Advancing wellbeing research: would Americans be happier if they lived like Australians?
Patulny, Roger ; Fisher, Kimberly
Introduction
Every happiness is but lent by chance for an uncertain time, and
may therefore be demanded back the next hour.
--Arthur Schopenhauer
Subjective wellbeing increasingly is recognised as a key indicator
of quality of life. Positive psychologists have spent much of the last
decade promoting the emotional qualities of subjective wellbeing such as
happiness or satisfaction as critical components of good mental health
(Seligman et al. 2005). Such emotions may complement--or even provide a
credible alternative to--income in national economic measures of social
progress (Diener & Seligman 2004; Layard 2005). Many of these
studies are limited, however, by their focus on measuring the qualities
associated with generalised domains, such as satisfaction with life as a
whole, or with work, family or leisure. We argue that these broad
assessments miss the variation in emotional experiences associated with
daily activities and the consequences for wellbeing when routines
change.
Measures of 'emotional wellbeing'--which we define here
as the continuous (or hedonic) flow of emotions and enjoyment derived
from the activities performed in the specific sequences that fill a
day--often provide better indicators of subjective wellbeing than more
conventional life-domain measures. Kahneman and Krueger find through
experiments and reviews of experimental research (Kahneman et al. 1997,
2006; Krueger & Schkade 2007) that life-domain assessments are
over-weighted by extreme and recent experience, and by moods. They
observe question-order effects and find indications that life-domain
measures asked two weeks apart are less reliable than diaries collecting
emotion data alongside time use information (thus capturing emotion in
context). Kahneman and Krueger thus distinguish between generalised
measures that ask 'How happy/satisfied are you in general?'
(Easterlin 1974; Heady & Wearing 1992; Oswald 1997; Helliwell 2003)
and hedonic measures that ask 'How much time do you spend doing
enjoyable activities?' (Juster & Stafford 1985; Robinson &
Godbey 1997; Kahneman & Krueger 2006; Gershuny 2011), and they
emphasise the importance of the latter measures.
A growing number of researchers in the field of time diary surveys
have started to collect data on emotions. These studies capture the
range and frequency of the daily activities that take place across a
society, and situate emotional responses in the contexts that shape
these emotions (Kahneman & Krueger 2006; Gershuny 2012; Koll &
Pokutta 2012). This approach has the analytical benefit of increasing
the variability of responses, in that the metric used is much wider than
a single 10-point scale of happiness, and it also captures data on
location, co-presence and enjoyment based over 1,440 minutes per day.
When conducted on a national scale, time and emotion diary surveys form
a useful compendium of measures of the subjective wellbeing of nations,
complementing national income and production measures (Kahneman &
Krueger 2006; Gershuny 2012; Koll & Pokutta 2012). They allow
analysis of variations in wellbeing across countries and regions
resulting from local policy settings that impact time-schedules, such as
business or school or service opening times, labour laws, child support
arrangements, transportation and service availability.
Time diary and affect measures also offer a unique opportunity to
address a dilemma concerning assessing the unintended consequences of
making policies. Daily activities take place in cycles bounded by the
biological needs, cultural conventions, and social structures of
communities and nations. Policies that change one element of behaviour
may have unintended consequences for other aspects of daily life. For
example, policies aimed at reducing time spent commuting can open space
in daily schedules that people can fill with other activities which
could be generally perceived as 'positive' (for example,
exercise) or 'negative' (for example, watching television).
Time use methods can assess the impact of policy and behavioural changes
and associated satisfaction in ways that more general questions about
wellbeing cannot. Simply asking people how satisfied they are with life
in general does not reveal whether it was the behaviour change inspired
by the policy that made them more or less satisfied, or whether this
resulted from other changes in activities that occurred as an unintended
consequence of the policy.
In this paper, we contribute to the growing literature exploring
the potential of time use data to model these potential consequences of
policy and behavioural change on emotions. We do this by comparing time
use patterns in two countries with relatively similar cultural and
institutional contexts, Australia and the United States, to see how
people in the United States might feel if they altered--or policy
encouraged them to alter--their routines to live more like Australians.
To date, nationally representative surveys of time use and emotion
only have been conducted in three countries, (1) which limits the direct
comparison of emotional wellbeing across countries. Nevertheless, we can
use the available affect profiles and associated time use patterns for a
country such as the United States counterfactually, to see how emotional
wellbeing there might be improved (or degraded) if people living in the
United States were to modify their routines to live more like people in
another country. This technique builds on counterfactual analytical
techniques widely used in political science and historical analysis
(Tetlock & Belkin 1996). However, rather than modelling changes in
daily routines from a consequent change in policy, we map the ordinal
emotion rating associated with activities in one time-pattern (that of
the United States) to estimate the emotional responses with the
alternative time-pattern (that of Australia). Our strategy is a modest
version of that adopted by Gershuny (2011, 2012), who first observed
similar emotion ratings associated with activities undertaken in the
1980s in the United States and United Kingdom. His approach was to map
an amalgamated emotion rating based on these affect ratings to the
contemporary time use patterns of 15 countries in order to compare the
emotional experiences that countries with different welfare regime types
might confer upon their national populations.
It is important to note that we are not directly comparing
Americans and Australians, nor do we assert that Australians would
respond in the same way as Americans to various activities. Rather, we
estimate how people in the United States might feel if they altered
their routines to live more like Australians, in order to investigate
the consequences of changing time-based behaviour patterns for emotional
wellbeing. In this context, cultural differences between the two
countries in how people feel about activities become irrelevant.
Nevertheless, as we will demonstrate, there are a sufficient number of
similarities between these two countries to make the use of Australian
patterns a realistic counterfactual to examine how changes in American
behaviours might impact American emotional wellbeing.
Key differences in policies affecting these two countries
(discussed below) include those of gender and wellbeing. Such
differences are likely to produce potentially large differences in
wellbeing worth investigating through survey research. Stevenson and
Wolfers (2009) use the American General Social Survey to examine
gendered wellbeing in America from the 1970s to 2006, and find that
women's happiness has declined absolutely and relative to
men's since the 1970s. Stevenson and Wolfers (2009) point to the
contradiction between the improvement in women's income and
employment situations over this time period, and a decline in their
wellbeing. However, they do not adequately identify why this pattern has
emerged, beyond suggesting only that women now compare their employment
success to their male peers rather than to the careers of women from
previous generations. They do not engage with time-based measures and
changes in behaviours.
Kahneman and Kruger (2006), on the other hand, bring a time-based
explanation to similar findings. They observe that the amount of time
women report that they have experienced negative emotions has remained
fairly constant since 1965, while men's reporting of negative
emotions has decreased, mainly due to men's reduced paid work time.
Such assessments suggest that the impact of policy on behaviours, and
thus on wellbeing, may impact the sexes differently.
This paper maps activity-associated emotion ratings from the 2006
Princeton Affect and Time Survey onto time-use patterns in the
Multinational Time Use Study (MTUS) version of the 2006 American Time
Use Survey and the 2006 Australian Bureau of Statistics National Time
Use Survey. Our purposes are to determine whether Americans would
experience a net gain or loss in 'pleasant' time from shifting
to Australian time patterns; to identify in what areas or activities
Americans would gain and lose from such a change; and to identify the
consequences for gender equity and men and women's emotional
wellbeing.
Time, emotional wellbeing and gender--United States and Australia
Both Australians and Americans report feeling generally happy in
international surveys of wellbeing (Leigh & Wolfers 2006). Given
relatively abundant material prosperity in both countries, this is
unsurprising. Heady and colleagues (2004) reveal through analysis of
five separate national panel datasets (including one from Australia)
that material resources have a much stronger impact upon life
satisfaction than income. Kahneman and Deaton (2010) assess satisfaction
and emotional wellbeing amongst nearly half a million responses to the
American Gallup-Healthways Well-Being Index. They find that positive
life evaluation (satisfaction) increases with income, though is subject
to diminishing returns. That is, the relationship is non-linear and
increases with the log of income, rather than income itself. However,
they also find that while emotional wellbeing increases with income, the
relationship diminishes and flat-lines at an annual income of
approximately (US) $75,000. Beyond this, increased income does not
improve emotional wellbeing. Emotions thus demonstrate a variability
that transcends purely material conditions, and given the cultural and
institutional differences between Americans and Australians, we would
expect the time use patterns of both countries to generate differences
in total pleasant and unpleasant time.
Table 1 displays the daily time allocation of working-aged women
and men in Australia and the United States, with reference to 21 other
countries. Australian men work relatively long hours in paid employment
(fifth of 23) and also spend the highest total time caring for children.
American men have similar patterns and spend only modestly less time in
these activities. They engage in marginally more housework than
Australian men, but rank similarly compared to men in many other
countries in terms of their commitment to paid and unpaid work. Women in
the United States and Australia have similarly high levels of time spent
caring for children. Australian women perform both housework and paid
work at a middle-range level, whereas American women work more hours for
pay and engage in relatively fewer hours of housework. As with men,
women's total commitment to paid and unpaid work is similar in both
countries. Women and men in the United States spend roughly 30 minutes
more per day socialising, and around one-quarter of an hour more on
weekdays and half an hour more on weekends in voluntary activities than
Australians. In sum, Australian and American men and women spend similar
and substantial amounts of time in paid and unpaid work, but Americans
spend more time in social and voluntary activities.
Australians' and Americans' emotional reactions to paid
work and child-care activities may be influenced by national policy
contexts (Hook 2010; Craig & Mullan 2010). National and State
governments in both countries have instigated policies to encourage
women's transition into the workplace, though Australian
legislators have shown greater willingness to regulate private child
care facilities (Craig & Siminski 2010; Craig & Mullan 2010),
and assist with childcare costs. Australia has also recently introduced
its first national paid maternity leave scheme, while the United States
lacks any such comprehensive national scheme. A further difference (and
possible consequence of the aforementioned polices) is that Australian
mothers with young children are more likely to take part-time jobs,
while a relatively high proportion of mothers with children aged less
than 12 in the United States work full-time, when compared with other
OECD countries (OECD 2011).
Working parents in Australia and the United States also tend to
work long hours, compared with working parents in other OECD countries
(OECD 2011) --and even though more women than men work for pay
part-time, women's overall lower time in paid work is
counter-balanced by their higher unpaid work time (Fisher & Robinson
2010). This means that working mothers and fathers face similar
constraints on the proportion of time they can allocate to the more
enjoyable activities. While the literature exploring the gendered
division of domestic labour by welfare regime type is extensive (Rice et
al. 2006; Drobnic & Guillen 2011), so far the time-use literature
has not investigated the gendered distribution of emotional experiences
across various daily activities.
Current research suggests that men and women tend to report
different levels of emotion during activities, and that different
profiles of time spent in such activities across countries will thus
contribute greatly to differences in emotional wellbeing (or lack
thereof). In general, women are more likely to enjoy social time and
activities that are oriented more towards informal support and emotional
relations rather than work-orientated connections (Emmerick 2006).
Compared to men, women report less satisfaction with (and interest in)
paid work (Sloane & Williams 2000), more stress at work (Hutri &
Lindeman 2002), and more frequent and longer pain durations,
particularly associated with intensity of work (Dao & LeResche 2000;
Burchell & Fagan 2004). In general, longer work hours also are
thought to cut into more traditionally feminised 'social'
activities, such as time looking after children while talking with other
parents (Maher et al. 2008; Gray 2009). As recent income support
policies in many countries encourage parents to work longer hours,
single parents, and single mothers in particular, not only cope with
more severe financial constraints than couple parents (De Vaus et al.
2009), they also face a greater likelihood of social exclusion from the
time binds imposed by care and total work responsibilities (Patulny
& Wong 2013).
In Australia, women report high levels of time-stress reconciling
paid work and unpaid care commitments (Craig & Sawrikar 2009), and
the role strain Australian mothers experience balancing work and care is
associated with increasing mental health problems (Bryson et al. 2007).
Similar findings emerge in research into stress experienced by women in
the United States (Epstein & Kalleberg 2004). Offer and Schneider
(2011) attribute the greater stress and work-family conflict experienced
by American women to their greater tendency to multitask across domains
of activity (for instance, child care overlapping with leisure, or
domestic work overlapping with personal care) compared to American men.
Men in general are likely to report more unhappy time at work
arising from working longer hours. Men are potentially more likely to
suffer greater tiredness from sleeping slightly less than women
(Robinson & Michelson 2010). Gershuny (2011) found that men in the
United States and the United Kingdom were less likely than women to
report feeling happy while working. Nevertheless, working fathers still
experience less work-related stress and burn-out than mothers when
children are young (Hill et al. 2008).
Leisure and social contact patterns may also differ more by gender
than by country. Australian men are more likely to report loneliness and
depression with regards to social activities, or the lack thereof
(Franklin & Tranter 2008). Both Australian (van Wanrooy & Wilson
2006) and American (Epstein & Kalleberg 2004) men are likely to have
seen their friendship networks and social time deteriorate as a
consequence of long working hours. Australian men (Patulny 2009) and
American women (Cornwell 2011) also are more likely to lose collegial
ties upon retirement.
Gershuny (2011) found that in both the United States and United
Kingdom in the 1980s, women tended to enjoy in-home leisure more than
men (excluding television, which both genders rate roughly equally).
Gershuny also observed that women rated shopping and travel more highly
than men, though for both sexes these activities are among the more
unpleasant of the activities that people undertake (2011).
Methods
Our procedure for estimating whether Americans would experience a
net gain or loss in 'pleasant' time from shifting to
Australian-style time-use patterns involved four stages: (1)selecting
time-use data (MTUS)and affect data (PATS); (2) using regression models
to estimate the likelihood of a PATS episode being
'unpleasant' based on the activities and demographic qualities
(explained further in Patulny & Fisher 2013); (3) imputing the
likelihood of episodes in the MTUS data being 'unpleasant' for
Americans based on PATS ratings by use of the models from the previous
step; and (4) aggregating the time in the 'unpleasant' MTUS
episodes to estimate 'unpleasant minutes per day'. We now
briefly summarise these stages.
1) Selecting data
The 2006 Princeton Affect and Time Survey (PATS) serves as our
source to generate activity-related affect scores. PATS is a nationally
representative telephone survey of 11,905 respondents in the United
States. Sample weights adjust the demographic distribution to match the
United States Current Population Survey. PATS asked participants to
describe their activities from the previous 24-hour day, then collected
six affect ratings for three randomly chosen 15-minute episodes during
that day (2). The survey asked diarists to rate their perceptions on a
0-6 scale of how much they felt happy, interested, in pain, sad,
stressed and tired during the time period. Table 2 displays a sample
diary, enabling readers to get a feel for the flexible structure of the
data.
The Multinational Time Use Survey (MTUS) offers a usefully
harmonised version of both the 2006 American Time Use Survey and the
2006 Australian National Time Use Survey episode data. In the MTUS file,
each row case represents a change in main activity, secondary activity,
location, mode of transport or who else was present. The main and
secondary activity codes are harmonised into 69 categories that appear
in a large majority of time use surveys. In this format, the surveys not
only are prepared for direct comparison but also organised in a similar
format to the PATS file.
2) Estimating the likelihood of an episode being
'unpleasant'
Kahneman and Krueger (2006) argue that measures of unpleasant time
yield more robust results than measures of positive enjoyment. We
repeated Kahneman and Krueger's (2006) approach for determining
which episodes are unpleasant, dividing activities into binary
categories, pleasant (0, the default), and unpleasant (1), representing
cases where the scores for any of the negative emotions outweigh the
score for the positive emotions for that particular episode. They note
that this is a conservative approach to measuring unpleasantness as
people are less likely to report intense negative as opposed to positive
emotions, and as a consequence produces fewer unhappy episodes, with
only 20 per cent of episodes found to be 'unpleasant' by their
measure. We analysed the PATS dataset to identify 'unpleasant'
episodes.
We next ran regression models to determine the impact of
demographic characteristics and activities on the likelihood of the
episode being 'unpleasant.' This approach mirrors that
undertaken by Gershuny (2011, 2012), who used activity-level emotion
information collected in time diary surveys in the United States in 1985
and in the United Kingdom in 1987 to make similar projections about
potential implications for national time accounts. In that paper,
Gershuny used much older time and affect data from full 24-hour diaries
rather than the more detailed utility measures for three random
intervals that we make use of from the PATS survey. Nevertheless, it is
worth noting that Gershuny found little difference in the utility
profiles for the British and Americans, and that the gender differences
were greater than the country differences (Gershuny 2011, 2012). We have
no data at present to say how closely Australian and American utility
profiles might map.
We used Linear Probability Regression (LPM) to predict
unpleasantness on the basis of the demographic qualities of the person
and the main activity that took place during the episode. The equation
used to impute the probability of an MTUS episode being unpleasant took
the following form:
[Pu.sub.e] = [a.sub.k]X + [b.sub.j]T
Here, [Pu.sub.i] is the probability of each episode e being rated
as unpleasant. T is a vector of j activity variables (such as
"sleep", "watching television" and so on) that
correspond to the "main activities" in the MTUS time-diary
datasets that may influence the unpleasantness of activities, while
[b.sub.i] is the set of coefficients relating this vector to each
episode. X is a vector of k (demographic) control variables that may
influence the enjoyment of activities, and [a.sub.i] is the set of
coefficients relating this vector to each to each episode. The results
suggest that the model provides a reasonable estimation of the affect
and the unpleasantness associated with activities in the United States
(see Patulny & Fisher 2012 for more detail).
3) Imputing the 'unpleasantness' likelihood of Australian
and American time diary episodes
Next we used the beta coefficients derived from the regression to
impute the likelihood of episodes in the Australian and American MTUS
data being 'unpleasant'. We uses the MTUS version of the 2006
ATUS data instead of the PATS in part to access a larger sample and in
part to compare two datasets into which we mapped utility profiles,
minimising the potential that the mapping exercise itself might impact
on the results. The results from the imputation suggest that the model
provides a reasonable estimation of the affect and the unpleasantness
associated with activities in the United States. The mean episode
unhappiness--what Kahneman and Krueger call u-index--estimated for the
American MTUS data was nearly identical to actual episode u-index in the
PATS data: around 0.18 (Patulny & Fisher 2013).
4) Estimating unpleasant minutes per day in Australian and American
MTUS data
The final step was to use the predicted affect scores in the MTUS
data to find 'unpleasant' and 'pleasant' episodes of
time in the American and Australian data subsets, and then to aggregate
these into 'unpleasant' and 'pleasant' time to
produce a sum total of 'pleasant' and 'unpleasant'
minutes per day that can be compared. Given that the modelling produced
estimates of the likelihood of episodes being pleasant or unpleasant
(rather than definitively as yes '1', or no '0') it
was necessary to choose a cut-off for defining episodes as one status or
another. We therefore coded episodes in MTUS as 'unpleasant'
if the predicted affect score for an episode was higher than the mean
u-index score in the PATS data of 0.18, or as 'pleasant' if
the score was equal to or lower than 0.18. Once episodes were defined as
'unpleasant' or 'pleasant', the time associated with
each episode was aggregated and summed into total 'pleasant'
and 'unpleasant' minutes per day. This measure is then
amenable to comparison across the time-schedules of different countries.
The procedure described above produces data amenable to descriptive
analysis and further regression modelling. As noted above, the
unpleasantness score is based on American data, so any derived results
will not say how Australians feel, but rather how people from the United
States would feel if they followed the time patterns of Australians.
From here we turn to our descriptive results.
Results
Unpleasant time
We first look at aggregate net unpleasant time if Americans were to
live like Australians (see Figure 1). These figures exclude time spent
asleep, as sleep time greatly exaggerates this difference. The broad
results show that Americans would have experienced a kind of double
'loss' if they lived like Australians, in that they would
experience both more unpleasant and less pleasant time. Adding these two
gaps together, Americans would have 45 minutes more net
'unpleasant' time per day if they transferred their affective
responses to an Australian time schedule.
[FIGURE 1 OMITTED]
Next we examine unpleasant time by activity category (see Figure
2). Americans would reduce their unpleasant time in paid work, and they
would gain more pleasant time in personal care. However, the shift to
Australian patterns would also substantially increase their time spent
in unpleasant unpaid domestic work, as well as increasing unpleasant
time spent watching television, which is one of the least valued leisure
activities, with decreasing pleasure the longer people remain in front
of the television. Americans also would lose a substantial amount of the
pleasant time they currently spend doing in-home free time and leisure
(mostly social) activities. In particular, Americans spend more
enjoyable time having friends over for dinner than do Australians, and
this would make a big difference to their overall level of wellbeing in
shifting between the two aggregate patterns of time use. Americans also
would stand to lose time in travel related to shopping (much of which
they enjoy). As things stood in 2006, American men spent less time in
unpleasant travel for work than they would have done living like
Australian men.
Overall, if Americans were to live like Australians, they would
gain a measure of 'happiness' in terms of slightly increased
pleasant time through less paid work and more pleasant personal care
activities. However, they would lose more than they gained in terms of
taking on more unpleasant domestic work, television and work-travel
time, and particularly in the loss of pleasant in-home social time and
leisure-travel time.
[FIGURE 2 OMITTED]
Unpleasant time by gender
When we separate women and men's experiences, large gender
differences appear. Whichever of the two patterns of time use they
followed, American men would enjoy more pleasant time and experience
less unpleasant time than women. The data show greater differences
between the emotional accounts of women and men than between the
lifestyles of Americans and Australians (if adopted by Americans). This
matches expectations from much of the research cited above. By shifting
to Australian time patterns, American women would have 3.5 hours more
net unpleasant time a day than American men. This would represent an
improvement over the actual American gender gap of 3.8 hours more
unpleasant time experienced by women. This is reflected in the
gender-country gap. There is a sizeable gap between the lifestyles of
American and Australian men of 57 minutes per day, such that American
men would be nearly an hour worse off per day if they lived like
Australian men. The gap in enjoyment between the lifestyles of American
and Australian women is smaller, but still substantial at 37 minutes per
day, or close to 4.5 hours per week. This suggests that a change to
Australian lifestyles would have a smaller negative impact upon American
women than men. While American women would lose out in terms of overall
emotional experience, they would benefit in a relative sense from a
reduced gender gap in unpleasant time (see Figure 3).
The differences noted above make more sense when examined in the
context of activities. In Figure 3, the striped bars represent what
Americans might feel living like Australians, while the solid bars are
the actual American patterns in 2006 (in each case, the lighter colours
are for men, the darker colours are for women). The gender gaps are
partially explained by some familiar time patterns. The vast bulk of
men's net unpleasant activity is in paid work. A small fraction
thereafter is located in education and unpaid domestic work. Every other
activity is, overall, pleasant for men. The bulk of women's
unpleasant time is spent in paid work and unpaid domestic work combined
together.
[FIGURE 3 OMITTED]
Most other activities are pleasant, but men spend more time than
women in pleasant personal care, civic activity, sports, in-home
leisure, media and computing, television-watching and travel. The
standout differences are in personal care, in-home leisure, travel, and
particularly television-watching. Unlike men, women find
television-watching to be a largely unrewarding activity. This may be
due to the quality of the time spent watching television. American men
not only watch television for 10-15 minutes longer per day than American
women, American men in couples also reported 'watching television
with their spouses' for an average of 10 minutes longer than
American women in couples reported doing the same activity with their
male partners. Men also are more likely to watch television for fewer
and longer spells than women (Fisher et al. 2007). These differences are
also more likely to arise in the early evening than at other times of
day, and are associated with women making more transitions between
leisure and paid and unpaid work than men. As women also multi-task more
than men (Epstein & Kalleberg 2004), this difference also may
reflect the stress of overlapping television-based leisure with (largely
unpaid) work.
The country-gaps involve some deviations from this pattern. Both
American men and women would reduce their unpleasant work time but
increase their unpleasant unpaid work time were they to adopt the
time-patterns of their Australian counter-parts. American men would gain
slightly more pleasant media time, but lose a larger amount of pleasant
in-home leisure, television-watching, and travel time if they lived like
Australian men. American women would gain some pleasant personal care
time, but lose pleasant in-home free time in particular, as well as
media, computing, and travel time. American men and women would thus
lose out from switching to Australian time patterns through loss of
pleasant in-home leisure and travel for both sexes, and loss of pleasant
television-watching time amongst men (see Figure 4).
[FIGURE 4 OMITTED]
Modelling pleasantness and unpleasantness in American and
Australian time patterns
We can now look at which emotions are associated with the most
unhappiness, comparing genders and controlling for a range of
demographic factors, using Ordinary Least Squares regression analysis.
Demographic controls are limited to those available in each of the PATS
and the MTUS datasets, and include country, sex, marital status, age,
employment status and education. All controls are coded as binary
dummies, and the consequent reference category is a fairly typical
mid-life American female--married with children, aged 35-65, not in the
labour force, and having completed secondary education but no
post-secondary education.
The models largely confirm the descriptive findings above that
American time schedules are more pleasant (for Americans) than
Australian time schedules, and reveal an important socio-economic
dimension to this picture. Table 3 shows the marginal effects of various
demographic characteristics on unpleasant and pleasant minutes per day.
The country and gender effects are as expected before we bring in
controls. American men have much less unpleasant and more pleasant time
than American women (the reference category). American men have more
pleasant activity patterns than women (whichever of the two patterns
they follow), though the model predicts that they would be less happy if
they followed the time schedules of Australian men. These effects are
consistent regardless of which control variables are introduced into the
model.
However, the model does reveal that socio-economic conditions have
an influence on the time and affective experiences of American women.
Once we control for high and low levels of education, Australian time
schedules shift from being significantly more unpleasant to
significantly less unpleasant for American women (the coefficient
switches from plus to minus). In other words, while American women of
lower and higher levels of education would experience more net
unpleasant time living like Australians, middle-educated women would
experience less net unpleasant time from such a behaviour shift. So it
seems that these American women (at least) would benefit from switching
to an Australian lifestyle.
Discussion and conclusion
In order to see clearly the factors that drive differences in the
experience of emotions, it is essential to understand the social context
and everyday activities associated with emotional states. This paper
demonstrates that time and affect diary surveys can help with this by
setting up counterfactual explorations of the potential changes in
emotional wellbeing from one country shifting its daily routine patterns
to more closely resemble those of another. Such an exercise complements
policy debates. Americans can test propositions suggesting that they
would enjoy their time more if only they had the labour market policies
of the Republic of Korea, or the family policies of Sweden, etcetera, by
combining time diary information from the favoured country with American
time and affect data. Such analysis would allow informed speculation as
to the impact on emotional wellbeing arising from time-based activity
changes that presumably result from policy. At a minimum, this technique
can show that a venerated pattern of life as an aspiration for policy
may not necessarily encourage change that would leave a population
better off.
The results here suggest there is fertile ground for cross-national
research once affect dimensions are added to the time use surveys of
more countries. This particular example suggests that Americans would
experience more unpleasant time and less pleasant time if they lived
like Australians. Primarily, these emotional 'losses' would
arise from more unpleasant unpaid domestic work and television-watching,
as well as a loss of pleasant in-home free-time and social leisure.
Americans would experience some emotional 'gains' that would
partly offset these losses--such as engaging in less unpleasant paid
work and more pleasant personal care--but these positive changes leave a
deficit of 45 minutes less enjoyable time per day.
Of particular significance, this research demonstrates the presence
of a gender gap in negative emotional experiences in the United States
which is associated with behaviour patterns, largely reflecting that men
enjoy more pleasant personal care, in-home leisure, travel and
television-watching time than women. This research also suggests that
adjustments in behaviour have potential to alter the depth of this gap.
In 2006, women in the United States lived with 3.8 hours of more
unpleasant time than men. Were Americans to shift to Australian
lifestyles, this gender gap would shrink slightly to women experiencing
only 3.5 hour of more unpleasant time than men. This adjustment would
occur as American men would lose out more--losing 57 minutes of pleasant
time shifting to an Australian lifestyle--while women would lose only 37
minutes of pleasant time. The gender gap finding has a qualifier;
middle-educated American women with secondary-school (but not tertiary)
education should feel happier leading an Australian lifestyle. We might
speculate on the role that the Australian welfare system plays in this,
given its propensity towards delivering a large number of family tax
benefits to Australian middle classes, though it is wiser here to simply
refer this finding as a topic for future research.
Though we use Australian data to set up a counterfactual analysis
of emotional wellbeing in the United States, this paper raises questions
to explore in Australia if and when emotion data becomes available in
this country. It highlights the importance of social connection and
isolation in improving wellbeing. In-home leisure is a great source of
pleasantness for Americans that some Australians (relatively) lack.
Social contact has been previously linked to subjective wellbeing, with
strong links recorded between subjective wellbeing and social capital,
as measured by ties to friends and neighbours, workplace ties, civic
engagement, trustworthiness and trust (Helliwell et al. 2010). Mellor
and colleagues (2008) find that loneliness is strongly associated with
the discrepancy between need to belong and satisfaction with personal
relationships, and with life in general. The finding that Australians do
not spend as much time in (potentially) pleasant social activity as they
(comparatively) might, and that this might impact upon their wellbeing
may indicate that it is time to start thinking of
'barbecue-starter' policies in Australia. Pending more
research on social connections and wellbeing in different urban
contexts, there are implications for policies aiming to support families
and for urban and suburban planning as well.
We emphasise that emotion matters in wellbeing research, but
further work needs to determine the best strategy for collecting these
data. Kahneman and Krueger (2006), Gershuny (2012), Robinson and Godbey
(1997), and others have demonstrated the limitations in conventional
wellbeing research, and the need to collect time-use and affect diaries
to measure the contextual mechanisms associated subjective wellbeing.
Nevertheless, the few surveys collecting these data adopt a variety of
approaches. The surveys collected in the United States and the United
Kingdom in the mid-1980s (summarised in Gershuny 2012) asked diarists to
complete one column rating their level of happiness during each entry in
the diary, though these columns employ a variety of scale points. The
2009-10 French national time use study collected by the official
statistical office INSEE adopted a similar approach. The Kahneman and
Krueger approach summarised in this paper collected ratings for six
emotions for three randomly selected 15-minute intervals. The 2010
American Time Use Survey collected by the United States Bureau of Labour
Statistics asked the same six items designed by Kahneman and Krueger,
but asked these questions of three randomly selected episodes. Future
research needs to compare results of these methods to test which
measures of emotion offer most analytic potential, and also to
investigate whether fewer questions asked of all episodes or more
questions asked of fewer episodes provides that optimal research
potential.
We conclude by suggesting future surveys should also be more
cognisant of sociological issues and approaches. Kahneman and
Krueger's interest is primarily in the psychological enjoyment of
activities, rather than in the social interactions that shape enjoyment
and emotional wellbeing. They are therefore interested in and include
only a limited number of emotions in their surveys, and do not measure
the prevalence and influence of the more 'socially
stigmatised' emotions such as anger, shame and envy. Similarly,
they do not account for Hochschild's key contention (1979, 1983)
that emotions can be 'managed', in that people adjust their
emotions according to the normative expectations of the people around
them in specific social contexts. This usually occurs in the workplace,
but might be endemic to family, friends and indeed every social
interaction. It is difficult to assess when those emotions most commonly
managed are the stigmatised emotions such as anger, shame and envy that
people often try to hide. We suggest then that there is also a need for
an 'audit' of emotions and emotion work, to be undertaken
through a range of methods including time-diaries. We encourage this as
a way forward for future research into time and emotion in Australia and
other countries. Further suggestions and details as to how this might be
achieved are outlined by the authors elsewhere (Patulny 2012).
References
Bryson, L., Warner-Smith, P., Brown, P. & Fray, L. (2007)
'Managing the work-life roller-coaster: private stress or public
health issue?', Social Science & Medicine, 65 (6), 1142-1153.
Burchell, B. & Fagan, C. (2004) 'Gender and the
intensification of work: evidence from the European Working Conditions
Survey', Eastern Economic Journal, 30 (4), 627-42.
Cornwell, B. (2011) 'Age trends in daily social contact
patterns', Research on Aging, 33 (5), 598-631.
Craig, L. & Mullan, K. (2010) 'Parenthood, gender and
work-family time in USA, Australia, Italy, France and Denmark',
Journal of Marriage and Family, 72 (5), 1344-1361.
Craig, L & Sawrikar, P. (2009) 'Work and family: how does
the (gender) balance change as children grow?', Gender, Work &
Organisation, 16 (6), 684-709.
Craig, L. & Siminski, P. (2010) 'Men's housework,
women's housework and second births in Australia', Social
Politics 17 (2), 235-266.
Dao, T.T. & LeResche, L. (2000) 'Gender differences in
pain', Journal of Orofacial Pain, 14 (3), 169-95
De Vaus, D. Gray, M. Qu, L. & Stanton, D. (2009) 'The
effect of relationship breakdown on income and social exclusion',
paper presented to the Australian Social Policy Conference, University
of New South Wales, Sydney, 8 July.
Diener, E. & Seligman, M.E.P. (2004) 'Beyond money: toward
an economy of well-being', Psychological Science in the Public
Interest, 5 (1), 1-31.
Drobnic, S. & Guillen, A.M. (eds) (2011) Work-life Balance in
Europe: The Role of Job Quality, Basingstoke, Palgrave Macmillan.
Easterlin, R.A. (1974) 'Does economic growth improve the human
lot? Some empirical evidence'. In A.D. Paus & W.R. Melvind
(eds), Nations and Households in Economic Growth: Essays in Honor of
Moses Abramovitz, New York, Academic Press, Inc.
Emerick, I.J. (2006) 'Gender differences in the creation of
different types of social capital: a multilevel study', Social
Networks, 28 (1), 24-37.
Epstein, C. & Kalleberg, A.L (eds) (2004) Fighting for Time:
Shifting Boundaries of Work and Social Life, New York, Russell Sage
Foundation.
Fisher, K., Egerton, M., Gershuny, J.I. & Robinson, J.P. (2007)
'Gender convergence in the American Heritage Time Use Study
(AHTUS)', Social Indicators Research, 82 (1), 1-33.
Fisher, K. & Robinson, J.P. (2010) Daily routines in 22
countries: diary evidence of average daily time spent in thirty
activities, Technical Paper 2010-01, Centre for Time Use Research,
University of Oxford.
Franklin, A. & Tranter, B. (2008) Loneliness in Australia,
Housing and Community Research Unit Paper 13, University of Tasmania.
Gershuny, J.I. (2000) Changing Times: Work and Leisure in
Postindustrial Society, Oxford, Oxford University Press.
Gershuny, J.I. (2011) Time-use surveys and the measurement of
national well-being, Centre for Time-use Research, Department of
Sociology, University of Oxford.
Gershuny, J.I. (2012) 'National utility: measuring the
enjoyment of activities', European Sociological Review, On-line
first, DOI 10.1093/esr/jcs077.
Gray, A. (2009) 'The social capital of older people',
Ageing and Society, 29 (1), 5-31.
Heady, B. & Wearing, A. (1992) Understanding Happiness: A
Theory of Subjective Well-being, Melbourne, Longman Cheshire.
Heady, B, Muffels, R. & Wooden, M. (2004) Money doesn't
buy happiness ... or does it? A reconsideration based on the combined
effects of wealth, income and consumption, Discussion Paper 1218, Bonn,
Institute for the Study of Labor.
Helliwell, J.F. (2003) 'How's life? Combining individual
and national variables to explain subjective well-being', Economic
Modelling, 20 (2), 331-60.
Helliwell, J.F, Barrington-Leigh, C.P., Harris, A. & Huang, H.
(2010) 'International evidence on the social context of
well-being'. In E. Diener, J.F. Helliwell & D. Kahneman (eds)
International Differences in Well-being, Oxford, Oxford University
Press.
Hill, J.E., Jacob, J.I., Shannon, L.L., Brennan, R.T., Blanchard,
V.L. & Martinengo, G. (2008) 'Exploring the relationship of
workplace flexibility, gender, and life stage to family-to-work
conflict, and stress and burnout', Community, Work & Family, 11
(2), 165-81.
Hochschild, A. R. (1979) 'Emotion work, feeling rules, and
social structure', American Journal of Sociology, 85 (3), 551-75.
Hochschild, A. (1983) The Managed Heart: The Commercialization of
Human Feeling, Berkeley, University of California Press.
Hook, J.L. (2010) 'Gender inequality in the welfare state: sex
segregation in housework, 1965-2003', American Journal of
Sociology, 115 (5), 1480-1523.
Hutri, M. & Lindeman, M., (2002) 'The role of stress and
negative emotions in an occupational crisis', Journal of Career
Development, 29 (1), 19-36.
Inglehart, R., Foa, R., Peterson, C. & Welzel, C. (2008)
'Development, freedom, rising happiness: a global perspective,
1981-2007', Perspectives on Psychological Science, 3 (4), 264-85.
Juster, F.T. & Stafford, F.P. (1985) Time, Goods, and
Well-being, Ann Arbor, Institute for Social Research, University of
Michigan.
Kahneman, D. & Deaton, A. (2010) 'High income improves
evaluation of life but not emotional well-being', Proceedings of
the National Academy of Sciences of the United States of America, 107
(38), 16489-16493.
Kahneman, D. & Krueger A.B. (2006) 'Developments in the
measurement of subjective well-being', The Journal of Economic
Perspectives, 20 (1), 3-24.
Kahneman, D., Wakker, P. & Sarin, R. (1997) 'Back to
Bentham? Explorations of experienced utility', Quarterly Journal of
Economics, 112 (2), 375-405.
Kroll, C. & Pokutta, S. (2012) 'Just a perfect day?
Developing a happiness optimised day schedule', Journal of Economic
Psychology, On-line first, DOI 10.1016/j.joep.2012.09.015.
Krueger, A. & Schkade, D.A. (2007) 'The reliability of
subjective well-being measures', Journal of Public Economics, 92
(8-9), 1833-1845.
Layard, P.R.G. (2005) Happiness: Lessons from a New Science, New
York, Penguin Press.
Leigh, A. & Wolfers, J. (2006) 'Happiness and the Human
Development Index: Australia is not a paradox', Australian Economic
Review, 39 (2), 176-84.
Maher, J.M., Lindsay, J. & Franzway, S. (2008) 'Time,
caring labour and social policy: understanding the family time economy
in contemporary families', Work, Employment and Society, 22 (3),
547-58.
Mellor, D., Stokes, M., Firth, L., Hayashi, Y. & Cummins, R.A.
(2008) 'Need for belonging, relationship satisfaction, loneliness,
and life satisfaction', Personality and Individual Differences, 45
(3), 213-18.
Multinational Time Use Study, Versions World 5.5.3, 5.80 and 6.0
(released October 2011). Created by Gershuny, J. & Fisher, K. with
Altintas, E., Borkosky, A., Bortnik, A., Dosman, D., Fedick, C.,
Frederick, T., Gauthier, A. H., Jones, S., Jun, J., Lai, A., Lin, Q.,
Lu, T., Lui, F. MacRae, L., Monna, B., Gimenez Nadal, J. I., Pauls, M.,
Pawlak, C., Shipley, A., Tinonin, C., Torres, N., Victorino, C. &
Yeung, O. Centre for Time Use Research, University of Oxford, United
Kingdom.
Offer, S. & Schneider, B.L. (2011) 'Revisiting the gender
gap in time-use patterns: multitasking and well-being among mothers and
fathers in dual-earner families', American Sociological Review, 76
(6), 809-833.
OECD (Organisation for Economic Co-operation and Development)
(2011). StatExtracts, complete databases available via the OECD library,
www.oecd.org (extracted 15 December 2011).
Oswald, A. (1997) 'Happiness and economic performance',
Economic Journal, 107 (445), 1815-1831.
Patulny, R. (2009) 'The golden years? Social contact amongst
retired men and women in Australia', Family Matters, (83), 39-47.
Patulny, R. (2012) 'The need for an emotion work survey',
paper presented to the Australian Sociological Association Conference,
Brisbane, 26-29 November.
Patulny, R. & Fisher, K. (2013) An implementation of an
experimental time use methodology: an 'Australianised
counterfactual' estimate of national subjective wellbeing in the
USA, Technical Paper 2012-01, Centre for Time Use Research, University
of Oxford.
Patulny, R & Wong, M. (2012) 'Poor mothers and lonely
single males: the 'essentially' excluded women and men of
Australia', Social Policy and Society, Available on CJO
doi:10.1017S1474746412000449.
Rice, J.M, Goodin, R.E & Parpo, A. (2006) 'The temporal
welfare state: a cross-national comparison', Journal of Public
Policy, 26 (3), 195-228.
Robinson, J. & Godbey, G. (1997) Time for Life: The Surprising
Ways Americans Use Their Time, University Park, Pennsylvania State
University Press.
Robinson, J.P. & Michelson, W. (2010) 'Sleep as a victim
of the "time crunch" --a multinational analysis',
Electronic International Journal of Time Use Research, 7 (1), 61-72.
Seligman, M.E.P., Steen, T.A., Park, N. & Peterson, C. (2005)
'Positive psychology progress: empirical validation of
interventions', American Psychologist, 60 (5), 410-21.
Sloane, P.J. & William, H. (2000) 'Job satisfaction,
comparison earnings, and gender', Labour, 14 (3), 473-502.
Stevenson, B. & Wolfers, J. (2009) 'The paradox of
declining female happiness', American Economic Journal: Economic
Policy, 1 (2), 190-225.
Svanberg-Miller, A. (2004) 'Gender differences in subjective
wellbeing: A cross-cultural perspective', Proceedings of the 6th
Australian Conference on Quality of Life, Deakin University.
Tetlock, P.E. & Belkin, A. (Eds.) (1996) Counterfactual Thought
Experiments in World Politics: Logical, Methodological, and
Psychological Perspective, Princeton, Princeton University Press.
Van Wanrooy, B. & Wilson, S. (2006) 'Convincing the
toilers? Dilemmas of working long hours in Australia', Work,
Employment and Society, 20 (2), 349-68.
Endnotes
(1) Time and affect studies have been conducted in the United
States (University of Michigan 1985 American's Use of Time Project,
Kahneman and Krueger's 2006 PATS--Princeton Affect and Time Survey,
a supplement of the Bureau of Labor Statistics 2010 American Time Use
Survey); United Kingdom (1987 Unilever Time Diary Survey and 2011 Five
Cities Project conducted by Trajectory with input from the Centre for
Time Use Research); and France (a supplement of one sub-sample of the
INSEE 2009-2010 Emploi du Temps, the French contribution to the second
round of the Harmonised European Time Use Study).
(2) There is contention over whether Kahneman and Krueger's
Day Reconstruction Method (DRM) of getting participants to recall and
complete a diary of the previous days' activities and emotions is
as robust as 'beeper studies', where participants
electronically record their activities and emotions in real time when
prompted by a beep or mobile phone app signal. Nevertheless, the DRM
does collect emotion data very close to the actual experience, and also
collects the full context of activities in sequence, unlike beeper
surveys. As the DRM involves a series of prompts which enable
participants to remember a great deal of contextual detail--where, what
and with whom they were doing things--which serve as useful prompts for
reducing ambiguity and increasing recall about how they kit. Further, a
key strength of the DRM is that is not contemporaneous. It is therefore
not subject to emotional response bias in recording emotions--such as
being too upset or angry to bother recording that one feels upset or
angry when randomly 'beeped'.
Table 1: Average minutes per day * spent by women and
men aged 18-64 in Australia and the USA on selected
activities on weekdays and weekend days, compared to
21 other countries **
Men
Australia USA
Paid work (5 of 23) (8 of 23)
Weekdays 7 hours 28 min 7 hours 20 min
Weekend days 2 hours 22 min 2 hours 7 min
Child care (1 of 23) (3 of 23)
Weekdays 29 min 25 min
Weekend days 40 min 28 min
Housework (8 of 23) (=6/7 of 23)
Weekdays 1 hour 45 min 1 hour 51 min
Weekend days 2 hours 51 min 3 hours 2 min
Volunteering (20 of 23) (=1/2 of 23)
Weekdays 8 min 20 min
Weekend days 12 min 46 min
Socialising (19 of 23) (15 of 23)
Weekdays 48 min 1 hour 19 min
Weekend days 2 hours 0 min 2 hours 35 min
Women
Australia USA
Paid work (12 of 23) (6 of 23
Weekdays 4 hours 20 min 5 hours 9 min
Weekend days 1 hour 8 min 1 hour 21 min
Child care (1 of 23) (3 of 23)
Weekdays 1 hour 28 min 1 hour 3 min
Weekend days 1 hour 16 min 47 min
Housework (=9/10 of 23) (20 of 23)
Weekdays 3 hours 39 min 3 hours 1 min
Weekend days 4 hours 13 min 4 hours 5 min
Volunteering (13 of 23) (2 of 23)
Weekdays 14 min 29 min
Weekend days 13 min 52 min
Socialising (17 of 23) (8 of 23)
Weekdays 56 min 1 hour 28 min
Weekend days 1 hour 55 min 2 hours 33 min
Source: Fisher & Robinson 2010
* These average time figures are weighted to
account for sample errors and to evenly distribute
days of the week. These means reflect time spent
per person across the whole population small time
in volunteering reflects the low participation
rate on any given day the original source contains
participation rates as well as average time per
day in activities.
** Countries covered in tables: Australia,
Belgium, Brazil, Bulgaria, Canada, Estonia,
Finland, France, Germany, Italy, Latvia,
Lithuania, Netherlands, Norway, Poland, Republic
of Korea, Slovenia, Spain, Sweden, Turkey, United
Kingdom, United States--the number of 23 in the
top of each cell reflects the rank in total time
commitment 1 represents the country where people
of that gender spend the most time in that
activity.
Table 2: Example PATS diary
Time Main Activity Where Who with
04:00-07:00 Sleep Home Partner
07:00-07:30 Shower, dress Home Alone
07:30:08:00 Eat breakfast Home Partner
08:00-09:00 Commute Train Strangers
09:00-12:00 Work Work Colleagues
12:00-13:00 Eat lunch Work Alone
13:00-17:00 Work Work Colleagues
17:00-18:30 Drink, socialise Pub Friends
18:30-19:00 Cook Home Partner
19:00-19:30 Eat dinner Home Partner
19:30-21:00 Watch TV Home Partner
21:00-21:30 Read Home Partner
21:30-04:00 Sleep Home Partner
Time Happy Sad Stressed U-index
04:00-07:00
07:00-07:30 4 2 2 0
07:30:08:00
08:00-09:00
09:00-12:00
12:00-13:00 5 1 1 0
13:00-17:00 3 2 4 1
17:00-18:30
18:30-19:00
19:00-19:30
19:30-21:00
21:00-21:30
21:30-04:00
Note: proportion of PATS episodes where negative emotion rating >
positive emotion rating > around 20%"
Table 3: Weighted OLS Regressions on unpleasant and pleasant time
Unpleasant Time
Country Country+ Gender
+ Gender + Socio-demographics
B S.E Sig B S.E Sig
Intercept 563.8 720.5
US men -106.0 4.7 *** -97.4 4.0
Australian women 23.0 4.5 *** 21.5 4.1 ***
Australian men 3.0 6.5 14.0 5.6 **
Single, no child 121.8 3.7 ***
Single, has child 19.6 6.2 ***
Couple, no child -17.3 3.6 ***
Aged 35 years -25.9 3.3 ***
or less
Aged 65 years 197.5 6.6 ***
or more
Employed 109.7 4.1 ***
Student -106.5 6.0 ***
Retired -132.6 7.3 ***
No secondary 119.0 3.9 ***
education
Post secondary -137.1 3.3 ***
education
N 24,996 24,996
Adjusted [R.sup.2] 0.04 0.31
Pleasant Time
Country Country+ Gender
+ Gender + Socio-demographics
B S.E Sig B S.E Sig
Intercept 354.1 171.4
US men 121.5 4.4 *** 107.4 3.6 ***
Australian women -14.1 4.3 *** 38.3 3.6 ***
Australian men -16.9 6.2 *** -28.1 4.9 ***
Single, no child 132.3 3.3 ***
Single, has child -29.5 5.4 ***
Couple, no child 7.4 3.1 **
Aged 35 years 3.1 2.9
or less
Aged 65 years 190.7 5.8 ***
or more
Employed 153.7 3.7 ***
Student 139.5 5.3 ***
Retired 156.5 6.4 ***
No secondary -129.9 3.4 ***
education
Post secondary 156.8 2.9 ***
education
N 24,996 24,996
Adjusted [R.sup.2] 0.05 0.41
Reference Category--US women, in couple with dependent child,
aged 35 to 65, working age but not in labour force, completed
secondary education but no further qualifications
* P < 0.05 ** P < 0.01 *** P < 0.001
