Association between Personality Traits and Phubbing: The Co-Moderating Roles of Boredom and Loneliness

Round 1

Reviewer 1 Report

The research investigates the interaction of marital and
parent-adolescent subsystem in the overall family structure.
Specifically the study focus on how phubbing ( the action of
neglecting one's physical interlocutor to frequently, more or less compulsively consult one's cell phone or other interactive device) affects familiar relationships, both intra-person ( how mothers’ phubbing conditions mothers-children relationship and how fathers’ phubbing conditions fathers-children relationship) and inter-person
(how mothers’ phubbing conditions fathers-children relationship and how fathers’ phubbing conditions mothers-children relationship).
We consider the topic original and relevant in the field for several reasons: the lack of scientific literature that has previously deepened this theme, especially taking a specific point of view on the relationship binding every element of the familiar system; the use of smartphone is a contemporary factor having recently overturned our
daily life, so we can consider it something on which focus our attention and study to well understand how our relationships are evolved, how they could keep changing and how we can impact on this, giving a psychodynamic and relational point of view to scientific protocol. For the relevance that we attribute to the topic and for the
potential that we see in this research to lead the way to additional insights, we believe some expedients could give greater emphasis to this work and help his comprehension. For example we suggest to switch the expression “wives’ partner phubbing” with “husband's phubbing” and
so “husbands' partner phubbing” with “wive's phubbing”; for the same reason, we suggest to use a more immediate and easier expression to indicate how a phubber mother conditions her relationship with the son  and how a phubbing father conditions his relationship with the
children (intra-person effect). We also suggest to clarify what you mean with the expression “inter person effect”. Another suggestion could be to directly interview children about their relationships with parents instead of exploiting parents’ only point of view. Although the chance to improve features concerning methodology, we found the references are appropriate and tables and figures really useful to
make more intuitive research results, besides the statistical analysis is well compelling.
Finally we estimate that the conclusions are consistent with the evidence and arguments presented and they addressed the main question posed, even if it could be interesting and a subject for further researches, to evaluate the possible underlying factors of these evidences.  

 

Author Response

We would like to thank the reviewers for their time. We provided a response to each comment as illustrated in the attached file

Reviewer 2 Report

Dear authors

 

Thank you for the opportunity to read your paper. I find it a valuable contribution to the current state of knowledge. However, I list some issues you can consider addressing in your revision to improve the paper. I do look forward to seeing the paper published. 

 

• Abstract: “The interaction extraversion by boredom was significantly associated with phubbing at moderate and high levels of boredom” I don't understand this sentence, is extraversion interacting with boredom?

• Abstract: “a new rising phenomenon in our society.” Redundant, I would delete this part of the sentence.

• Abstract: “Phubbing should be addressed in schools, universities and social media to raise awareness about the factors correlated with this behavior and its relations to social interactions.” I don't think a normative implication on school and pedagogy should be part of a medical/psychological study, and you don't mention this in the paper (which I also would not recommend). 

• Manuscript, theoretical part: “where 86% of the population owns a smartphone”. Is this true? Including infants? Or 86% of the adults?

• Manuscript, theoretical part: “This has led to an increase in phubbing as smartphone addictions and phubbing are correlated”. I don't follow the causality, having a smartphone does lead to addiction, does lead to phubbing? I am unsure if the concept of addiction must be introduced, or if so, it has to be argued and contextualized.

• Manuscript, theoretical part: I don't quite follow, that if a partner phubbs the phubbee also phubbs (okay), but that this shall unequivocally explain frequency. I could argue it rather points to a reciprocal dynamic.

• Conclusion: “We were able to establish a significant correlation between marital status” I would reformulate: show/reveal/demonstrate 

• Discussion: The discussion could be richer, when discussing what the findings mean for the subject and/or society. What does it mean that lonely people phubb more, probably they become more lonely etc…

Author Response

We would like to thank the reviewers for their time. We provided a response to each comment. 

Reviewer 3 Report

My comments are mainly methodological. Overall, the manuscript explores an innovative and important topic.

The sample is intentional, not probability, so it is impossible to use statistical significance and, therefore, not regression. Moreover, the sample is significantly skewed, with students and people with higher education predominating. Does this correspond to the structure of the population at this age?

A cronbach's alpha greater than 0.9 indicates that some items in the index are redundant and measure the same thing. In the case of this scale, it would be appropriate to reduce the items that enter it. Leaving aside the unsuitability of the data for regression analysis, the results are poorly described. What is important is the substantive significance. I recommend that the manuscript be revised so that it does not violate the rules for the use of statistical significance. Even descriptive analyses can provide substantial results.

 

Author Response

We would like to thank the reviewers for your time. We provided a detailed response to each comment. 

Author Response File: Author Response.pdf

Reviewer 4 Report

1. In several places you use the term 'correlation' that refers to 'lineal association between variables', while you are building a model, that is, trying to represent causal relations. I suggest to change the 'correlation' to 'relation', when you are referring to these causal relations, and use 'correlation' when you are giving measures based on correlation's coefficients, as correlations do not imply causality, although causal related variables are usually correlated, if the interdependence is linear.  

2. In table 1 there are phubbing means against marital status, but in the description of the variables (table 1), you do not inform about how the sample was distributed between single and married; surely, the second group has a quite low number of cases.

3. In table 3 you indicate 'continuous' variables; really they are indices obtained aggregating Lickert, that is ordinal, variables. Even if these could be approximated by a continuous variable, in fact they are discrete.

4. Again, in the model (table 4),there is a dummy variable related to the marital status; but the reader ignores the frequencies of each group in this variable as there is no previous description.

5. Section 3.2 is titled 'Multivariable analysis', when it is a regression model presented; it is somewhat misleading. 6. Also, some explanatory variables present p-values leading to non detection of the influences on phubbing of the corresponding exogenous variables; if these are excluded, the model would change; but the final model is not presented.

6. Neither is analyzed a probable problem that can affect the conclusions: the multicollinearity, evident in several explanatory variables.

7. In 3.3 section, interactions between several explanatory variables are discussed, but these interactions are not included in the (inexistent) final model (or models); just models on subsamples. But these interactions should be present in a global model, introducing the corresponding additive and multiplicative effects, and selecting the variables finally found significant.

8. In the regression model, no measures of fit are presented; it is necessary; nor a comment about diagnostic checking of the model finally selected, which is absent.

Having your file with a fairly high number of cases, the main changes in the model and in the model with the interactions related to the moderating effects should be easily implemented, and, then, the conclusions should be checked that they are inline with your proposals.

Author Response

We would like to thank the reviewers for your time. We provided a detailed response to each comment. 

Author Response File: Author Response.pdf

Round 2

Reviewer 3 Report

Although the authors have substantially revised the manuscript, my fundamental methodological complaint remains: The sample is intentional, not probability, so it is impossible to use statistical significance and, therefore, not regression. This is because any regression is inference from the sample to the population and is therefore not appropriate for non-probability samples. Similarly, any significance tells whether it is likely that the same results as in the sample will be found in the population. Moreover, the paper remains heavily technical, with little emphasis on the substantive significance of the results.

Author Response

We thank the reviewer for their comments. 

Author Response File: Author Response.pdf

Reviewer 4 Report

In the abstract you start describing the data used; and even some results; this, obviously is not the place to do it; even there is a reference about more research 'in Arab and Western countries'; one can not understand this categorization when studying phubbing in relation to personality. In any case, future work intentions can not be part of the abstract that should focus on what the manuscript is about.

Also you keep talking about 'correlation' between phubbing and personality. Surely your aim is to detect some relations that could be interpreted as causal. A correlation is only a measure of linear interdependence, with no causal implications deduced.

In the title of table 3 appears 'continuous' as a property of the variables used; these variables can not be continuous; in fact they are discrete (as aggregate of ordinal Likert variables). The title 'bivariate analysis' is not informative as they are just Pearson correlations; an appropriate title would be 'Correlation matrix between the variables in the regression model'.

Table 4: again the title 'Multivariate analysis' is not informative; it is just a regression model that is presented, using as dependent variable the phubbing variable (which is used as numerical, as done in previous tables); there is no such thing as 'ENTER model'; maybe the authors did use an automatic selection of explanatory variable using an stepwise algorithm; also they talk about selecting as explanatory variables those who have tested for correlation, in table 3, with p-values under 20%; but this is not appropriate for selecting variables in a regression model; the 'Extroversion' variable in the model has p = 0.445, thus it should be eliminated in the model (as, using a stepwise method implies in relying on statistical testing for including/extracting exogenous variables); even 'Conscientiousness' shows a p of 0.221 (it could change when omitting 'Extroversion', as well as the tests associated to other variables). A model with significant variables should be presented, and not the first obtained.

What is strange is that the authors report a Nagelkerke R², more common in logistic models and, even, not recommended in many statistical environments; on the other hand, the determination coefficient, which has a clear interpretation, is omitted.

The interactions (table 5) between explanatory variables in the model to explain the variability in phubbing should be analyzed using the model with the corresponding interaction terms; not using tests not taking into account the other variables.  The same can be said about table 6, when only couples of variables are used to explain the conditional effects.

If the moderator effects are to be estimated within the model (not the model presented, but the final model with the significant variables and interaction terms to estimate the moderating effects) this has not been done; so the conclusions are based on partial analysis of data, and not, as with multivariate approach, although this word appears in some table.

In the limitations, there is a sentence that is controversial: that a cross-section data can not be used to infer causal relations. Then, what is the purpose of the regression model you postulate? It should be to detect relations that can be interpreted as causal. As you introduce 'Marital status' as explanatory variable of phubbing, it seems adequate to conclude that married people are less prone to phubbing that single; thus concluding that marriage can diminish this problem. You don't need to carry out a temporal study, analysing how phubbing evolves from higher level to lower status when cases of singles analyzed are followed into marriage.

Author Response

We thank the reviewer for their comments. 

Author Response File: Author Response.pdf

Round 3

Reviewer 3 Report

The reworked manuscript remains very technical. The authors insist on using regression and statistical interference despite the non-probability sampling. Moreover, the stepwise regression method is one of the least recommended. I understand the authors' comment, but this is simply a limitation of convenience sampling, and I believe that descriptive statistics have a place in publications. I recommend considering using a different regression method and at least modifying the language of the conclusion to be more concerned with substantive meaning.

 

 

Author Response

We re-did the linear regression using the ENTER model, and we elaborated on that point in the limitations section. We moved the clinical implications paragraph to the conclusion for it to have a substantive meaning as requested by the reviewer. We moved the idea that was in the conclusion to the end of the limitations paragraph (the idea was suggested by Reviewer #1 in the first revision).

Reviewer 4 Report

Just some additional comments.

In table 4 you present a regression model, with 4 explanatory variables, all with low p-values. The confidence interval for the regression coefficients is usually not reported, but, this is minor. The VIF statistics show that the (negative) influence of multicollinearity is very low. And there are points worth of commenting: the determination coefficient is very low, as only 25.2% of the dependent variable is explained by the model, and, thus, nearly 3/4 is not. In the title, 'Multivariate analysis' is redundant (if you are using 5 variables, it is already shown, and regression models, are not usually included in what is kwon as 'Multivariate Statistics'), but you do not include in the table's title what is the dependent variable, that is the main object of building this model. The method of selecting explanatory variables (stepwise) is not to be put in the title (and it is not a recommended method for this task, as you can omit variables relevant, just by the order in which the variables are introduced). Also, the sentence 'variables that showed a p < 0.2 in the bivariate analysis' is out of question: this is not a criteria of a pre-selection of the x variables in any regression model. Then, after this table, the comment 'Numbers in bold indicate significant p-values' is to be ommited.

In the interaction analysis, why the interacting variables are not included as they should in the model?

Author Response

In table 4 you present a regression model, with 4 explanatory variables, all with low p-values. The confidence interval for the regression coefficients is usually not reported, but, this is minor. The VIF statistics show that the (negative) influence of multicollinearity is very low. And there are points worth of commenting: the determination coefficient is very low, as only 25.2% of the dependent variable is explained by the model, and, thus, nearly 3/4 is not.

The VIF values were added to show that there was no multicollinearity between variables (which is something good in our case).

A residual confounding bias is possible since not all factors associated with phubbing were included in the questionnaire; the determination coefficient was low indicating that the variables entered in the model explained 25% of the dependent variable only.

Yes, we have already added in the limitations paragraph the possibility of residual confounding bias since there are other factors that might be associated with phubbing.

In the title, 'Multivariate analysis' is redundant (if you are using 5 variables, it is already shown, and regression models, are not usually included in what is kwon as 'Multivariate Statistics'), but you do not include in the table's title what is the dependent variable, that is the main object of building this model. The method of selecting explanatory variables (stepwise) is not to be put in the title (and it is not a recommended method for this task, as you can omit variables relevant, just by the order in which the variables are introduced).

Also, the sentence 'variables that showed a p < 0.2 in the bivariate analysis' is out of question: this is not a criteria of a pre-selection of the x variables in any regression model. Then, after this table, the comment 'Numbers in bold indicate significant p-values' is to be ommited.

We rephrased the title as follows:

Table 4. Linear regression of factors associated with the generic scale of phubbing score taken as the dependent variable.

The dependent variable is included in the title. We removed the “multivariate analysis” and “stepwise” terms, as well as the sentence about independent variables.

The comment 'Numbers in bold indicate significant p-values' was omitted.

In the interaction analysis, why the interacting variables are not included as they should in the model?

They were actually included; in table 4, we show the model without any interacting variables. In table 5, we show the model with the interacting variables.