Factor Structure and Psychometric Properties of Brief COPE in Russian Schoolteachers

Round 1

Reviewer 1 Report

First of all, let me say thank you for this valuable study and interesting manuscript. Although I am convinced about its relevance to the readership I see a high need for improvement. Let me go into more details:

*)  I highly agree with you that it is important to provide such a scale for Russian school teachers. Please comment in short, on why you did use only school teachers for validating your scale.

Looking forward to the revised version!

*) On the first page, you write: Recently, however, it was noticed that a model with only two categories... then you cite [7], which is Carver, Scheier, and Weintraub from 1989. This is not recently but more than 30 years old. So I ask you to update your literature research and integrate real recent studies on the Brief COPE.

*) Last paragraph of the introduction: Thus, some subscales consist of only two items. In the Brief COPE, ALL subscales consist of two items, with 14 scales. Please explain how you could use 32 questions. I understood that you used a mixture of COPE [7] and Brief COPE [6]. But why?

*) Results: Please explain from a theoretical perspective, why you anticipated six factors.

*) Table notes and descriptions are missing.

*) Text after table 1: items 20 and 22 represent reframing, an important scale in the COPE and Brief COPE. Please explain from a theoretical perspective, why you put them to positive coping. All in all, I cannot follow the argument why you dissolved the 14-factor structure of the Brief COPE. All of the scales are theoretically derived by Carver and colleagues and the 14-factor structure is the most stable found in different studies.

*) I understand the argument that drug use among teachers is difficult to measure. But, in anonymous data collections, I do not see it as a solution to skip the whole scale because of this point. 

*) Avoidance scale: It is clear that Model 2 fits better than Model 1 because it is simpler. However, the difference is marginal. Moreover, the scale is present in all subsequent analyses. Is it taken out or not?

*) Please explain from a theoretical perspective, why the PSS is suitable for displaying the construct validity.

*) Page 11: Retest-validity: Did you mean retest-reliability?

*) Unfortunately, since I don't understand Russian, I can't check if the translation is correct.

Author Response

Response to reviewer 1 comments

Thank you very much for your review! Your comments help us to revise our research and extend its theoretical background. We take all your comments into consideration and provides the response below. We hope that the quality of the manuscript increases significantly.

Point 1.  Please comment, in short, on why you did use only school teachers for validating your scale.

Response 1. We use only school teachers, because that professional group have specificity, regarding their social function: they do not only educate children, but also translate to them some behavioral norms and values. Because of that, some particular coping strategies may seemed inappropriate for school teachers (and it has to be taking into consideration in the questionnaire design). We add the explanation to the manuscript (Lines 38-45)

Point 2. On the first page, you write: Recently, however, it was noticed that a model with only two categories... then you cite [7], which is Carver, Scheier, and Weintraub from 1989. This is not recently but more than 30 years old. So I ask you to update your literature research and integrate real recent studies on the Brief COPE.

Response 2. Indeed, reсent studies is in scarce in our literature review. We integrated some new studies on the brief COPE in the introduction section (Numbers 2, 35, 38, 50, 57, 39, 46; Lines 60-63)

Point 3. Last paragraph of the introduction: Thus, some subscales consist of only two items. In the Brief COPE, ALL subscales consist of two items, with 14 scales. Please explain how you could use 32 questions. I understood that you used a mixture of COPE [7] and Brief COPE [6]. But why?

Response 3. The purpose of the study was not just a validation of existing brief COPE version, but the developing of a new version for school teachers, based on Rasskazova and colleague’s COPE version. Because of that, we did not apply the Brief COPE, where all subscales consist of two items, with 14 scales. We combined some of the scales based on the size of their correlations in a correlation matrix presented in [39]. If the correlations were large enough, we combined the scales. We added the explanation to the manuscript (Lines 84-88)

Point 4. Results: Please explain from a theoretical perspective, why you anticipated six factors.

Response 4. Our questionnaire is based on a Russian COPE version, developed by Rasskazova and colleagues. According to their research, some scales of the COPE are highly correlated. Thus, as it is seen from the empirical data, the scales are integrated in six factors (each factor presents intercorrelated scales).

The development of our instrument is based on the idea that COPE consists of first-order factors, that may be integrated in second-order factors. Carver proposed 4 second-order factors in the initial COPE version: 1) problem-focused 2) emotion-focused 3) disengagement 4) acceptance [Carver et. al., 1989]. However, further studies on the COPE factor structure demonstrates that the COPE has three or five second-order factors [Sica et. al., 1997; Stowell et. al., 2001; Litman, 2006]. Regarding the brief COPE, different studies distinguish from 2 to 15 second-order factors [Solberg et. al., 2021]. The number of factors derived depend particularly on the language of the instrument [Solberg et. al., 2021]. Thus, the six subscales proposed in the current study may be described as second-order factors.

In our study, preliminary PCA analysis confirms the 6-factor structure. It can be clearly seen from the scree plot on eigenvalues (Figure 1). We added a short clarification to the manuscript (line 175)

Point 5. Table notes and descriptions are missing.

Response 5. We added table notes for each table (lines 183, 209-210, 218-219, 227-228, 237-238, 248, 255, 351)

Point 6: Text after table 1: items 20 and 22 represent reframing, an important scale in the COPE and Brief COPE. Please explain from a theoretical perspective, why you put them to positive coping. All in all, I cannot follow the argument why you dissolved the 14-factor structure of the Brief COPE. All of the scales are theoretically derived by Carver and colleagues and the 14-factor structure is the most stable found in different studies.

Response 6. The “positive coping” subscale was designed based on Rasskazova’s questionnaire and intercorrelations among subscales in her analysis. The “positive reframing” and “humor” subscales are highly intercorrelated in Rasskazova’s study, that is way we integrated them to the “positive coping” subscale. The strong intercorrelations between different subscales in Rasskazova’s study is a sign, that Russian version of a COPE has a structure, distinguished from a classic one with 14 scales. The detailed argumentation for our brief COPE structure is provided below.

We combined some of the scales based on the size of their correlations in a correlation matrix presented in Rasskazova and colleague’s research [39]. If the correlations were large enough, we combined the scales. Of course, from a theoretical standpoint, all scales can be important in certain scenarios: we may be able to make interesting conclusions, while taking very little examinee time, if we assume that their scores represent their intended constructs. In case we make this assumption, it is really enticing to shorten scales to two items, just like we use the number of books an examinee owns as a marker of their SES. However, with COPE, which measures complex psychological constructs, we had concerns about the validity of such conclusions. If we take just two samples of self-reported behavior, is it really all it takes to represent a latent construct? If so, why not just take one item per scale and interpret the behavior directly? In our opinion, psychological constructs should be observed in varied behaviors, otherwise we are not measuring a latent trait at all.

Another point of view for this study was looking at the usability of this questionnaire as a tool to determine individual differences. Two-item scales have a lot of problems from a psychometric standpoint. Scales like that can't have high reliability. In fact, we believe the original measure had inflated reliability. If we take from factors in a FA only the highest loading items, it will not lead to increased validity, instead the constructs will become shallow. Consider the «Seeking social support for instrumental reasons» scale: it consisted of very similar items in the original: «I talk to someone to find out more about the situation», «I talk to someone who could do something concrete about the problem». They are all worded the same with a minor alteration. This is a factor that inflates reliability, without actually sampling different behavior. We could just inquire about getting help from "those close to me" and get the same information without the inflated item correlations. If «Seeking social support for instrumental reasons» was indeed a latent construct and not just one behavior, wouldn't it be more varied in its indicators?

So, coming around to the original point, with only the items that we had available, we chose to combine some of the scales and create more varied constructs, to make our version of COPE a reliable monitoring tool.

To support our questioning the reliability of 2-item scales, we provide arguments from the article:  Yuan, W., Zhang, L., & Li, B. (2017): Adapting the Brief COPE for Chinese Adolescents with Visual Impairments. Journal of Visual Impairment & Blindness, 111(1), 20–32. doi:10.1177/0145482x1711100103: «In their study on psychological adaptation to visual impairment, Bergeron and Wanet-Defalque (2013) utilized the Brief COPE to assess adults’ coping with acquired visual impairments. However, they merely reported the internal consistencies of two out of the 14 subscales; that is, denial (.54) and acceptance ( .57). It is necessary to further examine the internal psychometric properties of the Brief COPE among visually impaired persons. Limitations of the Brief COPE regarding the low reliability coefficients and unstable factor loadings of some sub scales may be attributed to the fact that each subscale of the Brief COPE only contained two items. Brevity is a merit of the Brief COPE; however, it may also have sacrificed the rigor of its psychometric properties to some extent».

Point 7.  I understand the argument that drug use among teachers is difficult to measure. But, in anonymous data collections, I do not see it as a solution to skip the whole scale because of this point.

Response 7. The scale was excluded primarily due to its poor psychometric properties. The absolute value of the PCA loading for it was below the threshold chosen (0.3). We added the clarification to the manuscript (line 192)

Point 8.  Avoidance scale: It is clear that Model 2 fits better than Model 1 because it is simpler. However, the difference is marginal. Moreover, the scale is present in all subsequent analyses. Is it taken out or not?

Response 8. The Avoidance scale demonstrates comparatively low reliability. Correspondingly, the model without that scale fits better. Nevertheless, we propose the initial model (Model 1) as a final one, because we consider avoidance coping as a significant construct to measure. Moreover, as you have mention, the difference between two models is negligible, and, overall, the initial model also shows acceptable psychometric properties. We added the clarification in the manuscript (lines 292-298)

Point 9.  Please explain from a theoretical perspective, why the PSS is suitable for displaying the construct validity.

Response 9. The previous studies demonstrate correlations between different coping strategies and PSS. We added the information in a section, where instruments are described (lines 146-149)

Point 10.  Retest-validity: Did you mean retest-reliability?

Response 10. Yes, we mean retest-reliability. The term was corrected (line 319)

Point 11. Unfortunately, since I don't understand Russian, I can't check if the translation is correct.

Response 11. We use the translation performed and tested on Russian-speaking sample by Rasskazova and colleagues. We add that clarification in the manuscript (lines 120-122)

Reviewer 2 Report

The manuscript has a good internal structure. However, there are a number of aspects that should be revised:

- The abstract should contain the percentage of women or men, as well as the Mean Age and standard deviation.

- The keywords could be in alphabetical order and one of them should refer to the study design.

- The introduction provides interesting information but is scarce. The number of references used in the introduction should be significantly increased, focusing especially on those references from the last five years and in English.

- It is recommended to avoid the first person plural (e.g., "we propose..." at the end of the introduction. Please revise the entire manuscript. Instead, the third person singular should appear. 

- At the end of the introduction should be clearly defined not only the general objective, but also the specific objectives -numbered and related later to the Results section-. Hypotheses could also be put forward, although this is only a recommendation in this case.

- The method is well structured. The manuscript stands out in this section.

- The link in the Procedure does not work. Please check that it works in case you want to expose it. 

- The Results provide relevant information. 

- The Discussion provides useful information but should discuss more studies than were used in the Introduction. 

- There should be a Conclusions section, different from the Discussion.

Thank you for your attention.  

Author Response

Thank you very much for your review! We highly appreciate your comments, that help us to uncover some considerable shortcomings, unrevealed during the manuscript preparation. Below you can find our responses on your comments.

Point 1. The abstract should contain the percentage of women and men, as well as the mean age and standard deviation.

Response 1. The information required was added to the abstract (Line 7)

Point 2. The keywords could be in alphabetical order and one of them should refer to the study design.

Response 2. The keywords were rearranged in alphabetical order. A keyword “psychometric properties” was added as one, referred to the study design (Line 16).

Point 3. The introduction provides interesting information but is scarce. The number of references used in the introduction should be significantly increased, focusing especially on those references from the last five years and in English.

Response 3. We extended the introduction section and integrated some new studies on the brief COPE (Numbers 2, 35, 38, 50, 57, 39, 46; Lines 38-45, 60-63, 82-91)

Point 4. It is recommended to avoid the first person plural (e. g. “we propose…” at the end of introduction). Please revise the entire manuscript. Instead, the third person singular should appear.

Response 4. We changed the first person plural on the third person singular in the entire manuscript.

Point 5. At the end of the introduction should be clearly defined not only the general objective, but also the specific objectives – numbered and related later to the Results section. Hypotheses could also be put forward, although this is only a recommendation in this case.

Response 5. We provided the description of our specific objectives: 1) to test brief COPE psychometric properties, 2) to analyze a factor structure, 3) to evaluate the external validity on a sample of Russian schoolteachers. (Lines 89-91)

Point 6. The link in the Procedure does not work. Please check that it works in case you want to expose it. 

Response 6. Unfortunately, the online platform for data collection is not available at the moment. Thus, it is impossible to provide a direct link. You can find the detailed description of the platform in the following article:

Nikulchev, E., Ilin, D., Silaeva, A., Kolyasnikov, P., Belov, V., Runtov, A., ... & Malykh, S. (2020). Digital Psychological Platform for Mass We b-Surveys. Data, 5(4), 95. https://0-www-mdpi-com.brum.beds.ac.uk/2306-5729/5/4/95

Point 7. The Discussion provides useful information but should discuss more studies than were used in the Introduction.

Response 7. We extended the discussion section and discussed additional studies (Number 43, 48, 31, 19, 10; Lines 259-267)

Point 8. There should be a Conclusions section, different from the Discussion.

Response 8. We added the short conclusion section, where the objectives of the study and the main results are summarized (Lines 327-333). 

Round 2

Reviewer 1 Report

Thank you for the revised manuscript. The quality of the paper is much higher now. The points I mentioned in my review are more or less fully met by the authors. However, I ask once again for intensive and professional proof before publishing, also things like AE/BE.

All in all, I estimate the paper as an important contribution to doing research in the school system. The paper does not aim to discuss theoretical concerns about how to define dimensions, scales, or factors for measuring coping behaviors, although this is an open question in coping research. Here, the focus lies in providing a tool for research. Since it is mentioned in this context and linked to previous research that coping dimensions are influenced by language and culture, among other factors, I would suggest the publication of the manuscript after proof.

Author Response

Thank you very much for your comments! Our manuscript was corrected by MDPI experts in terms of English language. We hope that now the language quality of the paper is much higher

Author Response File: Author Response.docx