| ObjectiveUnder the guidance of the core thought of the theory of seven Emotions of traditional Chinese medicine,with the characteristic concept of traditional Chinese medicine,based on a variety of qualitative methods and modern methods and technologies,and in accordance with the international scale development process and the latest development concept,the"seven Emotions Scale" and the computer adaptive version are developed to provide scientific quantitative tools for the comprehensive emotional evaluation of modern people.At the same time,the suitability and application process of different qualitative methods and technologies in the scale development process were explored,and a scale mixed methodology system including appropriate qualitative technologies was established.Methods1.Establishment of research working group:It is mainly divided into two parts:academic advisory group and core group.The academic advisory group consists of six experts,including scale scholars,psychologists,statisticians,qualitative research experts,computer intelligence experts and traditional Chinese medicine clinical experts with titles of associate high school or associate professor or above,who are responsible for the guidance of core technologies and specific contents of this study.The core group consists of 10 researchers,supervisors and team members,and is mainly responsible for the design,implementation,management and statistical analysis of this study.2.Basic characteristics and framework model of preset scale:the applicable population,scale purpose,number of items,completion time,response scale,scoring rules,etc.of preset scale are specified.3.Construct the scale conceptual framework and item library:On the basis of integrating the earlier conceptual framework and item library,two members of the core group independently searched all scales of Mental Disorders on the ePROVIDE platform,and consulted team psychologists and psychiatrists.Sorted out the commonly used clinical psycho-psychological assessment scales of the Department of TCM Psychology of the First Affiliated Hospital of Guangzhou University of Chinese Medicine,the Department of Psychiatry of the Affiliated Brain Hospital of Guangzhou Medical University and the Department of Psycho-psychology of the Third Affiliated Hospital of Sun Yat-Sen University,and sorted out the items of the existing psychological assessment scales with excel.Meanwhile,from April 13,2021 to October 24,2021,patients and healthy people who meet the diagnostic criteria of seven conditions determined by experts in the Department of psychology and psychiatry were selected for outpatient treatment and ward treatment in the First Affiliated Hospital of Guangzhou University of Chinese Medicine and the Affiliated Brain Hospital of Guangzhou Medical University,and semi-structured interviews were conducted in strict accordance with the qualitative research process.At the same time,four commonly used qualitative research paradigms,namely descriptive phenomenology,descriptive nature research,grounded theory and frame analysis,are used to analyze the interview data,and compare and evaluate the similarities and differences and suitability of descriptive phenomenology and descriptive nature research in constructing items,and grounded theory and frame analysis in constructing conceptual framework.4.Qualitative optimization screening of conceptual framework and item library:First,core group members applied warehousing method and selection method to conduct preliminary screening of items and classification of seven emotions;Subsequently,6 experts in TCM clinical medicine,TCM psychology,psychiatry and scale research were invited to conduct cognitive interviews on the overall design of the scale,the importance of item content,the suitability of item and the expression of item.Then,three rounds of Delphi survey were carried out by inviting well-known experts in the fields of TCM internal medicine,TCM psychology,Western medicine psychology,psychiatric internal medicine and scale assessment from third-class hospitals or universities in the country.Finally,the conceptual framework and item library are qualitatively optimized and sereened.5.Quantitative screening of items and evaluation of scale measurement performance:A cross-sectional survey was conducted between May 2022 and December 2022 on 300 patients and 500 healthy people who met the inclusion criteria of seven-emotion scale using the initial version.The paper version of Seven Emotions Scale was formed by quantitative screening of 300 randomly selected samples based on exploratory factor analysis method,correlation coefficient method,Kronbach coefficient method and discrete trend method.A total of 800 cases of data were used to analyze the feasibility,reliability and validity of the paper scale.6.Development of computer adaptive assessment scale:Using item response theory,one-dimensional test,local independence test,formal test of item characteristic curve and item information curve,difference analysis of item response theory model parameters and item function,and test information volume analysis were carried out on the paper version of "Seven Emotions Scale".The items were screened according to IRT related criteria,and the parameters of each item were estimated to construct the computer adaptive scale item library.Then,the computer adaptive simulation was carried out for items in various fields to test the measuring accuracy of the computer adaptive version scale for different potential traits,and the computer adaptive version of the Seven Emotions Scale was developed.At the same time,the research and development of the computer adaptive software platform was preliminarily explored.Results1.Item library construction1.1 ePROVIDE platform Retrieval A total of 106 scales of Mental Disorders were searched in the ePROVIDE system,and literatures related to the development purpose,application conditions,target population,item content and scale development process were manually read one by one.93 scales that did not meet the requirements were screened out according to the exclusion criteria,and 13 scales included in the study were sorted and analyzed.HIGH-R(hypomania interview guide(including hyperthymia)-retrospective assessment)covers HIGH-C and high-C-SR.HIGH-R is selected as the representative;Mood and Anxiety Symptom Questionnaire MASQ(Mood and Anxiety Symptom Questionnaire)covers Mini-MASQ and MasQ-SF;MASQ is taken as the representative;no specific items can be queried on sheehan-panic disorder scale(SPS).Therefore,8 scales were finally included in the study:GHQ-20(general health questionnaire),HADS,HAMA,4DSQ-50(Four-Dimensional Symptom Questionnaire),HIGH-R,SIS(sheehan)irritability scale),MASQ,8-MINI,269 entries in total.1.2 Collection of commonly used clinical psychological assessment Scales:A total of 16 commonly used psychiatric and psychological assessment scales were collected,including 4 anxiety scales(SAS,HAMA,GAD-7,BPRS and STAI,85 items in total)and 6 depression scales(CES-D,HAMD,PHQ-9,BDI,MADRS and SDS).104 items in total),2 panic scales(excluding social fear and place fear,PDSS and PASS,with a total of 20 items),4 mania scales(HCL-32,MDQ,YMRS,BRMS,with a total of 69 items),and 1 common general scale(SCL-90,with a total of 90 items).The HAMA scale has been included in the ePROVIDE search above,so it is not included in this step.Finally,a total of 16 scales commonly used in clinical psychiatry,with a total of 368 items,are included in this step.Through the above two methods,a total of 24 existing psychological and emotional scales with 637 items were selected.The core group conducted preliminary screening and discussed the deletion of items obviously unrelated to the seven emotions(mainly quantitative classification and ranking scale items dependent on doctors’ judgment,such as YMRS-11,BRMS-13,PDSS-7,SCL-23).Finally,583 items were included in the entry database.1.3 Qualitative interview:Interviews with 30 patients and 15 healthy subjects were included.In the process of item construction,it is found that descriptive phenomenology is not suitable for the construction of scale items.A total of 306 items of understanding and description about seven emotions are successfully obtained from the interviews of healthy people by using the summative content analysis of the descriptive nature of the research:63 items of sadness,58 items of fear,71 items of anger,62 items of sorrow,52 items of happiness;From the interview data of patients with seven emotions,476 items were obtained,of which 148 were sad,100 were afraid,88 were angry,96 were thinking and 44 were happy.In addition to 192 items formed in the previous research,the final construction of the seven emotions scale item database containing 1557 items;In the construction of conceptual framework,it is found that both frame analysis and grounded theory are applicable to the construction of conceptual framework of the scale:frame analysis is used to construct four levels of the scale:"physical symptoms,psychological feelings,behavioral manifestations,emotional cognition",and grounded theory is used to construct five levels:"physical symptoms,external manifestations,psychological feelings,behaviors,and emotional control".However,considering the convenience of researchers,the operability of research process and the practicability of forming themes,frame analysis is more suitable for the construction of scale frame than grounded theory.2.Qualitative optimization screening of item library2.1 Core Group optimization screening:According to the established standards and work manual,a total of 167 items are retained after screening by the core group,including sadness-45,thought-27,panic-43,anger-28 and joy-26.Then a group meeting was held to discuss whether the expression of each item is easy to understand and whether there is a better expression that is more consistent with clinical routine.In addition,the contents of the items are evaluated again,the meaning of each key expression word is analyzed,the expression meaning of each item is evaluated whether there is repetition and ambiguity,and the items are deleted,modified,merged and split.After discussion by the core group,the items with poor specificity were deleted,such as the items with similar expression contents,and the items from patient interviews were preferred.A total of 94 items entered the cognitive interview,including sadness-23,thinking-15,panic-21,anger-18,joy-17.2.2 Optimization and screening of cognitive interviews:Six experts from the First Affiliated Hospital of Guangzhou University of Chinese Medicine,Guangdong Hospital of Traditional Chinese Medicine,School of Basic Medicine of Guangzhou University of Chinese Medicine,and Brain Hospital Affiliated to Guangzhou Medical University were interviewed for 13 times,totaling 473 minutes.A total of 94 entries were added to the Delphi survey,including sorrow-24,Thought-17,Fear-21,anger-17,and happy-19.2.3 Delphi Survey Screening Optimization Results Between December 2021 and April 2022,a total of three rounds of Delphi survey were conducted,a total of 20 experts participated in the first round of Delphi survey,16 experts successfully completed three rounds of Delphi survey,the authority coefficients of the three rounds of Delphi survey were 0.868,0.850,0.850 respectively.2.3.1 The first round of Delphi survey:two items with repeated contents proposed by experts were deleted,the expressions of eight items were modified,one item was added,and the item was added as "frequently having nervous or fearful nightmares at night" in the field of panic.At the same time,the items in each domain are divided into important and consistent items,unimportant items,and important but inconsistent items according to the established criteria.Kendall coefficient of each field is between 0.203 and 0.403,P is less than 0.05,a total of 95 items enter the second round of Delphi survey.2.3.2 The second round of Delphi survey:A total of 15 "important and consistent items" have been examined and approved by experts and confirmed to be directly included in the final scale;A total of 11 "less important items" were deleted after expert analysis;For the analysis of "important but inconsistent items",the Kendall coefficient of each field is between 0.176～0.309,which is lower than the Kendall coefficient of each field in the first round(between 0.203 and 0.403),indicating that the expert opinions of this round of survey are more divergent than that of the last round of survey.Finally,a total of 15 items directly included in the final scale were determined by this round of survey analysis,and 13 items were deleted.A total of 67 items were entered into the third round of survey.2.3.3 The third round of Delphi survey:a total of 28 items entered into the final scale were determined,and 15 items were directly included into the final scale in the second round.The initial version of Seven Emotions Scale containing 43 items was formed after the screening by Delphi method,including joy domain 8,sorrow sorrow-10,anger-8,thought-7 and panic-10.Kendall harmony coefficient of the third round of expert consultation ranges from 0.215 to 0.477,with P all less than 0.05,indicating that expert opinions tend to be consistent.3.Quantitative screening of entries according to the size of exploratory factor loading,the factor loading of entries X1(I feel in a good mood)and X8(I feel crazy due to excitement)are 0.486 and 0.454 respectively,less than 0.5,to be deleted;According to the discrete trend method,the standard deviations of X1,X2 and X3 are 0.842,0.843 and 0.793,respectively,all of which are smaller than the deletion standard 0.85,indicating poor differentiation ability,which is to be deleted.According to Kronbach coefficient method,after deleting X8 entries,Kronbach coefficient is 0.869,which is larger than the original Kronbach coefficient of 0.842 in this field,so it is intended to be deleted.According to the correlation coefficient method,the correlation coefficients of items X3 and X8 and their fields are 0.579 and 0.191,respectively,less than 0.6.The representativeness of items is poor,and they are to be deleted.Combining the above four methods,entries X1,X3 and X8 are deleted by both methods at the same time,and finally deleted.4.Scale measurement performance evaluation4.1 Feasibility analysis:According to the established sample size,a total of 310 questionnaires were sent out in the clinic and 310 were recovered,with a recovery rate of 100%.Among them,300 were effective,10 invalid questionnaires were excluded,and the effective rate was 96.77%.A total of 535 questionnaires were received in the network survey of healthy people,35 invalid questionnaires were excluded,and the effective rate was 93.46%.It takes most patients about 5 minutes to complete the paper version of the scale,and the average filling time of the network survey displayed on the star background of the questionnaire is 4.6 minutes,indicating that the assessment burden of patients with this scale is small and has good feasibility.4.2 Reliability analysis:The Seven Emotions Scale has good reliability.Cronbach,sαand split half reliability analysis of the total volume scale and all fields are above 0.85.4.3 Validity analysis:When fitting the model for the first time,the model fitting results(2/df=4.981,CFI=0.891,IFI=0.891,TLI=0.884,NFI=0.868,RMSEA=0.071)were within the acceptable range,but the model fit was average.After adjusting the correlation between error covariances(e6 and e7,e31 and e32,e38 and e39),the fitting index of the model can be considered good(2/df=4.144,CFI=0.914,IFI=0.915,TLI=0.908,NFI=0.890,RMSEA=0.063).Therefore,the five-factor model of seven-emotion scale is considered to have good structural validity.4.3 Validity analysis:The paper version of Seven Emotions Scale has good content validity,structure validity,convergence validity and differentiation validity.The fitting index of confirmatory factor analysis model was good(x2/df=4.144,CFI=0.914,IFI=0.915,TLI=0.908,NFI=0.890,RMSEA=0.063).The standardized factor loading of each item corresponding to the five fields ranges from 0.624 to 0.896,the average extraction variance AVE value of each field ranges from 0.607 to 0.650,the combination reliability CR value of each field ranges from 0.884 to 0.943,and the hetero-elemental HTMT values of each field are within the standard range.The largest was 0.816.5.Item library construction based on item response theory5.1 One-dimensional test:In each field,the first characteristic root value of the exploratory factor is more than 4 times greater than the second characteristic root value,and the percentage of explanatory variance of the first characteristic root is more than 65%,significantly greater than the standard value 20%,indicating that the scale conforms to the one-dimensional hypothesis.5.2 Local independence test:The Q3 value between X4 and X7 is 0.406,and the Q3 value between B9 and B10 is 0.47,indicating that there is a certain dependence between X4 and X7,and there is a certain dependence between B9 and B10.According to relevant requirements,delete the one which is larger than the other entries in Q3 value.Two entries,X7 and B9,were deleted.The Q3 between the other entries is less than 0.36 and is independent of each other.5.3 Formal test of project characteristic curve and project information curve:each item basically conforms to the ideal shape of "monotone on both sides,normal in the middle",and the options of each item are well set;Almost all the information curves of the items show a "plateau shape",high in the middle and low on both sides,indicating that for most people within the range of ability value,the item measurement reliability is better and the measurement is more accurate.5.4 Difference analysis of project response theoretical model parameters and project functions:the minimum value of item difficulty coefficient is-2.046,the maximum value is 3.328,and the difficulty coefficient of all items is in the ideal range of-4 to 4,bl to b4 is monotonically increasing,there is no reverse threshold,indicating that the difficulty of all items is moderate and the setting is reasonable.The differentiation degree a is in the ideal range.Combined with the differentiation degree and difficulty parameters,it is considered that most items in the scale have good overall quality.When individuals showed different genders or ages,none of the items showed DIF when using the scale for assessment(R2<0.02).5.5 Test information content analysis:the average information content of each item is greater than 0.625,indicating that the item quality is good,and no item is deleted by this method.Finally,in addition to the partial independence test of the entries,there is a certain dependence between X4(I am more passionate than before)and X7(I have become very active),and a certain dependence between B9(I feel bored living)and B10(I have suicidal thoughts),which does not meet the established criteria.Other analyses showed that the scale items had higher quality.After discussion by the core group,in accordance with the principle of preserving items with good distinguishing ability,the item with less distinguishing degree among the two interdependent items was deleted.The parameter of distinguishing degree of X4 was 2.814,smaller than that of X7(5.127),so it was deleted.The discrimination parameter of B10 is 2.638,smaller than that of B9(2.957),so it is deleted.Therefore,Finally,items were analyzed based on item response theory.Except for the local independence test of items,items X7 and B9 did not meet the established standards and were deleted,other analysis showed that the scale items had high quality.A library of 38 items of high quality computer adaptive scale was constructed,including 4 items of happiness,9 items of sorrow,8 items of anger,7 items of thought and 10 items of panic.6.Computer Adaptive ScaleIn addition to the area of the area of the area,most of the potential traits measured by the potential characteristics of 72.05%are only 0.4,and the Pearson correlation coefficient of the potential trait estimates and the true value is 0.93(P<0.001).In each of the other areas,the CAT scale has significantly reduced the number of questions that have been tried,and The Pearson correlation coefficient of the potential trait estimates and the true value of the CAT simulation in each field is greater than 0.95(P<0.001),there is a very significant high positive correlation,and most of the calculated measurements of the test are less than 0.3.Overall,the CAT scale can be accurately estimated to be tested for potential quality,which is feasible and is significantly efficient and accurate in measuring the paper scale.ConclusionThe paper version of the Seven Emotions Scale developed in this study has good feasibility,reliability and validity,and can be used as a comprehensive emotional assessment tool for modern people.On this basis,the computer adaptive version of the Seven Emotions Scale can measure the seven emotions of modern people efficiently and accurately.However,in order to develop a high quality computer adaptive emotional assessment tool that can be applied and popularized in practice,we should continue to update,optimize and maintain the computer adaptive question bank in the later stage.At the same time,this study evaluated qualitative methods and technologies in all aspects of TCM scale research,clarified the similarities and differences and suitability of different qualitative methods in the scale development process through actual research,and preliminarily established a scale development mixed methodology system based on the integration of various qualitative methods and technologies and modern quantitative evaluation technologies.It fills the gap of mixed methodology in TCM scale research and is expected to promote the clinical application of TCM scale and the development of TCM efficacy evaluation. |
PDF Full Text Request