Indian Journal of Psychological Medicine
  Home | About Us | Editorial Board | Search | Ahead of print | Current Issue | Archives | Instructions | Contact | Advertise | Submission | Login 
Users Online: 2079 
Wide layoutNarrow layoutFull screen layoutHome Print this page Email this page Small font sizeDefault font sizeIncrease font size


 
 Table of Contents    
COMMENTS ON PUBLISHED ARTICLE
Year : 2019  |  Volume : 41  |  Issue : 6  |  Page : 598-599  

Comments on “correlation of cognitive resilience, cognitive flexibility and impulsivity in attempted suicide”


Clinical Psychologist, Department of Clinical Psychology, PsyClinic, Paschim Vihar, Delhi, India

Date of Web Publication11-Nov-2019

Correspondence Address:
Dr. Tarun Verma
B-3/141, Paschim Vihar, Delhi - 110 063
India
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/IJPSYM.IJPSYM_355_19

Rights and Permissions

How to cite this article:
Verma T. Comments on “correlation of cognitive resilience, cognitive flexibility and impulsivity in attempted suicide”. Indian J Psychol Med 2019;41:598-9

How to cite this URL:
Verma T. Comments on “correlation of cognitive resilience, cognitive flexibility and impulsivity in attempted suicide”. Indian J Psychol Med [serial online] 2019 [cited 2019 Dec 13];41:598-9. Available from: http://www.ijpm.info/text.asp?2019/41/6/598/270662



Sir,

A correlational study [1] recently reported that impulsivity, cognitive flexibility (CF), and cognitive resilience (CR) were linked in a sample of 270 people who attempted suicide. The authors, however, have erroneously mentioned regression analysis as correlation. Nowhere in the text does one find correlation coefficients (r); on the contrary, regression coefficients (β), calculated through multiple regression analysis, have been described as equivalent to correlation coefficients, even when there are several independent variables (IVs) in the regression model. This flaw is reflected all over the text, which does not give an accurate interpretation of the analysis results. Moreover, the results in [Table 2], [Table 3], [Table 4][1] consist of many biased and questionable values, which puts the validity of the research under serious doubt.

In a simple linear regression with only one IV, the value of β is equal to the value of r between the two variables. [Table 4] has β = 1.00, which implies that CF and CR are perfectly correlated. Even the standard error is also shown to be 0, which is only theoretically possible, this is impossible on practical grounds because errors in measurements are likely to be there due to known or unknown sources. However, the authors have defended these results in their text, which seems unreasonable. On the contrary, the authors should have looked into this error more carefully as the two variables of CF and CR do not seem to share an absolute relationship theoretically or empirically. These two variables are reported with contradictory relationships based on unstandardized coefficient (B) (10% variance in CR by CF) and β values, along with the unlikely value of R 2 = 1.00, which can range between 0 and 1.0 only. Standardized coefficient (β) is interpreted to account for the variance in the dependent variable (DV) by the IV, which as the study indicates is 100%, but that is impossible because there is only one IV. It is possible to get significant β values greater than 1, but such values are comprehensible when there are multiple IVs.

The most surprising findings are perceived in [Table 2] and [Table 3]. The values of all parameters are the same, except in the column of unstandardized values, which in [Table 3] are 10 times the values in [Table 2]. Even the values of R 2 and F are similar for both regression models with different DVs. Do the authors mean to imply that in their study CR and CF are the same variables? And if they have the same values by chance, as indicated by their perfect association in [Table 4], how come it is even remotely possible that these two distinct variables came out to be exactly the same? The mean values of the two scales measuring CR and CF are 4.49 and 44.93, respectively, which again represent a 10-fold difference between the two scores. The same pattern is observed for standard deviations of the two variables. These results require further exploration. How come the distributions of CR and CF appear to be so similar in this study that is unlikely to occur in any research? The variance and other statistical values for these two variables should be different enough such that the actual variation in the data is reflected in the table values.

It is also not clear why the authors have put CR and CF as DVs. Regression analysis is meant to predict the variation in DV by an IV or a set of IVs, not to measure correlation. β weights determine relative variances (amount of explained variation) in the DV by different IVs, and are not limited to any range, unlike r, which can range between −1.0 and 1.0 only. Correlation, on the other hand, measures covariation or the extent of dependency between only two variables. R 2 is the variation in DV explained by the model which includes some set of IVs, where some variation in DV is left unexplained by this model making the value of R 2 less than 1. The whole rationale behind the application of statistical tests seems unreasonable and unclear in this study. Why didn't the authors simply use Pearson's product-moment correlation method? If the study is correlational then using the Pearson method is sufficient to find the underlying relationships among variables. Regression analysis could have been used if they had to statistically control some variable(s) in order to understand the hierarchical relationships among variables.

Research quality is greatly affected by the inaccurate interpretation of statistical analyses and results. Wrong or unclear findings in studies mislead the readers and future researchers.[2] Researchers may take the help of statisticians to give them guidance for proper analysis and interpretation of data.[3] Expert statisticians can also be consulted to verify the inferences drawn from the results.[4] Insufficient analysis reduces its validity and the readers may not show any interest in the aforementioned implications of the study.[5] In this study,[1] the results look spurious and extremely theoretical, with no practical validity. The authors should have used appropriate tests to make conclusions from their data because various errors have been overlooked, which reduces the utility of this study.

Financial support and sponsorship

Nil.

Conflicts of interest

There are no conflicts of interest.



 
   References Top

1.
Ram D, Chandran S, Sadar A, Gowdappa B. Correlation of cognitive resilience, cognitive flexibility and impulsivity in attempted suicide. Indian J Psychol Med 2019;41:362-7.  Back to cited text no. 1
[PUBMED]  [Full text]  
2.
Goodman SN, Altman DG, George SL. Statistical reviewing policies of medical journals: Caveat lector? J Gen Intern Med 1998;13:753-56.  Back to cited text no. 2
    
3.
Altman DG, Bland JM. Improving doctors' understanding of statistics. J R Stat Soc A 1991;154:223-67.  Back to cited text no. 3
    
4.
Wulff HR, Andersen B, Brandenhoff P, Guttler F. What do doctors know about statistics? Stat Med 1987;6:3-10.  Back to cited text no. 4
    
5.
Stavale R, Ferreira GI, Galvão JA, Zicker F, Novaes MR, Oliveira CM, et al. Research misconduct in health and life sciences research: A systematic review of retracted literature from Brazilian institutions. PLoS One 2019;14:e0214272.  Back to cited text no. 5
    




 

Top
 
 
  Search
 
  
    Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
    Access Statistics
    Email Alert *
    Add to My List *
* Registration required (free)  

 
  In this article
    References

 Article Access Statistics
    Viewed106    
    Printed0    
    Emailed0    
    PDF Downloaded13    
    Comments [Add]    

Recommend this journal