You are on page 1of 5

Journal of Global Communication

Vol. 9, No. 2, July-December 2016 : 113-117 IndianJournals.com


A product of Diva Enterprises Pvt. Ltd.
DOI: 10.5958/0976-2442.2016.00019.7

Research Article

Assessing the Reliability of Attitude Scale by Cronbach’s Alpha

Sagar K. Wadkar1*, Khajan Singh2, Ritu Chakravarty3 and Shivaji D. Argade4


1
Assistant Professor – Prime Minister’s Rural Development Fellows Scheme (PMRDFs), Tata Institute of Social Sciences,
Mumbai, Maharashtra, India
2
Principal Scientist, 3Scientist (SS), Dairy Extension Division, ICAR-National Dairy Research Institute, Karnal, Haryana
4
Scientist, ICAR -Central Institute for Women in Agriculture, Bhubaneswar, Odisha, India
*Corresponding author email id: sagarkwadkar@gmail.com

Date of Receipt : 23-12-2015; Date of Acceptance : 25-09-2016

ABSTRACT
Downloaded From IP - 14.139.94.113 on dated 1-Dec-2016

Likert’s summated rating scale is often used in social science researches as an instrument to assess psychological constructs.
Reliability is a vital element in the evaluation of a measurement instrument. The present article highlighted the procedure
Members Copy, Not for Commercial Sale

and application of Cronbach’s alpha coefficient of reliability for assessing the internal consistency of the attitude scale.
www.IndianJournals.com

The list of 54 statements indicating the attitude was prepared and sent to the 60 judges for rating on three point’s
continuums. The ‘t’ values for each statement was found out and 22 statements were retained in the final scale with above
1.75 ‘t’ value. The study revealed that the strong internal consistency ( = .91) among the selected items of the scale. The
study suggested the use of the Cronbach alpha in the social science researches is a good means to estimate the reliability
coefficient of all the items and/or combination of them for the accuracy of measuring instrument.

KEYWORDS: Attitude scale, Cronbach’s alpha, e-Agriservice, Internal consistency, Likert scaling technique, Psychological
construct, Reliability

INTRODUCTION favourable attitude will have the highest scores whereas


individuals with the least favourable (or unfavourable)
Most of the research studies in social sciences, especially
attitudes will have the lowest scores. While not all summated
related to psychological and social aspects such as attitudes,
scales are created according to Likert’s specific procedures,
emotions, opinions and behaviour use scaling technique to
all such scales share the basic logic associated with Likert
quantify these constructs which are not directly measurable.
scale (pp. 22–23). The range of Liker scale captures the
They frequently use multiple-item Likert’s summated rating
intensity of their feelings for a given item. However, the
scale to quantify the construct(s). One benefit of the
result of analysis of multiple items reveals a pattern that
summated scale is its ability to represent the multiple aspects
has scaled properties (Likert, 1932; Jamieson, 2004; Carifio
of a concept in a single measure (Hair et al., 2006). Likert
and Rocco, 2007).
scale is a psychometric response scale, primarily used in
questionnaires to obtain participant’s preferences or degree The development of the scale is not the end of the research,
of agreement with a statement or set of statements. It is a but rather a means to collect data of research variables.
non comparative scaling technique and is unidimensional However, the question of reliability rises as the function of
(only measures a single trait) in nature (Bertram, 2009). scales is stretched to encompass the realm of prediction.
Respondents are asked to indicate their level of agreement While using these types of scales, it is imperative to calculate
with a given statement by way of an ordinal scale, that is reliability coefficient for internal consistency among items
strongly agree (5), agree (4), undecided (3), disagree (2) or subscales, one may be used for the evaluation of the
and strongly disagree (1). The specific responses to the instrument. Cronbach’s coefficient alpha is widely used as
items are combined so that individuals with the most an index of reliability and frequently reported in social and

113 Vol. 9, No. 2, July-December 2016


Sagar K. Wadkar, Khajan Singh, Ritu Chakravarty and Shivaji D. Argade

behavioural studies (Cronbach, 2004; Zumbo and Rupp, judges with their total individual scores were considered
2004; Sijtsma, 2009). It is an index of reliability associated as high group and the bottom 25 per cent as the low group
with the variation accounted by the true score of the so that these two groups provided criterion groups to
‘underlying construct.’ Construct is the hypothetical variable evaluate the individual statements. Thus, out of 40 judges
that is being measured (Hatcher, 1994). to whom the statements were administered for the item
analysis, 10 judges from, each with highest and lowest
Ritter (2010) explained most frequently used reliability
scores were used to evaluate the individual statement. The
estimate, coefficient alpha, so that the coefficient’s
critical ratio, and that is the ‘t’ value which is a measure of
conceptual underpinnings will be understood. Gliem and
the extent to which a given statement differentiates between
Gliem (2003) reported that single-item questions pertaining
the high and low groups of the respondents for each
to a construct are not reliable and should not be used in
statement was calculated by using the formula given by
drawing conclusions by comparing the reliability of a
Edwards (1957). Table 1 shows the procedure for
summated, multi-item scale versus a single-item question.
computing ‘t’ value, and in the same manner, the ‘t’ value
Wuensch (2013) also reported that the Cronbach alpha have
of other statements were worked out.
undertaken the problems of split half and Spearman–Brown
reliability estimate by computing the Spearman–Brown Where,
corrected split-half reliability coefficient for every one of
the possible split-halves, and then find the mean of those XH –XL
t
2 2
coefficients. Therefore, an attempt was made to study and
X   X 
Downloaded From IP - 14.139.94.113 on dated 1-Dec-2016

H –XH L –XL n  n  1
evaluate the reliability of developed attitude scale through
Members Copy, Not for Commercial Sale

Cronbach’s alpha by using Statistical Package for Social 2


 X 
www.IndianJournals.com

Sciences (SPSS). 2
 X   X 2 H
H – XH H 
MATERIALS AND METHODS n
2
The attitude, in this study, was operationalised as the degree 2  X 
 X   X 2 L
of positive or negative feeling of the farmers towards the a L – XL L 
AQUA (Almost All Questions Answered) e-Agriservice. The n
e-Agriservice is operationalizedas “the services (like weather X H = mean score on a given statement for the high group,
and marketing services, input and output price information, X L = mean score on a given statement for the low group,
good agricultural and dairy farming practices, inputs
XH2 = sum of squares of the individual score on a given
availability, and government subsidies and schemes)
statement for high group, XL2 = sum of squares of the
provided by ICT tools via internet in online as well as
individual score on a given statement for low group, XH =
offline mode to the target group” (Wadkar et al., 2015).
summation of scores on given statement for high group,
Understanding the processes of human attitude is crucial
XL = summation of scores on given statement for low
to understanding behaviour. The method of summated
group, n = number of judges in low and high groups, t =
rating suggested by Likert (1932) and Edwards (1957) was
extent to which a given statement differentiates between
followed in the development of scale. A list of 54 items
the high and low groups,  = summation.
indicating the attitude of farmers towards the e-Agriservice
was prepared and sent to the 60 judges for rating on three The statements with more than 1.75 ‘t’ value were retained
point’s continuums. The term item is used throughout this for further investigation. Thus, selected 22 statements were
article, but items could be anything – statements, questions included in the final scale (as mentioned in Table 2).
– indicators of which one might ask to what extent they
A scale is reliable when it gives consistently the same results
measure the same thing. Out of 60 judges, 40 judges had
when applied to the same sample. The Cronbach alpha
returned the same set of statements after duly recording
coefficient of reliability test was employed by using the
their judgements in a stipulated span of 2 months. These
SPSS software. Thus, final set of the 22 statements was
responses were considered for the item analysis.
administered on five-point continuums to a group of 40
Based upon the total individual scores, the 40 judges score users of the e-Agriservice from non-sample area and which
was arranged in descending order. The top 25 per cent of was not included in the actual sample size of the study.

114 Vol. 9, No. 2, July-December 2016


Assessing the Reliability of Attitude Scale by Cronbach’s Alpha

Table 1: Example to show the procedure of computing ‘t’ value


Statement Response Category Highest Lowest
2
X F fX fX X f fX fX2
The aAQUA e-Agriservice Agree 3 8 24 72 3 3 9 27
provides answers to the Uncertain 2 2 4 8 2 7 14 28
query within time Disagree 1 0 0 0 1 0 0 0
10 28 80 10 23 55
nH XH XH 2 nL XL XL 2

Table 2: A list of selected statements with their respective ‘t’ values


S.No. Statements t value
1 The e-Agriservice empowers me to have control over works 3.59
2 The e-Agriservice improves efficiency of experts and extension workers in reaching a large number of 3.21
farmers with less effort
3 It enhances users effectiveness about dairy farming 2.69
4* The services provided by the e-Agriservice are not realistic and worthwhile 2.59
5* The e-Agriservice is more of propaganda & less usage for dairy farmers 2.30
Downloaded From IP - 14.139.94.113 on dated 1-Dec-2016

6* I could have contacted other source for dairy-related queries 2.30


7 The aAQUA e-Agriservice is alternative to the present dairy extension system 1.80
Members Copy, Not for Commercial Sale

8* The e-Agriservice does not improve the knowledge regarding different aspects of dairy farming 1.76
www.IndianJournals.com

9 It is not just the agro-advisory service but also develops my capability in dairy farming 3.64
10* The e-Agriservice cannot meet location specific needs of the farmers 3.25
11 The aAQUA e-Agriservice provides answers to the farmers’ queries within time 2.47
12* Availing the e-Agriservice facility is a time-consuming activity 2.40
13 The service provider helps to retain and attract new users with the efficient mobilisation of its activities 2.40
14* The internet unavailability obstructs the access and utilisation of the e-Agriservice by the farmers 2.09
15* The techno-savvy people can benefit more from the aAQUA e-Agriservice 4.43
16 It helps to generate employment opportunities among farming community 3.28
17 The e-Agriservice helps to develop self-reliance among farming community 3.17
18* The aAQUA e-Agriservice should be stopped 3.15
19 It aids to increase income which leads to enhance standard of living 2.68
20* The aAQUA e-Agriservice alone would solve the problems of farmers 2.09
21 It is the best means to collect information on market prices of agricultural and non-agricultural products 1.90
22 The weather services provided by the e-Agriservice are satisfactory 1.76
*Negative statements

The total individual score of each user was calculated by ∑ =1 2


summing up the responses given to all the statements and = 1− 2
total item variance was calculated by summing up of all −1
users’ responses to the particular item. Where,  = Cronbach’s alpha reliability coefficient, K =
RESULTS AND DISCUSSION Number of items,  yi = the variance of item i for the
current sample of persons,  y i = the variance of the
Cronbach’s Alpha: An Index of Reliability observed total test scores.
The collected data tabulated and analysed to estimate the The statistics of respondents were calculated, which
alpha value. The alpha value calculated by using formula includes the mean of users score (45.80), variance of score
as follows: (  x ) 92.27 and sum of item variance (  y i) 12.15.

Journal of Global Communication 115


Sagar K. Wadkar, Khajan Singh, Ritu Chakravarty and Shivaji D. Argade

Cronbach’s alpha was found to be excellent .910, which is Item22). However, the perusal of Table 3 indicated that the
very high and indicates strong internal consistency among Item8 and Item22 had a weak correlation for item-analysis
the 22 items. Essentially, this means that respondents who purposes (r = .291 and r = .297, respectively).
tended to select high scores for one item also tended to
select high scores for the others; similarly, respondents Table 3 also highlights the Cronbach’s alpha that would
who selected a low score for one item tended to select low result if a given item was deleted. It shows that, the alpha
scores for the other attitude statements. Thus, knowing value if the given item were not included among a set of
the score for one attitude statement would enable one to items. For example, for Item1, the Cronbach’s alpha if
predict with some accuracy the possible scores for the Item1 was deleted would drop from the overall total of
other attitude statements. .910 to .903. It explains that, since the alpha would drop
with the removal of first attitude statements (Item1), which
Table 3: Item total statistics appears to be useful and contribute to the overall reliability
Scale Scale Corrected Cronbach’s of the attitude scale. However, the Item8 and Item22 had a
Mean Variance Item- Alpha if weak correlation with the combine score of remaining
if Item if Item Total Item items, the alpha does not increase by a large degree from
Deleted Deleted Correlation Deleted
deleting both these items (i.e.  = .911 in both the cases).
Item1 43.90 82.621 .630 .903 Therefore, it can be concluded that, there is no need to
Item2 43.70 85.905 .393 .909 eliminate these two items from a total set of 22 items for
Item3 43.40 84.674 .570 .905
Downloaded From IP - 14.139.94.113 on dated 1-Dec-2016

further data collection from the actual respondents of the


Item4 43.95 83.629 .594 .904 study area, to assess the attitude of the farmers towards
Members Copy, Not for Commercial Sale

Item5 44.15 86.239 .396 .909 the e-Agriservice. George and Mallery (2003) provided the
www.IndianJournals.com

Item6 43.45 85.418 .517 .906 following rules of thumb: as the value of Cronbach’s alpha
Item7 43.80 81.853 .681 .902 >.9 – excellent, >.8 – good, >.7 – acceptable, >.6 –
Item8 43.95 88.155 .291 .911 questionable, >.5 – poor and <.5 – unacceptable (p. 231).
Item9 44.00 81.789 .649 .903 While increasing the value of alpha is partially dependent
Item10 43.70 81.484 .715 .901
upon the number of items in the scale, it should be noted
that this has diminishing returns. In present developed
Item11 44.00 86.316 .423 .908
attitude scale, the alpha value was found excellent, which
Item12 43.25 87.671 .454 .908
indicates the strong internal consistency among the set of
Item13 43.60 85.411 .400 .909
items. Thus, Cronbach’s alpha found that items used in
Item14 43.35 84.134 .543 .906 scale for data collection were appropriate and reliable. Barua
Item15 44.30 83.379 .670 .903 (2013) also suggested that Cronbach’s alpha can be used
Item16 43.90 79.568 .788 .899 for setting up a cut-off point to arrive at a diagnosis in a
Item17 43.65 86.029 .370 .910 newly developed instrument with ordinal variables which
Item18 43.65 81.397 .697 .902 does not have any gold-standard instrument for comparison.
Item19 43.45 87.734 .325 .910
CONCLUSION
Item20 43.60 82.779 .637 .903
Item21 43.65 83.924 .744 .902 This paper has demonstrated the procedure for calculating,
Item22 43.40 87.516 .297 .911 interpreting and reporting the reliability of developed attitude
scale by using Cronbach’s alpha. This measure of reliability
Table 3 highlights the column containing the ‘Corrected focuses on the internal consistency of the set of items
Item-Total Correlation’ for each of the items. It indicates forming the scale. Cronbach’s alpha is the average of all
the correlation between a given attitude item and the sum possible split-half coefficients resulting from different ways
score of the remaining items. For example, the correlation of splitting the scale items. It emphasised that reliability
between attitude Item1 and the sum of items2 to 22 is r = tests are especially important when derived variables are
.630. This indicated that, there was average and positive intended to be used for subsequent predictive analyses.
correlation between the scores on the one item (Item1) The present developed scale shows better reliability and
and the combined score of the remaining items (Item2 to has strong and positive correlation between all the items,

116 Vol. 9, No. 2, July-December 2016


Assessing the Reliability of Attitude Scale by Cronbach’s Alpha

so there is no need to re-examine and modify the individual Gliem JA and Gliem RR, 2003. Calculating, interpreting, and
items for further investigation. The developed attitude scale reporting Cronbach’s alpha reliability coefficient for Likert-type
scales. Paper Presented at the Midwest Research-to-Practice
would be useful to explore farmers’ reactions on other Conference in Adult, Continuing, and Community Education, 8–
Information and Communication Technology (ICT) based 10 October 2003. Columbus, OH: The Ohio State University.
e-Agriservices. The study suggested to use alpha value of Hair JF, Black WC, Babin BJ, Anderson RE and Tatham RL,
the reliability coefficient for scales used in the social 2006. Multivariate data analysis (6th ed.). Upper Saddle River,
sciences researches for its validity and accuracy for the New Jersey: Pearson Education, Inc.
appropriate data collection and subsequent interpretation Hatcher L, 1994. A step-by-step approach to using the SAS(R) system
of the results. for factor analysis and structural equation modeling. Cary, NC:
SAS Institute.
ACKNOWLEDGEMENT Jamieson S, 2004. Likert scales: How to (ab) use them. Medical
Education, Vol. 38, No. 12, pp. 1217–1218.
The author would like to thank the University Grant
Commission (UGC), Government of India for providing Likert R, 1932. A technique for the measurement of attitudes. Archives
financial help to carry out Ph.D. research work. of psychology. New York: Columbia University Press.
Ritter NL, 2010. Understanding a widely misunderstood statistic:
REFERENCES Cronbach’s alpha. Paper Presented at the Annual Meeting of the
Southwest Educational Research Association, 18 February 2010.
Barua A, 2013. Methods for decision-making in survey questionnaires New Orleans.
based on Likert scale. Journal of Asian Scientific Research, Vol. 3,
Downloaded From IP - 14.139.94.113 on dated 1-Dec-2016

No. 1, pp. 35–38. Sijtsma K, 2009. On the use, the misuse, and the very limited usefulness
of Cronbach’s alpha. Psychometrika, Vol. 74, pp. 107–120.
Bertram D, 2009. Likert scales. CPSC 681 – Topic Report, The Faculty
Members Copy, Not for Commercial Sale

of Mathematics University of Belgrad. Wadkar SK, Singh K, Mohammad A, Malhotra R and Kale RB,
www.IndianJournals.com

2015. Identifying the factors governing attitude towards the e-


Carifio J and Rocco J, 2007. Ten common misunderstandings, Agriservice among dairy farmers in Maharashtra, India, Journal
misconceptions, persistent myths and urban legends about Likert of Agriculture and Rural Development in the Tropics and
scales and Likert response formats and their antidotes. Journal of Subtropics, Vol. 117, No. 1, pp. 1–10.
Social Sciences, Vol. 3, No. 3, pp. 106–116.
Wuensch KL, 2013. Cronbach’s alpha and maximized Lambda4.
Cronbach LJ, 2004. My current thoughts on coefficient alpha and Available on [http://core.ecu.edu/psyc/wuenschk/MV/Alpha.docx].
successor procedures. Educational and Psychological
Measurement, Vol. 64, pp. 391–418. Zumbo BD and Rupp AA, 2004. Responsible modelling of measurement
data for appropriate inferences: Important advances in reliability
Edwards AL, 1957. Techniques of attitude scale construction. Ballard and validity theory. In: Kaplan D, ed. The Sage Handbook of
Estate, Mumbai: Vakils, Feffer and Simons Private Ltd. Quantitative Methodology for the Social Sciences. pp. 73–92,
George D and Mallery P, 2003. SPSS for Windows step by step: A Sage, Thousand Oaks, CA.
simple guide and reference. 11.0 update (4th ed.). Boston: Allyn &
Bacon.

Journal of Global Communication 117

You might also like