Journal of Economic Psychology 28 (2007) 283–313

www.elsevier.com/locate/joep

Does watching TV make us happy?

a,*
Bruno S. Frey , Christine Benesch a, Alois Stutzer b

a
Institute for Empirical Research in Economics, University of Zurich, Winterthurerstrasse 30,
CH-8006 Zurich, Switzerland
b
Department of Economics, University of Basel, Petersgraben 51, CH-4003 Basel, Switzerland

Received 20 January 2006; received in revised form 5 July 2006; accepted 4 August 2006
Available online 14 February 2007

Abstract

Watching TV is a major human activity. Because of its immediate benefits at negligible immediate
marginal costs it is for many people tempting to view TV rather than to pursue more engaging acti-
vities. As a consequence, individuals with incomplete control over, and foresight into, their own
behavior watch more TV than they consider optimal for themselves and their well-being is lower
than what could be achieved. We find that heavy TV viewers, and in particular those with significant
opportunity cost of time, report lower life satisfaction. Long TV hours are also linked to higher
material aspirations and anxiety.
! 2007 Elsevier B.V. All rights reserved.

JEL classification: D12; D91; I31; J22

PsycINFO classification: 3000; 3740; 3900

Keywords: Life satisfaction; Mispredicting utility; Revealed behavior; Self-control problem; TV consumption

1. Introduction

Watching TV is a very important activity, carried out by most people in the majority of
countries. On average, people in Europe spend 226 min watching TV a day, in the United

*
Corresponding author. Tel.: +41 44 634 3730; fax: +41 44 634 3599.
E-mail address: [email protected] (B.S. Frey).

0167-4870/$ - see front matter ! 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.joep.2007.02.001
284 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

States TV viewing, on average, amounts even to 297 min per day (IP Germany, 2005). In
many countries nowadays, watching TV occupies on average almost as much time as work-
ing. As it is a totally voluntary, freely chosen activity, it seems obvious that people enjoy it,
because they would not do it otherwise. They are more satisfied with having the opportu-
nity to watch TV to the extent they do rather than watching less TV or none at all.
This implication is shared by standard neoclassical economic theory. Individuals are
assumed to know best what provides them with utility and are free to choose the amount
of TV consumption that suits them best. By revealed preference, it follows from the fact
that individuals watch so much TV as has been empirically observed that it provides them
with considerable utility.
Recent developments, particularly in behavioral economics, cast doubt on this conclu-
sion. The theory of revealed preference has been questioned (see, for instance, Sen, 1982,
1995): it is, in general, not possible to infer the utility produced by observing behavior,
because individuals do not always act rationally. More concretely, anomalies and biases
in behavior have been identified (e.g. Thaler, 1992), which undermine the direct link
between observed behavior and the utility gained. Individuals may also be subject to hab-
its which they do not have fully under control. They may consume some goods, such as
drugs, alcohol or tobacco to a greater extent than they find to be good for themselves.
They are subject to a self-control problem (e.g. Schelling, 1984), again interfering with
the direct relationship proposed by revealed preference theory. As Gruber and Mullaina-
than (Gruber & Mullainathan, 2005) empirically show, (predicted) smokers, according to
their own evaluation, consider themselves to be better off if smoking was restricted by a
tax. Finally, individuals may systematically mispredict the utility derived from future con-
sumption (e.g. Loewenstein, O’Donoghue, & Rabin, 2003; Loewenstein & Schkade, 1999).
In particular, happiness research (for a survey, see Frey & Stutzer, 2002a, 2002b; Layard,
2005) has empirically shown that individuals overestimate the utility of future income (e.g.
Easterlin, 2001), at the same time as they underestimate the utility of personal interactions
(Frey & Stutzer, 2004). The consumption decisions made by individuals are systematically
distorted according to their own evaluations.
This paper studies the strong notion that TV viewing is a case in which the theory of
revealed preference does not fully apply: many people watch more TV than they consider
good for themselves. The extent of TV viewing is not generally utility maximizing.
The case for mistakes in TV consumption choice can be formulated within a systematic
framework: individuals are subject to a self-control problem, mainly induced by the fact
that watching TV offers immediate benefits (e.g. entertainment and relaxation) at very
low immediate marginal costs. Many costs (e.g. not enough sleep, underinvestment in
social contacts, education or career) are only experienced in the future. Individuals with
time inconsistent preferences are therefore unable to adhere to the amount of TV viewing
they planned or which, in retrospect, they would consider optimal for themselves. This ten-
dency is aggravated when people mispredict future costs because they underestimate utility
from socializing and neglect changes in preference due to TV consumption. Extensive TV
viewing is, according to this alternative view, the result of mispredicting utility and a self-
control problem, leading to a lower level of individual utility than what could be achieved.
It is very difficult to discriminate between the view of optimal TV watching and the view
of over-consumption based on observed behavior. How is it possible to assess from the
outside whether 4 h TV viewing a day are too much and are actually regretted by the con-
sumer? Neither is there conclusive information about optimal consumption behavior in
 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313 285

TV viewers’ reaction to price changes. Whether or not people consistently allocate their
time and income, they will react to relative price changes in the consumption of TV watch-
ing in the direction predicted by standard economics. Only if people adhere to external
commitment devices that limit their future choice sets, their observed behavior indicates
that they are subject to self-control problems. Yet, there exists only anecdotal and no sys-
tematic evidence for the use of such self-commitment devices (e.g. people put an uncom-
fortable chair in front of their TV or cancel their cable subscription in order not to watch
too much TV or even get rid of their TV).
We pursue a completely different approach and propose data on subjective well-being
to study whether people make systematic mistakes in their choice of time devoted to TV
watching. Life satisfaction or reported subjective well-being can serve as a proxy for expe-
rienced utility as suggested by happiness research. Based on this methodology, it is in prin-
ciple possible to study whether higher TV consumption lowers an individual’s utility or
well-being as suggested by the pessimistic view about people’s TV watching. In our anal-
ysis, we explore to what extent standard information on individual TV consumption and
subjective well-being can inform the debate.
The empirical analysis studies data from the first wave of the European Social Survey.
This is an exceptionally rich data set providing information for more than 42,000 people
from 22 different countries for 2002/03. The baseline econometric estimate lends support
to the hypothesis of over-consumption: excessive TV viewers, on average, report lower life
satisfaction. This negative correlation holds even after controlling for a large number of
covariates of individual well-being.
We are aware that this correlation does not imply causation. The causality issue can
neither be resolved with an extensive set of control variables in a multiple regression ana-
lysis nor with panel data. Instead one would want to study large scale changes in people’s
opportunities to watch TV that are set from outside. We are not aware that suitable data,
such as a natural experiment, exist. Instead two other aspects of the interrelation between
TV consumption and subjective well-being are studied: (i) Whether the utility costs of
extensive TV consumption depend on the opportunity cost of time. (ii) Whether TV view-
ing affects people’s preferences and beliefs.
We find that particularly individuals with time constraints, who watch TV for many
hours, report lower life satisfaction and that watching TV is positively related to people’s
material aspirations, as well as to anxiety levels, and negatively related to their financial
satisfaction, to their trust in others, as well as to the perceived relative frequency of social
activities. Section 2 of this paper discusses TV viewing as a major activity in modern life.
Section 3 develops the basic testable hypothesis. The following Section 4 presents the
data and Section 5 gives the results of the basic econometric estimates and integrates them
into the existing literature on TV and happiness. The next section deals with the possibility
of reverse causation and addresses the role of opportunity costs of time and of changes in
preferences and beliefs. Section 7 concludes.

2. TV viewing

Leisure activity today is dominated by television. The reduction in (paid and unpaid)
working hours achieved over the past decades, resulting in more leisure time, has to a
large extent been replaced by watching television. According to time use studies (Robinson
& Godbey, 1999, p. 338–347, see also Aguiar & Hurst, 2007; Bittman, 1998; Gershuny,
286 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

2000; Goodin, Rice, Bittman, & Saunders, 2005), the average leisure time of adult
Americans (19–64 years of age) over the period 1965–1995 rose by 6.2 h from 34.8 to
41 h.1 In the same period of time, TV viewing time rose by 6 h. In 1995, the average Amer-
ican spent 16 h a week, or 2 14 h a day, in front of the TV.
Similar trends can be observed for other industrialized countries.2 Between the 60s and
90s, leisure time in those countries (for adults aged 20–59) rose, on average and controlling
for structural changes, by well over 6–36 h per week (Bittman, 1998).3 At the end of the
90s, according to time use studies, TV viewing time in European countries4 averaged
between almost 2–2 34 h a day, or between 14 and 19 h per week (Aliga & Winqvist,
2003). 20% of the respondents in the European Social Survey 2002/03 indicated that they
watch TV for more than 3 h per day.
Television rating agencies report even higher average viewing times than time use stud-
ies: on average, in 2004, Americans (age 16 and older) watched close to 5 h, and Europeans
(about age 15 and older)5 about 3 34 h a day (IP Germany, 2005). This is due to the fact that
other age groups are represented (especially the retired devote more than average time to
TV watching) and that the electronic measuring instruments (so-called people’s meters)
cannot distinguish between primary, secondary and tertiary activities. It is known that
many people engage in multi-tasking, for instance watching TV at the same time as cook-
ing. But exclusive TV viewing is still a major activity taking up almost 70% of total TV
time (Grahn, Lehmann, Schmitz, & Breinker, 2003; Robinson & Godbey, 1999).

3. Basic hypothesis: Watching too much TV

Two opposite views are possible when qualifying the huge amount of leisure time spent
in front of the TV. At first glance, it might seem obvious that watching TV produces high
individual utility. TV watching is a voluntary activity and people can freely decide how
much time they want to devote to it. However, an alternative view does not take consumer
sovereignty as given but takes the possibility of a systematic error in TV consumption into
account. In fact, studies suggest that TV viewing is only rated as below average or just as
average enjoyment. American women, surprisingly enough, even rate it behind cleaning
(Robinson & Godbey, 1999, p. 250). Moreover, TV is identified as that activity which
would be given up first if another activity would require more time (Robinson & Godbey,
1999, pp. 238–239).
We therefore want to study the second view: individuals tend to watch too much TV in
the sense that they are afterwards sorry that they devoted so much time to viewing. More-
over, it seems difficult to overcome this weakness of will; many people are dissatisfied with

1
Schor (1991) shows that the number of working hours in America has risen, and the amount of leisure time
has fallen. But the data supporting this view has been criticized by authors using more accurate diary-based time
use data, e.g., Robinson and Godbey (1999, pp. 49–53), Kubey and Csikszentmihalyi (1990, p. 22) and Sullivan
and Gershuny (2001).
2
Denmark, The Netherlands, Finland, Italy, Belgium, Sweden, Germany (East and West), Czechoslovakia,
Great Britain, Norway, Poland, France, Yugoslavia, Bulgaria, Hungary, Canada, USA and Australia.
3
Gershuny (2000), for the same time period, shows an overall increase in leisure time without controlling for
shifts in structural variables. The countries investigated are Canada, Denmark, The Netherlands, Norway, UK,
USA, Hungary, and Finland.
4
Denmark, Finland, Belgium, Sweden, Great Britain, Norway, France, Hungary, Estonia, Slovenia.
5
Age categories vary by country.
 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313 287

their own past behavior, but nevertheless again and again devote more time to watching
TV than, according to their own evaluation, is good for them. The basic hypothesis is thus
that TV lends itself to over-consumption.
The main reason is that TV viewing is characterized by immediate benefits and negligi-
ble immediate marginal costs. One just has to push a button. In contrast to going to the
cinema, the theater or any outdoor activity, there is no need to be appropriately dressed
before leaving the house, there is no need to buy a ticket or to reserve a seat in advance.
Watching TV does not require any special physical or cognitive abilities (Kubey &
Csikszentmihalyi, 1990, p. 173). Unlike other leisure activities, TV viewing does not need
to be coordinated with other persons. It is quite possible to sit alone in front of the TV,
while other leisure activities, such as tennis or golf, require a partner with similar time
availability and similar preferences. As a consequence, watching TV has, compared to
other leisure activities, an exceedingly low or nonexistent entry barrier. At the same time,
it offers entertainment value and is considered to be one of the best ways of reducing stress.
Moreover, while watching TV, immediate marginal costs are even lower and having a
remote control is an invitation to ultra short-term optimization (zapping). Many of the
costs resulting from such consumption behavior are not experienced immediately, or
not predicted at all. The negative effects of not enough sleep, for example, only arise
the next day, and the consequences of underinvestment in social contacts, education or
career take much longer to appear. An increase in one’s material aspirations might not
be foreseen at all. These characteristics of the consumption good induce many individuals
to fall prey to excessive TV viewing. Some studies discuss if television can become a habit
(Christakis & Zimmerman, 2006) or even an addiction in a medical or psychological sense
(Kubey & Czikszentmihalyi, 2002; McIlwraith, 1998).
In economic theory, addiction is not necessarily considered to be irrational or suboptimal.
In the model of rational addiction (Becker & Murphy, 1988), addicts maximize their current
and future utility under stable preferences. More recent work questions this rationality
assumption in the case of addictive goods. Addictive behavior has, for example, successfully
been modeled with time-inconsistent preferences (e.g. Gruber & Koszegi, 2001; O’Donog-
hue & Rabin, 1999a, 2002). In these models, individuals, due to their hyperbolic discounting,
put more emphasis on the present as compared to all other periods of time and tend to grab
immediate rewards. Gruber and Mullainathan (2005) test empirically if smokers have time
inconsistent preferences and are therefore subject to self-control problems. They show that
predicted smokers (i.e. people who would smoke with some probability given some tax level)
would be happier if cigarette taxes were higher. This result is inconsistent with models of
rational addiction in which higher prices reduce utility.
Time inconsistent preferences and self-control problems have been confirmed in many
laboratory experiments (for an overview see e.g. Frederick, Loewensteinm, & O’Donog-
hue, 2002), and they have been applied to other areas than addiction.6 Recent empirical
evidence from the field is presented for saving decisions (Angeletos, 2001), food consump-
tion (Cutler, Glaeser, & Shapiro, 2003; Shapiro, 2005), job search (DellaVigna & Paser-
man, 2005), labor supply (Fang & Silverman, 2006) or health club visits (DellaVigna &
Malmendier, 2006).

6
For formal models of time inconsistent preferences, see e.g. Laibson (1997) and O’Donoghue and Rabin
(1999a, 1999b) and references mentioned therein.
288 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

Regarding television consumption, there is some (anecdotal) evidence that individuals

may have self-control problems. 40% of US adults and 70% of US teenagers admit that
they watch too much TV (Kubey & Czikszentmihalyi, 2002). Another interesting observa-
tion is that short and long term evaluations of TV consumption tend to diverge – or, as
Robinson and Godbey (1999, p. 299) put it: ‘‘We may not enjoy television in general,
but the programs we saw last night were pretty good.’’ Some individuals totally abstain
from watching TV because they know that they would not otherwise be able to control
their television viewing behavior. They cancel their subscription for cable TV in order
not to ‘‘zap’’ too much, lock their TV set away in a cupboard or place an uncomfortable
chair in front of it. Such self-control mechanisms are not necessary for time consistent
individuals. To lower the utility, or raise the cost, of an undesired alternative would be
irrelevant and unnecessary.
Here, the role of self-control problems and of mispredicting future costs and benefits in
TV viewing is addressed with regard to consumers’ utility. It is hypothesized that, for peo-
ple facing similar restrictions, heavy TV viewing indicates impeded self-control rather than
a taste for TV. Accordingly, heavy TV consumption is expected to result in lower utility.7
This hypothesis has, of course, to be understood ceteris paribus. Individual utility
depends on a large number of other factors, which have to be taken into account in order
to identify the specific effect of TV on utility. In the empirical analysis, utility is proxied by
life satisfaction, and the effects are partial, controlling for major socio-demographic fac-
tors normally included in a happiness equation (see e.g. Frey & Stutzer, 2002a, 2002b).

4. Data on TV consumption and life satisfaction

In order to empirically address the basic hypothesis on TV over-consumption, the first

wave of the European Social Survey (ESS) is used. The ESS is a survey that was carried
out in 22 European countries in 2002/2003. In each country, about 1200–3000 people were
interviewed. For 338 observations, data for life satisfaction or television viewing is miss-
ing, resulting in a sample of 42,021 observations.
In addition to life satisfaction and television viewing time, the ESS includes a large
number of socio-demographic characteristics. Control variables to be used are household
income (adjusted for comparative price level at current US prices8), gender, age, marital
status, employment status, education, working time, nationality and type of location.
The dependent variable life satisfaction is the response to the question: ‘‘All things con-
sidered, how satisfied are you with your life as a whole nowadays?’’ Answers are given on
an 11-point scale ranging from 0 ‘‘extremely dissatisfied’’ to 10 ‘‘extremely satisfied’’
(Fig. 1). The average life satisfaction amounts to 7.0 (standard deviation 2.7). This average
varies considerably between countries, and ranges from 5.6 in Hungary up to 8.4 in Den-
mark (Fig. 2). It is sometimes doubted that such life satisfaction data are comparable
between nations, as the answers might partly reflect cultural differences. To control for
such unobserved differences, country fixed effects are included in the regression analysis.
Television consumption is also captured by one single question: ‘‘On an average week-
day, how much time do you spend watching television?’’ Answers are given in eight

7
See O’Donoghue and Rabin (1999b) for a discussion of the problem of welfare comparisons for people with
time-inconsistent preferences.
8
Data on comparative price levels are from the World Bank’s World Development Indicators.
 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313 289

30%

25%

20%

15%

10%

5%

0%
0 1 2 3 4 5 6 7 8 9 10
extremely extremely
unsatisfied satisfied

Fig. 1. Reported life satisfaction in 22 European countries 2002/2003. Data source: European Social Survey.

Denmark
Switzerland
Finland
Sweden
Luxembourg
Norway
Netherlands
Austria
Ireland
Belgium
United Kingdom
Spain
Italy
Germany
Slovenia
Israel
France
Czech Republic
Greece
Poland
Portugal
Hungary

4 5 6 7 8 9 10

Fig. 2. Average reported life satisfaction in 22 European countries 2002/2003. Note: Average reported life
satisfaction on a scale from 0 ‘‘extremely unsatisfied’’ to 10 ‘‘extremely satisfied’’. Data source: European Social
Survey.
290 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

categories, ranging from ‘‘no time at all’’ to ‘‘more than 3 h’’. About 3% of respondents do
not watch any television at all, while over 20% spend more than 3 h a day in front of their
TV set (Fig. 3). This percentage varies considerably between countries. While only about
10% of respondents in Switzerland watch more than 3 h TV a day, more than 38% do so in
Greece (Fig. 4).
Subjective time use data are often criticized as being inaccurate or biased. ‘‘Watching
television’’ might not be understood in the same way by all respondents, and they might
not differentiate between television viewing as primary, secondary or even tertiary activity.
Respondents might not even correctly remember all the times they were watching tele-
vision, or they might revert to social norms or images they would like to have of them-
selves. Many also seem to translate the ‘‘average day’’ into ‘‘the average day that you
watch television’’ (Robinson & Godbey, 1999, p. 60). Nevertheless, answers to such ques-
tions seem to be a reliable measure for general television viewing behavior. A comparison
of US data shows that different measurement methods give similar average results. In the
1992 SPPA National Survey with nearly 6000 respondents (in which a single question, sim-
ilar to the one in the European Social Survey, was asked about television viewing), the
average television viewing time of 3 h was close to time use data based on much more
detailed diary surveys (Robinson & Godbey, 1999, p. 152).
Based on the data described, a microeconomic happiness function is specified. The Life
satisfactioni of individual i depends on his or her television consumption TVi and on per-
sonal characteristics Xi, as well as on country specific effects Dl in country l:
Life satisfactioni ¼ b0 þ b1 TVi þ c1 X i þ c2 Dl þ ei
As the extent of television consumption is captured in a categorical variable with an open-
ended category for those spending a lot of time watching TV, TV consumption cannot be
included as a continuous variable. Instead, we include dummy variables in the regression
equation. Those who watch less than half an hour of television a day form the reference
group. The 6 categories for people who watch more than half an hour of TV a day are
combined into 3 categories.

25%

20%

15%

10%

5%

0%
No time at all Less than 0.5 0.5 hour to 1 More than 1 More than More than 2 More than More than 3
hour hour hour, up to 1.5 hours, up hours, up to 2.5 hours, up hours
1.5 hours to 2 hours 2.5 hours to 3 hours

Fig. 3. Reported television consumption in 22 European countries 2002/2003. Data source: European Social
Survey.
 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313 291

Greece
United Kingdom
Czech Republic
Netherlands
Spain
Ireland
Luxembourg
Hungary
France
Italy
Denmark
Israel
Portugal
Belgium
Germany
Poland
Austria
Finland
Norway
Sweden
Slovenia
Switzerland

0% 10% 20% 30% 40%

Fig. 4. Heavy television viewers in 22 European countries 2002/2003. Note: Fraction of respondents with more
than 3 h reported TV viewing on an average weekday. Data source: European Social Survey.

5. Basic estimation results

5.1. Partial correlation between TV viewing and life satisfaction

Table 1 reports the partial correlation between TV consumption and reported life sat-
isfaction. In the first specification, a ordinary least squares estimator is applied. In view of
the categorical nature of the dependent variable, a second specification is added, estimat-
ing an ordered probit. The similarities in the relative size of the coefficients indicate that
the least squares estimator delivers a satisfactory approximation of the partial correlation.
As the least squares results are easier to interpret, they are discussed primarily.
People who watch less than half an hour of TV a day are more satisfied with their life,
ceteris paribus, than people who choose any other level of TV consumption. For those
watching TV for anything between half an hour and 2.5 h, average reported life satisfac-
tion is about 0.l0 points lower than in the reference group of people watching for less than
half an hour. The estimated effect is even larger for people watching for more than 2.5 h a
day. On average, they report a 0.18 points lower life satisfaction than people in the refer-
ence group. All the differences are statistically significantly different from zero, at least at
 292
B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313
Table 1
Television consumption and life satisfaction
Dependent variable: OLS Ordered probit
Life satisfaction Coefficient t-Value Coefficient t-Value
No TV at all #0.110 #1.56 #0.013 #0.36
Less than 0.5 h TV Reference group
0.5–1.5 h TV #0.101* #2.18 #0.058* #2.49
More than 1.5–2.5 h TV #0.101* #2.14 #0.067** #2.86
More than 2.5 h TV #0.183** #3.84 #0.091** #3.82
Working hours 0.005 1.43 0.001 0.86
Working hours, squared #0.00008* #2.33 #0.00003(*) #1.70
ln (household income, PPP) 0.362** 21.36 0.160** 18.89
Age #0.068** #16.72 #0.033** #16.37
Age, squared 0.0007** 16.99 0.0003** 16.76
Male Reference group
Female 0.136** 6.16 0.081** 7.36
Foreigner Reference group
Citizen of country 0.332** 6.04 0.158** 5.74
Married, living with partner Reference group
Married, not living with partner #0.914** #5.64 #0.409** #5.08
Separated, living with partner #0.407* #2.09 #0.212* #2.20
Separated, not living with partner #0.950** #10.17 #0.466** #10.06
Divorced, living with partner #0.337** #4.07 #0.180** #4.37
Divorced, not living with partner #0.605** #12.15 #0.291** #11.77
Widowed, living with partner #0.363* #2.32 #0.172* #2.20
Widowed, not living with partner #0.539** #11.98 #0.264** #11.80
Never been married, living with partner #0.262** #5.48 #0.135** #5.64
Never been married, not living with partner #0.528** #13.99 #0.275** #14.61
No children at home Reference group
Children living at home #0.200** #6.44 #0.103** #6.65
Not completed primary education #0.479** #8.06 #0.223** #7.57
Primary or first stage of basic education Reference group
Lower secondary or second stage of basic education 0.097** 2.59 0.045* 2.44
Upper secondary education 0.191** 5.13 0.075** 4.01
Post-secondary, non-tertiary education 0.333** 6.73 0.130** 5.27
First stage of tertiary education 0.325** 7.32 0.126** 5.70

B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

Second stage of tertiary education 0.400** 7.27 0.141** 5.16
Paid work, employed Reference group
Paid work, self-employed 0.034 0.80 0.023 1.10
Unemployed, looking for a job #1.112** #11.19 #0.485** #9.82
Unemployed, not looking for a job #0.658** #5.68 #0.292** #5.07
Education 0.341** 3.73 0.140** 3.08
Permanently sick or disabled #1.066** #9.99 #0.478** #9.02
Retired 0.024 0.28 0.013 0.30
Community or military service 0.344(*) 1.75 0.164(*) 1.67
Housework, looking after children #0.014 #0.17 0.002 0.04
Big city Reference group
Suburbs #0.020 #0.56 #0.002 #0.11
Town, small city 0.037 1.20 0.025 1.64
Country village 0.151** 4.85 0.080** 5.16
Farm, home in countryside 0.232** 4.62 0.128** 5.09
Country-fixed effects Yes Yes
Constant 5.854** 28.42
Observations 42,021 42,021
R2/pseudo-R2 0.18 0.04

Notes: (1) **Significant on 99% level, *significant on 95% level, (*) significant on 90% level.
(2) Variable for household size and dummy variables for highest income category and for the different countries are not shown. Dummy variables for missing
observations for income, household size, working hours, gender, marital status, children, education, employment status, and citizenship are neither shown.
Data source: European Social Survey, World Development Indicators.

293
294 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

the 95% level. The general finding is thus consistent with the basic hypothesis that exten-
sive TV watching makes people worse off, because it indicates over-consumption due to a
self-control problem and misprediction of future costs.
The partial correlation between TV consumption and life satisfaction is estimated for
the whole population and is thus representing an average effect of TV viewing across peo-
ple. It is most likely that some groups of people suffer higher disutility from extensive TV
consumption than others. In the next section, ex ante hypotheses (in contrast to ex post
rationalization) are formulated about people who are expected to lose the most if they
watch TV extensively.
The partial correlation cannot easily be explained as spurious, simply reflecting some
specific individual characteristics of people who spend a lot of time in front of the TV.
A large set of socio-demographic characteristics that are systematically related with
reported life satisfaction and might as well be with TV consumption is taken into consid-
eration. These characteristics include, e.g., respondents’ age, sex, nationality, marital sta-
tus, household income, level of education, and employment status. The correlation of
these variables with reported life satisfaction is discussed in various literature surveys
(see, e.g. Frey & Stutzer, 2002b). Specifically for the ESS, the determinants of life satisfac-
tion are discussed in Lelkes (2006).
Partial correlations of other factors with life satisfaction allow for the assessment of the
size of the effect of TV consumption in relative terms. For example, the difference in life
satisfaction between those watching more than 2.5 h and those watching less than half
an hour (0.18) is more than one third of the difference in life satisfaction between people
who have never been married and are without a partner and married people (0.53). The
difference is about the same as the one between people having upper secondary education
and those who simply completed primary school or the first stage of basic education (0.19).

5.2. Integration into the literature and discussion

The basic results supplement the existing studies looking at the utility of TV consump-
tion. Two approaches can be distinguished. The first captures the short run or instant
effects by measuring ‘‘Activity Enjoyment Ratings’’. In the context of time use studies,
individuals are asked to rate TV viewing compared to other leisure time activities. In
the United States, in 1985, with a rating of 7.8 on a scale from 0 to 10, it proved to be
valued somewhat higher than the average enjoyment of 7 derived from other activities.
Nevertheless, it ranks lower than most other activities undertaken in leisure time (Robin-
son & Godbey, 1999, p. 243). On the index of positive affect of 900 Texan women con-
structed by Kahneman, Krueger, Schkade, Schwarz, and Stone (2004), TV ranks with
4.2 (on a scale from 0 to 6) roughly in the middle of all activities. With 2.2 h per day, it
is one of the most time consuming activities of these women. With the Experience Sam-
pling Method, participants are randomly asked how they feel at a particular moment in
time, using a beeper or a hand-held computer. On the affect scale, composed of cheerful-
ness, friendliness, happiness, and sociability, TV viewing is located in the lower part of the
scale and can hardly be distinguished from reading, working, hobbies and idling. Eating,
social contacts, sports and sex, on the other hand, are clearly higher ranked. Nevertheless,
individuals have little inclination to do anything else (Kubey & Csikszentmihalyi, 1990).
This short-run evaluation captures the momentary affect, but it is difficult to determine
the utility individuals would have derived had they done something else.
 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313 295

In a second approach, long-run aspects of the utility derived from TV consumption are
captured. In a survey on the general satisfaction with TV undertaken in the United States
in 1975, TV was given an average rating of 5.9 points on an enjoyment scale ranging from
0 to 10. It ranks considerably behind most other leisure time activities and below the aver-
age of 6.8 of all rated activities. In 1995, TV viewing with 4.8 points ranked even lower
when compared to all other leisure time pursuits (Robinson & Godbey, 1999, pp. 243,
250). In these surveys, TV viewing is generally rated much lower than in surveys capturing
instant utility hinting at possible time-inconsistent preferences. Yet, such surveys are faced
with the problem that watching TV is associated with a low (‘‘couch potato’’) image, and
there is a general consensus that many programs are stupid. For that reason, the answers
given may reflect what is taken to be socially desirable. It should be noted that surveys on
general life satisfaction (as used in our study) are not affected by this bias.
Several studies relate TV viewing with global measures of subjective well-being and are
thus closest to our design. In a study of roughly 3000 Americans in 1979 (Morgan, 1984),
people watching a lot of TV considered their life to be more ‘‘lousy’’ on an index consisting
of the aspects lonely, boring, depressing, unsatisfying, uneventful and unhappy, and less
‘‘great’’ on an index consisting of the aspects interesting, active, meaningful, fun, fulfilling,
stimulating and exciting, compared with people watching less TV. In a random survey of
1000 West Germans, there is also a negative correlation between the duration of TV view-
ing and general life satisfaction, controlling for size of household, education and age (Espe
& Seiwert, 1987). TV viewing has also been found to have a negative effect on life satisfac-
tion due to reducing time spent in relational activities (Bruni & Stanca, 2007; for a broader
argument see also Corneo, 2005; Putnam, 1995).
So far, the negative partial correlation between TV consumption and subjective well-
being has been interpreted in terms of over-consumption leading to a lower utility level.
However, the partial correlation could well be the result of reverse causation. It is quite
plausible that unhappy people watch more TV than happy ones. In fact, Espe and Seiwert
(1987) postulate a causal influence of dissatisfaction with life on TV consumption, but
offer no corresponding evidence. Controlling for as many situational factors as possible
in the regression equation can attenuate the problem. However, it cannot be resolved, nei-
ther with an extensive set of control variables in a multiple regression analysis nor with
panel data.9 Ideally, one would need information about exogenous changes in the oppor-
tunities for TV consumption, e.g. due to satellite TV being shut down or due to TV being
introduced in a new place because of technical innovation. We are not aware of any such
event that could be connected to survey data on reported subjective well-being.
We propose a different approach for further analysis of the effect of TV consumption on
subjective well-being and for shedding light on the issue of causality. First, additional
hypotheses are formulated, exploiting the heterogeneity in the expected effect of TV
over-consumption for different groups of people. Second, evidence is collected that exten-
sive TV consumption is related to systematically different preferences and beliefs. This
would support the view that there are long-term costs of TV consumption that are very
difficult to foresee.

9
Panel data allow for the control of unobserved time-invariant individual specific factors that affect reported
subjective well-being, as well as the explanatory variable of interest. This is not helpful in our analysis, because the
entire theory is based on some people suffering from an unobserved time-invariant self-control problem.
296 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

6. The role of opportunity costs of time and changes in preferences and beliefs

This section presents two extensions of the basic analysis on the statistical relationship
between TV viewing and life satisfaction: (a) Opportunity costs of time are taken into account
to assess the reduction in well-being due to self-control problems. (b) A path analysis is con-
ducted to get an idea of the different ways that TV consumption can affect life satisfaction.

6.1. Opportunity costs of time and the effect of watching TV

So far, it has been assumed that self-control problems in TV consumption affect every-
body alike. Additional tests of the hypothesis that self-control problems in TV consumption
reduce people’s utility are possible if different types of individuals can be identified who suf-
fer to a different extent from over-consumption. We emphasize that in particular individuals
with high opportunity costs of time can use time more profitably when not watching TV. This
includes, for instance, the self-employed (e.g. craftsmen, lawyers, architects or artists) or per-
sons in high positions (e.g. managers, top bureaucrats or politicians), who can freely transfer
time from leisure to work. For this group of individuals, the self-control problem of watch-
ing too much TV generates considerable costs. Their utility is lower due to their inability to
fully control themselves. In contrast, individuals with low opportunity costs of time suffer lit-
tle, if any, disutility when they fail to watch the amount of TV they would consider optimal
for themselves. Accordingly, it is hypothesized that TV consumption significantly lowers the
life satisfaction of individuals with high opportunity costs of time, while it has a smaller neg-
ative effect on the life satisfaction of individuals with low opportunity costs of time.
Opportunity costs of time cannot be measured directly in our data. Therefore we use
different indicators to distinguish between individuals with high and low opportunity costs
of time:

(i) People who can freely transfer time between work and leisure tend to have higher (mon-
etary) opportunity costs of time compared to people with fixed working hours. Hence,
respondents are assigned to the two groups according to the flexibility of their working
hours. Answers to the question ‘‘[P]lease say how much the management at your work
allows you to be flexible in your working hours?’’ are given on an 11-point scale ranging
from 0 ‘‘I have no influence’’ to 10 ‘‘I have complete control’’. Respondents who
indicate a value between 0 and 5 constitute the group with low opportunity costs of
time, while those who give an answer between 6 and 10 form the other group. Only
individuals who are employed (at least part time) are included in these sub-samples.
(ii) As a second indicator, employment status and profession are used. Retirees and the
unemployed tend to have lots of free time and therefore form the group with low
opportunity costs of time. On the other hand, working people, especially those
who are self-employed and those in high positions and professions (legislators, senior
officials, managers and professionals according to ISCO-88 classification) are
assigned to the group with high opportunity costs of time.

Table 2 reports summary statistics for the different groups, as well as for the whole pop-
ulation. The groups divided according to flexibility of working hours do not differ much in
most socio-demographic characteristics. The group with high opportunity costs of time
watches a bit less TV (e.g. 20% of respondents watch more than 2.5 h TV a day compared
Table 2
Summary statistics
Mean values Whole Flexibility of working hours as distinction criteria Employment status/profession as distinction criteria
population
Group with high Group with low Group with high Group with low

B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

opportunity costs opportunity costs opportunity costs opportunity costs
of time of time of time of time
Life satisfaction 7.03 7.51 6.94 7.28 6.73
No TV at all 3.25% 3.00% 2.79% 3.73% 2.72%
Less than 0.5 h TV 5.61% 6.86% 6.20% 8.66% 2.82%
0.5–1.5 h TV 29.21% 38.59% 32.98% 40.71% 18.35%
More than 1.5–2.5 h TV 29.34% 31.47% 32.44% 27.90% 26.95%
More than 2.5 h TV 32.59% 20.08% 25.59% 19.01% 49.15%
Working hours 18.87 39.20 38.35 41.03 0
Household income 1956.56 2919.68 2065.86 2879.43 1313.70
(EUR, PPP)
Age 46.18 40.96 39.44 43.50 64.32
Not completed primary 3.91% 0.43% 1.19% 1.09% 8.76%
education
Primary or first stage of basic 13.27% 4.66% 8.09% 7.08% 22.44%
education
Lower secondary or second 22.68% 15.96% 22.34% 12.50% 24.86%
stage of basic education
Upper secondary education 32.48% 35.45% 37.75% 24.29% 27.03%
Post-secondary, non-tertiary 8.30% 10.88% 10.24% 9.53% 5.14%
education
First stage of tertiary 12.92% 21.30% 13.21% 28.92% 8.08%
education
Second stage of tertiary 5.72% 11.01% 6.68% 16.22% 3.26%
education
Observations 42,021 6460 7062 5950 8974
Data source: European Social Survey, World Development Indicators.

297
298 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

to 26% in the other group) and has a somewhat higher income, as well as a higher level of
education. People in this group report, on average, half a point higher life satisfaction
(7.51 compared to 6.94) than people in the other group. The groups divided according
to employment status and profession differ more with regard to their socio-demographic
characteristics. The group with high opportunity costs of time includes more young people
and more males, who watch considerably less TV (e.g. only 19% of respondents watch
more than 2.5 h TV a day compared to 49% in the other group) and have a much better
education (see also Table A1).
Table 3 reports the results of linear regression estimates for the different groups accord-
ing to the different criteria. In order not to overload the table, the regression coefficients
for the control variables are not explicitly shown (they are presented in Table A2).
Columns 1 and 2 show the estimation results for sub-samples with high and low oppor-
tunity costs of time (according to flexibility of working hours). Individuals in the group
with high opportunity costs of time, who watch more television than the reference group,
report lower life satisfaction ceteris paribus. The effects are quite considerable. The subjec-
tive well-being of viewers who watch half an hour and more television a day is between
0.33 and 0.38 points lower than that of light viewers spending less than half an hour
watching television a day. The effects are statistically significant at the 99% level. The mag-
nitude of the coefficients corresponds to almost two third of the difference in life satisfac-
tion between people who are divorced and without a partner and married people (0.61),
and amounts in the highest category to over one third of the effect of unemployment
(#1.11) (both estimates are taken from the full sample). In contrast, for people with
low opportunity costs of time, the coefficients of all television viewing categories are smal-
ler (between 0.04 and 0.14) and not statistically significant.
A similar picture emerges when comparing the self-employed, managers, senior officials,
legislators and professionals with retirees and unemployed persons (columns 3 and 4 in
Table 3). Coefficients in the high opportunity costs of time sub-sample are considerable
in size. The average life satisfaction of people watching more than 1.5 h TV a day is between
0.23 and 0.39 points lower compared to people who watch less than half an hour a day.
These effects are statistically significant at the 95–99% level. In contrast, for the group with
low opportunity costs of time, no correlation between television consumption and reported
life satisfaction is visible. The coefficients are not statistically significant and very small. The
coefficient for watching 0.5–1.5 h TV is not statistically significant in either group.10
Of course, the question still arises whether the negative correlation for people with high
opportunity costs of time is a causal relationship and, if so, in which direction the causality
goes. However, it is difficult to understand why dissatisfied people, who have high oppor-
tunity costs of time, resort to TV viewing, while dissatisfied people with low opportunity
costs of time do not.

6.2. TV viewing and differences in preferences and beliefs

When people make decisions about watching TV, they are expected (and assumed to be
able) to assess and to adequately take into consideration for themselves the long-term costs
of TV viewing. Recent research on the prediction of future utility challenges the assumption

10
Qualitatively the same results are obtained for both groups if the samples are pooled and interaction terms
with the proxy for cost of time are estimated.
Table 3
Television consumption and life satisfaction: opportunity costs of time
Dependent variable: Flexibility of working hours as distinction criteria Employment status/profession as distinction criteria
Life satisfaction Group with high opportunity Group with low opportunity Group with high opportunity Group with low opportunity
costs of time costs of time costs of time costs of time
No TV at all #0.355* 0.056 #0.238 #0.251

B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

(#2.33) (0.33) (#1.58) (#1.24)
Less than 0.5 h TV Reference group
0.5–1.5 h TV #0.328** #0.072 #0.074 #0.105
(#3.59) (#0.69) (#0.80) (#0.70)
More than 1.5–2.5 h TV #0.339** #0.041 #0.229* 0.052
(#3.62) (#0.38) (#2.38) (0.35)
More than 2.5 h TV #0.377** #0.140 #0.389** #0.012
(#3.78) (#1.28) (#3.76) (#0.08)
Socio-demographic Yes Yes Yes Yes
characteristics
Country-fixed effects Yes Yes Yes Yes
Constant 6.203** 5.410** 6.204** 6.919**
(12.39) (10.45) (10.94) (11.93)
Observations 6460 7062 5950 8974
R2 0.14 0.15 0.16 0.22
Notes: (1) Group 1 contains respondents with flexibility of working hours of 6 and higher on a scale from 0 to 10 and group 2 contains respondents with flexibility of
working hours of 5 and lower. In both groups, respondents’ employment status is ‘‘paid work, employed’’ and nothing else. Group 3 contains the self-employed, as
well as managers, senior officials, legislators, professionals (according to ISCO-88 classification) with employment status ‘‘paid work’’ (employed or self-employed)
and nothing else, and group 4 contains retirees and the unemployed with no other employment status.
(2) Ordinary least squares estimations.
(3) **Significant at 99% level, *significant at 95% level, (*) significant at 90% level.
(4) t-values in brackets.
(5) Variables for working hours, household income, age, gender, citizenship, marital status, children, education, employment status, area of living and household size,
and dummy variables for highest income category and for the different countries are not shown. Dummy variables for missing observations for income, household
size, working hours, gender, marital status, children, education, employment status, and citizenship are neither shown.
Data source: European Social Survey, World Development Indicators.

299
300 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

underlying the revealed behavior approach of human well-being (for a survey see Wilson &
Gilbert, 2003). People systematically underestimate that their preferences change due to
processes of adaptation (Loewenstein et al., 2003). Moreover, misprediction of utility is
asymmetric, whereby the positive affects of a high material standard of living are overesti-
mated and the positive affects of activities with strong intrinsic attributes, like socializing,
are underestimated (Frey & Stutzer, 2004). These aspects of mispredicting utility are
directly relevant for TV consumption choice. People are expected to watch too much TV
if they underestimate the future costs of TV viewing, due to neglecting social contacts
and rising material aspirations. We are aware that it is not possible to discriminate between
limited self-control with perfect and with imperfect projection of future beliefs in the cur-
rent empirical framework. However, we are convinced that it is important to take mispre-
diction of utility into account when refining the analysis of TV consumption.
Previous research has looked at the differences in beliefs and preferences between heavy
and light TV viewers, induced by the fact that life portrayed on TV differs systematically
from real life. Television programs contain much more violence and chaotic relationships
and show many more affluent people and more luxury than exist in real life (e.g. Lichter,
Lichter, & Rothman, 1994). People who spend a lot of time watching TV therefore tend
to overestimate crime rates, to show more anxiety (Gerbner, Gross, Morgan, Signorielli,
& Shanahan, 2002) and less trust in others (e.g. Gerbner, Gross, Morgan, & Signorielli,
1980; Signorielli, Gerbner, & Morgan, 1995). They overestimate the affluence of others
(O’Guinn & Shrum, 1997), report higher material aspirations (e.g. Bruni & Stanca, 2006;
Richins, 1987; Sirgy et al., 1998; Shrum, Burroughs, & Rindfleisch, 2005) and rate their
own relative income lower which is related to lower subjective well-being (Layard, 2005).
We perform a path analysis to shed some light on the long-term consequences of TV
consumption and how they correlate with subjective well-being. An analysis is made as
to whether people who watch more TV report lower financial satisfaction (keeping house-
hold income constant) and tend to believe that it is important to be rich, whether they feel
less safe or trust others less and whether they think that they participate less in social activ-
ities. Financial satisfaction is captured with the question ‘‘[. . .] how do you feel about your
household’s income nowadays?’’ Answers are given on a scale from 1 ‘‘living comfortably
on present income’’ to 4 ‘‘very difficult on present income’’, which is reverted for the
empirical analysis. Respondents indicate on a scale from 1 to 6 how important it is for
them to be rich. They are asked how safe they feel walking in the local area after dark.
Answers range from 1 ‘‘very safe’’ to 4 ‘‘very unsafe’’, and this scale is reverted again in
order to let higher values reflect a better feeling of safety. As a proxy for trust serves
the answers to the question ‘‘Generally speaking, would you say that most people can
be trusted or you can’t be too careful in dealing with people?’’ Answers range on an 11-
point scale from 0 ‘‘you can’t be too careful’’ to 10 ‘‘most people can be trusted’’. Finally,
respondents are asked: ‘‘Compared to other people of your age, how often would you say
that you take part in social activities?’’ Answers range from 1 ‘‘much less than most’’ to 5
‘‘much more than most’’.
Table 4 presents the ordinary least square estimates11 of the partial correlations
between TV viewing and the different measures capturing people’s beliefs and preferences.

11
In view of the categorical nature of the dependent variables, ordered probit estimates would be more accurate.
However, the results of such estimates differ very little from the OLS estimates. As the latter are easier to
interpret, they are presented here. The ordered probit estimates can be obtained from the authors on request.
Table 4
Television consumption and people’s beliefs and preferences
Dependent variable (I) Financial (II) Importance (III) Feeling of (IV) Trust in (V) Perceived social

B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

satisfaction to be rich safety people activities
No TV at all #0.050* 0.022 0.012 #0.081 #0.145**
(#2.18) (0.48) (0.46) (#1.03) (#4.58)
Less than 0.5 h TV Reference group
0.5–1.5 h TV 0.005 0.092* #0.060** #0.040 #0.016
(0.34) (3.10) (#3.65) (#0.79) (#0.74)
More than 1.5–2.5 h TV 0.004 0.163** #0.061** #0.044 #0.010
(0.27) (5.44) (#3.69) (#0.84) (#0.49)
More than 2.5 h TV #0.037* 0.218** #0.092** #0.164** #0.106**
(#2.35) (7.16) (#5.49) (#3.11) (#4.93)
Socio-demographic Yes Yes Yes Yes Yes
characteristics
Country-fixed effects Yes Yes Yes Yes Yes
Observations 41,520 37,360 41,644 41,865 41,247
R2 0.36 0.12 0.20 0.17 0.07
Notes: (1) Dependent variables: Financial satisfaction on scale from 0 to 3, importance to be rich on scale from 0 to 5, feeling of safety on scale from 0 to 3, trust in
people on scale from 0 to 10, perceived relative frequency of social activities on scale from 0 to 4.
(2) Ordinary least squares estimations.
(3) **Significant on 99% level, *significant on 95% level, (*) significant on 90% level.
(4) t-values in brackets.
(5) Variables for working hours, household income, age, gender, citizenship, marital status, children, education, employment status, area of living and household size,
and dummy variables for highest income category and for the different countries are not shown. Dummy variables for missing observations for income, household
size, working hours, gender, marital status, children, education, employment status, and citizenship are neither shown.
Data source: European Social Survey, World Development Indicators.

301
302 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

In all five estimations, the coefficients for watching more than 2.5 h TV show the expected
signs and are statistically significant at the 95–99% level: Heavy TV viewers report lower
satisfaction with their financial situation, place more importance on affluence, feel less
safe, trust other people less and think that they are involved in less social activities than
their peers. The effects are sizeable and (in relation to the respective scale of the dependent
variable) are the largest for ‘importance of being rich’ and ‘feeling of safety’. For interme-
diate levels of TV consumption, there are also positive effects on the importance of being
rich and negative effects on the feeling of safety (statistically significant at the 95–99%
level). There is no statistically significant correlation between intermediate levels of TV
consumption and financial satisfaction, trust and perceived relative frequency of social
activities.12
In the next step, an analysis is made of whether those preferences and beliefs influenced
by TV affect reported life satisfaction. Accordingly, the five variables are included in the
life satisfaction-television equation. For simplicity, least squares estimates are presented,
and the intervening variables are included cardinally.13 Column I presents the results with-
out including any intervening variable (corresponding to the results in Table 1). In column
II, all five intervening variables are included in the regression. The partial correlations with
life satisfaction of all of the variables have the expected signs and are statistically signifi-
cant at the 99% level. The coefficient for people watching more then 2.5 h TV is halved and
drops from #0.18 in the regression without any intervening variable in column I to #0.09.
The coefficients for TV viewing between 0.5 and 2.5 h are also decreased. Including the
intervening variables one by one into the regression equations (columns III to VII in Table
5) shows that the indirect effect of TV consumption on life satisfaction is smallest for
‘importance of being rich’ and is of about the same size for the other four variables (look-
ing at the differences in coefficients for more than 2.5 h TV viewing). According to the
Sobel’s test,14 differences in coefficients between regression with and without intervening
variables are all statistically significant for the category of heavy TV consumption.
The results of the path analysis show that the negative relationship between TV con-
sumption and life satisfaction can partially be explained by differences in beliefs and pref-
erences of people watching more TV. This finding corroborates the hypothesis that there
are long-term consequences – or negative internalities – of TV consumption. If these con-
sequences are not completely foreseen, people overestimate the utility from TV consump-
tion and end up at a lower utility level. We are aware that the path analysis does not
exclude reverse causation. However, it supports a richer picture of psychological processes
involved in people’s demand for TV consumption that might help to understand any sys-
tematic errors in TV consumption choice.

12
Absolute frequency of social contacts does not depend statistically significantly on the extent of watching TV.
13
The respective estimates, with the intervening variables included ordinally, do not differ qualitatively from the
ones presented here. They can be obtained from the authors on request.
14
The Sobel’s test shows the statistical significance of the indirect path from an explanatory variable (TV
consumption) over the intervening variable to the dependent variable (life satisfaction). For OLS, the multiplied
coefficients of this path correspond to the difference in coefficients of the explanatory variable in the regressions
with and without intervening variable (MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002; MacKinnon,
Warsi, & Dwyer, 1995).
Table 5

B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

Television consumption and life satisfaction: intermediate processes
Dependent variable: (I) Excluding (II) Including (III) Including (IV) Including (V) Including (VI) Including (VII) Including
Life satisfaction channels all channels financial satisfaction importance to be rich feeling of safety trust in people social activities
No TV at all #0.110 #0.036 #0.075 #0.109 #0.111 #0.096 #0.066
(#1.56) (#0.53) (#1.10) (#1.56) (#1.59) (#1.38) (#0.95)
Less than 0.5 h TV Reference group
0.5–1.5 h TV #0.101* #0.084(*) #0.106* #0.100* #0.084(*) #0.095* #0.096*
(#2.18) (#1.90) (#2.36) (#2.14) (#1.83) (#2.07) (#2.09)
More than 1.5–2.5 h #0.101* #0.080(*) #0.104* #0.096* #0.083(*) #0.093* #0.097*
TV (#2.14) (#1.81) (#2.29) (#2.06) (#1.78) (#2.03) (#2.09)
More than 2.5 h TV #0.183** #0.090* #0.155** #0.177** #0.156** #0.156** #0.151**
(#3.84) (#1.99) (#3.37) (#3.71) (#3.29) (#3.34) (#3.20)
Channels:
Financial satisfaction 0.678** 0.759**
(47.31) (52.05)
Importance to be rich #0.038** #0.040**
(#4.74) (#4.74)
Feeling of safety 0.180** 0.298**
(13.46) (21.57)
Trust in people 0.129** 0.163**
(30.19) (37.46)
(continued on next page)

303
 304
Table 5 (continued)
Dependent variable: (I) Excluding (II) Including (III) Including (IV) Including (V) Including (VI) Including (VII) Including

B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

Life satisfaction channels all channels financial satisfaction importance to be rich feeling of safety trust in people social activities
Social activities 0.232** 0.307**
(21.97) (28.07)
Socio-demographic Yes Yes Yes Yes Yes Yes Yes
characteristics
Country-fixed effects Yes Yes Yes Yes Yes Yes Yes
Observations 42,021
R2 0.18 0.26 0.22 0.18 0.19 0.20 0.20
t-test Sobel:
0.5–1.5 h TV 0.34 #2.65* #3.60** #0.78 #0.74
More than 1.5, up to 0.27 #3.72** #3.64** #0.84 #0.49
2.5 h TV
More than 2.5 h TV #2.35* #4.15** #5.32** #3.10** #4.86**
Notes: (1) Channels: Financial satisfaction on scale from 0 to 3, importance to be rich on scale from 0 to 5, feeling of safety on scale from 0 to 3, trust in people on
scale from 0 to 10, perceived relative frequency of social activities on scale from 0 to 4.
(2) Ordinary least squares estimations.
(3) **Significant on 99% level, *significant on 95% level, (*) significant on 90% level.
(4) t-values in brackets.
(5) Variables for working hours, household income, age, gender, citizenship, marital status, children, education, employment status, location and household size, and
dummy variables for highest income category and for the different countries are not shown. Dummy variables for missing observations for financial satisfaction,
importance of being rich, feeling of safety, trust in people, social activities, income, household size, working hours, gender, marital status, children, education,
employment status, and citizenship are neither shown.
(6) Sobel’s t-test for the difference in coefficients with and without intervening variable.
Data source: European Social Survey, World Development Indicators.
 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313 305

7. Concluding remarks

Hardly anybody would deny that watching TV provides pleasure, at least part of the
time, and that TV programs create focal points for personal discussions. However, many
people report that they would like to spend less time watching TV. Observed consumption
behavior might thus be a weak indicator for individuals’ pleasure from TV viewing.
This paper addresses the issue that long hours of TV viewing may indicate imperfect self-
control, as well as misprediction of the long-term costs of TV consumption, reducing indi-
viduals’ well-being. Specifically, people with significant opportunity costs of time are
expected to regret the amount of their own TV viewing. These are primarily persons with
flexible working hours, who can freely transfer time between leisure and work. People
with low opportunity costs of time, such as retired or unemployed people, or individuals
with fixed working hours, are expected to be little burdened by their weak willpower, and
therefore experience no significant utility loss, even if they spend many hours in front of
the TV. In our empirical analysis, we apply data on subjective well-being as a methodolog-
ical tool to directly evaluate whether some behavior is positively or negatively related to indi-
vidual well-being, ceteris paribus. We find that people who spend a lot of time watching TV
report, on average, lower life satisfaction, ceteris paribus. This negative effect is much larger
for people with high opportunity costs of time than for those with low opportunity costs of
time.
In a path analysis, some light is shed on the long-term consequences of TV consumption.
We find that heavy TV viewers report lower satisfaction with their financial situation, place
more importance on affluence, feel less safe, trust others less and think that they are involved
less in social activities than their peers. The effects themselves can explain about half of the
negative correlation between TV consumption and life satisfaction. The observed correla-
tions are strong and do not disappear once a large set of individual characteristics are con-
trolled for. Moreover, there is systematic structure in the data suggesting that in a major
human activity in modern life individuals have systematically imperfect foresight and con-
trol over their own behavior. The utility gained is lower than what could be achieved. This
shortcoming in human decision-making is reflected in efforts to reduce this utility loss. Time-
constrained individuals resort to all kinds of rules designed to restrict their TV viewing. For
instance, they make it a rule to only watch the news, place uncomfortable chairs in front of
the TV set, locate the TV set in an unattractive room, or even decide not to have a TV set at
all. Despite their efforts, our empirical results suggest that some of the individuals in ques-
tion are unable to fully compensate for their self-control problem.

Acknowledgements

We are grateful for helpful remarks from Matthias Benz, Giacomo Corneo, Carol Gra-
ham, Jonathan Gruber, Silke Humbert, Simon Lüchinger, Susanne Neckermann, Martha
Starr, Fred Zimmerman and the co-editor Daniel Read.

Appendix A

See Tables A1 and A2.

 306
B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313
Table A1
Summary statistics
Mean values Whole Flexibility of working hours as distinction criteria Employment status/profession as distinction
population criteria
Group with high Group with low Group with high Group with low
opportunity costs opportunity costs opportunity costs opportunity costs
of time of time of time of time
Life satisfaction 7.03 7.51 6.94 7.28 6.73
No TV at all 3.25% 3.00% 2.79% 3.73% 2.72%
Less than 0.5 h TV 5.61% 6.86% 6.20% 8.66% 2.82%
0.5–1.5 h TV 29.21% 38.59% 32.98% 40.71% 18.35%
More than 1.5–2.5 h TV 29.34% 31.47% 32.44% 27.90% 26.96%
More than 2.5 h TV 32.59% 20.08% 25.59% 19.01% 49.15%
Working hours 18.87 39.20 38.35 41.03 0
Household income (EUR, PPP) 1956.53 2919.68 2065.86 2879.43 1313.70
Age 46.18 40.96 39.44 43.50 64.32
Male 47.37% 58.22% 55.10% 65.04% 50.91%
Female 52.51% 41.75% 44.80% 34.89% 49.03%
Foreigner 2.76% 2.43% 4.66% 3.64% 2.36%
Citizen of country 96.00% 95.63% 94.21% 96.27% 97.47%
Married, living with partner 53.89% 58.28% 56.23% 63.75% 52.51%
Married, not living with partner 0.39% 0.42% 0.40% 0.39% 0.59%
Separated, living with partner 0.27% 0.43% 0.31% 0.45% 0.13%
Separated, not living with partner 1.22% 1.24% 1.36% 1.31% 1.01%
Divorced, living with partner 1.53% 2.31% 1.90% 2.13% 1.17%
Divorced, not living with partner 5.00% 5.63% 5.69% 4.54% 5.64%
Widowed, living with partner 0.42% 0.14% 0.08% 0.10% 1.04%
Widowed, not living with partner 8.58% 1.33% 1.53% 1.85% 24.41%
Never been married, living with partner 5.94% 10.31% 8.51% 7.55% 1.48%
Never been married, not living with partner 22.28% 19.50% 23.45% 17.48% 11.72%
Children living at home 40.07% 49.94% 49.25% 51.82% 20.60%
Not completed primary education 3.91% 0.43% 1.19% 1.09% 8.76%
Primary or first stage of basic education 13.27% 4.66% 8.09% 7.08% 22.44%
Lower secondary or second stage 22.68% 15.96% 22.34% 12.50% 24.86%
of basic education
Upper secondary education 32.48% 35.45% 37.75% 24.29% 27.03%

B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

Post-secondary, non-tertiary education 8.30% 10.88% 10.24% 9.53% 5.14%
First stage of tertiary education 12.92% 21.30% 13.21% 28.92% 8.08%
Second stage of tertiary education 5.72% 11.01% 6.68% 16.22% 3.26%
Paid work, employed 40.47% 100.00% 100.00% 50.99% 0.00%
Paid work, self-employed 8.18% 0.00% 0.00% 49.01% 0.00%
Unemployed, looking for a job 2.90% 0.00% 0.00% 0.00% 11.52%
Unemployed, not looking for a job 1.43% 0.00% 0.00% 0.00% 5.95%
Education 8.16% 0.00% 0.00% 0.00% 0.00%
Permanently sick or disabled 2.02% 0.00% 0.00% 0.00% 0.00%
Retired 20.12% 0.00% 0.00% 0.00% 82.53%
Community or military service 0.32% 0.00% 0.00% 0.00% 0.00%
Housework, looking after children 11.76% 0.00% 0.00% 0.00% 0.00%
Big city 18.51% 17.35% 19.54% 20.30% 18.25%
Suburbs 14.36% 18.68% 13.04% 15.48% 13.62%
Town, small city 29.20% 28.31% 31.56% 26.49% 30.30%
Country village 31.04% 28.95% 29.89% 28.50% 31.28%
Farm, home in countryside 6.43% 6.42% 5.54% 8.99% 6.21%
Observations 42,021 6460 7062 5950 8974
Data source: European Social Survey, World Development Indicators.

307
 308
Table A2
Television consumption and life satisfaction: opportunity costs of time

B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

Dependent variable: Flexibility of working hours as distinction criteria Employment status/profession as distinction criteria
Life satisfaction Group with high Group with low Group with high Group with low
opportunity costs opportunity costs opportunity costs opportunity costs
of time of time of time of time
No TV at all #0.355* 0.056 #0.238 #0.251
(#2.33) (0.33) (#1.58) (#1.24)
Less than 0.5 h TV Reference group
0.5–1.5 h TV #0.328** #0.072 #0.074 #0.105
(#3.59) (#0.69) (#0.80) (#0.70)
More than 1.5–2.5 h TV #0.339** #0.041 #0.229* 0.052
(#3.62) (#0.38) (#2.38) (0.35)
More than 2.5 h TV #0.377** #0.140 #0.389** #0.012
(#3.78) (#1.28) (#3.76) (#0.08)
Working hours 0.004 0.008 0.014*
(0.72) (1.17) (2.46)
Working hours, squared #0.00005 #0.0002* #0.0002**
(#0.81) (#1.94) (#3.26)
Household income (EUR, PPP) 0.326** 0.363** 0.385** 0.339**
(7.69) (8.27) (8.77) (8.52)
Age #0.062** #0.067** #0.083** #0.092**
(#4.45) (#4.33) (#5.22) (#8.27)
Age, squared 0.0007** 0.0007** 0.0008** 0.0008**
(4.14) (4.00) (4.73) (8.90)
Male Reference group
Female 0.034 0.190** 0.154** 0.130*
(0.71) (3.71) (2.87) (2.55)
Foreigner Reference group
Citizen of country 0.320** 0.241* 0.227(*) 0.275(*)
(2.71) (2.15) (1.69) (1.71)
Married, living with partner Reference group
Married, not living with partner #0.971** #0.574 #0.314 #1.112**
(#2.84) (#1.50) (#0.80) (#3.26)
Separated, living with partner #0.110 #0.500 #0.274 0.155
(#0.33) (#1.17) (#0.76) (0.24)
Separated, not living with partner #0.667** #0.978** #1.004** #0.999**
(#3.30) (#4.67) (#4.60) (#4.13)
Divorced, living with partner #0.113 #0.421* #0.259 #0.530**
(#0.76) (#2.37) (#1.52) (#2.39)

B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

Divorced, not living with partner #0.306** #0.549** #0.641** #0.501**
(#2.94) (#4.96) (#5.09) (#4.35)
Widowed, living with partner #0.388 #1.714* #2.180** #0.230
(#0.66) (#2.10) (#2.86) (#0.98)
Widowed, not living with partner #0.541** #0.940** #0.544(*) #0.407**
(#2.73) (#4.68) (2.89) (#5.19)
Never been married, living with #0.102** #0.105 #0.198(*) #0.270
partner (#1.21) (#1.08) (#1.92) (#1.32)
Never been married, not living with #0.373** #0.470** #0.535** #0.691**
partner (#4.72) (#5.92) (#6.20) (#7.17)
No children at home Reference group
Children living at home #0.069 #0.169* #0.131(*) #0.291**
(#1.01) (#2.30) (#1.71) (#3.38)
Not completed primary education 0.020 #0.149 #0.110 #0.433**
(0.06) (#0.63) (#0.44) (#4.39)
Primary or first stage of basic Reference group
education
Lower secondary or second 0.133 0.142 0.202(*) 0.075
stage of basic education (1.10) (1.35) (1.69) (0.97)
Upper secondary education 0.147 0.198(*) 0.263* 0.160*
(1.27) (1.92) (2.30) (1.98)
Post-secondary, non-tertiary 0.195 0.277* 0.292* 0.415**
education (1.50) (2.24) (2.17) (3.34)
First stage of tertiary education 0.175 0.435** 0.256* 0.290**
(1.44) (3.72) (2.12) (2.71)
Second stage of tertiary education 0.209 0.442** 0.167 0.600**
(1.62) (3.28) (1.27) (4.07)
(continued on next page)

309
Table A2 (continued)

310
Dependent variable: Flexibility of working hours as distinction criteria Employment status/profession as distinction criteria
Life satisfaction
Group with high Group with low Group with high Group with low
opportunity costs opportunity costs opportunity costs opportunity costs
of time of time of time of time
Paid work, employed Reference group
Paid work, self-employed #0.131*

B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

(#2.09)
Retired Reference group
Unemployed, looking for a job #1.476**
(#12.37)
Unemployed, not looking for a job #1.015**
(#7.75)
Big city Reference group
Suburbs 0.064 0.149(*) #0.048 0.045
(0.84) (1.70) (#0.57) (0.52)
Town, small city 0.047 0.230** #0.033 0.082
(0.69) (3.20) (#0.44) (1.13)
Country village 0.145* 0.266** 0.105 0.285**
(2.05) (3.58) (1.39) (3.91)
Farm, home in countryside 0.368** 0.288* 0.281* 0.401**
(3.45) (2.32) (2.54) (3.35)
Country-fixed effects Yes Yes Yes Yes
Constant 6.203** 5.895** 6.204** 6.919**
(12.39) (10.22) (10.94) (11.93)
Observations 6460 7062 5950 8974
R2 0.14 0.15 0.16 0.22

Notes: (1) Group 1 contains respondents with flexibility of working hours of 6 and higher on a scale from 0 to 10 and group 2 the people with flexibility of working
hours of 5 and lower. In both groups, respondents’ employment status is ‘‘paid work, employed’’ and nothing else. Group 3 contains the self-employed, as well as
managers, senior officials, legislators and professionals (according to ISCO-88 classification) with employment status ‘‘paid work’’ (employed or self-employed) and
nothing else and group 4 contains retirees and the unemployed with no other employment status.
(2) Ordinary least squares estimations.
(3) **Significant at 99% level, *significant at 95% level, (*) significant at 90% level.
(4) t-values in brackets.
(5) Variable for household size and dummy variables for highest income category and for the different countries are not shown. Dummy variables for missing
observations for income, household size, working hours, gender, marital status, children, education, employment status, and citizenship are neither shown.
Data source: European Social Survey, World Development Indicators.
 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313 311

References

Aguiar, Mark, & Hurst, Erik. (2007). Measuring trends in leisure: The allocation of time over five decades.
Quarterly Journal of Economics, 122(3), in press.
Aliga, Christel, & Winqvist, Karin, (2003). How women and men spend their time, Results from 13 European
countries. Statistics in Focus Theme 3 – 12/2003, Eurostat.
Angeletos, George-Marios (2001). The hyberbolic consumption model: Calibration, simulation, and empirical
evaluation. Journal of Economic Perspectives, 15(3), 47–68.
Becker, Gary S., & Murphy, Kevin M. (1988). A theory of rational addiction. Journal of Political Economy, 96(4),
675–700.
Bittman, Michael (1998). The land of the lost long weekend? Trends in free time among working age Australians,
1974–1992. Society and Leisure, 21(2), 353–378.
Bruni, Luigino, & Stanca, Luca (2006). Income aspirations, television and happiness: Evidence from the world
values survey. Kyklos, 59(2), 209–225.
Bruni, Luigino, & Stanca, Luca. (2007). Watching alone: Relational goods, television and happiness. Journal of
Economic Behavior & Organization, in press.
Christakis, Dimitri A., & Zimmerman, Frederick J. (2006). Early television viewing is associated with protesting
turning off the television at age 6. Medscape General Medicine, 8(2), 63.
Corneo, Giacomo (2005). Work and television. European Journal of Political Economy, 21(1), 99–113.
Cutler, David M., Glaeser, Edward L., & Shapiro, Jesse M. (2003). Why have Americans become more obese?
Journal of Economic Perspectives, 17(3), 93–118.
DellaVigna, Stefano, & Malmendier, Ulrike (2006). Paying Not to Go to the Gym. American Economic Review,
96(3), 694–719.
DellaVigna, Stefano, & Daniele Paserman, M. (2005). Job search and impatience. Journal of Labor Economics,
23(3), 527–588.
Easterlin, Richard A. (2001). Income and happiness: Towards an unified theory. Economic Journal, 111(473),
465–484.
Espe, Hartmut, & Seiwert, Margarete (1987). television viewing types, general life satisfaction, and viewing
amount: An empirical study in West-Germany. European Journal of Communication, 13, 95–110.
Fang, Hanming, & Silverman, Dan (2006). Time-inconsistency and welfare program participation: Evidence from
the NLSY. Mimeo: Yale University & University of Michigan.
Frederick, Shane, Loewensteinm, George, & O’Donoghue, Ted (2002). Time discounting and time preference: A
critical review. Journal of Economic Literature, 40(2), 351–401.
Frey, Bruno S., & Stutzer, Alois (2002a). Happiness and economics: How the economy and institutions affect
well-being? Princeton, NJ: Princeton University Press.
Frey, Bruno S., & Stutzer, Alois (2002b). What can economists learn from happiness research? Journal of
Economic Literature, 40(2), 402–435.
Frey, Bruno S., & Stutzer, Alois. (2004). Economic consequences of mispredicting utility. IEW Working Paper
No. 218, University of Zurich.
Gerbner, George, Gross, Larry, Morgan, Michael, & Signorielli, Nancy (1980). The mainstreaming of America:
Violence profile no. 11. Journal of Communication, 30(3), 10–29.
Gerbner, George, Gross, Larry, Morgan, Michael, Signorielli, Nancy, & Shanahan, James (2002). Growing up
with television: Cultivation processes. In Jennings Bryant & Dolf Zillmann (Eds.), Media effects: Advances in
theory and research (pp. 43–67). Associates, Mahwah, NJ: Lawrence Erlbaum.
Gershuny, Jonathan (2000). Changing times: Work and leisure in postindustrial society. Oxford: Oxford
University Press.
Goodin, Robert E., Rice, James M., Bittman, Michael, & Saunders, Peter (2005). The time pressure illusion:
Discretionary time versus free time. Social Indicators Research, 73(1), 43–70.
Grahn, Almut, Lehmann, Andreas, Schmitz, Edgar, & Breinker, Carsten (2003). MIT – Medien im Tagesablauf
2003. Cologne, Germany: IP Germany.
Gruber, Jonathan, & Koszegi, Botond (2001). Is addiction rational? Theory and evidence. Quarterly Journal of
Economics, 116(4), 1261–1303.
Gruber, Jonathan, & Mullainathan, Sendhil (2005). Do cigarette taxes make smokers happier? B.E. Journals in
Economic Analysis and Policy: Advances in Economic Analysis and Policy, 5(1), 1–43.
IP Germany (2005). Television 2005: International keyfacts. Cologne, Germany: IP Germany.
312 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313

Kahneman, Daniel, Krueger, Alan B., Schkade, David A., Schwarz, Norbert, & Stone, Arthur A. (2004). A
survey method for characterizing daily life experience: The day reconstruction method. Science, 306(5702),
1776–1780.
Kubey, Robert, & Csikszentmihalyi, Mihaly (1990). Television and the quality of life. How viewing shapes
everyday experience. Hillsdale, NJ: Lawrence Erlbaum.
Kubey, Robert, & Czikszentmihalyi, Mihaly (2002). Television addiction is no mere metaphor. Scientific
American, 286(2), 74–80.
Laibson, David (1997). Golden eggs and hyperbolic discounting. Quarterly Journal of Economics, 112(2),
443–477.
Layard, Richard (2005). Happiness: Lessons from a new science. New York, NY: Penguin.
Lelkes, Orsolya (2006). Knowing What Is Good for You. Empirical Analysis of Personal Preferences and the
‘Objective Good’. Journal of Socio-Economics, 35(2), 285–307.
Lichter, S. Robert, Lichter, Linda S., & Rothman, Stanley (1994). Prime time: How TV portrays American
culture. Washington, DC: Regnery.
Loewenstein, George, O’Donoghue, Ted, & Rabin, Matthew (2003). Projection bias in predicting future utility.
Quarterly Journal of Economics, 118(4), 1209–1248.
Loewenstein, George, & Schkade, David A. (1999). Wouldn’t it be nice? Predicting future feelings. In Daniel
Kahneman, Ed Diener, & Norbert Schwarz (Eds.), Well-being: The foundation of hedonic psychology
(pp. 85–105). New York, NY: Russell Sage Foundation.
MacKinnon, David P., Lockwood, Chondra M., Hoffman, Jeanne M., West, Stephen G., & Sheets, Virgil (2002).
A comparison of methods to test mediation and other intervening variable effects. Psychological Methods,
7(1), 83–104.
MacKinnon, David P., Warsi, Ghulam, & Dwyer, James H. (1995). A simulation study of mediated effect
measures. Multivariate Behavioral Research, 30(1), 41–62.
McIlwraith, Robert D. (1998). ‘‘I’m Addicted to Television’’: The personality, imagination, and TV watching
patterns of self-identified TV addicts. Journal of Broadcasting & Electronic Media, 42(3), 371–386.
Morgan, Michael (1984). Heavy television viewing and perceived quality of life. Journalism Quarterly, 61(3),
499–504.
O’Donoghue, Ted, & Rabin, Matthew (1999a). Addiction and self control. In Jon Elster (Ed.), Addiction: Entries
and exits (pp. 169–206). New York, NY: Russell Sage Foundation.
O’Donoghue, Ted, & Rabin, Matthew (1999b). Doing it now or later. American Economic Review, 89(1), 103–124.
O’Donoghue, Ted, & Rabin, Matthew (2002). Addiction and present-biased preferences. Working Paper E02-
312, Economics Department, University of California, Berkeley.
O’Guinn, Thomas C., & Shrum, L. J. (1997). The role of television in the construction of consumer reality.
Journal of Consumer Research, 23(4), 278–294.
Putnam, Robert D. (1995). Tuning in, tuning out – the strange disappearance of social capital in America.
Political Science & Politics, 28(4), 664–683.
Richins, Marsha L. (1987). Media, materialism, and human happiness. Advances in Consumer Research, 14,
352–356.
Robinson, John P., & Godbey, Geoffrey (1999). Time for life: The surprising ways americans use their time (2nd
ed.). University Park, PA: Pennsylvania State University Press.
Schelling, Thomas C. (1984). Self-command in practice, in policy, and in a theory of rational choice. American
Economic Review, 74(2), 1–11.
Schor, Juliet B. (1991). The overworked American: The unexpected decline of leisure. New York, NY: Basic Books.
Sen, Amartya (1982). Choice, welfare and measurement. Oxford, UK: Blackwell.
Sen, Amartya (1995). Rationality and social choice. American Economic Review, 85(1), 1–24.
Shapiro, Jesse M. (2005). Is there a daily discount rate? Evidence from the food stamp nutrition cycle. Journal of
Public Economics, 89(2–3), 303–325.
Shrum, L. J., Burroughs, James E., & Rindfleisch, Aric (2005). Television’s cultivation of material values. Journal
of Consumer Research, 32, 473–479.
Signorielli, Nancy, Gerbner, George, & Morgan, Michael (1995). Violence on television – the cultural indicators
project. Journal of Broadcasting & Electronic Media, 39(2), 278–283.
Sirgy, M. Joseph, Lee, Dong-Jin, Kosenko, Rustan, Meadow, H. Lee, Rahtz, Don, Cicic, Muris, et al. (1998).
Does television viewership play a role in the perception of quality of life? Journal of Advertising, 27(1),
125–142.
 B.S. Frey et al. / Journal of Economic Psychology 28 (2007) 283–313 313

Sullivan, O., & Gershuny, J. (2001). Cross-national changes in time-use: Some sociological (hi)stories
re-examined. British Journal of Sociology, 52(2), 331–347.
Thaler, Richard H. (1992). The winner’s curse. Paradoxes and anomalies of economic life. New York, NY: Free
Press.
Wilson, Timothy D., & Gilbert, Daniel T. (2003). Affective forecasting. In M. Zanna (Ed.). Advances in
experimental social psychology (vol. 35, pp. 345–411). New York: Elsevier.