Validation of Abbreviated Science Anxiety Scale in the Indian Context

: The abbreviated science anxiety scale 1 was validated in the Indian context with 290 secondary school students of grade 8 as the sample subjects. Exploratory factor analysis revealed the original two factors explaining 54.778 % of total variance in science anxity construct. Confirmatory factor analysis validated the factor structure of the construct with excellent goodness of fit estimates like CMIN/DF = 1.830, CFI = 0.969, TLI = 0.957, RMSEA = 0.054 and SRMR = 0.0392, conducted using the estimator maximum likelihood (ML). The floor and ceiling effect estimation for content validity showed that both the effects are absent with estimates way lesser than the benchmark of 15 % at 6.55 % and 0 % respectively. The internal consistency reliability estimation using Cronbach’s alpha found that the five items of the first factor “Learning science anxiety” and four items of the second factor “Science evaluation anxiety” had this estimate at 0.803 and 0.678 respectively, both of which fairly indicate good measurement of reliability of the scale. The education implications of the study are discussed.


INTRODUCTION
The seminal works of Pekrun [2][3][4] on academic emotions and their regulation strategies established the role of various classroom related academic emotions on the performance of the students in the literature.One of the prominent and potent academic emotions which undesirably play out in the classroom is the subject related anxiety in the students 5 .In the context of STEM education and specifically in the field of science education, the academic parallel of this variable is science anxiety.It is defined as "a debilitating combination of fearful negative emotion and cognition in the context of science learning" 6 .A recent study

METHODOLOGY Participants
The subjects of the sample comprised of 290 students of 8 th grade from the Army Public School (APS), Jalandhar Cantonment, from the Indian State of Punjab.

Procedure
The investigator conducted this cross-sectional descriptive design study as part of an action research, which itself was one of the several activities she was required to perform, besides teaching, during her semester long teaching internship in the mentioned school.The school authorities provided the formal permission to conduct such study on their students.Before administering the study, the purpose of the study, anonymity of the data gathered of the sample subjects and their consent to be part of the work was taken verbally during her regular teaching sessions supervised by an in-service teacher of the school.The students of the study were selected using purposive sampling in this study from multiple sections of grade 8.

Instrument
The scale adapted in the present study was itself adapted from the Abbreviated Math Anxiety Scale (m-AMAS) 9 .It has two factors, namely, the Learning science anxiety measured with five items and the Science evaluation anxiety comprised of the remaining four items of the scale.The responses of the students on how much anxiety they experienced on nine specific activities related to science learning and evaluation are registered on a five point Likert scale where 1=low anxiety and 5=high anxiety.The statements of the nine items are seen in Table 1.1."Taking a science test"

Descriptive Statistics
The measures of centrality tendency, dispersion and symmetry in the form of mean, standard deviation, skewness and kurtosis are seen in Table 1.

Absent
The percentages of total 290 subjects having the lowest score of 9 and the highest score of 45 of the scale were estimated and found to be at 6.55 % and 0 % respectively, which are way below the benchmark of 15 % 10 .These results establish the content validity of the scale.The exploratory factor analysis was conducted using SPSS Statistics Ver.23.0 on the sample size of 290 using Principal Component Analysis (PCA) extraction method under direct oblimin rotation, for a threshold inter item correlation fixed at 0.32.The determinant obtained was 0.08 which was way higher than the benchmark of 0.00001 indicating that the quality of the data was good enough to go ahead with the further analysis of factor extraction (see Table 1.5).Also, the Keiser Meyer Olkin sampling adequacy estimate was high enough at 0.872 above the benchmark of 0.6 indicating the sample size collected for the present study to be sufficient, along with a desirably significant Barlett's test of sphericity result (see Table 1.6).Two factors were extracted akin to the original study, which together explained 54.778 % of total variance in the construct (see Table 1.7).The first factor had the initial five items and the second factor had the remaining four items with factor loadings ranging from 0.571 to 0.896 (see Table 1.8).The reliability estimation was conducted by finding the internal consistency reliability of the items of first and second factors, which were found to be acceptable at 0.803 and 0.678 respectively (see The path diagram of the factor structure of the abbreviated science anxiety scale displayed to inter-related factors of the construct associated with learning and evaluation of science in the secondary students with the correlation coefficient high at 0.77 (see Figure 1.1).Also the items loaded very highly and precisely on their respective factors only.The goodness of fit estimates were excellent with CMIN/Df less than 3 at 1.83, both CFI and TLI greater than 0.95 at 0.969 and 0.957 respectively, SRMR less than 0.08 at 0.0392 and RMSEA slightly higher than its benchmark of 0.05 at 0.054 as per 11 (see Table 1.11).The composite reliability of the first factor "Learning Science Anxiety" was 0.805, which is greater than the benchmark of 0.6 and its average variance explained is at 0.452 which is lesser than the benchmark of 0.5.The composite reliability of the second factor "Science Evaluation Anxiety" was 0.691, which is greater than the benchmark of 0.6 and its average variance explained is at 0.362 which is lesser than the benchmark of 0.5 12 .Also, the square root of the average variance explained for the two factors are 0.672 and 0.601 respectively, lesser than the correlation coefficient between the factors of 0.77.These results indicated that there is scope of improvement for the convergent and divergent validities of the scale.The item-total correlation of the nine items ranged from 0.53 to 0.604 higher than the benchmark of 0.3 to indicate content validity [13][14][15][16][17] (see Table 1.12).Using the values obtained in Table 1.3, the Heterotrait-Monotrait HTMT test of discriminant validity was conducted, and its estimate was found to be at 0.758, which is lesser than the benchmark of 0.85 or 0.9 18 , indicating the presence of divergent validity of the scale in this study.

DISCUSSION
The seriousness of the science anxiety related issues is contextualized from that fact that India is a young nation and its youth requires source of employment at their disposal to lead a quality life.There are a lot of employment opportunities in STEM related professions and the demand of quality STEM professionals by the industry can be met when there is a supply of quality students from the school to the colleges and universities for graduation into the job market.Such a logistics cannot get manifested until science stream students with high science self-efficacy can be developed from upper primary and secondary levels of school education system in the country.Unrealistic parental pressure, lower recruitment of quality science teachers in school institutions and poor perception of science as a subject among the young adolescents owing to the presence of higher level of science anxiety levels in them, are some of the serious challenges which require immediate addressing through scientific approach of research.
Trading on these lines, the present study tried to adapt a brief, robust and a cross-cultural tool to measure science anxiety scale among secondary school students of India.The tool was found to extract the same two factors namely Learning science anxiety and Science evaluation anxiety with the nine items of the scale loading highly on their respective factors, with excellent content, construct and discriminant validities and acceptable reliability estimates in the form of Cronback alpha and composite reliability.However, the scale's convergent validity can improve.
Considering the prevalence of science anxiety levels among secondary school students of India, the teaching practitioners can take cognizance of this state of affair and indulge in

1 Floor and Ceiling Effects of Content Validity EstimationTable 1 .
4 Floor and Ceiling Effects of the Items of Science Anxiety Scale

3 Estimation of Construct Validity of the Factor Structure of Abbreviated Science Anxiety Scale using Confirmatory Factor Analysis (CFA) Figure
1.1 Path Diagram of the Factor Structure of Science Anxiety Construct in Indian Secondary School Students.Table 1.11Goodness of Fit Estimates of Abbreviated Science Anxiety Scale

Table 1 .
12 Estimation of Composite Reliability, Average Variance Explained and Item-Total Correlation of the Abbreviated Science Anxiety Scale.