You are on page 1of 53

Chapter 8 Page 335

CHAPTER 8: DATA ANALYSIS OF THE ISLAMIC ACCOUNTING


QUESTIONNAIRE

8.0 INTRODUCTION

This chapter consists of two main parts. The underlying conditions for applying statistical

techniques especially considerations of the research design, the scales used, the

independence of the sample groups, and the distribution characteristics of the underlying

populations, are discussed first in section 8.1. A discussion of the validity and reliability of

the Islamic Accounting Questionnaire in section 8.2 follow this. In the second part of this

chapter, starting with section 8.3, an analysis of the dependent variables of the Islamic

Accounting Questionnaire (IAQ) is presented. Interpretations of the results of this analysis

are presented in section 8.4. Section 8.5 presents some descriptive statistics on the

independent variables while the results of Kruskal-Wallis tests of significance are presented

in section 8.5. The chapter is concluded in section 8.6 with a summary of the main findings.

8.1 RESEARCH DESIGN, VARIABLES, SCALES AND STATISTICAL


TECHNIQUES

8.1.1 Research Design

Bryman & Cramer (1997) outline two types of research designs: the experimental and the

survey/correlational design. In the experimental research design, the researcher can

control and manipulate the variables of interest by having a control group and varying the

treatment or stimulus. He can also control the sequencing of the treatment or stimulus i.e.

their time order. This is important to establish cause effect relationships, which can not only

be used to explain phenomenon but to predict and control future occurrences.

In the survey/correlational design, which is adopted in this research, the researcher is

unable to control and manipulate the variables, for example, it is impossible for the
Chapter 8 Page 336

researcher to control the social accounting education or the working experience variables of

the professional accountants and academics surveyed in this study. What concerns us

here is that “the research design characteristics permeate and inform a number of stages of

the research process” and “has implications for the kinds of statistical manipulation (i.e the

type of tests) which can be performed on the resulting data” (Bryman & Cramer, 1997, p 5).

The experimental design is said to be especially strong in respect of internal validity. An

internally valid study is said to provide firm evidence of cause and effect. According to

Bryman & Cramer (1997), three criteria have to be fulfilled to establish cause and effect

namely; apparent relationships, non-spuriousness and time order of the two related

variables to establish which is the cause and which is the effect.

Experimental designs achieve a high degree of internal validity because these allow the

researcher to eliminate contaminating factors through the use of a control group and

randomly assigning cases to the test and control groups. The survey design used in this

research, however, cannot be said to have the same degree of internal validity as an

experimental design, as in this case, data on a large number of variables is collected

concurrently and the independent variables cannot be manipulated. The concurrent

collection of data also makes it impossible to establish a time order to the variables in

question as compared to the experimental design where manipulated independent variables

can be directly observed. However survey design can reveal relationships between variables

although it is limited in its capacity to elucidate causal processes. These relationships can

however be spurious as another unidentified variable may cause both these variables to

move with each other, which cannot be controlled for as in the experimental design, or the

relationship can be purely circumstantial. The spuriousness or otherwise of relationships

between variables can be tested by using multivariate techniques, which will not be

undertaken in this research as it is exploratory. However, establishing the putative cause

and effect may be very difficult in survey design as we cannot establish which variable is the

cause and which variable is the effect as the relationship can be in either way. For example,

it is difficult to establish whether a favourable attitude to Islamic accounting leads to a


Chapter 8 Page 337

disaffection with conventional accounting or whether disaffection with conventional

accounting causes a favourable disposition to Islamic accounting.

In many cases, however, the direction of cause and effect is not a problem, especially when

individual characteristics of respondents are concerned. For example, if it is found that

religion of respondent and favourable disposition to Islamic accounting are related, then it is

reasonable to assume that it is religion which is the cause and the disposition to Islamic

Accounting which is the effect and not the other way round!

This research is mainly exploratory in nature and seeks to explore the attitudes and

perceptions of Malaysian accountants and academics towards Islamic and conventional

accounting and their ethical stance regarding Muslim business organisations. It is more

concerned to see the majority opinion or perception rather than cause effect relationships of

the independent variables themselves.

8.1.2 Variables

Variables are those measures that indicate the extent of the concept. It is named as such

because they vary with respect to the case or units of analysis i.e. not all accountants think

that maximising profits is desirable. The may also vary with respect to the same case, over

time. If there is no variation in the measure, it is a constant. Constants are very scarce in

the social sciences because the human dimension is very varied in its behavioural

responses. In social science research, we try to measure the variation that variables exhibit.

In univariate analysis, one variable at a time is analysed to find out how the cases or

persons in this research are distributed in relation to a single variable. In multivariate

analysis, we analyse more than one variable at a time.

Variables can be classified into dependent and independent variables. Dependent variables

are those that vary because of the causative effect of independent variables i.e. independent

variables cause the variation in the dependent variable. These are the variables that are of

primary interest to the researcher as he is trying to predict or explain this variable. Solutions

to problems may be obtained by trying to analyse the variation of the dependent variable
Chapter 8 Page 338

(Sekaran, 1992). Independent variables are those that influence the dependent variable

positively or negatively.

Sekaran (1992) also defines two other variable types i.e. Moderating and intervening

variables. A Moderating variable is one that has a strong contingent effect on the

independent-dependent variable relationship. The presence of this variable modifies the

originally expected relationship between the independent and dependent variables

(Sekaran, 1992). For example, in this study it could be hypothesised that being Muslim (the

independent variable) may cause the person to be more disposed towards Islamic

accounting. However, social and environmental education may make him much more so.

Intervening variables are those that arise from the independent variable between the time

the independent variable act to produce the effect on the dependent variable. It introduces

a time dimension and better explains cause and effect relationships. For example, social

accounting education (the moderating variable) may sharpen, refine and broaden the Islamic

ethical concepts of Muslim accountants (the intervening variable) to be more disposed

towards business ethics of Muslim business organisation. (See figure 8 -1).

Sharpened ethical Favourable disposition to


concepts Islamic Accounting
Muslim
(intervening variable)

Social Accounting Education.


(moderating variable)

FIGURE 8-1:MODERATING AND INTERVENING VARIABLES

8.1.3 Scales or levels of measurement

The variables discussed above can take certain scale types or levels of measurement

(Siegel & Castellan, 1988). They inform us that behavioural scientists, taking physics as a

model, assign numbers to behavioural or social variables to enable data analysis and

manipulation. He warns us that in order to perform certain arithmetic operations such as

adding and squaring, the structure of the method of mapping numbers (scoring) to social
Chapter 8 Page 339

observations should be isomorphic to some numerical structure which includes these

operations. The interpretable operations on a given set of scores are dependent on the level

of measurement achieved (Siegel & Castellan, 1988). This has also implications for the

techniques that can be used to analyse data.

Many statisticians have divided the level of measurement scales into four categories;

nominal or categorical, ordinal, ratio, and interval scales. The weakest level of measurement

is the nominal or categorical scale, which uses numbers to classify an object, person or

characteristic into different mutually exclusive categories. For example, 1= Muslim, 0= Non

Muslim is used in section 4 of the Islamic Accounting Questionnaire. Since any set of

symbols may equally well represent the classification, the symbols can be interchanged, if

done so consistently, without altering the information in the scale. For example, religious

classification in the IAQ could easily be reclassified into 1 for Muslim and 2 for Non Muslim.

This means that the only kinds of statistics admissible are descriptive statistics, which would

be unchanged by such a transformation, such as mode and frequency counts. In certain

cases non-parametric tests such as χ2 can be used to test hypotheses regarding distribution

of cases among categories. However mathematical operations such as additions cannot be

performed, as it would be meaningless.

The ordinal or ranking scale is the next level of measurement. In this scale objects in one

category of the scale are not only different from other categories but have some kind of

relation, such as magnitude or order, among them. The categories have relations of ‘more

than’ or ‘less than’ to other categories. However, the difference between the scales may not

be equal. For example in the Likert scale Strongly Agree although given a scale of 5

compared to Agree (given a 4) does not mean that SA shows twice the agreement of Agree,

as different unequal numbers could easily have been given such as 5 for agree and 7 for

strongly agree.

The ordinal scale is order preserving if transformed as such does not change the information

contained in the ordinal scale. The statistic most appropriate for this is the median as it is not

changed by changes in scores above or below it. With ordinal scaling, hypotheses can be
Chapter 8 Page 340

tested by a large group of non-parametric or distribution-free tests. For the application of

parametric tests, there needs to be an assumption of normality of the underlying population.

The Interval scale is one where the different objects, not only has an ordering between

themselves, but the differences or distances between any two numbers on the same scale

has meaning. This scale has no absolute zero. In this scale, the ratio of any two intervals is

independent of the unit used and the zero point. An example of an interval scale is

temperature be it Celsius or Fahrenheit. It can be shown that the ratio of the differences

between readings one scale is equal to the ratio of the equivalent differences on the other

scale. In the case of temperature the ratio between 30 and 10 and 10 and 0 degrees Celsius

is the same as the ratio between the equivalent 86 and 50 and 50 and 32 degrees

Fahrenheit1.

Since any change in the numbers associated with the positions of the objects in a interval

scale must preserve, not only the positions, but also the relative differences between the

objects. This scale is not affected by multiplying by a positive constant and adding a

constant to this product. Since this is a truly quantitative scale, all the common parametric

statistics such as means and standard deviations can be applied to data measured on this

scale. If the data meets two further conditions i.e. normally distributed population and if the

observations are independent then the researcher can utilise normal parametric tests such

as the t test and F test. Using non-parametric tests in such cases would not utilise all the

information contained in the data set.

The Ratio Scale has all the characteristics of the interval scale, but in addition, has an

absolute zero, for example, age and mass. The numbers used in this scale are true

numbers with a true zero. The ratio between any two numbers are preserved when

multiplied by any positive constant and such transformation does not alter the information

contained in the scale. Any parametric test can be used when this scale is achieved but it is

rare in the social sciences.

1
30-10/(10-0)=86-50/50-32
Chapter 8 Page 341

Bryman & Cramer (1997) argue that, although the 1 to 5 scale and the Likert scale (such as

that used in the IAQ for sections 1 to 3) are, strictly speaking, ordinal scales, many

researchers treat them as interval scales. “Thus, there seems to be a trend in the direction of

this more liberal treatment of multiple-item scales as having the qualities of interval variables

although many purists would demur from this position. Moreover, there does not appear to

be a rule of thumb which allows the analyst to specify when a variable is definitely ordinal

and when interval.” In this research, it is proposed to follow current practice and to treat

multiple-item measures such as those used in the IAQ as though they were interval scales.

Labovitz (1970) also argued for the treatment of ordinal data as interval in view of the

considerable advantages that can accrue to the researcher as a result of using analytical

techniques such as correlation and regression which are both powerful and easy to interpret.

Further he argued that the amount of error that can occur is minimal.

Bryman & Cramer (1997) further argue that this process of treating ordinal data as interval

becomes more compelling when the number of categories is considerably greater. Further

Sekaran (1992) informs us that research already undertaken indicates that a slight increase

in the number of items in a multiple-item scale does not improve the reliability of the ratings.

Thus the Likert and 1 to 5 scale, used in this research, is treated as interval to allow the

researcher to perform non-parametric statistics. However, to ensure this does not result in

very much difference, parametric tests are also carried out.

8.1.4 Measurement scales

Measurement is the attribution of numbers to the units of analysis, which represents the

concept or phenomena, the researcher is trying to discern. Measurement assists in the

specification differences in the level of the concept found by allowing systematic differences

between the object such as people studied (Bryman & Cramer, 1997). However, concepts

such as the perceptions of ‘ethical position of Muslim Business Organisations’, usually need

several indicators to reflect the concept fully, as there are usually many facets through
Chapter 8 Page 342

which a concept might be expressed. A single indicator may leave out these other

dimensions that together measure the concept.

In the Islamic accounting questionnaire, three main concepts are measured viz. the ethical

position of business in section 1, the perceptions on conventional accounting in section 2

and the perceptions on Islamic accounting in section 3. Section 3 may be divided into 3 sub

concepts i.e. the objectives of Islamic accounting, the importance of various stakeholders

and the nature and content on Islamic accounting. To measure these concepts various

indicators (each represented by a separate group of questions) are used.

In section 1, indicators regarding the Shari’ah indicate the ethical position of Islamic

business i.e. the rules/framework under which Muslim organisations operate, investment

policy, belief in the ability of Islamic principles to be developed and operational policies of

Muslim business organisations. These are in turn subdivided into more refined indicators by

means of statements. Likert scaling is used in some of these and a 1 to 5 scaling is used in

other questions. By using aggregate scores of the related items, a more comprehensive

integrated measure can be obtained for individual cases to better discriminate the

participants as opposed to a situation in which fewer indicators are used to measure the

concept. It also reduces the chances of a misunderstood question (as happened in this

research) invalidating the research as the other indicators may save the day.

8.1.5 Statistical Techniques

Statisticians, mathematicians and even social scientists have developed hundreds of

statistical tests, which one can apply in data analysis. Computer statistical packages such as

SPSS, which have been developed especially for the social sciences makes the number

crunching easy and fast. However the bewildering assortment of tests available have made

the researcher spoilt for choice. However, not all tests are equal, neither are they equally

applicable in all circumstances. Statistical tests can be basically divided into parametric and

non- parametric tests. The former tests are used when the populations from which the

sample is derived, have normal or approximately normal distributions. Further if two or more
Chapter 8 Page 343

samples are being compared, then in order to use parametric tests, not only should both the

populations from which the samples are derived be normal but also their variances should

be equal. These restrictive assumptions are not required for Non-parametric tests. These

tests (sometime referred to as ‘distribution-free’ tests) require only that the distributions of

the populations from which they are derived are similar in shape (Siegel & Castellan, 1988).

Inferential statistics are based on probability. Here we try to deduce characteristics of the

populations based on the characteristics of the sample. The way this is done is to set up the

hypothesis we want to reject as the H0, the null hypothesis and the hypothesis we want to

accept as the alternative hypothesis, H1. The decision is to reject Ho in favour of H1 when a

statistical test yields a value whose probability of occurrence, if Ho is true is below a given

small probability level, known as the significance level and denoted by α. This is usually .05

or .01 representing 5% or 1%. There is always a chance that we are wrong because we can

never be sure that the sample represents the population exactly.

There are two types of errors which one can make in testing hypotheses using statistical

techniques; Type I and Type II error. Type I error (denoted as α) is the probability of

rejecting the null hypothesis when in fact it is true. This is the same as the significance level

above. Type II error (denoted as β) is the probability of accepting the null hypothesis when it

is false. There is an inverse relationship between Type I and Type II errors. This means that

a decrease in α will increase β for any given sample size. The only way to decrease both

errors will be to increase the sample size (Siegel & Castellan, 1988). The researcher may

have to balance the probabilities of making both types of errors. Different statistical tests

give different balances between committing these two errors. Here is where the power of a

test becomes important.

The power of a test is defined as the probability of rejecting Ho, when it is in fact false i.e.

Power = 1-β. Therefore the more powerful the test, the better. The power of a test

increases with the size of the sample, and also depends on the type of test. Besides the

power, the choice of statistical test used depends on the research design, the scales used,
Chapter 8 Page 344

the distribution and variance of the populations from which the sample is derived. These will

be discussed beginning in the next section. Finally, a more difficult part of data analysis is

the interpretation of the results. It is only too easy in this age of computerised statistical

packages to compute statistics, which are totally meaningless such as the median colour of

a car. The interpretations the researcher gives to the results of the data analysis depend, as

for the choice of the statistical technique to be used, on the research design, the statistical

model, the scale and measurements used and the statistical technique applied.

8.2 VALIDITY AND RELIABILITY

Whenever measurement takes place, there is always the possibility of errors. These errors

could be procedural errors, where the way in which the research was carried out was not

proper or there might be problems with the measuring instrument itself. In order for research

to be rigorous i.e. its results acceptable to the academic community, the validity and

reliability of the research results must be demonstrated as far as possible.

When the error which occurs with measurement is random i.e. varies randomly between

cases, in some case positive and some cases negative then the observed mean score is an

unbiased estimate of the true mean despite the measurement errors (Summers 1977, p 15).

However, if the same measure is given repeatedly to the same person, it may produce a

different set of scores. This is said to affect the reliability of the measurement. Reliability of a

measuring instrument refers to the consistency of the scores produced on repeated

measurement. Thus, the presence of random errors affects the reliability or consistency of

the measuring instrument.

However, errors need not be random. If an observed score is systematically affected by

another variable to which the instrument is not sensitive, it is said to affect the validity of the

measurement procedure. Hence measurement of validity of an instrument measures both

random and constant errors whereas reliability measurements only measure random errors.

8.2.1 Validity
Chapter 8 Page 345

Validity means that the findings of the research are in agreement with the theoretical or

conceptual values. In other words, it is the ability produce results that are accurate and to

measure what is supposed to be measured (Sarantakos, 1998). For example, if we find that

80% of Malaysian Accountants strongly agree on the need for an alternative Islamic

accounting, then it is actually 80% who are in agreement and not, say 30%.

Validity indicates the degree to which an instrument measures the construct which is under

investigation (Bohrnstedt, 1977). According to Sekaran (1992) content validity ensures that

“the measure includes an adequate and representative set of measures that would tap the

concept. The more the scale items represent the domain or universe of the concept, the

greater the content validity” (p 171). Content validity measures how well the dimensions and

elements of a concept have been delineated. He suggests that face validity is a minimum

index of content validity i.e. whether the items look like they are measuring what they are

supposed to measure.

Measuring face validity is a subjective procedure wherein experts give their judgements on

whether the items are valid. As mentioned in chapter 7, the IAQ went through eight revisions

and a pilot study. The staff and postgraduate students both Muslim and Non Muslim, of the

Department of Accountancy of the University of Dundee were consulted in the process of

designing the questionnaire and it went through a metamorphosis. As a result, the face

validity of the IAQ was judged to be adequate.

Bohrnstedt (1977) suggests the use of cluster analysis to analyse the content validity of the

items in the scale, after dividing the items into strata. If the items in a stratum correlates

higher than with the items in other strata, then the items have content validity for that

dimension. The theory is that if different items are measuring the same attitude or trait, then

the underlying trait should be causing the covariance among the items, The higher the

correlation, the better the items are measuring the construct.

This was undertaken in the IAQ as follows; for each section of the IAQ, a cluster analysis

was performed using SPSS to see how well the questions fit together. A dendrogram was

printed to indicate which clusters hang together with others to identify the various strata.
Chapter 8 Page 346

The dendrogram in figure 8-2 shows that the items in section 1 of the IAQ can be classified

into 3 clusters below 5 distance cluster level. Questions 1.5a-d, Q.1.7a – d and Q1.3a-c,

Q1.2, and Q1.6 fit very well together, as they all associate below the 5 distance cluster level.

. Further Q1.1 and 1.4 seem to be related but not that closely2. The findings are line with the

concepts behind the Questionnaire and hence suggest that the questions in section 1 do

have content validity.

2
A principal component analysis show that Q1.1 and Q1.4 measure two different concepts. A reliability
analysis gives an alpha value of –0.0124 indicating very insignificant negative correlation.
Chapter 8 Page 347

* * * * * * H I E R A R C H I C A L C L U S T E R A N A L Y S I S
Rescaled Distance Cluster Combine

C A S E 0 5 10 15 20 25
Label Num +---------+---------+---------+---------+---------+

Q1.7A 12 -+
Q1.7B 13 -+
Q1.7D 15 -+-----+
Q1.7C 14 -+ +-----+
Q1.2 2 -+-+ I I
Q1.3B 4 -+ +---+ I
Q1.3A 3 ---+ +-----------------------------------+
Q1.6 11 ---+ I I
Q1.3C 5 ---+ I I
Q1.1 1 -------+-----+ I
Q1.4 6 -------+ I
Q1.5C 9 -+-+ I
Q1.5D 10 -+ I I
Q1.5A 7 ---+---------------------------------------------+
Q1.5B 8 ---+

FIGURE 8-2 :DENDROGRAM USING WARD METHOD FOR SECTION 1 OF THE IAQ

For section 2, of the questionnaire, the dendogram in figure 8-3 shows that only Q2.1-Q2.3

and Q2.6a-Q2.6b fit together below the 5 distance cluster level. However principal

component analysis of Q2.1-Q2.4, Q2.6a-Q2.6c and Q2.5b-Q2.5c shows only one

component above an eigenvalue of one. This means the questions in these three groups

measure the same concept within the group. However, rQ2.5a seem to measure a different

concept and hence is not combined together for hypothesis testing.

* * * * * * H I E R A R C H I C A L C L U S T E R A N A L Y S I S
Dendrogram using Average Linkage (Between Groups)
Rescaled Distance Cluster Combine

C A S E 0 5 10 15 20 25
Label Num +---------+---------+---------+---------+---------+
Q2.2 2 -+-+
Q2.3 3 -+ +---------+
Q2.1 1 ---+ +-----+
Q2.4 4 -------------+ +-----+
Q2.5C 6 -------------------+ +-----+
Q2.5B 5 -------------------------+ +-----------------+
RQ2.5A 10 -------------------------------+ I
Q2.6A 7 -----+---------------------+ I
Q2.6B 8 -----+ +---------------------+
Q2.6C 9 ---------------------------+
FIGURE 8-3: DENDROGRAM USING WARD METHOD FOR SECTION 2 OF THE IAQ
Chapter 8 Page 348

For section 3, on the objectives and characteristics of Islamic accounting, the dendogram in

figure 8-4 shows that Q3.2a and b are grouped together (the financial objectives) are

distinguished from Q3.2c, d, e and f (non-financial objectives, which also form a group).

Q3.3a-e appear to be related at around 5-distance cluster level. Questions Q3.4a, b,c,d and

e also form a close group near the 5-distance cluster level while Q3.4f seems to be related

to the group at slightly above 5-distance cluster level. Perhaps this is reflective of the fact

that nationality issues are not considered Islamic issues. Q3.6 to Q.3.13 can be taken

together as except for Q3.10, q3.7 and Q3.11are related much less closely at 8 or 10

distance level. However, the reliability test in table 8-1 shows all the questions Q3.4a-f and

Q3.6 –Q3.13 measure the same concept. Hence, as a whole, this section can be said to

have content validity.

* * * * * * H I E R A R C H I C A L C L U S T E R A N A L Y S I S
Dendrogram using Average Linkage (Between Groups)
Rescaled Distance Cluster Combine

C A S E 0 5 10 15 20 25
Label Num +---------+---------+---------+---------+---------+

Q3.2D 2 -+-+
Q3.2E 3 -+ +-+
Q3.4D 13 ---+ +-+
Q3.2C 1 -----+ I
Q3.12 23 -+-+ I
Q3.13 24 -+ I I
Q3.6 17 ---+-+ I
Q3.8 19 ---+ +-+-------+
Q3.4E 14 ---+ I I I
Q3.9 20 -----+ I I
Q3.2F 4 -------+ +-+
Q3.10 21 -------+ I I
Q3.4B 11 -----+-+ I I
Q3.4C 12 -----+ +-----+ I I
Q3.4A 10 -------+ +-+ I
Q3.4F 15 -------------+ +-------------------------------+
Q3.3D 8 -----+-----+ I I
Q3.3E 9 -----+ I I I
Q3.3A 5 -----+---+ +---+ I I
Q3.3B 6 -----+ I I I I I
Q3.3C 7 ---------+-+ +-+ I
Q3.7 18 ---------+ I I
Q3.11 22 ---------------+ I
RQ3.2B 25 -----------+---------+ I
RQ3.2A 26 -----------+ +---------------------------+
Q3.5 16 ---------------------+

FIGURE 8-4: DENDROGRAM USING WARD METHOD FOR SECTION 3 OF THE IAQ
Chapter 8 Page 349

8.2.2 Tests for Reliability

Reliability refers to the consistency of the test to produce the same results on repeated use,

otherwise the measuring instrument would not be accurate over repeated use. Reliability

can be divided into external and internal reliability. External reliability refers to the

consistency of the measure over time. To test for such reliability, the same test could be

administered twice to the same group of participants after some time lapse for test-retest

reliability. This was not an option in this research. Hence the external reliability could not be

tested. However, the internal reliability of the items in the questionnaire was tested using

SPSS alpha model method. This tests gives us an idea of how reliable a group of questions

measure the same concept when they are added together to get an index or aggregate

score, as is undertaken in this research. However, since there are many ways in which we

can split the items in a scale, there may be different measures of reliability depending on

how we split the items of the scale. To get over this, Cronbach’s Alpha, which generally

measures the ‘average’ of all possible split-half reliability coefficients, is calculated. If this

alpha measure results in is 0.8 or above (as a rule of thumb) (Bryman & Cramer, 1997), then

the items are said to be internally reliable.

From table 8-1 it can be seen that most of the groups of Questions from the IAQ which are

used to compute aggregate scores measures the concept reliably when the questions are

combined together. This is evidenced by the Cronbach’s alpha figure of around 0.8.

Remembering that 0.8 is a rule of thumb, the question groupings measuring the concepts

are highly valid since all the groups except for ETH1 and CAINF are higher than or around

0.8 do not seem to reliably measure the concept. A principal components analysis of the

questions constituting ETH1 and CAINF showed that there was only one principal

component with an eigen value of 1 or more. Hence these questions as a group a reliable3.

3
A Pearson correlation matrix also show that the correlations between the questions are significant except
between Q1.3c and 1.6 which is significant only at the 10% level.
Chapter 8 Page 350

TABLE 8-1
Reliability Analysis for Islamic Accounting Questionnaire
Question Group Aggregate Hypothesis Cronbach’s Alpha
Score
Q1.2,Q1.3a-b, rQ1.3c,Q1.6 ETH1 H1 0.27724
rQ1.5a-rQ1.5d ETH2 H2 0.8466

Q1.7a-d QSREL H3 0.9465

Q2.1-2.4 UNISL H4 0.7611


Q2.5b and Q2.5c CAINF H5 0.57265
Q2.6a-c CAP H6 0.7570
Q3.2a-Q3.2b MFOB,
H8 0.7364
Q3.2c-Q3.2f MNFOB
Q3.3a-e STIMP H9 0.7659
Q3.4a-Q3.4f and Q3.6-Q3.13 SEI H10 0.8509

8.2.3 Sampling Error

In order to enable generalisations to the populations, the characteristics of the sample must

represent the characteristics of the population. In this research, the

respondents for the IAQ were from education (accounting academics) and professional

accountants working public practice, commerce and industry, and government. 270

questionnaires (90 each to accountants in the public sector, in public practice and in

commerce & industry respectively) were mailed through the MIA. Another 157

questionnaires were hand delivered to universities while 35 were delivered to accountants in

audit firms and 12 to accountants in companies making a total of 474 questionnaires

4
On an initial analysis of the responses to Q 1.4 were being answered in the “wrong direction”. The researcher
suspected that the question was not understood. This was confirmed by some participants’ requesting
clarification to the same questions when the questionnaire was administered to some academics. The phrase
“technical interpretation” (see figure 7-2, page 309) was not understood properly and the word its objectives
(the Shari’ah’s) was misinterpreted to be the company’s. Many responded with strongly agree answers. Some
of the respondents who indicated their address were contacted to explain what was meant by Q1.4. As a result
about 70% of the contacted participants changed their answers while others maintained their original response.
As all participants could not be contacted because of the address of all respondents were not known, it was
decided to leave Q1.4 out of the aggregate scores and hypothesis testing. Reliability analysis conducted with
Q1.4 included in this group of question gave a Cronbach’s alpha of 0.1421. Hence, the reliability of including
question 1.4 in this grouping was suspect. Hence, Q1.4 is left out of this grouping and the score ETH1. A
principal components analysis of this group of questions including Q1.1 produces two components with eigen
values above 1, while if Q1.1 is left out of the score, there is only one component with eigen value above 1.
Hence, Q1.1 does not measure the same concept and is left out of the group.
Chapter 8 Page 351

distributed. Of this 206 Questionnaires were returned representing a response rate of about

43%. Of the responding accountants, 46 responses from commerce and industry, 34

responses from public practice and 25 responses were from the government sector. The

other responses were from 101 accounting academics in the Universities. Some of the

questions were partially completed but most were usable for many of the questions. The

sample size of each question therefore varies up to a maximum of 206. Missing values are

taken into account when computing aggregate scores.

Table 8-2 shows the number sent and received, which is used as the basis for the sampling

error analysis:

T ota l in S e c tor T ota l T ota l R e sp on se


S e c tor P op u la tion % Q . Sent R e c e ive d r a te %
E d u c a t io n 441 28 157 101 64
C om m e r c e & In d ustr y 558 36 102 46 45
P u b lic P r a c tic e 231 15 125 34 27
G overn m en t 325 21 90 25 28
1555 100 474 206 43

TABLE 8-2:DISTRIBUTION AND RECEIPT OF ISLAMIC ACCOUNTING QUESTIONNAIRES


From the total population of Muslim accounting academics and accountants6, it can be seen

that the education sector is over represented while the pubic practice and government sector

are underrepresented. A χ2 test (see table 8-3) shows that the differences between actual

responses and the expected responses for the working sector is significantly different.

Hence, the sample composition is not reflective of the population composition. This means

that inferences cannot be made as a whole to the actual population although point estimates

of the population parameters can be made by weighting the means of each working sector

according to the sectoral composition of the population.

χ2 DF P-Value

5
Including rQ2.5a in CAINF gives a Cronbach’ alpha of 0.3888 making the group more unreliable.
6
There were 1840 Muslim accountants registered with MIA in October 1998 when the survey was carried out.
Of these 726 were not working in any of the three sectors tested. They might include retired, dead and other
members who might not be working as accountants but who are still registered with MIA. This group was
overlooked by the researcher and was not sampled in the research. Hence, any results would not be applicable
to them. However, if the assumptions that these members are not active in the accounting field is true, then their
opinion as accountants may not matter. The hypotheses tested would only be applicable to the accountants
working in the three sectors.
Chapter 8 Page 352

50.401 3 0.0000

TABLE 8-3: χ2-TEST FOR SAMPLE REPRESENTATION OF POPULATION SECTORAL


CHARACTERISTICS

Since the researcher wants to hypothesise regarding two groups, the academics and the

professional accountants in the three sectors, it has to be done for each sector. Since the

academics are presented as a separate group, no problem arises from this sampling error.

However, for the professional accountants, the analysis has to be conducted by sector and

each statistic by sector has to be significant before the null hypothesis for the whole

population of accountants can be rejected or otherwise.

8.3 DEPENDENT VARIABLES: FINDINGS

Tables 8-4 to 8-6, shows the response frequencies, the frequency percentages and

descriptive statistics for each question (in sections 1,2 and 3 of the questionnaire), in the

IAQ. The questions in the table are arranged according to the scale used. Table 8-4 shows

the questions using the Likert Scale while Table 8-5 shows the questions using the ranking

1-5 scale. Questions, which have fixed responses, are shown separately in table 8-5. The

questions, which have an opposite direction (i.e. where the researcher expects a low score

to show a favourable disposition to Islamic accounting), are shaded. The unshaded

questions indicate favourable dispositions towards the alternative hypotheses if responses

tend towards agree. The shaded questions indicate favourable dispositions towards the

alternative hypothesis if responses tend towards disagree. In order to make analysis by

sections easier, different sections within the same table are separated by a bold line.

8.3.1 Analysis & Findings of Section 1 of Islamic Accounting Questionnaire (IAQ).

From table 8-4, it can be seen the mean response for section 1 of the IAQ varies between

1.98 to 4.70. Questions 1.3c to Question 1.5d should have low scores for a favourable

disposition to Islamic business ethics. It can be seen that the mean scores for this group of

questions range from 1.98 to 3.84 and are indeed lower compared to the others, generally
Chapter 8 Page 353

supporting the hypothesis No. 1 and 2 see chapter 7) of the researcher. Specifically, the

respondents believed that Islamic business organisations should not involve themselves in

prohibited interest based investments and short selling, and ‘speculative’ investments, such

as options and contra trading, which are ‘gray’ areas in terms of Islamic law, contradicting

the spirit of the Shari’ah. Q1.5a to d covers these areas.

Referring to the response frequencies, more than 73% of valid participants did not agree to

Islamic business organisations participating in short selling and interest based debt market.

As for contra trading and participating in futures/options, about 59% and 40% respectively

disagreed.

This is not surprising as the position of these under Islamic law is still being debated

although a majority of scholars consider short selling prohibited because under Islamic law,

one cannot sell what one does not possess. Futures and options are looked upon with

concern in terms of uncertainty (gharar) and gambling. However, the scholars are do not

want to prohibit it totally because there might be a necessity for futures in case of real need

as opposed to speculative activity in the market. Kamali (1997b) for example has argued that

futures and options may be permissible whereas Usmani (1999) has argued otherwise. Only

about 11% strongly support participation in options markets and about 22% neither agree

nor disagree. However, the researcher’s hypothesis that Islamic business organisations

should not aim to maximise profit for shareholders is not supported by this research. The

response frequencies for this question show about 74% do not support the hypothesis, while

only 11% do. Both a t test for mean=3.84 and sign test of median=4 versus the alternative

hypotheses of mean <3.84 and median <4 show p values of .51 and 0.3085 meaning that

the alternative hypotheses have to be rejected. However, in many cases the participants

indicated (by means of comments in the questionnaire) that maximisation of profits of

shareholders should be subject to the Shari’ah constraint. The overwhelming support for

social welfare as indicated by the response to q1.2 (98% agreed or strongly agreed to this)

with a mean of 4.7 and, a median and mode of 5, shows that the respondents believe that

maximisation of profits should not be at the cost of these other objectives.


Chapter 8 Page 354

The group of questions with the highest favourable frequencies is that concerned with the

perception of the participants with regard to the capability of the socioeconomic principles to

be developed into a modern business and a legal framework for various type of

organisations. The agreement for this varied between 97% to 99% of which about 58%

indicated strong agreement. Although this might reflect the religious beliefs of the

participants rather than any proven capability, religion does have affect on values and

perceptions and subsequently financial transactions (Sa’ada, 1997). This at least indicates

that attempts at secularisation of Muslim society may have failed. There is a yearning and

strong belief in the capability of Islam to meet modern business legislation without losing

sight of ethics.
Chapter 8 Page 356

TABLE 8-4: RESPONSE FREQUENCIES FOR DEPENDENT VARIABLES (LIKERT SCALE) - IAQ
Str. Disagree Disagree SD & Disagree NAD Agree Str. Agree SA & Agree Std. SE Valid
Count % Count % Count % Count % Count % Mean Median Mode Dev. of Mean Cases

Q1.1 30 15.1 12 6.0 42.0 21.1 14 7.0 63 31.7 80 40.2 143 71.9 3.76 4 5 1.42 0.10 N=199
Q1.2 2 1.0 0 0.0 2.0 1.0 2 1.0 50 24.5 150 73.5 200 98.0 4.70 5 5 0.59 0.04 N=204
Q1.3a 0 0.0 6 3.0 6.0 3.0 10 4.9 79 38.9 108 53.2 187 92.1 4.42 5 5 0.72 0.05 N=203
Q1.3b 3 1.5 2 1.0 5.0 2.5 7 3.4 68 33.3 124 60.8 192 94.1 4.51 5 5 0.75 0.05 N=204
Q1.3c 3.0 1.5 20 9.9 23.0 11.3 30 14.8 103 50.7 47 23.2 150 73.9 3.84 4 4 0.94 0.07 N=203
Q1.4 31.0 16.3 31 16.3 62.0 32.6 20 10.5 68 35.8 40 21.1 108 56.8 3.29 4 4 1.39 0.10 N=190
Q1.5a 87.0 45.5 56 29.3 143.0 74.9 21 11.0 19 9.9 8 4.2 27 14.1 1.98 2 1 1.16 0.08 N=191
Q1.5b 35.0 19.0 39 21.2 74.0 40.2 40 21.7 50 27.2 20 10.9 70 38.0 2.90 3 4 1.30 0.10 N=184
Q1.5c 73.0 38.8 65 34.6 138.0 73.4 23 12.2 20 10.6 7 3.7 27 14.4 2.06 2 1 1.13 0.08 N=188
Q1.5d 50.0 27.2 58 31.5 108.0 58.7 34 18.5 31 16.8 11 6.0 42 22.8 2.43 2 2 1.22 0.09 N=184
Q1.6 0 0.0 10 5.0 10.0 5.0 17 8.4 92 45.5 83 41.1 175 86.6 4.23 4 4 0.80 0.06 N=202
Q1.7a 0 0 0 0 0.0 0.0 2 1.0 80 39.8 119 59.2 199 99.0 4.58 5 5 0.51 0.04 N=201
Q1.7b 0 0 0 0 0.0 0.0 4 2.0 79 39.3 118 58.7 197 98.0 4.57 5 5 0.54 0.04 N=201
Q1.7c 0 0 0 0 0.0 0.0 6 3.0 81 40.5 113 56.5 194 97.0 4.54 5 5 0.56 0.04 N=200
Q1.7d 0 0.0 1 0.5 1.0 0.5 5 2.5 77 38.3 118 58.7 195 97.0 4.55 5 5 0.57 0.04 N=201
Q2.1 3 1.5 54.0 27.4 57.0 28.9 40.0 20.3 78 39.6 22.0 11.2 100 50.8 3.31 4 4 1.0 0.07 N=197
Q2.2 4 2.0 22.0 11.0 26.0 13.0 18.0 9.0 103 51.5 53.0 26.5 156 78.0 3.60 4 4 1.0 0.07 N=201
Q2.3 2 1.0 51.0 26.4 53.0 27.5 49.0 25.4 69 35.8 22.0 11.4 91 47.2 3.30 3 4 1.0 0.07 N=193
Q2.4 4.0 2.0 37.0 18.4 41.0 20.4 31.0 15.4 92 45.8 37.0 18.4 129 64.2 3.90 4 4 1.0 0.07 N=200
Q3.6 1.0 0.5 8.0 4.0 9.0 4.5 12.0 6.0 111 55.5 68.0 34.0 179 89.5 4.19 4 4 0.8 0.05 N=200
Q3.7 7.0 3.6 28.0 14.6 35.0 18.2 38.0 19.8 84 43.8 35.0 18.2 119 62.0 3.58 4 4 1.1 0.08 N=192
Q3.8 2.0 1.0 1.0 0.5 3.0 1.5 14.0 7.0 112 56.0 71.0 35.5 183 91.5 4.25 4 4 0.7 0.05 N=200
Q3.9 1.0 0.5 5.0 2.5 6.0 3.0 34.0 17.1 113 56.8 46.0 23.1 159 79.9 3.99 4 4 0.7 0.05 N=199
Q3.10 2.0 1.0 8.0 4.0 10.0 5.0 32.0 16.1 105 52.8 52.0 26.1 157 78.9 3.99 4 4 0.8 0.06 N=199
Q3.11 6.0 3.0 44.0 22.1 50.0 25.1 45.0 22.6 82 41.2 22.0 11.1 104 52.3 3.35 4 4 1.0 0.07 N=199
Q3.12 2.0 1.0 1.0 0.5 3.0 1.5 7.0 3.4 82 40.4 111.0 54.7 193 95.1 4.47 5 5 0.7 0.05 N=203
Q3.13 0.0 0.0 3.0 1.5 3.0 1.5 12.0 6.0 113 56.2 73.0 36.3 186 92.5 4.27 4 4 0.6 0.05 N=201
Chapter 8 Page 357

TABLE 8-5: RESPONSE FREQUENCIES FOR INDEPENDENT VARIABLES (1-5 RANKING SCALE) - IAQ
Not at all 2 (3)= (1) + 3 4 Very much so(6)
(1) (2) (2) (4) (5) (5)+(6)
Count % Count % Count % Count % Count % Count % Count % Mean Median Mode Std S.E.rror Valid N
Deviation of Mean
Q2.5a 31 16.1 61 31.6 92 47.7 67.0 34.7 26 13.5 8 4.1 34 17.6 2.58 3.00 3.00 1.04 0.08 N=193
Q2.5b 23 12.3 49 26.2 72 38.5 57 30.5 37 19.8 21 11.2 58 31.0 2.91 3.00 3.00 1.18 0.09 N=187
Q2.5c 14 7.8 45 25.0 59 32.8 51 28.3 43 23.9 27 15.0 70 38.9 3.13 3.00 3.00 1.18 0.09 N=180
Q2.6a 19 10.2 27 14.5 46 24.7 45 24.2 47 25.3 48 25.8 95 51.1 3.42 4.00 5.00 1.29 0.09 N=186
Q2.6b 15 7.9 22 11.5 37 19.4 39 20.4 55 28.8 60 31.4 115 60.2 3.64 4.00 5.00 1.25 0.09 N=191
Q2.6c 18 9.9 30 16.5 48 26.4 46 25.3 45 24.7 43 23.6 88 48.4 3.36 3.00 3.00 1.28 0.09 N=182
Q3.2a 0 0.0 11 5.4 11 5.4 28 13.7 67 32.7 99 48.3 166 81.0 4.24 4.00 5.00 0.88 0.06 N=205
Q3.2b 8 4.0 26 12.9 34 16.9 73 36.3 56 27.9 38 18.9 94 46.8 3.45 3.00 3.00 1.06 0.07 N=201
Q3.2c 1 0.5 4 2.0 5 2.5 22 10.9 55 27.2 120 59.4 175 86.6 4.43 5.00 5.00 0.80 0.06 N=202
Q3.2d 3 1.5 3 1.5 6 2.9 19 9.3 50 24.4 130 63.4 180 87.8 4.47 5.00 5.00 0.84 0.06 N=205
Q3.2e 1 0.5 5 2.5 6 3.0 12 5.9 41 20.3 143 70.8 184 91.1 4.58 5.00 5.00 0.76 0.05 N=202
Q3.2f 0 0.0 12 6.1 12 6.1 28 14.1 71 35.9 87 43.9 158 79.8 4.18 4.00 5.00 0.89 0.06 N=198
not important at less important as important more much more
all (1) (2) (3)= (1) + (2) (4) important(5) important (6) (4)+(5)+(6)
Count % Count % Count % Count % Count % Count % Count % Mean Median Mode Std Std. Valid
Deviation Error N
of
Mean
Q3.3a 3 1.5 21 10.2 24 11.7 115 56.1 38 18.5 28 13.7 181 88.3 3.33 3.00 3.00 0.89 0.06 205
Q3.3b 2 1.0 29 14.3 31 15.3 85 41.9 62 30.5 25 12.3 172 84.7 3.39 3.00 3.00 0.91 0.06 203
Q3.3c 5 2.5 33 16.2 38 18.6 76 37.3 51 25.0 39 19.1 166 81.4 3.42 3.00 3.00 1.05 0.07 204
Q3.3d 3 1.5 17 8.3 20 9.8 81 39.5 55 26.8 49 23.9 185 90.2 3.63 4.00 3.00 0.98 0.07 205
Q3.3e 6 2.9 27 13.2 33 16.1 72 35.1 65 31.7 35 17.1 172 83.9 3.47 3.00 3.00 1.02 0.07 205
Q3.4a 4 2.0 22 10.7 26 12.7 50 24.4 57 27.8 72 35.1 129 62.9 3.83 4.00 5.00 1.09 0.08 205
Q3.4b 10 5.0 26 12.9 36 17.9 49 24.4 72 35.8 44 21.9 116 57.7 3.57 4.00 4.00 1.12 0.08 201
Q3.4c 9 4.5 24 11.9 33 16.4 48 23.9 62 30.8 58 28.9 120 59.7 3.68 4.00 4.00 1.14 0.08 201
Q3.4d 1 0.5 6 3.0 7 3.5 11 5.5 31 15.4 152 75.6 183 91.0 4.63 5.00 5.00 0.76 0.05 201
Q3.4e 3 1.5 7 3.5 10 5.0 32 15.9 76 37.8 83 41.3 159 79.1 4.14 4.00 5.00 0.91 0.06 201
Q3.4f 13 6.5 26 12.9 39 19.4 59 29.4 65 32.3 38 18.9 103 51.2 3.44 4.00 4.00 1.13 0.08 201
Chapter 8 Page 358

TABLE 8-6:RESPONSE FREQUENCIES OF DEPENDENT VARIABLES (CATEGORICAL SCALE)-IAQ

Profit& Stewardship Own choice Zakat Islamic


Cashflows Accountability Accountability
Count % Count % Count % Count % Count % Mean Median Mode Std. Standard Valid
Deviation Error of N
Mean
Q3.1 3 1.5 7 3.4 4 2.0 41 20.0 150 73.2 4.60 5.00 5.00 0.81 0.06 N=205

Sundry Interest
Prohibited Mean Median Mode Std Standard Error Valid N
Income Income Transactions Deviation of Mean
- Int. Income
Count % Count % Count %
Q3.5 14 7.3 29 15.0 150 77.7 2.7 3.0 3.0 0.6 0.04 N=193
Chapter 8 Page 359

Responses to Q1.3a and b and Q1.6 indicate strong support for the view that Islamic

business should be concerned with fair treatment of all stakeholders, safeguarding the

environment and payment of sufficient wages to employees even at the expense of lower

profits to shareholders. The agreement for these issues varied from 86.6% to 94.1% of the

participants. A mean response of 4.23 to 4.51 indicates fairly strong agreement. For the

environment and fairness to stakeholders issues, the median and mode responses were 5

indicating strong agreement. In the case of paying sufficient wages to employees, the

median and mode was 4 indicating agreement. Some respondents were of the opinion that

Islamic Business organisations should pay competitive wages, which may not necessarily

be a low wage policy.

As stated earlier, question 1.4 seems to have been misunderstood and accordingly many

participants answered in the opposite direction to the hypothesis. Hence, although more

than 50% of the participants seem to have favoured a technical interpretation of the

Shari’ah, no firm conclusions can be made. The question seems to have been interpreted

as meaning that an Islamic business organisation should follow a technical interpretation of

the Shari’ah rather than one in line with the organisation’s objectives and intent, as

opposed to a question on the objective and intent of the Shari’ah. About 57% responded

with agreement indicating a non-favourable response. Of the participants contacted (how

many) to give a correct interpretation of the question about 70% agreed to change their

responses from Agree to Disagree. If this were true of the all the respondents, the results

would have supported the researcher’s contention. Since the results are doubtful, this

question has been excluded from the calculation of the aggregate scores and is not used

to test any hypothesis.

8.3.1.1 Aggregate Sc
cores for Section 1 of IAQ

An overall score for this section; ISBUSETH (for Islamic business ethics) was computed

using, SPSS mean function as follows:


Chapter 8 Page 360

ISBUSETH= mean.47(BUSOB, ETH2, QSREL, ESWELF) (8.1)

where

BUSOB8 = mean.3 (Q1.3a, Q1.3b, rQ1.3c) (8.2)

which is the mean of Questions 1.3a, Q1.3b and Q1.3c. The r in rQ1.3c means that it is a

re-coded version of Question 1.3c. This is undertaken to reverse the direction of scoring as

some questions were in the opposite direction to the hypothesis. Hence, in a re-coded

question a Strongly Disagree response would score a 5, whereas the normal scoring is a 5

for Strongly Agree. The purpose of this re-coding exercise was to ensure consistency in

the interpretation of the aggregate scores so that a higher score always indicates a

favourable disposition towards the researcher’s alternative hypothesis, whereas a lower

score would indicate otherwise. The BUSOB score uses the questions 1.3a to 1.3c, which

ask the participants about the objectives of Islamic Business organisations.

To ensure that sectional and aggregate scores incorporate the various dimensions of the

concepts measured, these scores are computed so as to require all the components of the

respective formulae to be non-missing. In order to obtain a reasonable total number of valid

participants (at least two third of all respondents) for the sectional and aggregate scores,

the minimum number of questions for sub-score to be validly computed was set by trial and

error until the aggregate and sectional scores were associated with a reasonable total

number of participants. If the computation of the lowest sub-scores had been set so as to

require all questions in the sub-score to be answered, the sub-scores, for many

participants would not have been computed, as certain participants would have not

answered certain questions. This would then cascade into the calculation of the sectional

and aggregate scores. These latter scores would not be computed for many participants

7
Mean.4 ( the .4) is a program instruction to ensure all four sub-scores must are valid and present for an
individual respondent before the overall score can be computed. If for example, there were too many
participants not answering a particular question, then the score would not be computed for that participant.
This is the reason for the different numbers of total valid participants for each question although the total
number of responses were 206. Mean.2 ( as in BUSOB) would ensure that at least two questions are valid
before the score is computed.
8
Business Objectives of Islamic business organisations)
Chapter 8 Page 361

thus associating them with a lower number of total participants. Hence, interpretations of

sectional and aggregate score would not be possible for many participants in the sample.

The second sub-score,

ETH2 (ethical investment) = mean.2(rQ1.5a,rQ1.5b,rQ1.5c,rQ1.5d) (8.3)

which is the mean of the questions Q1.5a to Q1.5d re-coded to reverse the direction of

scoring as was done for q1.3c for BUSOB. Here, since there are four questions, responses

to at least 2 of the 4 questions must be non-missing before the score is computed. The

researcher decided to reduce it to at least 2 as making it more stringent (e.g. insisting on 3

or 4 non-missing questions), the number of valid participants for the sectional and overall

aggregate scores became very small losing valuable information.

The third sub-score for ISBUSETH is QSREL (relevance of Qur’anic and Sunnah principles

for modern organisations). The series of questions, which asks the participants’ responses

to these, are Q1.7a-d. Hence, this score is computed, as the mean of these four questions

subject to at least three responses is non-missing.

Hence, QSREL= mean.4(Q1.7a,Q1.7b,Q1.7c, Q1.7d) (8.4)

Here even with the requirement of all the 4 questions to compute the score, the valid

number of participants was 200 and as such, it was felt that there was no need to reduce

the number of questions.

The final sub-score for ISBUSETH is ESWELF (employee & social welfare orientation of

Islamic/ Muslim Business Organisations) made up of mean of Questions 1.2 and 1.6

subject to the requirement that both questions should have responses. Hence,

ESWELF=MEAN.2(Q1.2,q1.6) (8.5)

We can see that Questions 1.1 and 1.4 has been completely left out in computing

ISBUSETH because of the confusion in the interpretations of question 1.4 and the fact that

the reliability analysis indicated that Q1.1 (and Q1.4) did not fit in with the rest. As such the

researcher felt that it was safer to leave out the responses to these questions in the

computation of the ISBUSETH score and consequently in the overall IA score discussed in

section 8.4.
Chapter 8 Page 362

8.3.1.2 Interpretations of aggregate scores

The aggregate scores will have a value between 1 and 5. This means there are only 4

classes of scores as scores of 0 to 1 and above 5 are not possible. Some aggregate

scores involve questions using mixed scales. As such a score of 1 may not mean “strongly

disagree” or “not at all” but a point on a scale indicating disposition to what the scores

measure. For easier interpretation, the raw aggregate score has been converted to 0 to

100% score. Thus, we can take 50% and more to indicate a favourable disposition towards

the concept measured by the score and vice versa. In order to do this, the following

formula is used.

% score = (Rawscore –1) x 100/4= (Rawscore-1) x 25. (8.6)

For example a raw score of 3 will be (3-1) x 25%= 50% on the percentage scale.

8.3.1.3 Aggregate scores for Section 1 of the IAQ.

Table 8-7 shows that the BUSOB score has a mean raw score of 3.8 with a median and

mode of 3.7. This converts to mean% of 68.9% and median% of 66.8% respectively

indicating strong support for ethical objectives of Islamic business such as, fairness of

stakeholder treatment, safeguarding the organisation and the relative unimportance of

maximising shareholder profits. A table of summarised frequency (Table 8-8) shows that

90% of the participants scored 56% or more for Islamic business objectives.

From table 8-7, we it can also be seen that the raw mean scores for ETH2 (i.e. ethical

investment decisions of Islamic business organisations) is 3.8 and mode and median of 4.0

and 5.0 respectively. This converts to mean score of about 70%, a median of 75% and

mode of 100%. This means that the most frequent score was 100% and at least 50% of the

participants were 75% in favour of ethical investment. Table 8-8 shows that 82% of the

participants responded with a mean of 50% or more for this group.

For QSREL (the relevance of the socio-economic principles contained in the Qur’an and

Sunnah to the development of a business/legal framework for various modern

organisations), the mean score was 4.6, with both the median and mode being 5. This
Chapter 8 Page 363

translates to about 90% for the mean, and 100% for the mode and median. The frequency

table shows 100% of the participants scored 50% and above. This shows overwhelming

faith in the relevance of the Qur’an and Sunnah for modern Muslim organisations and

perhaps shows that respondents felt that there should be a Muslim contribution to the

legal/ethical framework of organisations in future.

The mean score for ESWELF is 4.5 translating to a percent score of 86.5% with the same

median and mode. The frequency table (table 8-8) shows that 72% of the participants

scored 60% and above whereas 99% of the valid participants scored 50% and above. This

shows that respondents believed that Islamic business organisations should take into

account a decent living wage for its employees, even if shareholder profits were reduced

thereby. Similarly both Islamic and Muslim business organisations should promote social

welfare in the Islamic context rather than just concentrate on profits. Although maximising

profits for shareholders was considered a legitimate objective of such organisations, the

response to ESWELF indicates that it is not their only objective.

Finally an overall score for this section ISBUSETH (Islamic business ethics for

Muslim/Islamic business organisations) indicates both a mean and median of 4.2 and a

mode of 4.7. This corresponds to an overall mean% and median% of about 79% with a

mode% corresponding to 92%. The frequency table shows that 67.5% of the valid

participants (N=200 out of total of 206) scored 75% for this score and 100% of them scored

mean of 54.25% or more. The above score shows overwhelming support for the hypothesis

that Islamic business organisations should conform to Islamic business ethics and law.

8.3.2 Analysis & Findings of Section 2 of the Questionnaire

The mean score for the first part of section 2 of the questionnaire ranges from 3.31 to 3.9

(see table 8-4) indicating weak agreement with the hypothesis (Hypotheses No. 4 and 5,

chapter 7) that conventional accounting is inappropriate for Muslim organisations, due to its

ill effects. The median and mode were 4.0 (except for question 2.3 which had a median of

3.0) indicating agreement on this issue. However this agreement was not as strong as in
Chapter 8 Page 364

the previous section. The strongest agreement was on question 2.2 where 78% of the valid

participants agreed or strongly agreed that conventional accounting fosters materialism,

individualism and competition among Muslim users and these traits are not Islamic. About

64% of the valid participants agreed or strongly agreed that conventional accounting

needed major modification to provide the right kind of information for Muslim users.

However only about 51% agreed or strongly agreed that conventional accounting

principles, with their roots in capitalist cultural values, are inappropriate for Islamic

business organisations. Although about 28% disagreed on this issue, only 1.5% strongly

disagreed compared to 11.2% who strongly agreed.

Table 8-5, shows that about 48% of the valid participants’ (193) felt that conventional

financial statements provided information which was inappropriate for Islamic business

organisations. Only about 17% felt that it provided appropriate information. About 34%

thought that it provided fairly appropriate information. The mean, median and mode

responses to question 2.5b of section 3 suggest that the participants believed that

conventional accounting impeded the fair allocation of wealth between stakeholders.

However a closer look at the frequency section of the table 8-5, reveals that about 31 –

38% of the valid participants believed that it impeded fair allocation of wealth very much

and hindered much or very much the ability to make appropriate decisions in line with

Islamic objectives.
Chapter 8 Page 365

TABLE 8-7: IAQ AGGREGATE SCORES – DESCRIPTIVES


9
Aggregate Mean Mean% Median Median% Mode Mode % Maximum Minimum Percentile Percentile Standard Std Valid
Scores. 25 75 Error of Deviation N
Mean
BUSOB 3.8 68.9 3.7 66.8 3.7 66.8 5.0 2.00 3.3 4.0 0.0 0.5 203

ETH2 3.8 69.7 4.0 75.0 5.0 100.0 5.0 1.00 3.3 4.8 0.1 1.1 178

QSREL 4.6 88.9 5.0 100.0 5.0 100.0 5.0 3.00 4.0 5.0 0.0 0.5 200

ESWELF 4.5 86.5 4.5 87.5 4.5 87.5 5.0 1.50 4.0 5.0 0.0 0.6 201

ISBUSETH 4.2 78.7 4.2 79.2 4.7 91.7 5.0 3.17 3.8 4.5 0.0 0.4 171
UNISL 3.5 63.3 3.5 62.5 4.0 75.0 5.0 1.00 3.0 4.0 0.1 0.8 199

FACON 3.1 53.5 3.0 50.0 3.0 50.0 5.0 1.00 2.5 3.7 0.1 0.8 189

CAPRIN 2.5 38.2 2.3 33.3 2.0 25.0 5.0 1.00 1.7 3.0 0.1 1.1 190

UNCONAC 3.1 51.6 3.1 51.4 2.9 47.9 4.7 1.50 2.6 3.5 0.1 0.6 171

OBIA 3.7 68.3 3.7 66.7 3.7 66.7 5.0 2.2 3.5 4.0 0.0 0.5 204

STIMP 3.5 61.4 3.4 60.0 3.0 50.0 5.0 1.2 3.0 4.0 0.1 0.0 202

INF 4.0 74.2 4.0 75.0 5.0 100.0 5.0 1.4 3.6 4.6 0.1 0.7 203

CHIS 4.0 75.3 4.0 75.0 4.0 75.0 5.0 1.6 3.7 4.4 0.0 0.7 202

ONIA 4.0 74 4.0 75.1 2.9 46.3 4.7 2.3 3.8 4.2 0.0 0.5 194

IA 3.8 68.9 3.7 68.2 3.6 66.1 4.6 2.8 3.5 4.0 0.0 0.4 139

9
Mean%, Mode% and Median% are calculated using the formula: (M-1)*25 to convert them into % scale, where M= mean, mode or median respectively.
Chapter 8 Page 366

TABLE 8-8: IAQ AGGREGATE SCORES – RESPONSE FREQUENCIES

BUSOB ETH2 QSREL ESWELF ISBUSETH UNISL FACON CAPRIN UNCONAC

Freq. Cum Freq. Cum% Freq. Cum% Freq. Cum% Freq. Cum% Freq. Cum% Freq. Cum% Freq. Cum% Freq. Cum%
%
1 & above 7 3.90 1 .05 3 1.5 27 14.20
1.33 1 4.50 1 0.50 2 3 59 18.90 2 1.2
1.67 2 .15 13 25.80 5 4.1
2.00 1 .5 11 10.70 1 1.00 10 6 22 14 32 42.60 10 9.9
2.33 2 1.48 5 13.50 20 16 24 27 20 53.20 27 25.7
2.67 7 4.93 7 17.40 11 22 17 36 10 58.40 33 45.0
3.00 11 10.34 17 27.00 3 1.5 6 4.00 4 2 34 39 32 53 32 75.30 40 68.4

3.33 33 26.60 22 39.30 1 1.5 10 9.00 17 12 22 50 33 70 13 82.10 26 83.6


3.67 64 58.13 4 41.60 3 3 38 34 26 63 14 78 8 86.30 13 91.2
4.00 57 86.21 42 65.20 67 37 37 27.40 48 62.5 45 86 21 88 10 91.60 12 98.2
4.33 6 89.16 15 73.60 4 39 76 65.20 23 76 17 94 14 96 11 97.40 1 98.8
4.67 7 95.07 3 75.30 12 45 23 89 3 96 1 96 2 100
5.00 10 100 44 100 101 100 70 100 18 100 8 100 6 100 5 100 100
Valid 203 178 200 201 171 199 189 190 171
Participants
Missing 3 28 6 5 35.00 7 17 16 35
Chapter 8 Page 367

TABLE 8-9:IAQ AGGREGATE SCORES – FREQUENCIES (CONTINUED)

OBIA STIMP INF CHIS ONIA IA


Freq. Cum% Freq. Cum% Freq. Cum% Freq. Cum% Freq. Cum% Freq. Cum%

1 & above 1 0.50


1.33 1 0.50 1 0.50
1.67 1 1.00
2.00 2 1.00 3 2.50 4 2.50 2 1.00
2.33 3 2.50 21 12.90 8 6.40
2.67 3 3.90 11 18.30 3 7.90 2 1.50 6 4.10 4 2.9
3.00 21 14.20 55 45.50 23 19.20 11 6.90 1 4.60 13 12.2
3.33 43 35.30 39 64.90 29 33.50 33 23.30 26 18.00 41 41.7
3.67 59 64.20 11 70.30 22 44.30 41 43.60 60 49.00 48 76.3
4.00 59 93.10 33 86.60 44 66.00 60 73.30 63 81.40 25 94.2
4.33 8 97.10 18 95.50 32 81.80 34 90.10 34 99.00 8 100
4.67 2 98.00 3 97.00 8 85.70 17 98.50 2 100.00
5.00 4 100 6 100 29 100 3 100 139
Valid 204 202 203 202 194 139
Missing 2 4 3 4 12 67
Chapter 8 Page 368

Responses to Q2.6a to c on the appropriateness of the historical cost, prudence and

money measurement concepts seem to indicate that the participants consider that these

principles are appropriate for Islamic organisations. The mean score for these questions

ranges from 3.35 to 3.64 (see table 8-5). The median score of 4.0 (except for Q2.6c which

was 3.0) indicating that at least 50% of the participants are of this opinion. Hence, the

realisation that conventional accounting needs major modification may have arisen out of a

fear that it leads to a way of life not in line with Islamic teachings rather than from a deep

insight into the problems of conventional accounting principles, which were discussed in

chapter 3. This means that probably accounting education has succeeded in the

acceptance of these principles as Islamic and further research needs to be carried out in

detail to ascertain whether the participants are aware of the unislamic consequences of

these principles.

8.3.2.1 Aggregate scores for Section 2 of the IAQ.

An overall score for this section, UNCONAC (unsuitability of conventional accounting to

Islamic/Muslim Business organisations) was computed as follows:

UNCONAC=mean.3(UNISL,FACON,CAPRIN) (8.7)

where

UNISL10= mean.3(q2.1 to q2.4) (8.8)

FACON11 =mean.2(rq2.5a,q2.5b,q2.5c) (8.9)

CAPRIN12= mean.2(rq2.6a-rq2.6c) (8.10)

Again Q2.5a, Q2.6a-c had to be re-coded to reverse the direction of scoring and

appropriate minimum of two non-missing questions had to be incorporated.

Table 8-6 shows that the raw mean score for UNISL is 3.53, which translates to 64.25%.

This shows that overall the participants feel that, conventional accounting lead to improper

10
UNISL=unIslamic behavioural consequence of conventional accounting.
11
FACON=Financial statements prepared under conventional accounting is not suitable.
Chapter 8 Page 369

attitudes and cannot lead to Islamic social welfare; it needs major modification to be in line

with Islamic principles. However this feeling is not as strong as that on Islamic business

ethics discussed in section 8.3.2.1 where the mean score ranged from 70% to 90%. The

mean score for FACON is 53.5% (3.14-1 x 25) which is even lower than the score for

UNISL. This shows a low level of support for the hypothesis that the financial statements

prepared under conventional accounting are not suitable for Islamic business

organisations.

The raw mean score for CAPRIN was 2.55 with a median 2.33 and a mode 2.0. These

scores translate to a mean percentage score of 38.75%, a median of 33.25% and a mode

of 25.00%. The results for this score indicates that the hypotheses (No. 4 and 5, see

chapter 8) of the researcher, that conventional accounting principles are not suitable for

accountants, is not supported by the participants in this research. This may be due either

to a genuine feeling that they are suitable and that as on the surface, historical cost,

prudence and monetary measurement concepts do not appear to contradict Islamic

principles. Otherwise professional accounting education may have taken root and

succeeded in instilling a perception that these concepts are appropriate, in the participants.

The overall score for section 2 of the IAQ is represented by UNCONAC. From table 8-7, it

can be seen that the mean and median scores for UNCONVAC are both 3 and the mode is

2.9. This translates to percentage scores of around 52% for both the mean and median,

and 48% for the mode respectively. This shows there is a low level of support to the

hypothesis that conventional accounting is unsuitable for Muslim users and organisations.

However, with 171 valid participants out of 206 and from the median score, about 83% of

total respondents support the hypothesis. Even here, the accounting principles and the

technical aspects of conventional accounting that are not suitable seem to be in doubt.

Hence further research needs to broaden this area.

8.3.3 Analysis & Findings of Section 3 of the Questionnaire

12
CAPRIN= unsuitability of conventional accounting principles.
Chapter 8 Page 370

In Table 8-6, the responses to Q3.1 shows that 73.2% favoured “Islamic accountability” to

be the main objective of Islamic Accounting with 20% opting for Islamic accountability

through Zakat accountability. The traditional objective of financial statements offered by the

FASB (1978) for cashflows and investment decisions for stock market players only got

1.5% and the stewardship objective was favoured by 3.4% 13 . This is an overwhelming

change in the direction of Islamic accounting from conventional accounting and shows

tremendous support for hypothesis number (No.6) that Islamic accountability is the main

objective of Islamic accounting as compared to decision usefulness.

8.3.3.1 Objectives of Islamic Accounting

Despite this however, the participants ranked efficient allocation of capital highly with a

score of 4.2, 4.4 and 5 for the mean, median and mode of Q3.2a, with 80% of the

respondents scoring it 4 or 5 (see table 8-5). This suggests that decision usefulness with

the ultimate purpose of efficient allocation of resources, is perceived by the respondents to

have a high priority in Islamic accounting. This seem to be generally in line with answers to

question 2.3 where 48% of the participants felt that the efficient allocation of resources can

lead to Islamic welfare. The fact that 25% of the respondents were neutral or disagreed to

question 2.3 indicates a substantial number of participants are unsure about this question.

This area needs further probing to make sure that this is the case.

13
Only 4 out of 205 respondents indicated their own choice. Hence the scoring of own choice as 3 do not
present a problem.
Chapter 8 Page 371

The mean score for Q3.2b of 3.4 with both median and mode of 3 indicates that

maximising shareholder wealth is considered an objective of Islamic accounting which is of

average importance. About 17% of the participants felt that this was not important

whereas 36.3% of the participants scored a 3 which means of average importance.

Compared to the previous question where 80% scored 4 or 5 only about 48% scored a 4 or

5 for this question.

Question 3.2c to f asked the participants to rank the Islamic ethical objectives of Islamic

accounting regarding fair allocation of wealth, Zakat calculation and distribution, Shari’ah

compliance and a conducive co-operative environment for stakeholders. Table 8-4, shows

that between 79.8% and 91.1% of the participants scored these statements a 4 or 5

indicating that these objectives were important. The mean scores were 4.2 to 4.6 whereas

the mode and median were 5 except for the median of q2.2f, which was 4. As expected,

Shari’ah compliance and Zakat calculation and distribution seem to be the major objectives

of Islamic accounting.

An aggregate score, OBIA (Objectives of Islamic accounting) was calculated for this group

of questions using the formula.

OBIA =mean.4(rq3.2a,rq3.2b,q3.2c,q3.2d,q3.2e,q3.2f) (8.11)

This score measures the Objectives that Islamic accounting should pursue. Q3.2a and

Q3.2b were re-coded to reverse the direction of scoring. Here a minimum of 4 out 6

questions should be non-missing for this sub- score to be computed. Any further increase

in the minimum number of questions required to compute the score reduced the valid

number of participants by about 20% for each additional required question. Hence the

minimum required was kept to 4.

Table 8.7 shows that the mean for the OBIA score is 3.73with the median and mode both

being 3.7. This corresponds to a mean percentage score of 68.3% which shows a fairly

favourable disposition to the religious and ethical objectives of Islamic accounting

compared with a narrow profit and shareholder focus. The inter-quartile range 3.5 to 4

(62.5% and 75%) shows that 75% percentage of the respondents scored 62.5% or more.
Chapter 8 Page 372

8.3.3.2 Importance of other stakeholders

Q3.3a to Q3.3e asked to participants to rank the importance of other stakeholders as

compared to shareholders as users of Islamic accounting information. Most participants

(between 81.4 to 90.2%, see Table 8-5) suggested that employees, government,

community, benevolent loan creditors and consumers are at least, as important, if not more

important than shareholders. The mean score ranged from 3.3 to 3.6 and the mode and

median score was around 3. An exception to this general finding related to benevolent loan

creditors where the median was 4 meaning that this group was considered more important

than shareholders by 50% of the participants. This is understandable, as benevolent loan

creditors do not charge interest or get any return. They may lend interest free to an

Islamic voluntary or charitable organisation. As they are entitled to a return of their loan by

the Shari’ah, the only way they can monitor the usage of their funds is through accounting

information designed for them. Hence Islamic accounting, especially for non-business

organisations bears this stakeholder in mind.

An overall score for this group of questions, STIMP(importance of stakeholders) was

computed using the formula:

STIMP= mean.5(q3.3a,q3.3b,q.3.3c,q3.3d, q3.3e) (8.12)

Even with a minimum requirement of 5 questions for the score to be computed, the total

number of valid participants producing this score was 202. Hence all the questions were

retained for this aggregate score. Table 8-7 shows a mean score for STIMP of 3.5 with the

median and mode being 3.4 and 3.0 respectively. This produces a percentage mean score

of 62%, median of 60% and a mode of 50%. This shows that participants consider other

stakeholders are generally more important users of Islamic accounting than shareholders.

The inter-quartile range (3 to 4) shows that 75% of participants ranked other stakeholders

as important, more important or much more important than shareholders.

This is inline with the importance of different members of community in an Islamic society.

The rights of all sections should be taken into account when conducting business under an

Islamic ethos. The results therefore show strong support for the researcher’s hypothesis
Chapter 8 Page 373

(No.7, section 7.3.3.3 of chapter 7) that Islamic accounting should be geared to various

stakeholders in contrast to the shareholder/creditor emphasis of conventional accounting.

8.3.3.3 The Type Of Information Which Islamic Accounting Should Provide

Q3.4a to f asked the participants to rank the importance of different kinds of information

that should be provided by Islamic accounting. This includes the impact of organisational

activities on the environment, employee working conditions, distribution of salaries of

different groups of employees and managers, Shari’ah prohibited activities, social impact

on the community and allocation of wealth between National and Foreign interests. The

mean score for these groups of questions ranged from 3.4 to 4.6 with mode and median of

4 to 5 (see Table 8-5). This shows that the participants felt the information listed was

important or very important. Again as expected, the highest score was for information on

Shari’ah prohibited activities, although this hopefully would be redundant in an Islamic

organisation. In the real world however, Muslim and Islamic companies cannot be 100%

Islamic and some of their transactions with Non- Muslim countries may not be in line with

the Shari’ah. As such it is important to disclose this information, so that the user can

identify that portion of prohibited income and decide what to do with it.

The allocation of wealth between National and Foreign interest received the lowest score of

3.4 with only 51.2% considering it to be important or very important. Many of the

participants to whom the questionnaire was administered thought that a better question

would be between Muslim and Non Muslim interest would be more important as nationality

was considered not important as compared to a common faith.

An aggregate score for this group of questions, INF(information which IA should disclose)

was computed using the formula

INF= mean.4(q3.4a,q3.4b,q3.4c,q3.4d,q3.4e,q3.4f) (8.13)

The minimum required questions were set at 4 to obtain a reasonable count of valid

participants. Although an increase to the minimum number of questions to 5 only resulted

in about 7% decrease in valid participant count, It was decided to lose one question here to
Chapter 8 Page 374

increase participant count in the sectional and overall aggregate score (ONIA and IA)

respectively. Here the score is only computed if at least four of the five score were not

missing. Table 8-7 shows that the mean was 4.0, the median score was 4.0 with a mode of

5. This translates to mean percentage score of 75% for the mean, 75% for the median

and 100% for mode. The inter-quartile range of 3.6 to 4.6 indicates that at least 75% of the

participants scored 65% ((3.6-1) x25) or more indicating a highly favourable disposition

towards broader social, religious, employee and environmental concerns for Islamic

accounting.

8.3.3.4 The characteristics and nature of Islamic accounting

The last group of questions (Q3.6 to Q3.13) in this section asked the participants’ opinion

on the characteristics of Islamic accounting as opposed to conventional accounting. This

included whether Islamic accounting should be holisitic, de- emphasise the conventional

cash flow, profits and financial position information, whether it should take account of

externalities, unrealised profits, use current values and whether it should be Shari’ah

audited.

The participants scored a mean score of 3.35 to 4.47 to this question (see Table 8-4). This

means % score of 58.75% to 86.75% indicating favourable to highly favourable disposition

to the proposed content of Islamic accounting. The median and mode of 4 to 5 indicate

that a large number of participants supported these contents. As expected the Shari’ah

audit, integration of non-financial information and using current values for Zakat

calculations were supported by over 90% of participants. The only question which had a

low favourable response was the recognition of unrealised profits to ensure equitable

distribution of wealth between present and past shareholders which was supported by only

52.3% of the participants with 25 % totally disagreeing and another 23% unsure.

An aggregate score for this group of questions, CHIS (characteristics of Islamic

Accounting) was computed using the formula:

CHIS= mean.6(q3.6,3.7,3.8,3.9,3.10,3.11,3.12, q3.13) (8.14)


Chapter 8 Page 375

Here 6 out of the 8 questions must be non-missing before the score is computed. The

reduction of two questions, increase the valid participants scoring the variable by about

12%. Table 8-7 indicates that the mean for CHIS was 4.0 with the median and mode for

this score being 4. This works out to a mean, median and mode of 75%. These are high

scores indicating that the participants favoured Islamic accounting should be broader

based and revolutionary (emancipatory in the words of Tinker, 1985). This is in the sense

of having a different framework, using different values and broader than conventional

accounting. From the 25th percentile of 3.71, it can also be worked out that 75% of the

participants scored 67.5% or more for this aggregate score indicating a high level of

support.

8.3.3.5 Aggregate scores for Section 3 of the IAQ.

An overall score (ONIA-Objectives and Nature of Islamic Accounting) was computed for

this section as follows:

ONIA= mean.5(q3.1, OBIA, STIMP, INF,CHIS) (8.15)

The overall score ONIA has a minimum computational requirement of 5 which means all

the five components must be present and non-missing before the score is computed. From

Table 8-7, the mean score for ONIA was 4.0 with median 4 and mode of 2.9. This

translates to 75% for both the mean and median% and about 46.3% in percentage terms.

The mean and median show a high level of support for the objectives and characteristics of

Islamic Accounting proposed by the researcher. With a valid count of 194 out of 206 (or

94% of the sample), this support is also overwhelming.

8.4 INTERPRETATION OF THE FINDINGS AS A WHOLE

An overall Islamic accounting score (IA) was computed as follows:

IA = mean.3 (ISBUSETH, UNCONAC, ONIA) (8.16)

corresponding to the three sections of the Islamic Accounting Questionnaire. The equation

is the mean score of the three sub scores.


Chapter 8 Page 376

This mean score of the three sub scores is computed with the condition, all the three

sub-scores must not be missing. If any of the three sub-scores were missing, this overall IA

score would be computed as missing. In effect this, this takes into account almost the

entire questionnaire.

From table 8-7, it can be seen that the mean score for IA is 3.8, which works out to about

69%. From the inter-quartile range of the same table, it can be seen that at least 75% of

the valid respondents scored 3.5, which translates to 62.5%. This shows a high level of

support for the hypotheses on the need for the development of an Islamic accounting and

the problems with conventional accounting. However, with valid count of only 139

participants (about 67%), the response is not overwhelming but substantial. It must be

borne in mind that the criteria is very rigorous in that IA score takes into account almost all

the questions the researcher had asked in the Islamic accounting questionnaire and from

table 8-8, it can be seen that only 2.9% of the valid participants scored below 50%. This

means 97% of the valid 139 participants were favourably disposed to Islamic accounting

although their extent of support for the researcher’s contentions varied from 50% to

90.75%.
Chapter 8 Page 377

8.5 INDEPENDENT VARIABLES: FINDINGS

This section describes the findings on the independent variables given in section 4 of the

IAQ. Section 4 of the IAQ as discussed in chapter 7, asked the participants to fill in blanks

or tick boxes to certain information which was considered as having a bearing on the type

of responses given. This included the sector in which the participant worked and his or her

length of working experience. Also asked were the educational qualifications and place of

qualification, the Islamic qualification and the place this was acquired. The participant were

asked to indicate whether they received any formal education in social and environmental

accounting and they were asked for their religion. In the case of accountants in public

practice, another factor, which was thought to have some influence on the response, was

whether the local audit firm received technical assistance from overseas associates.

8.5.1 Working Sector

Sector Frequency % Valid % Cumulative


%
Government 25 11.6 11.6 11.6
Education 101 50.5 50.5 62.0
Commerce 46 21.8 21.8 83.8
Public Practice 34 16.2 16.2 100.0
Total 206 100.0 100.0

TABLE 8-10: PARTICIPANTS BY WORKING SECTOR

Table 8-10 shows that the largest number of participants was from the education sector

followed by forty-six from commerce and industry, thirty-four from public practice and

twenty-five from government.

8.5.2 Accountancy Related Experience:

The distribution of accounting related working experience (including teaching) of the

participants shows a mean experience was 8.2 years with a standard deviation of 7.08

years. The median experience is 5 years. The mode experience is 4 years (32 cases)

which also represents 15.5% of the sample. The next highest experience is 5 years with

12.1% of the sample. It can be seen that the age distribution is positively skewed with most
Chapter 8 Page 378

of the respondents having between 2 to 7 years of working experience (about 45% of the

respondents). Another 15% have between 7.5 to 12.5 years while about 10% of the

respondents have below 2.5 years of experience. This means that about 80% of the valid

sample respondents have less than 12.5 years experience. Considering that the retirement

age in Malaysia is 55, and the schooling starts at 7 and ends after 17 years of schooling to

degree level, then the maximum working life of graduates (and professionals) is 31 years

and the average is 15.5 years. Therefore, a suitably grouped frequency distribution would

be as the one shown in table 8-11. We can consider that the group more than 15 years as

a highly experienced category, 6 to 15 years as experienced, 2 to 5 years as moderately

experienced and below 2 as inexperienced. 20 of the participants did not reply to the

experience questions.

Working Experience Frequency Percent Valid % Cumulative %


Valid More than 15 yrs. 21 10.2 11.3 11.3
6 to 15 years 66 32.0 35.5 46.8
2 to 5 years 87 42.2 46.8 93.5
less than 2 years 12 5.8 6.5 100.0
Total 186 90.3 100.0
Missing System 20 9.7
Total 206 100.0
TABLE 8-11:GROUPED EXPERIENCE FREQUENCY TABLE

The data is consistent for a developing nation such as Malaysia, where the accounting

profession and capital market began only in the 1970’s. This is also especially true for

Muslims/Malays as these groups have been disadvantaged in terms of education and

social standing until the New Economic Policy implemented in the 1970’s produced a

generation of educated and experienced Muslims only in the 1990’s. Hence this can be

taken as evidence of the sample representing the population in terms of experience

distribution.

8.5.3 Education.

Education is very important in framing the values and attitudes of people. The type, the

place and level of education can determine attitudes. It is hypothesised that secular

education may lead to a norm of anti-religiousness among its graduates. This would be
Chapter 8 Page 379

especially true of a positivist/empirical/scientific education, which places value on doubt,

scepticism and empirical proof as the methodology and epistemology of knowledge.

However in recent years, especially in accounting a more critical and humanist approach

may have diluted this ‘science is all’ phenomenon. Hence the place and school from which

the education is obtained is important. It is hypothesised that if the education has been

from an Anglo-American school, positivism would be more apparent and the participants

might have a negative attitude towards Islamic accounting as they might be opposed to

anything which smacks of religion. Further the relatively permissive culture in the West may

have influenced the graduates somewhat into adopting values which are opposed to

Islamic or any religious beliefs. However, if graduates have been exposed to a more

broad-based or critical approach in their courses, this may make them more inclined

towards a social/Islamic form of accounting. Malaysian University graduates may have a

higher Islamic awareness level than those from other countries because of the introduction

of Islamic civilisation as a compulsory subject at university level in Malaysia. Further many

academics with Masters degrees may have received their basic degree qualifications from

Malaysia and this may have an effect on their attitude.

The level of education may have an effect on the attitude towards Islamic accounting. A

post graduate qualification may indicate a more favourable attitude whereas an

undergraduate qualification which emphasises technical or career oriented education or a

professional qualification alone may indicate a less favourable disposition. From the

perspective of the Muslim world, education has become dichotomous (see chapter 4) after

colonisation, secular education having overcome to a large degree the traditional religious

education system. Many scholars have also obtained their education from overseas.

Malaysia had adopted a policy of sending Malays to Western countries for their tertiary and

post secondary education since the 1970’s with scholarships and other incentives. There

were about 10,000 (mainly Malay/Muslim) scholars in the UK alone in 1998. Although the

policy of sending students overseas has gradually decreased especially with the economic

crisis beginning in May 1998, the Muslim professionals and academics who occupy
Chapter 8 Page 380

important positions in various sectors who form the populations were a result of this policy.

These conditions lead to the importance of education on attitudes towards a particular

subject.

HIGHEST COUNTRY FROM WHICH HIGHEST TOTAL %


QUALIFICATIONS QUALIFICATION
WAS OBTAINED
Anglo European Malaysia Other Non-
American Muslim countries
BA/BS 14 1 75 2 92 47.4
MA/MS 46 5 6 57 29.4
PhD/D 22 3 6 31 16.0
Professional 8 3 11 5.7
Diploma 2 2 1.0
Others 1 1 0.5
90 6 90 8 194 100.0

TABLE 8-12: QUALIFICATIONS/PLACE OF EDUCATION OF PARTICIPANTS

Table 8-12 shows that roughly an equal (47.4% and 45.4%) proportion of valid participants

had undergraduate and post graduate qualifications. However, while 81.5% of the

graduates obtained their qualifications from Malaysia, 88.7% or 79 participants obtained

their post-graduate qualifications from Non-Muslim countries mainly Anglo-American.

(81.5%obtained their education from Malaysia, About 6% had professional qualifications

alone. A cross-tabulation of education with sector shows that all except two of the

postgraduates are in the education (Universities) sector. Further 73.8% of the participants

majored in accounting in their highest qualifications.

8.5.4 Islamic Education

It is hypothesised that Islamic education would make the participants more favourably

disposed towards Islamic accounting. Islamic education could be of two types; a traditional

ritual education without its underlying objectives and philosophy or a more broad-based

one. Usually the higher the level, the broader the education, hence it is hypothesised that a

higher Islamic education level would lead to a more favourable attitude towards Islamic

accounting. Further if the Islamic education was obtained in the West, there might be a
Chapter 8 Page 381

more critical approach to religion and a more unfavourable attitude to Islamic accounting.

However one has to note that the Islamists do not normally have a traditional or degree

level Islamic education. Most of them have only basic religious education in Mosques or

secular schools and Islam is more often self-thought through reading, weekly study circles,

seasonal worship camps, attending talks in mosques and interaction with the local Islamic

community. Hence the level of formal Islamic education may not indicate Islamic

awareness.

Qualification Frequency % Valid % Cumulative %


Valid Ijazah (degree) 7 3.4 4.3 4.3
Sanawi (religious .A 1 .5 .6 4.9
Level)
Certificate(one year post 14 6.8 8.5 13.4
O level)
STPM ( A Level) 1 .5 .6 14.0
SPM (O Level) 127 61.7 77.4 91.5
PMR/SRP (Form 3) 6 2.9 3.7 95.1
Other 8 3.9 4.9 100.0
Total 164 79.6 100.0
Missing System 42 20.4
Total 206 100.0

TABLE 8-13: HIGHEST ISLAMIC QUALIFICATIONS

Table 8-13 shows only 7 participants (about 3.4%) had degree qualifications in Islam and

the overwhelming majority of participants namely 127 (about 62%) had only SPM or ‘O’

Level qualifications. Another 42 participants did not answer this question but we may

assume that they have none. This is consistent with the fact that those who are either

graduates in accounting/business or professional accountants usually followed a secular

education route and the only formal Islamic qualification available was a subject ‘Islamic

Religious Knowledge’ at SPM and STPM (O and A level).

For purposes of further analysis, the above qualifications are grouped into two categories;

(i) SPM and lower, and (ii) HSC and higher. Further, since all except two of these

participants obtained their Islamic qualifications from Malaysia, the place of Islamic

qualifications will not be analysed as a variable.


Chapter 8 Page 382

8.5.5 Social and Environmental Accounting Education

There have been calls for a more critical and ethical approach to teaching accounting (for

example, Gray et al., 1994). If the participants have had exposure to social and

environmental accounting education, this might act as a moderating variable to increase

their commitment for Islamic accounting.

Yes= have formal social accounting education; No= No formal social accounting education
Response Frequency Percent Valid % Cumulative %
Yes 71 34.5 34.5 34.5
No 135 65.5 65.5 100.0
Total 206 100.0 100.0

TABLE 8-14: FORMALLY STUDIED SOC & ENV ACCTNG

Table 8-14 shows that about one-third of participants have studied social and

environmental accounting. The effect of this on participants’ responses is further analysed

later.

8.5.6 Receipt of Foreign technical assistance by local firms

Lastly, one of the ways in which conventional accounting is globalised, is through the

activities of multinational accounting firms especially the big four and their erstwhile

competitors. Hence if local firms get technical assistance from their overseas

counterparts, then there might be a tendency to follow and defend Anglo-American ‘best’

practice and therefore a lower tendency towards an alternative Islamic accounting.

Yes=receive technical assistance; No= do not receive technical assistance


Frequency % Valid Percent Cumulative %
Valid Not applicable 165 80.1 82.5 82.5
yes 21 10.2 10.5 93.0
no 14 6.8 7.0 100.0
Total 200 97.1 100.0
Missing System 6 2.9
Total 206 100.0

TABLE 8-15: FIRM RECEIVES TECHNICAL ASSISTANCE

Table 8-15, reveals that about half the participants in public practice received technical

assistance from overseas countries. Since one of the principal avenues of globalisation of
Chapter 8 Page 383

Anglo-American accounting values is carried out through multi-national accounting firms,

this may lead to an unfavourable disposition towards Islamic accounting.

8.6 SIGNIFICANCE TESTS OF ASSOCIATION BETWEEN INDEPENDENT AND


DEPENDENT VARIABLES

In order to test for meaningful relationships between dependent and independent variables,

a statistical test that gives an indication of whether there are significant differences

between the means (or medians) of various groupings of respondents was undertaken.

The participants with different education and experience may be assumed to be from

different populations, which are independent of each other. In order to see whether there is

a significant differences in the characteristics of these various groups, two categories of

tests could have been undertaken. Since the normality of the various groups cannot be

assumed and there are more than two groups (represented by more than two independent

variables), a Kruskall- Wallis one way analysis of variance test of mean ranks for K

populations was undertaken using SPSS. This test gives the difference between the means

of ranks between groups and the associated p values (=α, the significance level).
Chapter 8 Page 384

TABLE 8-16: SIGNIFICANCE LEVELS; KRUSKAL-WALLIS TESTS OF DIFFERENCES OF RANKED MEANSBETWEEN


POPULATIONS GROUPED BY INDEPENDENT VARIABLES –IAQ

SOC &
PLACE ENV PLACE
WORKING PROF EXPERE HIGHEST OF ISLAMIC ACC TECH WORKING PROF EXPERE HIGHEST OF ISLAMIC SOC & ENV TECH
QUEST. SECTOR CATEG CATEG QUALIF QUAL. QUAL EDU. ASST. QUEST. SECTOR CATEG CATEG QUALIF QUAL. QUAL ACC EDU. ASST.
Q1.1 0.275 0.251 0.718 0.153 0.990 0.105 0.305 0.509 Q3.1 0.373 0.308 0.338 0.100 0.258 0.492 0.559 0.328
Q1.2 0.168 0.377 0.003 0.436 0.845 0.702 0.974 0.751 Q3.2a 0.044 0.646 0.370 0.009 0.074 0.452 0.300 0.198
Q1.3a 0.245 0.159 0.013 0.006 0.002 0.545 0.059 0.184 Q3.2b 0.932 0.942 0.932 0.239 0.478 0.746 0.429 0.602
Q1.3b 0.003 0.002 0.016 0.001 0.003 0.885 0.684 0.447 Q3.2c 0.115 0.122 0.140 0.029 0.564 0.841 0.898 0.704
Q1.3c 0.145 0.024 0.877 0.326 0.602 0.302 0.759 0.377 Q3.2d 0.462 0.546 0.287 0.052 0.602 0.810 0.839 0.316
Q1.4 0.137 0.162 0.373 0.006 0.011 0.576 0.040 0.155 Q3.2e 0.339 0.888 0.179 0.014 0.056 0.267 0.783 0.047
Q1.5a 0.018 0.058 0.377 0.080 0.261 0.527 0.729 0.897 Q3.2f 0.075 0.091 0.513 0.018 0.671 0.169 0.980 0.573
Q1.5b 0.586 0.535 0.517 0.070 0.707 0.741 0.307 0.793 Q3.3a 0.117 0.268 0.641 0.440 0.866 0.950 0.592 0.905
Q1.5c 0.590 0.300 0.587 0.083 0.949 0.575 0.260 0.686 Q3.3b 0.000 0.007 0.170 0.026 0.112 0.966 0.047 0.370
Q1.5d 0.068 0.032 0.846 0.000 0.499 0.227 0.253 0.229 Q3.3c 0.073 0.040 0.741 0.446 0.467 0.988 0.049 0.415
Q1.6 0.474 0.397 0.713 0.622 0.629 0.521 0.284 0.472 Q3.3d 0.242 0.190 0.347 0.538 0.150 0.255 0.033 0.066
Q1.7a 0.088 0.373 0.918 0.414 0.345 0.474 0.670 0.782 Q3.3e 0.165 0.132 0.339 0.089 0.236 0.811 0.031 0.526
Q1.7b 0.212 0.242 0.205 0.668 0.596 0.716 0.629 0.492 Q3.4a 0.000 0.000 0.105 0.000 0.005 0.461 0.228 0.768
Q1.7c 0.471 0.531 0.967 0.639 0.812 0.910 0.490 0.169 Q3.4b 0.000 0.000 0.278 0.011 0.011 0.363 0.394 0.140
Q1.7d 0.260 0.611 0.933 0.466 0.537 0.655 0.886 0.418 Q3.4c 0.020 0.007 0.122 0.008 0.307 0.653 0.117 0.093
Q2.1 0.031 0.054 0.676 0.004 0.588 0.389 0.216 0.482 Q3.4d 0.346 0.227 0.295 0.702 0.580 0.055 0.242 0.296
Q2.2 0.023 0.062 0.717 0.010 0.971 0.230 0.892 0.478 Q3.4e 0.050 0.008 0.568 0.056 0.392 0.497 0.287 0.672
Q2.3 0.101 0.567 0.767 0.016 0.716 0.535 0.688 0.474 Q3.4f 0.098 0.107 0.799 0.065 0.585 0.128 0.108 0.497
Q2.4 0.067 0.107 0.541 0.348 0.764 0.540 0.020 0.347 Q3.6 0.179 0.450 0.620 0.024 0.426 0.435 0.813 0.211
Q2.5a 0.441 0.809 0.532 0.911 0.479 0.558 0.524 0.385 Q3.7 0.001 0.800 0.278 0.016 0.586 0.292 0.499 0.983
Q2.5b 0.149 0.932 0.463 0.901 0.913 0.892 0.161 0.765 Q3.8 0.266 0.058 0.486 0.146 0.928 0.908 0.813 0.851
Q2.5c 0.066 0.149 0.327 0.395 0.720 0.491 0.183 0.099 Q3.9 0.000 0.001 0.164 0.001 0.087 0.527 0.252 0.494
Q2.6a 0.978 0.806 0.605 0.673 0.857 0.954 0.772 0.767 Q3.10 0.391 0.152 0.029 0.105 0.799 0.949 0.793 0.803
Q2.6b 0.668 0.675 0.831 0.609 0.644 0.984 0.704 0.687 Q3.11 0.730 0.652 0.184 0.113 0.983 0.711 0.391 1.000
Q2.6c 0.224 0.973 0.434 0.502 0.679 0.441 0.591 0.858 Q3.12 0.156 0.482 0.090 0.232 0.653 0.658 0.578 0.581
Q3.13 0.025 0.042 0.308 0.154 0.070 0.797 0.645 0.816
Chapter 8 Page 385

Total Number of Significant Variables out of 51 Variables 13 11 4 18 5 0 6 1


Chapter 8 Page 387

Table 8-16, gives the results of these tests for each dependent variable. The dependent

variables, which show significant differences between various groups (at the 5% level), are

shown in bold. It shows that the highest qualification, the working sector and the professional

category (whether professional, academic or others) of the participants, seem to be the most

significant variables affecting their responses. Social and Environmental education has very

little effect on the responses and these seem to be concentrated on those questions related

to the stakeholders that Islamic accounting should consider important as users of its

information. Similarly the place of education also seems to have little effect on the

responses, although the participants are almost evenly split between those educated in

Malaysia and those educated in Non-Muslim countries. This may be due to the fact the

participants who studied overseas for their post-graduate qualifications (see 8.5.2 above)

obtained a Master degree, which in the UK is only of one-year duration. This period may not

have been enough to influence their disposition towards Islamic accounting because they

may have obtained their undergraduate education in Malaysia over a period of four years.

The significant effect of place of education seems to be concentrated on information that

should be conveyed by Islamic accounting and the objectives of Islamic business

organisations. Islamic qualifications seem to have no effect at all on the responses.

However, this may be due to the fact that the sample only contained about 12 out of 186

participants with Islamic qualifications higher than school level, hence we may need a

sample with a broader spread of qualifications to see any effect. The effect of technical

assistance tells the same story that only one question on the objectives of Islamic

accounting seems to be significant. Here again the relevant sample is small - only 24 out of

206 participants, who responded, worked in public practice and out of this only 12 received

technical assistance. Hence, it is not possible to make any suggestions regarding the effect

of technical assistance on the disposition towards Islamic accounting. Due to the above

findings, the academics and accountants will be separately analysed. However, due to the

exploratory nature of the study, the cause-effect relationships of other independent variables

will not be analysed further.


Chapter 8 Page 388

8.7 CONCLUSION

In this chapter, the relationship of the research design, the scales used and the statistical

tests applied were discussed. The validity and reliability of the responses were tested and

discussed. Since the distribution of the responses did not follow a normal distribution, both

non- parametric and parametric tests will be applied in hypothesis testing in the next

chapter. The findings related to dependent variables, which consisted of questions in

sections 1 to 3 of the IAQ, were presented. The findings related to the aggregate scores,

consisting of groups of questions for each section, and the whole questionnaire were

discussed. The results indicate strong support for Islamic Accounting although some aspects

of conventional accounting such as those related to conventional accounting principles and

the traditional accounting objectives related to efficient allocation of resources and

maximisation of profits still find support among a substantial number of participants.

The findings related to the dependent variable were presented and regrouped for further

testing. The Kruskal-Wallis one way analysis of variance of ranked means shows that only

education, working sector and professional category seem to have significant effect on the

responses regarding disposition to Islamic accounting. Social and environmental accounting

education effects only a few of the variables. Since this is an exploratory study of the

subject, the cause effect relationships between dependent and independent variables will

not be analysed further. However, the hypotheses (hypotheses 1 to 10) regarding the

dependent variables as stated in the previous chapter, will be tested in the next chapter.

You might also like