Revising the Original Antonovsky Sense of Coherence Concepts: A Mixed Method Development of the Sense of Meaning Inventory (SOMI)

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

8 August 2024

 

The Editor

Sexes

 

The Authors

Revising the original Antonovsky Sense of Coherence concepts: A Mixed Method development of the Sense of Meaning Inventory (SOMI)

 

Thank you for the opportunity to provide some input into the above manuscript. The authors set out to revise the SOC-13, largely to deal with its problematic factorial structure which have not been supported in the literature. As a frequent user of SOC-13 I appreciate this endeavour. Overall, I found the manuscript admirable, especially phase 1. However, the statistical analysis in phase 2 requires some attention.

 

1.      While the manuscript is overall well written and the authors express themselves very well there are some niggling language and editorial issues that requires attention:

·        Page 1, line 40: “Sense of Coherence (SOC) is defined SOC as "a global…” I think the term SOC as underlined is misplaced.

·        Page 3, lines 101-102: “has been the development of the revised Sense of Coherence Scale (SOC-R) Bachem and Maercker (2016).” The referencing is awkward. The authors use APA format and it should thus be (SOC-R: Bachem and Maercker, 2016) or “(SOC-R) by Bachem and Maercker (2016).” As an aside I am aware that MDPI journals uses Vancouver formatting and if the paper is accepted that citation style would fix this issue.

·        Page 3, line 124: “SOMI” - prior to this the term has not been defined.

·        I am not familiar with the term “known group reliability”

·        I stand to be corrected but I do not think that it is “Physicians Health Questionnaire” but rather “Patient’s Health Questionnaire”.

·        Page 4, line 165: “Cronbach alpha reliability for our study is .85.” The “is” should be “was.”

·        If the authors are requested to revise the paper, I would suggest that the anonymizing is not required as MDPI journals do not use a blinded review process.

·        Page 5, line 235: “About two-thirds of the sample had some college, and half were working”. Is the word “education” missing after college?

·        There are instances where only the term “SOC” is used. For consistency the authors should clearly indicate that it refers to SOC-13.

·        In Table 2 the abbreviation CI should be defined. Perhaps include it in the header as Cognitive Interviewing (CI)

·        Page 12, lines 344 and 346: Perhaps complete the sentences e.g., “compared with those who were not depressed” (if that was meant as a comparison between depressed and not depressed).

·        Page 13, lines 377-378: 'semantic space (Heise, 1969).' Why is the citation also in the quote?

 

2.      The authors should be congratulated for the way in which they have linked trauma recovery to sense of coherence dimensions. In particular, the theorizing and definitions of manageability, comprehensibility, and meaningfulness on page 6 is excellent. Overall phase 1 of the study is impressive. A minor matter is that in the Table 2 it is indicated that item 17 was dropped. I might have misunderstood this but that would mean that after phase 1 there should be 22 items rather then 23?

3.      With respect to the instruments, the authors only use the total scales and I would suggest that it may not be necessary to name all the subscales as was done on page 4.

4.      It is also necessary to report the scoring of all the scales.

5.      Similarly, for the five-point Likert scale used for the SOMI, report the scale anchors as it can be 0-4 or 1-5. We should also know the format of the Likert scale (e.g Strongly Agree (1) to Strongly Disagree (5)”

6.      Please provide a reference for the cut-off score for the PHQ8 (page 4, line 154).

7.      In the Analyses section please provide the statistical package used for the analyses.

8.      In the Analyses section it is stated that RMSEA and CMIN/DF was used to determine acceptable fit. However, Table 4 includes many more fit indices. In this regard:

a.      I do not agree with the definitions of acceptable values and perfect value as specified in Table 4. For example, for GFI, AGFI, CFI etc. a perfect fit would be 1.00. I would like to refer the authors to the following paper that presents a review and recommendation of what is considered to be a fair fit as well as a good fit.

Hooper et al. (2008). Evaluating Model Fit: a Synthesis of the Structural Equation

Modelling Literature

b.      A non-significant chi-square indicates a perfect fit.

c.      All these indices and the criteria for a good fit should be discussed in the Analyses section with appropriate references.

9.      In Table 3 there are factor loadings above 1.00. This is unusual

10.   Figure 1 and Figure 2 have the same caption. I also think the Scree Plot does not add any value and can be omitted.

11.   In Table 4, in the header it states “Actual Values for Comprehension and Meaning Inventory”. What does this refer to?

12.   My biggest concern with the paper is that phase 1 concentrated on the three subscales, the three subscales had acceptable reliability, and the discussion clearly states “One impetus for this research was the need for stable factors that could be used in the clinical evaluation of trauma recovery.” Yet the validity analysis excludes the subscales and only focuses on the total scale.

13.   With respect to the confirmatory factor analysis:

a.      there are many factor loadings above 1.00. Are these unstandardized loadings or is this an example of a “Heywood case?”.

b.      It is typical to compare several models of the factor structure of an instrument to determine which model best fit the data. The authors only reported on a three-correlated factors model.

c.      The three-correlated factors model is not an accurate representation of the SOMI. In this regard the authors use the total scale in the validity analysis indicating that in addition to the three factors, there is also a total scale score. This indicates that a bi-factor or higher-order factor model is a better representation of the SOMI.

 

I trust the authors find my input useful if they are given the opportunity to revise the manuscript

 

Regards

Comments on the Quality of English Language

Minor editing

Author Response

Reviewer 1 responses:

• Page 1, line 40: “Sense of Coherence (SOC) is defined SOC as "a global…” I think the term SOC as underlined is misplaced. Revised
• Page 3, lines 101-102: “has been the development of the revised Sense of Coherence Scale (SOC-R) Bachem and Maercker (2016).” The referencing is awkward. The authors use APA format and it should thus be (SOC-R: Bachem and Maercker, 2016) or “(SOC-R) by Bachem and Maercker (2016).” As an aside I am aware that MDPI journals uses Vancouver formatting and if the paper is accepted that citation style would fix this issue. Revised
• Page 3, line 124: “SOMI” - prior to this the term has not been defined. Revised
• I am not familiar with the term “known group reliability” I added a definition
• I stand to be corrected but I do not think that it is “Physicians Health Questionnaire” but rather “Patient’s Health Questionnaire”. Of course you are correct. Changed. Thank you!
• Page 4, line 165: “Cronbach alpha reliability for our study is .85.” The “is” should be “was.” Revised

2

• If the authors are requested to revise the paper, I would suggest that the anonymizing is not required as MDPI journals do not use a blinded review process. Revised throughout
• Page 5, line 235: “About two-thirds of the sample had some college, and half were working”. Is the word “education” missing after college? Changed
• There are instances where only the term “SOC” is used. For consistency the authors should clearly indicate that it refers to SOC-13. I changed it to SOC-13 when I was referring to the SOC-13 scale.
• In Table 2 the abbreviation CI should be defined. Perhaps include it in the header as Cognitive Interviewing (CI). Added
• Page 12, lines 344 and 346: Perhaps complete the sentences e.g., “compared with those who were not depressed” (if that was meant as a comparison between depressed and not depressed). Changed
• Page 13, lines 377-378: 'semantic space (Heise, 1969).' Why is the citation also in the quote? The sentence was clarified and the quotes removed

2. The authors should be congratulated for the way in which they have linked trauma recovery to sense of coherence dimensions. In particular, the theorizing and definitions of manageability, comprehensibility, and meaningfulness on page 6 is excellent. Overall phase 1 of the study is impressive. A minor matter is that in the Table 2 it is indicated that item 17 was dropped. I might have misunderstood this but that would mean that after phase 1 there should be 22 items rather then 23? Correct. Changed. Thank you!
3. With respect to the instruments, the authors only use the total scales and I would suggest that it may not be necessary to name all the subscales as was done on page 4. Removed the subscales except in the Barriers to Help Seeking instrument since it was relevant to the calculation of the internal dimension.
4. It is also necessary to report the scoring of all the scales. Added these.
5. Similarly, for the five-point Likert scale used for the SOMI, report the scale anchors as it can be 0-4 or 1-5. We should also know the format of the Likert scale (e.g Strongly Agree (1) to Strongly Disagree (5)” Added.
6. Please provide a reference for the cut-off score for the PHQ8 (page 4, line 154). Done, and added to the reference list.
7. In the Analyses section please provide the statistical package used for the analyses. Added.
8. In the Analyses section it is stated that RMSEA and CMIN/DF was used to determine acceptable fit. However, Table 4 includes many more fit indices. In this regard:
9. I do not agree with the definitions of acceptable values and perfect value as specified in Table 4. For example, for GFI, AGFI, CFI etc. a perfect fit would be 1.00. I would like to refer the authors to the following paper that presents a review and recommendation of what is considered to be a fair fit as well as a good fit.

Hooper et al. (2008). Evaluating Model Fit: a Synthesis of the Structural Equation

Modelling Literature

1. A non-significant chi-square indicates a perfect fit.

Response: Our numbers are according to Byrne, 1994. This reference was added.

 

3

1. All these indices and the criteria for a good fit should be discussed in the Analyses section with appropriate references.
2. In Table 3 there are factor loadings above 1.00. This is unusual This was a typo. It has been changed
3. Figure 1 and Figure 2 have the same caption. I also think the Scree Plot does not add any value and can be omitted. Scree plot deleted.
4. In Table 4, in the header it states “Actual Values for Comprehension and Meaning Inventory”. What does this refer to? This was a typo. It has been changed
5. My biggest concern with the paper is that phase 1 concentrated on the three subscales, the three subscales had acceptable reliability, and the discussion clearly states “One impetus for this research was the need for stable factors that could be used in the clinical evaluation of trauma recovery.” Yet the validity analysis excludes the subscales and only focuses on the total scale. Exploratory factor analysis (EFA) was used to evaluate the construct validity of the SOMI with the first sample. Then, CFA confirmed the validity of the second sample. Cronbach alpha was included for all subscales. Additional subscale validity analysis of AVE and MSV were added.
6. With respect to the confirmatory factor analysis:
7. there are many factor loadings above 1.00. Are these unstandardized loadings or is this an example of a “Heywood case?”.

These are standardized loadings, and this notation has been added. These relationships are perfectly acceptable since they indicate a strong relationship between the variable and the factor.

1. It is typical to compare several models of the factor structure of an instrument to determine which model best fit the data. The authors only reported on a three-correlated factors model. We added a clarifying sentence that we ran the CFA on 2, 3 and 4 factor models. We only included the final model.
2. The three-correlated factors model is not an accurate representation of the SOMI. In this regard the authors use the total scale in the validity analysis indicating that in addition to the three factors, there is also a total scale score. This indicates that a bi-factor or higher-order factor model is a better representation of the SOMI.

Response:  We analyzed 2, 3 and 4 factor models and the 3-factor model was both consistent with the original theory as well as having the best psychometrics. We only present the final model, however we added a sentence explaining the multiple model evaluations. Moreover, as an experienced researcher and theorist, we do believe that each items and factors are appropriate and the item structure represents good fit with the factors. We also provided each factors Cronbach alpha reliability numbers also support these factors.

Reviewer 2 Report

Comments and Suggestions for Authors

 

Comments and Suggestions

I read the manuscript with happiness and believe that the study can provide further insights into the newly developed and validated tool on “Revising the original Antonovsky Sense of Coherence concepts: A Mixed Method development of the Sense of Meaning Inventory (SOMI)” in the US cultural context. The manuscript has two potential goals: (1) to preserve the three concepts of SOC and to create a measure that assumes that the factors have some interaction but are distinct, allowing researchers 113 to use them as subscales. and (2) approached the revision of the SOC13 scale from a different angle using both quantitative and qualitative aspects. This research has the potential to make significant contributions to psychology, health sciences and gender disciplines by validating Sense of Meaning Inventory (SOMI) and its potential protective role on GBV. As a reader, there are only few/minor points that should be addressed to learn more about the topic and methods as well as to ease the reading itself. All the comments and suggestions are found below:

 

A.    Over all merits of the manuscript

1.    The introduction section introduces the NOVEL and ORIGINAL issues by focusing on an important but often overlooked aspect of GBV, focusing on protective factors like Sense of Coherence (SOC) rather than just risk factors add value to the existing body of literature.

2.    This introduction section critically identifying and looking at the gaps that exist within the current understanding of SOC, specifically in its measurement and failures by researchers with the existing measuring scales. Therefore, the introduction sets a foundation for the research aims that follow and added a new input for further studies.

3.    The references in the introduction go as recent as 2022 to support the discussion on SOC, trauma recovery, and the challenges in measuring SOC. This means that this section is up-to-date and has involved the reader with the field.

4.    I appreciated the authors verified the scale using mixed method design (both qualitative and quantitative) processes are effectively harnessed in a 2-phased one proceeding which fully includes a qualitative phase of revision of items, and a second one for psychometric evaluation. Such a general structure increases validity and reliability of the final scale.

5.    The authors used two analytical techniques both Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA). Therefore, the exploration of the factor structure through an exploratory factor analysis and its confirmation through the confirmatory factor analysis, along with the necessary related statistical tests (discriminant validity, divergent and convergent validities), demonstrate an applied discipline in testing the psychometrics of the SOMI scale, and this well-applied discipline strengthens the credibility of this study's results.

B.   Areas for Improvement which needs revision:

1.    I suggest the authors strengthen more the introduction section related to SOC Psychometric properties which are found in MDPI journals and others.

2.    As far as I know, and evidence suggested that the new developed and constructed instrument sub scales discriminant validity should be assessed. For example, Manageability. Meaning and coherence sub scales. To clarify this construct validity refers to items that reflect the latest theoretical construct designed to measure. For example, for construct validity, Hair et al. (2019) suggested that individual standardised factor loadings (regression weights) should be with the minimum range of .5, and the best should be 0.7. Convergent validity is the relationship among the positive constructs. In addition, using the maximum shared variance (MSV) and the average variance extracted (AVE). The AVE values that exceed a threshold limit higher than 0.5 (AVE > 0.05) demonstrate good convergent validity.  Discriminant validity is the extent to which a construct is genuinely distinct from other constructs (Hair et al., 2019). Or When the MSV is lower than AVE are characterized by adequate discriminant validity (Hair et al., 2019). Please apply AVE, MSV and Square correlation.  

3.    Please follow the global acceptable standardized fitness of indices to report the CFA model and report in the manuscript the standardized model. For example, below example is the standardized and unstandardized models

Figure 1: See Figure 1 Unstandardized estimate of SOC measure

Figure 2: See Figure 1 Standardized estimate of SOC measure

 

4.    The validation analysis needs to consider a number of different groups. For example (gender identity, Gender based violence (GBV) history, age, education, and employment). Therefore, measurement invariance is important. For the measurement invariance (MI) testing the psychometric equivalence of the variables across various groups using CFA ( Putnick & Bornstein, 2016), the researchers followed well-established scientific procedures using single and multi-group CFA (Millsap, 2011; Putnick & Bornstein, 2016; Vandenberg & Lance, 2000), using the four MI stages (configural, metric, scalar and residual). However, the scalar invariance was not fitted well in this article (adolescent gender and developmental stages). The measurement invariance of the scientific studies to four-sequential analysis -stage used the multi-group CFA following Millsap (2011) and Putnick and Bornstein (2016) and arrived at the following recommendation criteria: ΔTLI, 0 = perfect and ≤ 0.01 = acceptable, ΔRMSEA, 0.015 for metric, scalar, and residual invariance (Chen, 2007; Putnick & Bornstein, 2016).

C.   General Comments

I believe that my suggestion, if added to this article, could make a significant contribution to the scientific community.

Good luck in the future!

Reference for comments

Hair, J., Black, W., Babin, B., & Anderson, R. (2019). Multivariate Data Analysis (Vol. 8). Annabel Ainscow.

Hair, J. F. J., Black, W. C., Babin, B. J., & Anderson, R. E. (2014). Multivariate Data Analysis (MVDA): Vol. Seventh Ed. Pearson Education Limited.

Kim, E. S., Cao, C., Wang, Y., & Nguyen, D. T. (2017). Measurement Invariance Testing with Many Groups: A Comparison of Five Approaches. Structural Equation Modeling, 24(4), 524–544. https://doi.org/10.1080/10705511.2017.1304822

Kline, R. B. (2016). Principles and practice of structural equation modeling. (Vol. 4). The Guilford Press, New York London.

Millsap, R. E. (2011). Statistical Approaches to Measurement Invariance. Taylor and Francis Group, LLC.

Putnick, D. L., & Bornstein, M. H. (2016). Measurement invariance conventions and reporting: The state of the art and future directions for psychological research. Developmental Review, 41, 71–90. https://doi.org/10.1016/j.dr.2016.06.004

 

 

Comments for author File: Comments.pdf

Author Response

I suggest the authors strengthen more the introduction section related to SOC Psychometric properties which are found in MDPI journals and others. A roundup of a recent critique was added.

As far as I know, and evidence suggested that the new developed and constructed instrument sub scales discriminant validity should be assessed. For example, Manageability. Meaning and coherence sub scales. To clarify this construct validity refers to items that reflect the latest theoretical construct designed to measure. For example, for construct validity, Hair et al. (2019) suggested that individual standardised factor loadings (regression weights) should be with the minimum range of .5, and the best should be 0.7. Convergent validity is the relationship among the positive constructs. In addition, using the maximum shared variance (MSV) and the average variance extracted (AVE). The AVE values that exceed a threshold limit higher than 0.5 (AVE> 0.05) demonstrate good convergent validity. Discriminant validity is the extent to which a construct is genuinely distinct from other constructs (Hair et al., 2019). Or When the MSV is lower than AVE are characterized by adequate discriminant validity (Hair et al., 2019). Please apply AVE, MSV and Square correlation.

We calculated the AVE and MSV for the subscales and placed them in table 4. An AVE greater than 0.5 suggests good construct validity. For the first factor the AVE is .68, for the second factor .75, for the third factor is .55. The MSV is the maximum correlation between the factors. The MSV is 0.5461 suggesting good discriminant validity. These psychometrics have been added to the analysis and findings sections.

 

Please follow the global acceptable standardized fitness of indices to report the CFA model and report the standardized model in the manuscript. For example, below example is the standardized and unstandardized models.  Following, Byrne, 1994, we used standardized fit indices, as these are widely recognized and used in confirmatory factor analysis (CFA) research.

 

The validation analysis needs to consider a number of different groups. For example (gender identity, Gender based violence (GBV) history, age, education,  and  employment).   Therefore,  measurement invariance  is important. For the measurement invariance (MI) testing the  psychometric equivalence of the variables across various groups using CFA ( Putnick & Bornstein,  2016),  the  researchers  followed  well-established  scientific procedures using single and multi-group CFA (Millsap, 2011; Putnick & Bornstein, 2016; Vandenberg & Lance, 2000), using the four MI stages. Response: While this is an excellent suggestion, we did not do this additional analysis and describe the importance of this for future studies in the discussion section.

Reviewer 3 Report

Comments and Suggestions for Authors

This is an interesting and innovative study that integrates the salutogenic approach in a very compelling way.
I would like to request the following revisions:

1. It is unclear why meaningfulness is connected to motivational aspects – this point is insufficiently substantiated and requires a more in-depth explanation.
2. I request that the background section conclude with research questions/hypotheses listed as separate bullet points.
3. Please create a separate heading for "Participants" and another for "Sampling," with each section standing alone.
4. The study is mixed methods, but this is not reflected in the methodology and findings. I ask that both the methodology and the findings be separated according to the different methods.     

   

Author Response

This is an interesting and innovative study that integrates the salutogenic approach in a very compelling way. I would like to request the following revisions:

1. It is unclear why meaningfulness is connected to motivational aspects – this point is insufficiently substantiated and requires a more in-depth explanation. More detail was added.
2. I request that the background section conclude with research questions/hypotheses listed as separate bullet points. This research did not have questions or hypotheses. It had aims which are already described in the aims section. These were added.
3. Please create a separate heading for "Participants" and another for "Sampling," with each section standing alone. 2.1 is about sampling, and 3.1 is about participants. I changed the title of 3.1 to participants. This was clarified.
4. The study is mixed methods, but this is not reflected in the methodology and findings. I ask that both the methodology and the findings be separated according to the different methods. I changed the headings to clarify this, and added a section explaining the design.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

22 September 2024

 

The Editor

Sexes

 

The Authors

Revising the original Antonovsky Sense of Coherence concepts: A Mixed Method development of the Sense of Meaning Inventory (SOMI)

 

Dear Editor and Authors

 

I provide my feedback on the revisions below. However, I would like to request the Editor not to consider me as a reviewer for any further revisions of this manuscript. I approach my reviews from the perspective of what I expect from reviewers of my own outputs, namely a critical and helpful review. I had thought that my first review was meant to be helpful. I certainly did not imply that the authors were not experienced researchers or theorists which seemed to be the gist of the statement “Moreover, as an experienced researcher and theorist, we do believe that each items and factors are appropriate and the item structure represents good fit with the factors”. I found this statement to be very dismissive of my inputs.

 

Herewith, further points for consideration:

·        I suggested that the scoring of all scales should be reported. And with that I meant the scoring format and not “sum scores” but if a 5-point scale, for example is it 0 to 4 or 1 to 5.

·        I want to once again suggest that all the fit indices used in Table 4 as well as acceptable fit criteria should be discussed in the Analyses section and not only RMSEA and relative chi-squared. I appreciate the reporting of AVE and MSV but this should be discussed in the Analyses section and also indicate how this relates to validity. For example, AVE > MSV indicates discriminant validity as it demonstrates that the latent variable has more in common with the items that contributes to its measurement (AVE) then it has with related concepts (MSV). I presume that MSV as used by the authors refer to maximum shared variance i.e., the squared correlation between the latent construct and the variable with which it has the highest correlation coefficient.

·        The results of AVE and MSV and the implications for discriminant validity should also be reflected under paragraph 3.3.3 and not only in the Discussion section.

·        I am slightly confused by the MSV value. In Table 3 only one MSV value is reported at the bottom of the table and after the reliability of the total scale. This leads to the conclusion that only an overall MSV value was computed and the AVE of the subscales was then compared to this overall value. This seems to be supported by the statement in the discussion “In our case, for factor 3, the AVE is 0.5461 and the MSV 0.54”.  If MSV is the squared correlation of factor 3 (manageability) it should be .685 (the highest correlation of manageability was with GBV-heal) squared which is .469. Which means AVE > MSV for factor 3.

·        The authors have added the three subscales to the table of intercorrelations which is appropriate. However, I do not understand why the intercorrelations between the total scale (SOMI) are listed above the diagonal and at odds with all the other intercorrelations which are listed below the diagonal.

·        Please note that there are still factor loadings of 1.00 in the EFA

·        The authors contend with regard to CFA factor loadings that “These are standardized loadings, and this notation has been added. These relationships are perfectly acceptable since they indicate a strong relationship between the variable and the factor.” Factor loadings are in reality the correlation between the factor and its indicators. As a correlation coefficient it is bounded by -1 and +1. A factor loading greater than 1 indicates that the factor accounts for more than a hundred percent of the variance. When a factor loading is greater than 1, and has an error variance that is negative, it is called a Heywood case. It’s not possible to see that in Figure 1 as the error variances are not clearly visible. There are only special instances, such as growth curve models where standardized factor loadings can be greater than 1, but in these instances, they are not interpreted as the correlation between factor and an indicator. If It is possible that factor loadings > 1 is permissible the authors should ideally explain why or provide a reference.

·        The authors state in their response: “We analyzed 2, 3 and 4 factor models and the 3-factor model was both consistent with the original theory as well as having the best psychometrics.” I would not be able to understand how they could arbitrarily create 2 and 4 factor models – on what grounds? Furthermore, the way in which the authors deal with the scale as consisting of a total scale as well as three factors clearly suggest that there is a total factor over and above the 3 factors. This is apparent in Table 3 where they reported a reliability coefficient for the total scale (15 items) and in Table 5 where they, in addition to the subscales report the correlation between the total scale and the other variables. This suggests that a model accounting for a total scale (i.e., a bifactor or second-order factor model) can usefully be explored. It will still be consistent with the “original theory”.

 

 

Thank you

 

Author Response

Response: Please accept our heartfelt apologies for the implication of this comment…what we were meaning is that we were viewing through the lens of our critical review of the SOC, and found them consistent with that view. We regret that we inadvertently caused affront.

 

• I suggested that the scoring of all scales should be reported. And with that I meant the scoring format and not “sum scores” but if a 5-point scale, for example is it 0 to 4 or 1 to 5. Response: Added

 

• I want to once again suggest that all the fit indices used in Table 4 as well as acceptable fit criteria should be discussed in the Analyses section and not only RMSEA and relative chi-squared. I appreciate the reporting of AVE and MSV but this should be discussed in the Analyses section and also indicate how this relates to validity. For example, AVE > MSV indicates discriminant validity as it demonstrates that the latent variable has more in common with the items that contributes to its measurement (AVE) then it has with related concepts (MSV). I presume that MSV as used by the authors refer to maximum shared variance i.e., the squared correlation between the latent construct and the variable with which it has the highest correlation coefficient. Response: Thank you for your support on this. These were added into the analysis section, the table and the discussion section.
• The results of AVE and MSV and the implications for discriminant validity should also be reflected under paragraph 3.3.3 and not only in the Discussion section. Response: This was added.

 

• I am slightly confused by the MSV value. In Table 3 only one MSV value is reported at the bottom of the table and after the reliability of the total scale. This leads to the conclusion that only an overall MSV value was computed and the AVE of the subscales was then compared to this overall value. This seems to be supported by the statement in the discussion “In our case, for factor 3, the AVE is 0.5461 and the MSV 0.54”. If MSV is the squared correlation of factor 3 (manageability) it should be .685 (the highest correlation of manageability was with GBV-heal) squared which is .469. Which means AVE > MSV for factor 3. Response: This was written incorrectly. We have recalculated the AVE and MSV for all subscales and the AVE for SOMI-CO is somewhat lower than the MSV. We have undated this in the table and discussed this in the discussion section.

 

• The authors have added the three subscales to the table of intercorrelations which is appropriate. However, I do not understand why the intercorrelations between the total scale (SOMI) are listed above the diagonal and at odds with all the other intercorrelations which are listed below the diagonal. Response: I moved these correlations into the other place on the table and removed references to the total SOMI score throughout.

 

• Please note that there are still factor loadings of 1.00 in the EFA. Response: Thank you for this comment. We reran the analyses, and decided to use direct Oblimin rotation and these factor loading were used, and the text is updated.

 

• The authors contend with regard to CFA factor loadings that “These are standardized loadings, and this notation has been added. These relationships are perfectly acceptable since they indicate a strong relationship between the variable and the factor.” Factor loadings are in reality the correlation between the factor and its indicators. As a correlation coefficient it is bounded by -1 and +1. A factor loading greater than 1 indicates that the factor accounts for more than a hundred percent of the variance. When a factor loading is greater than 1, and has an error variance that is negative, it is called a Heywood case. It’s not possible to see that in Figure 1 as the error variances are not clearly visible. There are only special instances, such as growth curve models where standardized factor loadings can be greater than 1, but in these instances, they are not interpreted as the correlation between factor and an indicator. If It is possible that factor loadings > 1 is permissible the authors should ideally explain why or provide a reference.

Response: We reran the CFA and these are unstandardized values. Regarding the Heywood case, in our CFA, we observed several unstandardized factor loadings exceeding 1, which raised concerns regarding the presence of a Heywood case. This anomaly may be attributed to the high correlations observed among certain indicators, suggesting potential redundancy in the measurement items associated with the factors. Given the theoretical underpinning of the factors, some indicators may inherently overlap, resulting in inflated loadings. The complexity of the factors measured, coupled with a limited sample size, could contribute to this phenomenon. Despite high factor loadings, the overall model fit indices remained within acceptable ranges, indicating that the model adequately represents the data. In response to the observed Heywood case, we thoroughly reviewed the measurement items and the literature, and consulted other statisticians. Items that demonstrated significant redundancy or ambiguity were considered for removal or revision, although the final model retained theoretically justified items. While the presence of Heywood cases suggests potential limitations in the measurement validity, future research could further explore the relationships among factors using larger samples or alternative measurement strategies to confirm these findings.

 

• The authors state in their response: “We analyzed 2, 3 and 4 factor models and the 3-factor model was both consistent with the original theory as well as having the best psychometrics.” I would not be able to understand how they could arbitrarily create 2 and 4 factor models – on what grounds? Furthermore, the way in which the authors deal with the scale as consisting of a total scale as well as three factors clearly suggest that there is a total factor over and above the 3 factors. This is apparent in Table 3 where they reported a reliability coefficient for the total scale (15 items) and in Table 5 where they, in addition to the subscales report the correlation between the total scale and the other variables. This suggests that a model accounting for a total scale (i.e., a bifactor or second-order factor model) can usefully be explored. It will still be consistent with the “original theory”. Response: The 2, 3 and 4 factor solutions were examined as part of the Exploratory factor analysis and were not examined in the CFA. Only the 3-factor solution was examined in the CFA. I moved that language into the EFA section. We removed references to a total score in the text and in the correlation table.