Ethics Essay

Effect of Self-Accountability on Self-Regulatory Behaviour: A Quasi-Experiment

Amit Dhiman1 • Arindam Sen2 • Priyank Bhardwaj3

Received: 7 August 2015 / Accepted: 10 December 2015 / Published online: 24 December 2015

� Springer Science+Business Media Dordrecht 2015

Abstract An individual’s accountability to oneself leads

to self-regulatory behaviour. A field experiment afforded

an opportunity to test this relation, given that external

accountability conditions were absent. A single group pre-

test/post-test design was used to test the hypothesis. A

group of full-time resident management students, n & 550, take four meals during the day in the institute mess. As a

part of the experiment, food wastage in the form of left-

overs on the plates of subjects was measured. As a pre-test,

the measurement occurred at two levels. Subjects could see

how much they are adding to the total waste by looking at a

weighing scale placed under a waste basket, and they could

also see the total waste data for each of the four meals for

the day and a day earlier displayed at a prominent place.

After 105 days, the weighing scale under the basket was

removed, and as a post-test measurement, the total waste

data for the four meals were noted down for another

72 days. A manipulation test indicated that the experiment

has had the desired effect of invoking self-accountability in

subjects during the pre-test phase, and diluting it during the

post-test phase. Time series analysis of pre-test and post-

test data indicated that the wastage data decreased in the

pre-test phase. However, the post-test waste data showed

an increase over a period of time. The results indicate that

accountability conditions like social norms invoke self-

accountability cognition leading to self-regulatory beha-

viours in individuals.

Keywords Internal accountability � Quasi-experiment � Self-accountability � Self-regulation � Deontological view

Introduction

Accountability to Whom: Others and/or Self

The decisions and actions people take affect the interests of

others at social and individual levels. Decision makers are

answerable to people who are affected by their actions and

decisions. This answerability is termed as the decision

maker’s accountability and is a universal feature of any

decision environment (Tetlock 1985). Frink and Klimoski

(1998) defined accountability as ‘‘perceived need to justify

or defend a decision or action to some audience(s) which

has potential reward and sanctions power, and where such

rewards and sanctions are perceived as contingent on

accountability conditions’’ (p. 9). Thus decision makers

justify their decisions and actions to the ‘audience’ which

evaluates them against certain standards and expectations.

These standards are determined by formal rules, or infor-

mal norms and values related to specific decisions or

actions. While formal rules guide an individual’s actions

and decisions in formal situations, it is impossible to frame

rules for every conceivable situation even in a relatively

closed social system like an organization. For the majority

of everyday situations in which individuals find them-

selves, certain unwritten norms, morals, or values form the

standards. Decision maker feels accountable because the

& Amit Dhiman amitdhiman@iimcal.ac.in

Arindam Sen

senarindam@hotmail.com

Priyank Bhardwaj

bhardwaj.priyank@gmail.com

1 Indian Institute of Management Calcutta, Kolkata, India

2 HSBC, Bengaluru, India

3 PWC, Mumbai, India

123

J Bus Ethics (2018) 148:79–97

https://doi.org/10.1007/s10551-015-2995-4

‘audience’ controls certain rewards and sanctions for one-

self based on the above evaluation. These rewards and

sanctions can take tangible and intangible forms such as

status, image, group-membership, and money. Tetlock

(1985) identified the following three motives for people to

take cognizance of accountability: to protect and enhance

one’s social image, and to secure control of desirable

resources. These motives are often complementary and

mutually reinforcing.

For a given action and associated consequences, people

are answerable to multiple ‘audiences’ constituting a web

of accountabilities. Typically, at the work place, a decision

maker is answerable to superiors, subordinates and peers;

in the family to the spouse, parents and children; and in the

society at large to neighbours and social institutions. This

set consists only of the external ‘audience’, but the decision

maker is answerable to ‘self’or the internal ‘audience’ as

well (Schlenker and Weigold 1989). In every situation,

internal and external accountabilities are at work. Different

stakeholders have different stakes in a given situation and

often these stakes congruent. In order to meet account-

ability conditions because of these stakeholders, decision

makers face a web of accountability forces often pulling

them in multiple and opposing directions. These multiple

forces are determined by structural, social, interpersonal,

and ethical contingencies embedded into the decision

making situation. Thus decision makers choose that action

which fulfils most salient and strong accountability con-

dition (Frink and Klimoski 1998).

While research on accountability in the disciplines of

management and psychology is scant, research on internal

accountability or self-accountability is almost non-existent.

This research studies self-accountability and its influence

on an individual’s behaviour in the absence of account-

ability to an external audience.

Past Research on Accountability

Frink and Klimoski (1998) found that the search for

accountability literature in management and psychology

resulted in fewer than fifty references. Lerner and Tetlock

(1999) were the first ones to comprehensively review

accountability literature across different fields. They noted

that in recent times accountability has been studied exten-

sively in other fields like health, education, politics, but not

extensively in psychology and management. Further, most

of the studies in psychology have been laboratory studies

which have limitations in terms of generalizability. In order

to advance accountability research, there is a need to con-

duct more field studies.

Empirical research has shown both the positive and

negative effects of accountability on people’s decision

quality and actions. Lab studies prove that it reduces

judgment biases such as primacy effects (Tetlock 1983),

reduces overconfidence in personality prediction (Tetlock

and Kim 1987), reduces sunk cost effects (Simonson and

Nye 1992) and leads to more accurate judgments and

decisions (Ashton 1992; Brtek and Motowidlo 2002; Mero

and Motowidlo 1995). On the negative side, it can inflate

sunk cost if the decision maker is already committed to a

decision (Tetlock et al. 1989), increases stereotyping

(Gordon et al. 1988) and impression management (Ferris

et al. 1997), shifts decisions towards undesirable prefer-

ences of strong constituencies (Adelberg and Batson 1978)

and forces the decision maker to even consider irrelevant

information compromising the decision quality (Tetlock

et al. 1989). But the extant research has focussed on the

effect of external accountability conditions while ignoring

its internal counterpart, accountability to oneself.

Self-accountability

As with external accountability, the decision makers’ self-

accountability influences their actions and decisions. Self-

accountability can be defined as the need to justify one’s

actions and decisions to oneself in order to confirm or

enhance a self-identity or image shaped by strongly held

beliefs and values. It has also been defined simply as a

‘‘desire to live up to a salient, internally held self-standard’’

(Peloza et al. 2013, p 104). It leads to the individual’s self-

regulatory behaviour, irrespective of whether external

regulatory and accountability conditions of reward or

punishment exist or not.

Standards, Values and Self-Accountability

Implicit in the definition of self-accountability is the

comparison of an individual’s ‘actual’ decision or an

action with the ‘personal standards’ they hold on that

decision or action. Such comparative cognitions lead

to certain emotional states and corrective measures to

remove discrepancy, if any. Higgins (1987) propounded

the self-discrepancy theory and identified three distinct

‘domains of self’: actual-self, ideal-self and ought-self.

From one’s own ‘standpoint’ (as distinct from the other’s),

actual-self constitutes of attributes/standards that one

believes they possess, ideal-self constitutes of attributes/

standards one would like to possess based on their hopes

and aspirations, and ought-self constitutes of attributes/s-

tandards one believes that one should possess based on a

sense of duty or obligation (Higgins 1987).

Actual-self refers to an individual’s present and repre-

sents their actual perception of themselves (Oppenheimer

1990). Actual-self is only one of the representations of the

80 A. Dhiman et al.

123

self, the self being a multidimensional construct (Markus

and Wurf 1987). Before being disentangled and identified

as a separate facet of broader construct of the concept of

the self, actual-self was subsumed under the general, uni-

dimensional construct of the concept of the self. As early as

1890, James (1890) identified three different parts of the

self ‘‘its constituents’’, ‘‘feelings and emotions they

arouse’’ and ‘‘the actions to which they prompt’’ (p. 292).

The constituents included the following two dimensions:

self- as -subject (or ‘I’ or self-as-knower) and self-as-object

(or ‘me’ or self-as-known). Baldwin (1897) considered the

notion of the self to be an interactive product of these two

selves, the two positioned at the two ends of a continuum

running from the inside to the outside of a person (Smith

2014). Mead (1934) considered the self as a social structure

made up of social constructions about oneself that indi-

viduals experience by being part of a group they belong to.

Extensive literature on the concept of the self has been

compiled in an excellent review by Wiley (1979). Actual-

self constitutes the above elements as seen from one’s own

point of view and also seen from significant others’ point

of view, termed as standpoints by Higgins (1987).

Markus and Nurius (1986) conceptualized the ideal and

ought selves as ‘‘possible selves’’, worthwhile to pursue as

future potentials. In an empirical study, Higgins (1987)

found that when subjects could not meet ‘ideal personal

standards’ seen from their own standpoint, they experi-

enced emotions of ‘dejection’ manifested as sadness and

disappointment; the psychological situation of an absence

of positive outcomes linked to one’s hopes and wishes.

And when they could not meet the ‘ought personal stan-

dards’ seen from their own standpoint, they experienced

‘agitation related’ emotions such as guilt, self-contempt

and the psychological situation of the presence of negative

outcomes. Such emotions trigger corrective behaviours or

self-regulation to remove the discrepancy (Higgins 1987;

Passyn and Sujan 2006). Passyn and Sujan (2006) also

found that ‘higher self-accountability’ emotions such as

guilt have a greater potential to affect compliant behaviour

compared to ‘lower self-accountability’ emotions such as

hope. The former are evoked as a consequence of ‘ought

standards’ and latter corresponds to ‘ideal standards’.

Individuals set these self-standards based on certain

beliefs and values they strongly hold. Values have been

defined as ‘‘general standards by which we formulate

attitudes and beliefs and according to which we behave’’

(Posner et al. 1987, p. 376). Rokeach (1973) argued that

values can be classified into terminal and instrumental

types. ‘‘Terminal values are values that lead to desirable

state of existence (e.g. a world of peace, wisdom), whereas

instrumental values describe preferred modes of conduct

(e.g. honesty, love)’’ (Finegan 1994). Further Rokeach

(ibid.) argued that the instrumental values more than the

terminal values, and within instrumental those concerning

morality more than those concerning competence, would

have ‘oughtness’. Rokeach (ibid.) in establishing the

supraordinate, almost universal nature of ‘oughtness’ cites

Heider (1958)—‘‘ought can be represented as a cognized

wish or requirement of suprapersonal objective order which

has an invariant reality, and whose validity therefore

transcends the point of view of any one person’’ (p. 222).

To summarize, it is evident that instrumental values like

honesty and self control, primarily concerning morality,

have the characteristics of ‘oughtness’, leading to a more

likely evocation of high ‘self -accountability’ when these

are salient as self-standards.

Social Norms and Self-Accountability

Values differ from social norms in that while the former

transcend specific situations, the latter are a prescription or

proscription to behave in a certain way in specific situa-

tions (Rokeach 1973). Thus the same value (s) can be the

guiding principle for different norms. As an illustration,

‘conservation of food’ as a social norm can be argued to be

based on ‘self control’ and ‘responsible’ instrumental val-

ues and ‘equality’ terminal value, three such sentinel val-

ues identified by Rokeach (ibid.). Thus social norms based

on ‘ought’ values themselves assume ‘ought’ characteri-

zation especially when there is a predominant consensus on

such norms in a specific society. ‘Conservation of food’ is

one such social norm in Indian society, as a few studies

reported that respondents working in the organizations,

similar to subjects in this study, had learnt conservation

early in their lives in the families they grew up in (Jain

et al. 2013; Shrinivasan et al. 2013).

In turn, social norms reinforce one’s values and beliefs

(Trevino et al. 1998). Wherever there is high social con-

sensus that an act or a decision is good or bad or a norm has

‘ought’ characterisation, individuals will tend to feel more

accountable internally and externally to such a norm.

Maheshwari and Ganesh (2004) studied successful imple-

mentation of a ‘code of ethics’ at Tata Steel, an organi-

zation reputed for its adherence to ethical practices. They

found that ethics got institutionalized in the organization

through a three-stage process: creating awareness and

building consensus about ethical practices, creating formal

or informal monitoring mechanisms (preferably the latter

e.g. family pressure), and rewarding or punishing ethical or

unethical behaviour respectively. The study illustrated the

importance of creating awareness and building consensus.

Outside such a formal setup, in society at large, there exist

norms and a consensus about certain universal values, e.g.

honesty. Creating awareness and making individuals con-

scious about these norms are essential, so that they feel

self-accountable and regulate their behaviour accordingly.

Effect of Self-Accountability on Self-Regulatory Behaviour: A Quasi-Experiment 81

123

Self-Accountability and Self-Regulation

The influence of self-accountability on an individual’s

behaviour can be understood by research in the domain of

self-regulation. Self-regulation is a conscious effort on the

part of individuals to align behaviours with established or

preferred standards (Vohs and Baumeister 2004). It

involves directing behaviour towards apriori goal states

considered necessary or appealing as standards. In this

study, ‘conservation of food’ is the social norm evoking the

self-regulatory effort of reducing food wastage.

In the current study, it is proposed that in the absence of

strong external accountability conditions, it is the individ-

ual’s self-accountability that affects their decisions and

actions. And individuals experience this accountability

more when certain ‘ought values’ and ‘ought norms’

become salient as self-standards. Meeting these standards

fulfils an individual’s need to maintain a self-image. In the

absence of any external accountability, perceived self-ac-

countability explains the cognition underlying the effect of

norm on behaviour. Carver and Scheier (1982) argued that

‘‘directing attention to self when a behavioural standard has

been evoked by the nature of one’s role or setting, engages

the comparator at the level of control that is superordinate.

The result is a tendency to compare one’s perceptions of

one’s present state or behaviour against the standard,

leading to a reduction of perceptible differences between

the two’’ (p. 120).

Further, the self-discrepancy theory postulates that ‘‘in-

dividuals are motivated to reach a condition where their self-

concept matches their personally relevant self guides’’

(Higgins 1987, p 321). Self-guides are ‘‘ideal-self’’ and

‘‘ought-self’’ ideas that form the standards individuals try to

compare their ‘‘actual-self’’ with and bridge the gap. Duval

and Wicklund’s (1972) theory of objective self awareness

also similarly argued that increasing attention on oneself, the

discrepancies between the real self and personal standards

become starkly important, leading to a motivation to remove

the gap. As already mentioned, evidence exists that in case of

self-discrepancy, negative feelings may result (Higgins

ibid). These feelings motivate individuals to bridge gaps

between actual behaviour and standards, as also predicted by

the control model of self-regulation (Carver and Scheier

1982) and the theory of cognitive and behavioural change

(Rokeach 1973). Such negative feedback operationalizes

self-accountability leading to corrective self-regulatory

behaviour. The present research tests ‘‘conservation of food’’

as an ought-to adhere to standard.

However, all extant theories—self-discrepancy theory,

theory of objective behaviour and the control model of self-

regulation—emphasize that the above comparison (self-

accountability) and corrective behaviour (self-regulation)

would only occur when self-discrepancy is available and

accessible to an individual (Higgins ibid). Self-discrepancy

is available more readily when the divergence is more

intense. It is more accessible if the activation of discrep-

ancy is recent or frequent, and the stimulus event is

meaningfully related to it (Higgins ibid).

The design of quasi-experiment in this study ensured

that the above three conditions are adequately met. In this

study, due to the repetitive nature of the stimulus in the

form of daily wastage data at the individual and collective

levels, the accountability cognition progressively influ-

ences the individual’s behaviour. The control model of

self-regulation (Carver and Scheier 1982) explains this

progressive behaviour. According to this model, the basic

unit of cybernetic control is the negative feedback loop,

which effects the reduction between the present condition

and the reference value. The process is cyclic and pro-

gressively achieves outputs closer to the reference value.

Therefore it is hypothesized that:

Hypothesis 1 For social norms that are ‘ought’ in nature,

in the absence of external accountability conditions, self-

accountability will regulate individual’s behaviour.

Before we discuss the study design, it needs to be

mentioned that we have considered a deontological view of

values, standards, and self-accountability i.e. individuals

hold certain values irrespective of their consequences or in

other words a worldview of ‘‘virtue as its own reward’’

(Turillo et al. 2002). This view maintains that an action’s

morality is independent of its consequences, e.g. criteria of

good for the maximum number of people. Behaviour is

assessed for morality by examining the rules and principles

which guide such behaviour. Thus we have not discussed

certain important variables which take into account the

consequences of behaviour and which have been shown to

affect moral decision making. For example, we have not

considered magnitude of consequences for victims (or

beneficiaries), probability of effect (probability of action

and its detection by others), proximity with victims (or

beneficiaries) and temporal immediacy of consequences

(Jones 1991). Our method ensures that consequences and

above factors are not salient for individual decision.

Relevance of this Research in Organizations

This research has significance for business organizations.

Peloza et al. (2013) showed that the activation of self-

accountability in case of consumers resulted in more pos-

itive response among them to buy products with ethical

appeals focussed on environmentally friendly and sustain-

able product features. To the knowledge of researchers,

there are not any experimental studies that investigate the

82 A. Dhiman et al.

123

self-accountability and self-regulation relationship. Self-

regulation as a means to sustain desirable employee

behaviours has become increasingly critical in innovation

driven and service driven industries. In these industries, as

organizations adopt flatter structures, empower and

enhance ownership among employees, make work more

meaningful to employees, the conventional means of con-

trol and behaviour regulation like supervision or perfor-

mance evaluations are being shunned (The Economic

Times, 27th July, 2015). Organizations dependent on a

high rate of innovation, e.g. Google (Garvin et al. 2013),

have increasingly facilitated employee’s self-regulation to

extract desirable behaviours. Even in India, in high-tech-

nology industries, new age startup organizations like

Inmobi are shunning conventional performance evalua-

tions, doing away with supervisory approvals for leaves,

removing upper limits for tour expenses (Goyal 2015).

Organizations are also encouraging self-regulation to mit-

igate misuse of resources like water, power and paper

inside organizations (Jain et al. 2013). Despite the growing

importance of self-regulation, its prevalence and imple-

mentation remain inconsistent. This study aims to investi-

gate whether important self-accountability conditions, in

the absence of external accountability conditions would

help. In other words, the study aims to demonstrate that

when self-discrepancies in the ‘‘ought-to’’ norm are avail-

able and accessible to individuals, is the norm more likely

to be adhered to? In doing so, this study brings out certain

conditions under which self-accountability may cause

desired self-regulation.

Method

Procedure and Design

As was stated earlier, ‘conservation of food’ is considered

as the social norm for testing the hypothesis. It almost

acquires the characteristic of an ‘ought to adhere’ norm in

the moral fabric of the Indian society in view of widespread

and visible poverty and hunger (Jain et al. 2013; Shrini-

vasan et al. 2013). So ‘wastage of food’ is an important

metric that provides feedback to individuals about the

extent of violation of the ‘zero’ food wastage standard.

A naturally occurring quasi-experiment afforded an

opportunity to study unobtrusively the hypothesized effect.

We studied student behaviour related to food wastage in a

student mess inside a management institute in Western

India. In this institute, 550 students stay inside the campus

and take their meals in the institute mess. Meals are served

four times: breakfast, lunch, evening tea with snacks, and

dinner. Meals are served at a buffet and students are free to

take as much as they like on their plates or help themselves

with as many servings as they like. Mess management

noticed that students often leave huge amounts of eat-

able meals on their plates apart from leftovers like banana

peels and egg shells. Interestingly, the wastage increases

considerably whenever there is a special meal on offer, e.g.

dinner on Friday. This wastage not only caused consider-

able loss to the mess contractor and institute, but also

caused problems in washing and disposing off the waste. In

order to improve the disposal system as well as monitor

and measure this wastage, the mess committee placed a

waste basket in the mess and instructed students to just

throw away the leftovers in it. Later on they placed this

basket on a weighing scale so that each individual can

notice the amount he/she is adding to the waste. Also the

mess workers started displaying the day’s total and meal

wise waste figures along with corresponding figures for the

day before on a board placed at a strategic location at the

mess hall entrance. In addition, the daily data was recorded

in a computer in an excel file.

Thus a condition was created where each student can

observe the amount he/she is wasting and the amount the

group is wasting. It is assumed apriori that ‘no food wastage’

is an important ‘ought’ instrumental value for these students.

As per the self-accountability conceptualization discussed

above, waste measurement and display created a condition

where individual may feel accountable to themselves for

wasting food. Also no individual student was placed in any

kind of external accountability condition because there was

no penalty or reward associated with individual or total

wastage. Individual students could observe the wastage they

are adding to. Sometimes a few friends may compare their

figures, but it was rarely observed by authors. Thus the

absence of external accountability can safely be assumed.

And so the treatment is construct valid. The assumption

about self-accountability will be checked with the students at

the end of the experiment. Wastage per person represents the

dependent variable.

In order to test hypothesis, there is a need to create

conditions of high and low self-accountability. Data are

available from day 1 of the intervention and it is expected

that in phase 1, the waste figures will come down pro-

gressively as more and more students feel the pressure of

accountability. But this alone does not vary the account-

ability conditions. In order to do so, after 107 days in phase

1, the weighing machine was removed in phase 2. Of

course, the total waste was weighed separately and it was

also displayed on the board. Thus the ‘self-accountability’

condition was removed in phase 2. But the other conditions

remained same in phase 2 as in phase 1. Data were col-

lected for another 97 days in phase 2. As per the self-

discrepancy argument, in phase 1 the conditions of avail-

ability and accessibility of self-discrepancy to individual is

well met while that it is removed in phase 2.

Effect of Self-Accountability on Self-Regulatory Behaviour: A Quasi-Experiment 83

123

This quasi-experiment being a pre- and post-test kind

without a control group, we need to rule out the plausibility

of threats to the internal validity of the study like matu-

ration and history (Shadish et al. 2002). Since in this study

the treatment is applied and then is removed, it may be that

post-test results may be influenced by maturation of sub-

jects during study. It is expected that in phase 2, the waste

figures will again start rising, if self-accountability had an

effect on students’ behaviours in phase 1. There exist

equally probable alternative expectations. After phase 1, it

may just happen that the waste figures may not rise at all

and remain at or near the stable low level achieved at the

end of phase 1. The explanation for two alternative

responses comes from behaviour modification literature

(Luthans 2005). Once the subjects are made aware of the

wastage they cause, each may become more conscious of

the food they take on their plates. This may develop into a

habit which is an automatic rather than a cognitively

controlled response. If figures reflect such a situation then it

will be checked at the end of experiment with the students.

Of course, it would not mean that self-accountability is not

working, rather that self-accountability being more cogni-

tive or controlled mind state, it may not be in operation in

case of effect of ‘habit’, and it may be a manifestation of

maturation. Thus in such a scenario, it can be concluded

that self-accountability acted initially, but once subjects

acquired a ‘habit’, it stopped influencing self-regulation.

To rule out history as a threat to validity, we tracked any

parallel event that might have coincided with the two

phases and might have influenced wastage figures. Some of

these events have been discussed and accounted for in the

next section. Other than those, we did not find any other

coinciding event. It needs to be mentioned that, though we

do not have figures of wastage before treatment, it was

already known that huge wastage was happening and thus

the intervention was employed initially to enhance

awareness and measure waste.

There are a few variables which need to be controlled to

test the hypothesis. First one is the preference for a dif-

ferent menu. It is expected that the waste will vary from

menu to menu because certain menus are liked more by the

majority of students compared to some other menus.

Menus, more or less repeat every week on the same day,

and thus a weekly fluctuation is expected.

Although self-accountability is an individual construct

and manipulation is also individually focussed, the mea-

surements are aggregated at the group level. The

researchers were aware of the ‘ecological fallacy’ they

would commit in making inferences about individual

behaviour from group level data. However, it was almost

impossible to collect individual level data without being

obtrusive and without producing ‘noise’ of visible identi-

fiability and external accountability among subjects.

Instead we would demonstrate that inference about indi-

vidual effects from group level data is not erroneous,

because the changes observed in the aggregate could not

have been achieved unless the large proportion of subjects

(as against statistically significant) had not brought down

their wastages. It is possible to follow this heuristic in this

phenomenon only because there is an upper bound on the

extent of wastage a subject reasonably would be

committing.

The large sample size of around 550 students makes it

more likely that the individual factors related to personality

are randomized, and so they do not have a systematic

influence on the dependent variable. Other factors such as

individual meal preferences are almost controlled because

the menu is decided by the mess committee, a student body

that consults the students about preferred menu. Moreover,

students face quite homogenous acculturation (e.g. both

academic/non-academic) at the institute and that may result

in nullifying significant influence of any unaccounted indi-

vidual factor. Also students fall in narrow range of age

groups, are predominantly males (88 %) and go through

highly homogenous daily routines. The sample constituted

12 % female students, the average age of the total group was

26.5 years with minimum and maximum of 22 and 32 years,

respectively. Males and females did not differ in educa-

tional, demographic, and socio-cultural background. Given

that females were far less in proportion, they were not

considered separately in the data measurement and analysis.

Manipulation Check Method

In order to check whether the intervention has had the

intended effect, i.e. subjects underwent self-accountability

cognitions; they were asked to fill up a short online ques-

tionnaire after the experiment (see Appendix 1 for the

questionnaire). It contained five items to check whether the

introduction of the weighing scale had any cognitive and

behavioural effect on the mess member’s food wastage

habits. It also checked whether the removal of the weighing

scale reversed this effect. All but two items were measured

on a 6-point Likert scale measuring responses ranging from

strongly agree to strongly disagree. It needs to be mentioned

here that ‘wastage’ here is a comparison between ‘actual’

and ‘ought’, and becoming conscious of the ‘wastage’ one

has caused implies self-accountability cognitions.

After the experiment, the questionnaire was displayed

on the two mess announcement notice boards in the stu-

dents’ intranet. After 4 days, a reminder was put on these

notice boards. The questionnaire remained active on these

notice boards for a period of 1 month. A total of 54

respondents answered, representing a 11 % response rate

based on the enrolment figure for the month of February.

84 A. Dhiman et al.

123

The reasons for lower response rate could be the

engagement of students in two critical activities during

this period: examinations and campus job interviews. In

fact after this period, second year MBA students (almost

50 % of the population) left the campus.

Data Organization and Treatment

We have plotted the time series data of wastage per person (in

g per person) for dinner, lunch and total meals (Figs. 1, 2, 3).

These plots show trends for two periods—wastage trends for

105 days after the introduction of the weighing machine and

wastage trends for 65 days after the weighing machine was

removed. For calculating wastage per person for a given

month, the number of persons eating during any meal was

taken to be the numbers enroled in the mess at the beginning

of every month. Due to limitations of resources, it was not

possible to physically check how many people actually ate

during every meal. But it is a safe assumption to make that

almost all students who enroled in a month would have eaten

most of the time at the mess. First reason is that the MBA

programme is residential and requires that students stay on

Total wastage/person

0

20

40

60

80

100

120

140

8/9 /20 06

8/1 6/2 00 6

8/2 3/2 00 6

8/3 0/2 00 6

9/6 /20 06

9/1 3/2 00 6

9/2 0/2 00 6

9/2 7/2 00 6

10 /4/ 20 06

10 /11 /20 06

10 /18 /20 06

10 /25 /20 06

11 /1/ 20 06

11 /8/ 20 06

11 /15 /20 06

11 /22 /20 06

11 /29 /20 06

12 /6/ 20 06

12 /13 /20 06

12 /20 /20 06

12 /27 /20 06

1/3 /20 07

1/1 0/2 00 7

1/1 7/2 00 7

1/2 4/2 00 7

1/3 1/2 00 7

Fig. 1 Wastage (in g) per person trend (actual and weekly moving average plot) for all meals combined (dinner, lunch, breakfast and high tea). Arrow points to removal of weighing scale

Dinner wastage/person

0

10

20

30

40

50

60

70

80

8/9 /20 06

8/1 6/2 00 6

8/2 3/2 00 6

8/3 0/2 00 6

9/6 /20 06

9/1 3/2 00 6

9/2 0/2 00 6

9/2 7/2 00 6

10 /4/ 20 06

10 /11 /20 06

10 /18 /20 06

10 /25 /20 06

11 /1/ 20 06

11 /8/ 20 06

11 /15 /20 06

11 /22 /20 06

11 /29 /20 06

12 /6/ 20 06

12 /13 /20 06

12 /20 /20 06

12 /27 /20 06

1/3 /20 07

1/1 0/2 00 7

1/1 7/2 00 7

1/2 4/2 00 7

1/3 1/2 00 7

Fig. 2 Wastage (in g) per person trend (actual and weekly moving average plot) for dinner. Arrow points to removal of weighing scale

Effect of Self-Accountability on Self-Regulatory Behaviour: A Quasi-Experiment 85

123

the campus during term time. Second, the authors, being

students themselves at the time of the study, verified that the

program is extremely intensive and students are required to

spend considerable time in studying and working in groups

during evenings and late at night, so that occasions to go out

are minimal. Last, students enrol and pay a lump sum amount

every month irrespective of the number of meals they eat,

and are conscious of the loss if they do not eat. However,

there are other known and random factors that influenced the

wastage data and threaten the internal validity of the study.

These need to be accounted for in the absence of a compar-

ison control group. These factors are discussed in the fol-

lowing sections and also discussed are the ways their effects

were accounted for.

Student Events

During the period of investigation, certain events took place

in the institute which resulted in additional people from

outside eating in the mess for a period of 3–4 days at a

stretch. These events included Chaos during 26th–28th

January, 2007, Confluence during November 23rd–26th,

2006 and Amaethon during 19th–21st December, 2006. We

have dropped the mess data for the first two events, because

there was no way to know how many people were eating in

the mess during these events. In any case these included a

considerably large number of additional people eating at the

mess, and since these were not part of the intervention, it is

suitable to drop these days from the analysis. For the last

event, since the numbers were far lower, we have used an

informed estimate about the number of additional people

eating during the event.

Term Off

Whenever a term of approximate 3 months ends for MBA

students, they have a week off and most of them travel to

their home towns. During these periods, we deducted

numbers equal to the batch size from the total number of

people enroled at the start of the month. Additionally, we

cross checked with the mess records the number of ‘mess

opt out’ forms filled up by such students for these periods.

Mess rules require that students moving out of campus for

a period of at least 1 week need to fill up a ‘mess opt out’

form in order to get a refund for that period. Since the sum

involved is considerable, it is assumed that most of those

who moved out filled up this form.

Specific Out of Campus Courses

Then there are periods during which the second year MBA

students were out of the campus as a part of academic

courses. For example, during the first week of February, 25

students were away from campus as part of the ERI course

conducted outdoors, and during February 12–18th, 30

students went out of campus as a part of the ‘khoj’ team for

a period of 10 days. These numbers were accounted for in

the calculations.

Other Programmes

Apart from the MBA programmes, the mess also catered

to the students enroled in other course. These students

were also present on the campus for most of the period

during the course of the intervention. These included 30

Lunch wastage/person

0

5

10

15

20

25

30

35

40

45

50

8/9 /20 06

8/1 6/2 00 6

8/2 3/2 00 6

8/3 0/2 00 6

9/6 /20 06

9/1 3/2 00 6

9/2 0/2 00 6

9/2 7/2 00 6

10 /4/ 20 06

10 /11 /20 06

10 /18 /20 06

10 /25 /20 06

11 /1/ 20 06

11 /8/ 20 06

11 /15 /20 06

11 /22 /20 06

11 /29 /20 06

12 /6/ 20 06

12 /13 /20 06

12 /20 /20 06

12 /27 /20 06

1/3 /20 07

1/1 0/2 00 7

1/1 7/2 00 7

1/2 4/2 00 7

1/3 1/2 00 7

Fig. 3 Wastage (in g) per person trend (actual and weekly moving average plot) for lunch. Arrow points to removal of weighing scale

86 A. Dhiman et al.

123

members of the faculty development programme who

stayed on campus for a period of more than 3 months

from October 31st–February 17th. There were 60 mem-

bers of the management programme for defence officers

who regularly ate in the mess for a period of 6 months

from Oct 4th–March 26th. Additionally, there are around

35 fellow program in management (FPM) students who

were also regular members of the mess during this

period. These numbers were taken into account, and term

breaks, wherever applicable, were also accounted for in

the calculations. The same mess rules applied to these

students as it applied to the 2-year MBA programme

students.

Specific Events

Last, there were certain events in which the students were

involved which would have affected the waste figures. It is

difficult to put exact numbers to such events, but while

making inferences and explaining the trends, these have

been taken into account. These included summer job

placements for the first year MBA students during the

month of October for 4 days, and pre-placement talks for

second year MBA students in the month of February. The

former event took place during the day and would not have

affected the dinner data that we considered to test our

hypothesis. The latter event, for most of the time, took

place in the evenings and involved students ranging from

40 to 60 members who frequently skipped dinner, if they

had snacks during such talks. In retrospect, this would not

have affected our phase 2 results adversely because, as

expected, the per person wastage data showed an upward

trend in this phase and subtracting these students from the

denominator of wastage per person data would have only

accentuated the hypothesized effect.

Focus on Dinner Data Only

In this paper, we have presented data for all meals—lunch,

dinner, breakfast and high tea. Data for the last two meals

has been included in the total meal wastage figure and is

not shown separately. Looking at the data we did not

consider it appropriate to analyse these two meals sepa-

rately because these represented only 25 % of the total

daily wastage and had lot of random noise due to a number

of unaccountable factors. Many students randomly skipped

breakfast or ate it quickly because they had to attend

classes in the morning, and given the late night working

habits at the institute, many struggled to reach classes in

the morning. Also, it was noticed that during high tea,

students visited the mess randomly to eat snacks. On the

other hand, dinner alone accounted for 42 % wastage and

lunch accounted for 33 %. We focussed on dinner for

testing our hypothesis because in the case of lunch, there is

much more random noise as compared to dinner. For

example, MBA first year students undergo unannounced

quizzes 2–3 times a week just after lunch time and these

are announced just before lunch. Students are not only

anxious, but many tend to skip lunch. Then there are other

visitors during working hours who visit the mess for lunch

on cash payment basis. On the other hand, during dinner,

the students are relaxed, there are practically no visitors,

and thus the data are a more valid representation of the

phenomenon under study. Additionally, special meals are

prepared mostly in dinners, and wastage data during these

meals is important because of higher wastage. Even then a

comparison of three data figures—total, lunch and dinner,

reveals that overall trends are similar for all meals. This

helped us in cross validating our inferences based on the

dinner data.

Actual and Moving Average Plots

In order to smoothen the variation and identify the trend,

we have used ‘moving average’ data in conjunction with

the actual data for drawing inferences. Each trend

chart shows two plots—one enumerating the actual

wastage and the other showing the weekly moving average.

We used a weekly moving average because many iterative

activities related to the mess happen over a cycle of a week.

For example, menu repeats (not exact replication) over a

week’s time, e.g. every Monday, Wednesday, and Friday

menu includes non-vegetarian dishes. Then the MBA and

other students had their term off for a period of 1 week.

Initially we had planned to account for people’s pref-

erences for certain menus over others as well as prepara-

tions. Since the menus were rarely replicated exactly, it

was difficult to get this data. Still we analysed wastage data

in conjunction with the meals menu to get better insights.

In any case, it may have only caused fluctuations in daily

data, but the overall trends have been unmistakable (refer

Figs. 1, 2, 3).

Results

Wastage Trends in Phase 1: Weighing Scale in Place

In Figs. 1, 2, and 3, series 1 corresponds to phase 1 of the

study, i.e. when the weighing scale was placed below the

waste basket for 105 days between August 9th and

November 21st ,

2006. The three charts, respectively, rep-

resent trends recorded for the total wastage per person data,

dinner wastage per person data and lunch wastage per

person data. A visual check on the wastage trends for series

1, especially the weekly moving average, across the three

Effect of Self-Accountability on Self-Regulatory Behaviour: A Quasi-Experiment 87

123

charts reveal a close match in the wastage data trends.

Although actual data fluctuations differ across the three, the

long-term decreasing trend is consistent. Interestingly,

from 22nd September to 4th October, the three show

similar, unexpected increases in the moving average and

absolute data. We searched for a plausible explanation for

this discrepancy. One plausible explanation could be rela-

ted to the 1 week vacation after the first term for the MBA

first year students. It was noticed that this glaring aberra-

tion lasted for a period of 2 weeks immediately after these

students returned from a 1 week vacation. Importantly, this

is the first vacation for the MBA first year students after

they had joined the institute and undergone the most

stressful MBA term. It seems that after spending 1 week

time at their homes, it takes some time for these students to

adjust to the mess food again. To check the validity of this

explanation, we compared wastage figures for a similar

period after the second 1 week vacation for these very

students. Thus looking at the 8th January to 15th January

2007 one week vacation, a similar increasing trend is

noticeable, though not a very significant one. But similar

effects were not evident for MBA second year students.

Therefore, one can only conjecture that over a period of

time students adjust better to such changes. No other

plausible explanation could be found.

Wastage Trends in Phase 2: Weighing Scale

Removed

In Figs. 1, 2 and 3, series 2 corresponds to the phase 2 of

the study, i.e. when the weighing scale was removed from

below the waste basket for 71 days between November

22nd and January 31st 2007. During this period two student

events—Confluence and Chaos—were organized which

brought a substantial number of visitors from outside who

temporarily ate in the mess. These events were organized

from 23rd to 27th November, 2006 and from 26th to 28th

January, 2007, respectively. In order to remove this

extraneous effect, we have dropped these periods from

analysis. Another event, Amaethon, was organized from

19th to 21st December, 2006, but we have kept the fig-

ures for this period in our analysis, because the number of

temporary visitors was relatively very small (6–9 %). After

removing these periods, it is evident from the three charts

that trends for per person wastage for dinner, lunch and

total are similar and increasing during phase 2.

Based on a visual comparison of actual and weekly

trends, we can infer that during phase 1, when the weighing

scale was placed, the wastage figures exhibited a decreas-

ing trend. But in phase 2, this trend reversed and the actual

wastage started increasing. Therefore, there exists prelim-

inary support for the self-accountability hypothesis.

Time Series Analysis

In order to test the statistical significance of the visual

trends noticed across the two phases, we conducted a time

series analysis for the two phases separately.

Series 1: Stationarity and Model Specification

Before testing any particular model, the series was checked

for stationarity using the Dickey–Fuller unit root test. In

order to incorporate distinct possibilities, three models, as

suggested by Gujarati (2003), were tested. These are given

below:

Model 1 : D yt ¼ a þ d yt�1 þ b2t þ e Model 2 : Dyt ¼ a þ dyt�1 þ e Model 3 : Dyt ¼ dyt�1 þ b2t þ e

The null hypothesis is d = 0, i.e. there is a unit root, and the series is non-stationary. According to the Dickey–Fuller

test statistic, if a t value for d is [ tcr (= f), then a null hypothesis is rejected and the series can be considered sta-

tionary. All three models show stationary properties based on

the Dickey–Fuller test statistics as given in Table 1.

In the next step, we plotted correlograms for the autocor-

relation function (ACF) and the partial autocorrelation func-

tion (PACF) of yt. These plots are shown in Fig. 4. It is evident

from these plots that yt is influenced by yt-2 and yt-7, i.e. a lag

effect of alternate day and 1 week, respectively. These effects

are not difficult to understand. First, the non-vegetarian menu

repeats every alternate day (i.e. Monday, Wednesday, and

Friday), except for the 2-day gap at the week end. Given that

almost 46 % of the students eat non-vegetarian meals, the

alternate day waste figures are expected to be related because

of the unavoidable, almost fixed weight of leftovers in the

form of bones. Weekly lag is also expected because the meal

menu repeats over a cycle of 1 week.

Along with these lags, we expected that yt will decrease

with time due to the hypothesized effect, as is evident in the

graphical plot shown in Fig. 4. We have also included one

period lag term in the model. Since the last day’s meal wise

and total wastage figure was displayed at a strategic location

in the mess, our expectation was that it will have a negative

Table 1 Dickey–Fuller test for checking stationarity of series 1

Models t value

for da Dickey–Fuller

tcr (=f) at 5 % Stationarity

(t [ tcr)

1 -12.2 -3.45 Yes

2 -9.31 -2.89 Yes

3 -3.38 -1.95 Yes

a yt-1 coefficient

88 A. Dhiman et al.

123

influence on the next day’s wastage figure. In addition, to

keep the model simple, we assumed a linear relation,

although intuitively yt may decay geometrically, with the

maximum fall witnessed in the initial period and the decay

rate falling down gradually. Therefore, we tested the fol-

lowing ARIMA (3, 0, 0) model to represent series 1:

yt ¼ a þ b0yt�1 þ b1yt�2 þ b2yt�7 þ b3t þ e: ð1Þ

The regression results (Table 2) show a moderate explana-

tory power of this model with R2adj ¼ 0:33 (F = 11.93; p\ 0.00). The ACF and PACF plots for studentized residuals (Fig. 5) fell within the 95 % confidence interval, thereby

proving that the model specification is adequate.

The regression coefficients reveal that yt-7 and t are the

significant and influential independent variables. As

expected, yt-1 and t have negative signs, whereas yt-2 and

yt-7 exert positive influence on yt.

Series 2: Stationarity and Model Specification

For series 2, we have used the same model as specified in

Eq. 1. Before that, as for series 1, we conducted D-F unit

Fig. 4 a Autocorrelation function (ACF) plots for series 1 (pre-test) data. b Parial ACF (PACF) plots for series 1 (pre-test) data

Table 2 Regression of dinner wastage (yt) for series 1

Model b t

(Constant) 4.02

Dinnerlag1 -0.07 -0.74

Dinnerlag2 0.07 0.72

Dinnerlag7 0.31** 3.21

Date -0.35** -3.01

** p \ 0.05; * p \ 0.10

Fig. 5 a ACF for studentized residuals of regression Eq. 1 for series 1. b PACF for studentized residuals Eq. 1 for series 1

Effect of Self-Accountability on Self-Regulatory Behaviour: A Quasi-Experiment 89

123

root test for series 2 using all three models as in series 1

above to test for stationarity of series. The results are given

in Table 3 below:

Based on the D–F test, we concluded that series 2 is

stationary, except when tested on model 3, which in any

case does not match the specified time series model in

Eq. 1 (Table 4).

In the next step, we plotted correlograms for the auto-

correlation function (ACF) and partial autocorrelation

function (PACF) related to yt. These plots are shown in

Fig. 6. For series 2, the first, second, and seventh lag values

are not that influential as in series 1. The 5th or 10th lag

values are comparatively more influential. However, none

of the lag effects fall outside the 95 % confidence interval.

Theoretically the logic of the 1st, 2nd, and 7th lag effects

still applies in case of series 2. Therefore, for series 2, we

regressed yt as per Eq. 1. The model explained the

insignificant proportion of variance in yt with R 2 adj ¼ 0:07

(F = 2.18; p = 0.082). The ACF and PACF plots for

studentized residuals (Fig. 7) fell almost within the 95 %

confidence interval, thereby proving that the model speci-

fication is adequate.

None of the independent variables emerged significant. 1

Some of the lag effect signs were also inconsistent with the

expectations. For example, we expected positive signs for

both yt-2 and yt-7, but we got -ve for first and ?ve for

second, respectively. This inconsistency is also reflected in

the ACF and PACF plots for the two series. While for

series 1, the first seven lag effects in ACF plot consistently

fell on one side of the mean line, for series 2 these fell on

both sides. As expected, time or date showed a positive

effect, thereby confirming the reversal of the series 1 trend

of temporally falling wastage figures. Thus we can infer

that, after removal of the weighing scale, the wastage data

again started increasing. It is also evident from the pro-

gressively increasing amplitude of variation in the wastage

data in the case of series 2. It again revealed a reversal of

the trend observed in series 1 wherein the variation in

wastage figures steadily fell after large initial variations.

These trends are visible in the ACF and PACF plots for the

two series in Figs. 4 and 6.

Table 3 Dickey–Fuller test for checking stationarity of series 2

Models t value

for da Dickey–Fuller

tcr (=f) at 5 % Stationarity

(t [ tcr)

1 -6.6 -3.45 Yes

2 -6.4 -2.89 Yes

3 -1.24 -1.95 No

a yt-1 coefficient

Table 4 Regression of dinner wastage (yt) for series 2

Model b t

(Constant) 2.44

Dinnerlag1 0.15 1.15

Dinnerlag2 -0.21 -1.61

Dinnerlag7 0.21 1.49

Date 0.17 1.17

** p \ 0.05; * p \ 0.10

Fig. 6 a ACF plots for series 2 (post-test) data. b PACF plots for series 2 (post-test) data

1 Based on ACF and PACF plots for dinner wastage figures for series

2, yt-5 and yt-10 also seem to have influence on yt. Although there

seems no logic for these effects, we included these lags also in Eq. 1

and ran regression again. Except that it improved R2adj to 17 %, and

ACF and PACF plots for residuals fell within 95 % limits, and time

(t) remained insignificant but in positive direction.

90 A. Dhiman et al.

123

It is our conjecture that progressively falling variations

in series 1 indicate increasing effects of self-regulation

exercised by subjects as the overall mean wastage fig-

ures also dropped. In series 2, the trend reversed with

variations amplitude again increasing.

Manipulation Check Results

As discussed in the section on methodology, the effect of

the weighing scale on students’ cognitive and behavioural

responses was checked using Likert scale items. 5 scale

items referred to phase 1, 3 to phase 2, and 1 was a general

question. Data (Panel 1) show that out of 54 respondents,

more than 70 % felt that the placement of the weighing

scale did make them conscious about the waste they are

adding, thus activating self-accountability. And they also

agreed that it is the main reason for the waste reduction. It

needs to be reiterated here that being ‘conscious of waste’

evokes self-accountability cognitions because ‘conserva-

tion of food’ is assumed to be an ‘ought to’ norm with an

almost universal nature, as argued earlier.

Almost 80 % of respondents felt that after the removal of

the weighing scale, they have stopped noticing the total waste

figures displayed on the board. Although they disagreed that

they have stopped noticing the food they waste on their plates.

They also agreed that the scale did have a major impact on

altering the wastage habits of the students. Almost 65 % of the

respondents replied that they did notice the removal of scale,

and 60 % noticed it in the month of December.

Therefore, it can be inferred that the placement of the

weighing scale did have the intended impact of evoking

self-accountability, and the waste trends and time series

results are not spurious. It is also evident that it led to a

change in waste producing behaviour.

Ecological Fallacy or Significant Result

As discussed earlier, we are using group level data to infer

about individual level phenomena. The experiment had to be

conducted in this manner partly in order to maintain its

internal validity, and partly due to the constraints of mea-

suring individual level information. However, a reduction of

dinner wastage data in phase 1 from 20 to 7.5 kg and

increase in phase 2 from 7.5 to 15 kg could not have been

achieved unless a large (against statistically significant)

proportion of subjects had not contributed to this reduction.

We are presenting a table below that presents the numbers

(of subjects) required to achieve the above wastage reduc-

tion under different average wastage reduction assumptions.

Fig. 7 a ACF plots for studentized residuals of regression Eq. 1 for series 2. b PACF plots for studentized residuals of regression Eq. 1 for series 2

-D average wastage/person

(g/person)

Wastage % of

average food intake

quantity (g) a

Number of

contributing

subjects (n)

500 100 25

100 20 125

75 16.7 167

50 10 250

25 5 500

a Average food intake has been taken at 500 g (As per report of

National Institute of Nutrition, ICMR, Hyderabad India, an adult

Indian consumes around 500 g of food in a meal. Reference: ‘‘Sample

meal plan for adult man’’ (p. 88). Dietary guidelines for Indians

(2011), National Institute of Nutrition, ICMR, Hyderabad, India.

Random measurement of food taken by subjects in the study also

revealed similar figure.)

Effect of Self-Accountability on Self-Regulatory Behaviour: A Quasi-Experiment 91

123

Further it needs to be pointed out that, at its lowest the

wastage per person was only 15 g, which would mostly be

accounted for by unavoidable left over like banana and egg

peals, fish/mutton/chicken bones and so on. It implies that

most of those who were in the habit of wasting food would

have reduced it in phase 1. Of course, many subjects might

not be wasting any food even before the intervention.

Based on the above arguments, we can confidently infer

that group level data do indicate reductions in wastage at

individual levels by a significant number of subjects.

Discussion

Evidence in Support of the Hypothesis

The overall wastage data trends, wastage data variation

trends, and time series analysis for the two series provide

support for the hypothesis. When the weighing scale was in

place, the moving average wastage data progressively

reduced from 40 to 15 g/person. After removal of the

weighing scale, it again increased to around 30 g/person.

Corresponding figures for the reduction of total wastage were

80 g/person initially to 40 g/person, and again an increase to

70 g/person. The time series analyses confirm these trends.

For phase 1, after controlling for lag effects, the effect size of

time on wastage data was negative and large at -0.354

(p \ 0.01). For phase 2, this size was smaller at 0.169, but positive. Therefore, we have reasonable confidence that

these trends are a consequence of self-accountability con-

ditions leading to self-regulatory behaviour.

The variation trends for the two series also indicate support

for the influence of self-accountability conditions on subjects’

wastage behaviour. It is expected that when students are not

put under self-accountability conditions, their food wastage

behaviour will be very erratic. But under accountability

Panel 1 Manipulation check results

No. Items N Mean SD Median Mode % C66th %*

Phase 1

1 After weighing scale was introduced, I keenly noticed scale

reading how much I am adding to the waste

54 3.94 1.63 4.5 5 70

2 After weighing scale was introduced, I keenly followed the

wastage statistics displayed on white board.

54 4.00 1.24 4 4 83

3 I think weighing scale made me more conscious about the

wastage

54 4.42 1.43 5 5 83

4 I think I consciously reduced wastage myself 54 4.87 1.06 5 5 91

5 I think overall mess members reduced wastage because they

became conscious of amount of wastage

54 4.24 1.09 4 5 83

Phase 2

8 Currently I have stopped following the wastage statistics

displayed on white board

53 4.36 1.34 5 5 77

9 After weighing scale was removed, I have stopped noticing

wastage (in my plate)

52 2.09 1.11 2 2 17

10 I do not think weighing scale can change individual wastage

habits

54 2.76 1.40 2 2 22

*C4

Phase 1 manipulation check response profiles Phase 2 manipulation check response profiles

6 Did you notice removal of weighing scale (basket remained)? Yes 34 64 %

7 When did you first notice the removal of weighing scale? Nov 2 4 %

Dec 18 42 %

Jan 12 28 %

92 A. Dhiman et al.

123

conditions, this behaviour may be more controlled. As is

evident from the figure, the high peaks at the beginning of

series 1 gradually tapered down to almost levelled data vari-

ation nearing 22nd November. But after the removal of the

weighing scale the variation seemed to grow again.

To validate our findings, we plotted per person total

and dinner wastage data only for Fridays across two ser-

ies (Ref. Fig. 8). Now Friday dinners are special both for

vegetarian and non-vegetarian mess members. The waste

figures for Friday invariably were higher and more erratic.

The trends clearly reveal that, even for Fridays, initially the

wastage data varied wildly and then steadied down along

series 1. The data again started showing higher variations

along series 2.

Therefore, considering the evidence and given that the

manipulation seems to have been effective, we can infer

that the self-accountability condition influenced the self-

regulatory behaviour of the subjects. The results also

indicate a progressively improving or deteriorating

response in series 1 and 2, respectively. It thus supports the

Carver and Scheier’s (1982) control model of self-regula-

tion. The results of series 2 also suggest that despite

undergoing behaviour modification conditions and a pro-

gressive change in behaviour, the changed behaviour did

not persist as a habit in series 2. The respondents of the

online survey also informed that the majority of them have

stopped noticing the cumulative waste figures displayed on

the board in post-test phase. It implies that the weighing

scale functioned as an individual stimulus that made stu-

dents conscious or self-accountable, making them more

observant about the wastage data as well as the size of

helpings they take and waste. Once the stimulus was

removed, the students seemed to have gone back to their

earlier careless food eating practices, less observant of their

food intake and waste behaviours. However, the majority

of them denied that they have stopped observing the

amount of wastage on their own plates.

Implication for Theory and Practice

In demonstrating the significant influence of self-ac-

countability on self-regulation in the absence of external

accountability conditions, the study seeks to establish one

of the key antecedents of self-regulation. The study adds

to the understanding of a relatively under-researched

construct of self-accountability. While any condition of

accountability to an external ‘audience’ entails a con-

comitant internal accountability, understanding the dis-

entangled influence of self-accountability on self-

regulation is this study’s contribution to the extant

WASTAGE ON FRIDAYS

0

20

40

60

80

100

120

140

8/ 11 /0 6

8/ 18 /0 6

8/ 25 /0 6

9/ 1/ 06

9/ 8/ 06

9/ 15 /0 6

9/ 22 /0 6

9/ 29 /0 6

10 /6 /0 6

10 /1 3/ 06

10 /2 0/ 06

10 /2 7/ 06

11 /3 /0 6

11 /1 0/ 06

11 /1 7/ 06

11 /2 4/ 06

12 /1 /0 6

12 /8 /0 6

12 /1 5/ 06

12 /2 2/ 06

12 /2 9/ 06

1/ 5/ 07

1/ 12 /0 7

1/ 19 /0 7

gm /p

er so

n

Fig. 8 Total and dinner wastage/per person trend only for Fridays across two series. Arrow points to removal of weighing scale

Effect of Self-Accountability on Self-Regulatory Behaviour: A Quasi-Experiment 93

123

literature. It also provides credence to the deontological

view on accountability that views adherence to certain

‘ought to’ values and norms as a reward in itself (Turillo

et al. 2002).

The practitioners may find useful insights from the

research about managing the contemporary work place

settings that are increasingly becoming flatter, freeing the

employees from conventional means of control like

supervision. Self-regulation has an important part to play in

such workplaces for getting the desired behaviour from

employees. This study indicates that unless in such cases

self-accountability is not evoked by providing feedback

information about ‘actual’—‘ought’ discrepancy, the

standards may not be adhered to and self-regulation may

fail. In other words, self-discrepancies need to be available

and accessible to individuals (Higgins 1987) for self-ac-

countability to function. However, for that to work,

employees need to know and imbibe the ‘‘ought to’’ stan-

dards followed by the organization. Organizations like

high-technology firms in Silicon Valley are dependent on

the contemporary knowledge of its employees, their inno-

vative ideas and intrapreneurship. In order to foster such

employee initiatives, organizations need to provide a cli-

mate of autonomy, risk taking, and ownership. Conven-

tional means of control on employees need to go as

exemplified by firms such as Google. Such firms are

increasingly dependent on normative controls enforced by

a conducive culture, peer and self-accountability and reg-

ulation. It would serve such organizations better if they

consciously invest in reinforcing ‘‘ought to’’ standards

through processes of employee initiation, managerial

coaching and feedback, day to day functioning, and the

leadership’s own demonstrative behaviours. Given the

results in this research, such standards, if well imbibed by

employees would function as self-guides to employees.

Then organizations need to put in place such formal and

informal systems and processes so that employees get

feedback themselves about any discrepancy between their

behaviour and such standards. It is already demonstrated by

some organizations like Google, South-West airlines and

Inmobi (Goyal 2015).

Limitations and Future Research

The pre-test/post-test design has generic limitations in terms

of confidence in inferences we make compared to a more

robust control group—treatment group design. Due to the

design limitations, it is necessary that other conditions which

could have an influence on students’ eating habit did not

change in the pre-test and post-test phases of the study. As

was discussed in earlier sections, the students’ experienced

similar conditions related to mess menu and work/time

schedules, across two phases. However, we cannot control

for validity threats like subject maturation, adaptation and

extraneous factors like change in season. On the other hand,

we believe that unobtrusive field design is the main strength

of this study. Therefore, results of manipulation check are

important which do indicate the intended influence of the

intervention. Second limitation relates to the manipulation

check itself. Due to the limitation of the online survey

method, we could not get a set of individual responses.

However, the cumulative response data on each question

indicated that the treatment has achieved its intended effect.

Despite these limitations, this study provides evidence that

in a field setting; conditions can be created that invoke

individuals’ self-accountability which results in their self-

regulatory behaviour, even though external accountability

conditions might be absent.

In future a more robust control group—treatment group/

pre and post-test research design can be adopted to still

enhance the internal validity. Future research can explore

the influence of gender differences on self-accountability

and self-regulation relations. As was argued, self-account-

ability cognitions are activated when self-discrepancies are

available and accessible. Dependent on which personal self-

guide or standard—ideal or ought to—is relevant and sali-

ent, the awareness of self-discrepancies may lead to dif-

ferential emotions. Comparison with ideal standards and a

resultant discrepancy may lead to dissatisfaction and dis-

appointment seen from one’s own standpoint but may lead

to shame and embarrassment seen from the other’s stand-

point. Comparison with ought standards and the resultant

discrepancy may cause guilt and self-contempt as seen from

one’s own standpoint but may lead to fear as seen from

others standpoint (Higgins 1987). These negative affect

states may be experienced differentially by males and

females in intensity and the motivations to address dis-

crepancies. For example, significant research has been

conducted on comparing guilt and shame as felt by males

and females (e.g. Ferguson and Crawley 1997; Lewis 1971).

One of the important theses and finding of this research has

been that males and females organize information about

self-discrepancies differently, with the former being more

guilt prone and the latter more shame prone (Ferguson and

Crawley 1997; Lewis 1971). Similarly, the defence mech-

anisms to handle these emotions for males and females also

wary with former found to be predominantly isolating affect

or externalizing it and latter mostly internalizing affect by

being self critical. Tagney (1990) reported slightly different

results with females reporting both guilt and shame more

than males. Given these gender related differences to affect,

it would be interesting to compare the self-accountability

cognitions, resultant affect and self-regulatory behaviour

94 A. Dhiman et al.

123

between males and females for ‘‘ideal’’ and ‘‘ought to’’ self

accountabilities.

The presented research could not include personality

variables and future research can address this gap. Self-ac-

countability, related self-discrepancy awareness and conse-

quent self-regulation would be influenced by an individual’s

self-consciousness (Fenigstein et al. 1975). The private self-

consciousness component (as distinct from public) is espe-

cially relevant for presented research due to its focus on

consciousness of one’s inner feelings and thoughts. Trapnell

and Campbell (1999) found that people high on neuroticism

and those high on openness to experience, respectively,

indulged in more rumination and reflection to be more self-

conscious. How does these modes influence self-account-

ability and regulation can be an interesting area to study. Other

relevant traits that can be studied include the locus of control,

and self-monitoring that focus on individual’s awareness and

sensitivity to internal and external cues, and likely adaptive

responses.

Appendix 1: Online Survey Questionnaire

A major endeavour of the outgoing messcom 2006–2007 has

been to reduce food wastage on all counts. One of the sources

identified by committee was the food unconsumed and leftover

by student members in their plates. One of the obvious reason

was lower quality of food on a given day, and uneatables like

peels and chicken bones. But the initial waste figures suggested

wastage much beyond estimates accounted for by these rea-

sons. To understand better as to why this is happening, we

started collecting waste, measuring it, and displaying it in the

mess. We are conducting a short survey related to our study. All

mess members are requested to respond to it online within a

period of next week. It will not take more than 5 min of your

time. It will be your contribution to a noble cause, as we may

leave a small legacy for future batches.

Kindly tick mark in one empty box against each ques-

tion on six-point scale ranging from ‘strongly agree’ to

‘strongly disagree’.

1 Strongly disagree

2 Disagree

3 Somewhat disagree

4 Somewhat agree

5 Agree

6 Strongly agree

1 2 3 4 5 6

After weighing scale was introduced, I keenly noticed scale reading how much I am adding to the waste After weighing scale was introduced, I keenly followed the wastage statistics displayed on white board. I think weighing scale made me more conscious about the wastage

I think I consciously reduced wastage myself.

I think overall mess members reduced wastage because they became conscious of amount of wastage.

Did you notice removal of weighing scale (basket remained)? Yes No

When did you first notice the removal of weighing scale?

Currently I have stopped following the wastage statistics displayed on white board. After weighing scale was removed, I have stopped noticing wastage (in my plate).

I do not think weighing scale can change individual wastage habits.

Effect of Self-Accountability on Self-Regulatory Behaviour: A Quasi-Experiment 95

123

References

Accenture too drops bell curve appraisals. Accessed at economic-

times.indiatimes.com/news/international/business/accenture-too-

drops-bell-curve-appraisals/on The Economic Times Business,

published on 27th July, 2015, at 06:55 AM IST.

Adelberg, S., & Batson, C. D. (1978). Accountability and helping:

when needs exceed resources. Journal of Personality and Social

Psychology, 36(4), 343–350.

Ashton, R. H. (1992). Effects of justification and a mechanical aid on

judgment performance. Organizational Behavior and Human

Decision Processes, 52(3), 292–306.

Baldwin, J. M. (1897). Social and ethical interpretations in mental

development. New York: Macmillan.

Brtek, M. D., & Motowidlo, S. J. (2002). Effects of procedure and

outcome accountability on interview validity. Journal of Applied

Psychology, 87(1), 185–191.

Carver, C. S., & Scheier, M. F. (1982). Control theory: A useful

conceptual framework for personality—social, clinical, and

health psychology. Psychological Bulletin, 92(1), 111–135.

Duval, S., & Wicklund, R. A. (1972). A theory of objective

awareness. New York: Academic Press.

Fenigstein, A., Scheier, M. F., & Buss, A. (1975). Public and private

self-consciousness. Journal of Consulting and Clinical Psychol-

ogy, 43, 522–527.

Ferguson, T. J., & Crawley, S. L. (1997). Gender differences in the

organization of guilt and shame. Sex Roles, 37(1/2), 19–44.

Ferris, G. R., Dulebohn, J. H., Frink, D. D., George-Falvy, J.,

Mitchell, T. R., & Matthews, L. M. (1997). Job and organiza-

tional characteristics, accountability, and employee influence.

Journal of Managerial Issues, 9(1), 162–175.

Finegan, J. (1994). The impact of personal values on judgments of

ethical behaviour in the work place. Journal of Business Ethics,

13(9), 747–755.

Frink, D. D., & Klimoski, R. J. (1998). Towards a theory of

accountability. In G. R. Ferris (Ed.), Research in personnel and

human resource management (Vol. 15, pp. 1–51). Stamford, CT:

JAI Press.

Garvin, D.A., Wagonfeld, A.B., & Kind, L. (2013). Google’s Project

Oxygen: Do Managers Matter? HBSP Case study.

Gordon, R. A., Rozelle, R. M., & Baxter, J. C. (1988). The effect of

applicant age, job level, and accountability on the evaluation of

job applicants. Organizational Behaviours and Human Decision

Processes, 41(1), 20–33.

Goyal, M. (2015). Why Inmobi may be India’s most innovative

company. The Economic Time Magazine August 02–08, 2015,

Bennett–Coleman India.

Gujarati, D. (2003). Basic econometrics. New York: McGraw Hill.

Heider, F. (1958). The psychology of interpersonal relation-

ships. New York: Wiley.

Higgins, E. T. (1987). Self-discrepancy: A theory relating self and

affect. Psychological Review, 94(3), 319–340.

Jain, M. et al. (2013). ‘‘We are not in the loop’’ Resource Wastage and

Conservation attitude of employees in Indian work place.

UbiComp’13, Sept 08–12, 2013, Zurich, Switzerland.

James, W. (1890). The principles of psychology (Vol. 1). New York:

Holt & Co.

Jones, T. M. (1991). Ethical decisions making by individuals in

organizations: An issue contingent model. Academy of Manage-

ment Review, 16(2), 366–395.

Lerner, J. S., & Tetlock, P. E. (1999). Accounting for the effects of

accountability. Psychological Bulletin, 125(2), 255–275.

Lewis, H. B. (1971). Shame and guilt in neurosis. New York:

International Universities Press.

Luthans, F. (2005). Organizational behavior. New York: McGraw

Hill.

Maheshwari, S.K. & Ganesh, M.P. (2004). Ethics in organizations:

The case of Tata Steel. W.P.NO. 2004-03-01, Working paper

series, Indian Institute of Management, Ahmedabad.

Markus, H., & Nurius, P. (1986). Possible selves. American

Psychologist, 41, 954–969.

Markus, H., & Wurf, E. (1987). The dynamic self concept: A social

psychological perspective. Annual Review of Psychology, 38,

299–337.

Mead, G. H. (1934). Mind, self, and society (Vol. 1). Chicago:

University of Chicago Press.

Mero, N. P., & Motowidlo, S. J. (1995). Effect of rater’s account-

ability on accuracy and favourability of performance rating.

Journal of Applied Psychology, 80(4), 517–524.

Oppenheimer, L. (1990). Introduction: Research on self-concept—the

state of art in Eastern and Western Europe. In L. Oppenheimer

(Ed.), The self-concept: European perspectives on its develop-

ment, aspects, and applications (pp. 1–7). Berlin: Springer.

Passyn, K., & Sujan, M. (2006). Self-accountability emotions and fear

appeals: Motivating behavior. Journal of Consumer Research,

32, 583–589.

Peloza, J., White, K., & Shang, J. (2013). Good and guilt-free: The

role of self-accountability in influencing preferences for products

with ethical attributes. Journal of Marketing, 77, 104–119.

Posner, B., Randolph, W., & Schmide, W. (1987). Managerial values

across functions: a source of organizational problems. Group

and Organization Studies, 12(4), 373–385.

Rokeach, M. J. (1973). The nature of human values. New York: The

Free Press.

Schlenker, B. R., & Weigold, M. F. (1989). Self-identification and

accountability. In R. A. Giacalone & P. Rosenfeld (Eds.),

Impression management in the organization (pp. 21–43). Hills-

dale, NJ: Lawrence Erlbaum.

Shadish, W. R., Cook, T. D., & Campbell, D. T. (2002). Experimental

and quasi-experimental designs for generalized casual inference.

NY: Houghton Mifflin Company.

Shrinivasan, Y. B., Jain, M., Seetharam, D. P., et al. (2013). Deep

conservation in urban India and its implications for the design of

conservation technologies. CHI, ACM, 2013, 1969–1978.

Simonson, I., & Nye, P. (1992). The effect of accountability on

susceptibility to decision errors. Organizational Behaviour and

Human Decision Processes, 51(3), 416–446.

Smith, J. D. (2014). Self-concept: Autopoiesis as the basis for a

conceptual framework. Systems Research and Behavioral

Science, 31, 32–46.

Tagney, J. P. (1990). Assessing individual differences in proneness to

guilt and shame: development of self-conscious affect and

attribution inventory. Journal of Personality and Social Psy-

chology, 59(1), 102–111.

Tetlock, P. E. (1985). Accountability: The neglected social context of

judgment and choice. In L. L. Cummings & B. M. Staw (Eds.),

Research in organizational behaviour (Vol. 1, pp. 297–332).

Greenwich: JAI Press.

Tetlock, P. E., & Kim, J. I. (1987). Accountability and judgment

processes in a personality prediction task. Journal of Personality

and Social Psychology, 52(4), 700–709.

Tetlock, P. E., Skitka, L., & Boettger, R. (1989). Social and cognitive

strategies for coping with accountability. Journal of Personality

and Social Psychology, 57(4), 632–640.

Trapnell, P. D., & Campbell, J. D. (1999). Private self-consciousness

and five factor model of personality: Distinguishing rumination

from reflection. Journal of Personality and Social Psychology,

76(2), 284–304.

96 A. Dhiman et al.

123

Trevino, L. K., Butterfield, K. D., & McCabe, D. L. (1998). The

ethical context in organizations: influences on employee atti-

tudes and behaviours. Business Ethics Quarterly, 8(3), 447–476.

Turillo, C. J., Folger, R., Lavelle, J. J., Umphress, E. E., & Gee, J. O.

(2002). Is virtue its own reward? Self-sacrificial decisions for the

sake of fairness. Organizational Behaviour & Human Decision

Processes, 89(1), 839–865.

Vohs, K., & Baumeister, R. F. (2004). Understanding self-regulation.

In R. Baumeister & K. Vohs (Eds.), Handbook of self-

regulation: Research, theory, and application (pp. 1–9). New

York: Guilford Press.

Wiley, R. (1979). The self-concept: Theory and research on selected

topics (Vol. 2). Lincoln: University of Nebraska Press.

Effect of Self-Accountability on Self-Regulatory Behaviour: A Quasi-Experiment 97

123

Journal of Business Ethics is a copyright of Springer, 2018. All Rights Reserved.