A flow chart should be reported to outline the number of participants recruited, excluded and finally included in the study.

- The statistical section for data analysis and processing is missing.

- The standard error does not appear in the model tables. Nor is it explained how the models were built.

- Material and methods. What do you mean by living standard (line 133) and by very religious (line 134-135)?

- It is not described what PM, NM or SD is (Line 186). It is not understood how the "z" scores were obtained (line 189).

Author Response

Dear Reviewer 2,

Please see the attachment bellow for all our answers regarding your insightful input.

Best wishes on behalf of all authors!

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Although a lot has improved in this version, the presentation of the analysis is still incomplete and at times, inaccurate. Please see detailed comments below. 

There seems to be a misunderstanding about what the step-wise models are telling us. First, there are two sets of step-wise models: one set includes only data from single participants, and the other set includes only data from married participants. Now, within each set, what the model summary in tables 6 and 7 are telling us is not the "full regression model"; Tables 6 and 7 give us "model comparison" output. That is, for example, the F change and p values corresponding to "step II" on the top panel of Table 6, are telling us that adding Attachment to a model that already includes Religiosity, Contraception, Age, and Sexual Orientation does not improve the fit of our model, suggesting that Attachment does not play a role in predicting variability in positive motivation among female students who are single. The logical next step for cases where the addition of a variable does not include the fit of the model is to drop that variable and then test another model with another variable; not, as the authors have done, to keep the variable that does not increase model fit (attachment) and then add other variables (emotional abilities). Therefore, the final model for positive motivations among single students should probably exclude attachment and include emotional abilities (if that increases the fit of the model compared to the original model that includes only religiosity, contraception, age, and sexual orientation. The same logic applies to all cases of step-wise regression analysis. For example, in the case of married/in a relationship students, for negative motivations, adding emotional abilities to the model that includes demographics and attachment does not significantly improve the fit of the model (look a the p value corresponding to step III); therefore, the second model (step II) is the one that conclusions should be based on. 

The second problem with the step-wise models is that they are sufficiently explained. For example, on line 267, the authors say "the regression model for the positive motivation criterion was statistically significant, as it explained 27.7% of the variance." This is already an inaccurate and incomplete statement for several reasons. 1. which model? there are three models for the positive motivation outcome and the authors don't make it clear which they think is the best model and the one to base conclusions on. 2. which subgroup? there are two subgroups (married/in relationship; single). Then, the authors go on to say "the most significant predictor variable was religiosity, and management of emotions was the only EI ability that altered the effect size". Again, here, the information is incomplete. I'm sure the authors have looked at model details and concluded that "religiosity" is the most important predictor. But we don't have that information because we don't have the full regression model results for the best-fitting model. We don't know, for example, what the coefficient, SE, t, and p is for "religiosity" in the model that includes religiosity, contraception, age, sexual orientation, emotional abilities.  Again, it's important to acknowledge that Tables 6 and 7 present "model comparison" output; not "model output".

 

I suggest the authors more carefully review the analyses and clarify the purpose of the step-wise regression models. Then, to support claims about specific predictors, the full model parameters should be included in addition to the model comparison output. 

Author Response

Dear Reviewer,

I hope our latest Manuscript version answers your doubts and clarifies the purpose of the step-wise regression models.

In this version, we conducted 4 hierarchical regression, forward selection method with three predetermined set of predictors: socio-demographic variables, attachment and emotional competence dimensions. In that way, adding the predictors, we are given an information on increment value of selected variables beyond the demographic data. Another advantage of stepwise method relies on its usage as an exploratory analysis whenever there is no existing theoretical foundation or sufficient theoretical models to draw from, such as this survey.

We highlighted the limitations, namely inflated Type I error (Harrell, 2015). The results are demonstrated in Tables 6 and 7, along with Tables A2-A5 in the Appendix section, displaying regression coefficients for each dependent variable and subsample. We opted for placement in the Appendix section in a desperate attempt to preserve out reader's attention, since this paper was already at 19-page mark.

I hope our response aligns with your expectations regarding the review, and I sincerely appreciate your valuable input. Thank you for your thoughtful feedback.

Author Response File: Author Response.pdf

Round 3

Reviewer 1 Report

I appreciate the efforts to clarify the models. Even though the tables look more clear now, the results still need more clarification and accurate descriptions. 

I appreciate the attempt to clarify the step-wise regressions. Even though the tables are now fixed, there is now zero information about what these models are telling us. In the results section, we just read that these models were carried out! But what are they telling us? Tables 6 and 7 need to be interpreted for the reader. 

Also, whenever a claim is made based on patterns of results, the specific piece of information that supports the claim needs to be added. For example, line 248, "the most significant predictor variable was religiosity....". The coefficient, p-value, and confidence interval for "religiosity" need to be added here (or alternatively, you can add "see Table 4). The rest of that sentence says, "...management of emotions was the only EI ability that altered the effect size". Then there is a regression coefficient that corresponds to Table 4 (linear regression on positive motivation). In the context of the linear regression, it is not clear why the authors are using the word "altered". What effect size was altered? And how was it altered? 

Author Response

Dear reviewer,

we have made the changes as you suggested. Namely, we have added β and p values for religiosity (lines 251 and 256).

We added a few more alterations, corrections and details that are marked red. We also added a few lines to elaborate added tables in the Appendix (see line 417-447).

Regarding you comment about the lack of interpretation at times, we emphasized on various different sections the absence of comparable studies and some conclusions were necessary made ''by proxy''. Also, we offered an extended future research ideas to further investigate this interdisciplinary social issue, as recommended.

Your guidance helped improve this Manuscript immensely, and for that we wanted to express our gratitude.

Sincerely,

Dario

Author Response File: Author Response.docx