Subjective Wellbeing: Time-Use or Personal Factors As Key Determinants. (Sociological Data Analysis)

Abstract

Subjective Wellbeing (SWB) is measured in relation to two categories of predictors in an attempt to discover which category is more useful in determining levels of SWB. The categories examined are time-use factors and personal factors. Time-use factors encompass specific uses of time for leisure, non-leisure, and other activities, whereas personal factors encompass enjoyment of non-leisure, happiness, stress. There is currently theoretical support for both, with some analyses finding that time-use is an excellent predictor of SWB whereas other studies support personal factors as the better predictor. My findings support personal factors as the superior predictor of SWB, outperforming time-use as a predictor across multiple measures and by large correlation coefficient differences.

Introduction

“Life, Liberty, and the pursuit of Happiness” is a phrase from the American Declaration of Independence that does well to explain the wishes and hopes of many people globally. The freedom to choose and interpret this statement in one’s own way also means that people in the same situations may have a different feeling on how that affects them personally. This subjectivity is an important thing to remember in any situation regarding implementation or abolishment of policies or laws. Countries can measure their development on a number of different scales in order to assess what areas can be improved upon in the future. One important measure which I focus on in this paper is the measure of subjective wellbeing (SWB).

SWB measures the overall wellbeing of an individual’s physical health, mental health, and sense of belonging to a community in their own opinion. Understanding what plays a role in forming this opinion is of great importance to anybody interested in improving the conditions of their compatriots through direct action or implementation of new policy. It is certainly easy to assume that less working hours and having more time for leisure would be beneficial, but is it certain that these play a larger role than perceptions of these activities for individuals?

SWB is a combination of many things, and it is uncertain as to what predicts high levels of it. On the one hand, there is the potential that time-use plays a large role in affecting SWB, but it remains equally possible that SWB is more affected about what a person feels about their time-use rather than the specific time-use itself. Hence, this is the research problem I intend to tackle. What are the key determinants of SWB, specific time-uses or the personal factors such as enjoyment of activities or feelings of stress?

Literature Review

It will be prudent to attain a broad conceptual frame prior to delving in to the possible relations of time-use, enjoyment, and SWB. As such, the following is an array of different theories and empirical research connected to the above topics.

Both leisure and non-leisure are possible modes of time-use, coming from certain activities that take place over time. How we think these uses of time affect SWB depends on our theoretical framework. Those coming from the school of critical thought see leisure time as part of the capitalist mode of production, where it is devoid of the inherent use-value it once had for being wasted (Baudrillard, 1998, p.154). Time is yet another precious commodity for exchange, such as when one pays the additional monetary price at a convenience store in order to save and earn more of the time-commodity. The act of wasting time is impossible, according to Jean Baudrillard, since it is in actuality an exchange of our human labour time for time consumption (1998, p.155). Following this, leisure time is actually a “pressured performance” (Baudrillard, 1998, p.155) to ‘make the most of it,’ such that a person must say what they accomplished even on their apparently ‘free time’. Theodor W. Adorno offered a similar approach when he describes the nature of leisure time activities that are no longer attended for enjoyment, but for an obligatory response to a rational decision that this outing will be of socioeconomic value in maintaining ones status or making new connections or the like (1975, p.66).

Unfortunately, I think this may be a case of armchair-theorizing rather than an accurate description of life’s drudgery. While it may be the case that some social functions are attended for rational interests in improving one’s standing, I do not think that leisure time is a direct commodity exchange in the capitalist mode of production. Baudrillard’s point relies upon the assumption that time is something we can possess, rather than something that passes independently (1998, p.151-2). If time passes independently of our ‘spending it’, and human labour time is still valuable as a mere act of labour over that time, then leisure time is not inherently commodified and alienated from its use value. While the time over which we take our leisure could be brought to the market of exchange, it can also be used directly as a use-value. Time is not ‘earned’ at a convenience store, less of it simply passes when we go there, leaving more time to either commodify ourselves as a labourer or relax and take leisure time.

At any rate, a theory of time use as a commodity is one way to perceive the possible relation between time-use, enjoyment, and SWB. Spending time on leisure may be like spending money on health care, one way of exchanging our goods to achieve greater wellbeing. Indeed, many studies have shown that time-use does have a large impact on SWB, positively or negatively depending on how time is used. In Time Use: A Report of the Canadian Index of Wellbeing (2010), data was collected from a plethora of aggregated empirical studies regarding the effects that time use has for wellbeing. Their primary research questions revolved around differences of age changing which time-uses were primary determinants of wellbeing. Some examples are that social leisure was found to not be associated with wellbeing for any age group (Time Use, 2010, p.37), whereas active leisure time, for seniors, was found to be a positive indicator of wellbeing (Time Use, 2010, p.39). Their main conclusion was that for working age adults, a decrease in working long hours, or for seniors, stable active leisure and increased rates of volunteering, and for children in preschool that are still being read to by their parents, all helped to increase wellbeing. They also concluded that there are negative trends for decreasing wellbeing for working age people experiencing more non-standard hours, providing more care for seniors, and experiencing time crunch. Children experienced wellbeing decreases due to excessive screen time and teenagers suffered due to having less meals with parents. An interesting perspective brought to light by this report is that different amounts and types of activity have different relations to wellbeing across people of diverse age groups. This may be informative in later analysis of the dataset if a trend relating to age appears.

One dichotomy important to understanding SWB is that of top-down and bottom-up influences. In Leonardi, Spazzafumo, and Marcellini’s (2005) research on the predictive power of these influences, “top-down” influences are explained as personality traits, whereas “bottom-up” influences are the events and circumstances a person experiences. Their primary research hypotheses included the thought that top-down influences will be significant predictors of life satisfaction for long periods, bottom-up influences will be weaker than top-down as predictors, and that top-down influences will be a better predictor of life satisfaction than actual life conditions. Their sample consisted of 297 Italian subjects in two age groups of 55-75 and over 75, and data was collected via two structured questionnaires separated by five years. They confirmed their hypotheses and found that top-down effects were significant predictors of life satisfaction, whereas bottom-up effects were only moderate predictors of life satisfaction. Most interesting, though, was the confirmation of the third hypothesis, which means that top-down factors are better predictors of life satisfaction than even the life conditions that are thought to constitute it.

Kuykendall, Tay, & Ng (2015) will help to inform my theoretical approach as it pertains to the relation between leisure and enjoyment data. They discuss that leisurely activities can be analyzed under two frameworks, the structural or subjective. In the structural framework, leisure activities are measured for frequency and variability, but these activities are rigidly defined by researchers (Kuykendall, Tay, & Ng, 2015, p. 365). Conversely, the subjective approach includes the same measurements, but for any and all activities that individuals view as leisurely as well. The subjective approach coincides with my interest in also using data about other activities for respondents to GSS-19 that were not part of leisure activities in order to discover a relation to SWB. Their analysis went further to suggest that leisure participation affects SWB in relation to the degree of satisfaction that the participant gets from doing it (Kuykendall, Tay, & Ng, 2015). Their data collection method was a meta-analysis and aggregation of many time-use, enjoyment, and SWB studies to form longitudinal data for coding and comparison. They hypothesized that leisure participation is positively associated with SWB, that subjective definitions of leisure will be better correlated with SWB than structural definitions, and that SWB will be more strongly associated with leisure engagement when measured as a frequency rather than a quantity. As predicted, higher levels of leisure participation was correlated with higher SWB (r = .22), as well as frequency measures of leisure participation having a stronger relation to SWB (r = .28) than quantity measures with SWB (r = .08). Conversely however, structural definitions of leisure showed a stronger relation to SWB (r = .27) than did subjective definitions (r = .19). While there was certainly a relation present in the findings, the researchers also noted that many external factors played a large part in determining the effect that leisure satisfaction had on SWB (Kuykendall, Tay, & Ng, 2015). I do not plan on adopting this further degree of analysis, as I think level of enjoyment in particular activities is too fluid a category to be an effective way of determining their possible SWB effects.

The controversy that this brings up is between the level of theory and the level of experimentation. At one end of the theoretical spectrum, there is the possibility that all leisure activities are just embedded in the capitalist mode of production, and their performance is no different from alienated labour, and thus it does not improve our SWB (Baudrillard, 1998; Adorno, 1975). At the other end, theory developed from research suggests that leisure time is strongly correlated with SWB (Time Use, 2010) and that what we need to do is figure out the amounts, variety, and external factors that contribute to its effectiveness in increasing SWB (Kuykendall, Tay, & Ng, 2015).

While there is an established link of relation between leisure activity and SWB, there seems to be a lack of information about SWB and specifically subjective leisure, or what is in the GSS-19 as levels of enjoyment for non-leisure activities. By looking to see the effect that enjoyment of non-leisure activities has on SWB, we can compare it to structural leisure activity and discover if the same SWB benefits of leisure time can be attained by creating enjoyment in all of a day’s activities. If Leonardi, Spazzafumo, and Marcellini (2005) are correct, these benefits may be largely dependent upon individuals to adopt a happier disposition rather than policy changes at government or corporate levels to introduce more enjoyment or leisure times to increase SWB.

Data Analysis

Critical Discussion of Data Collection and Sample Selection

I used the General Social Survey (GSS) Cycle 19 on Time Use as the data source for answers to my research questions. GSS-19 was designed to gather information on the population of Canada that was 15 years of age or older around topics such as general demographics, time-use, well-being, and activities of respondents. Notably, excluded from the population intentionally were residents of Canada living in the Yukon, Northwest Territories, and Nunavut, full-time residents of institutions, and respondents under 15 years of age (GSS-19, p.10). Reasoning for these exclusions is absent, but I believe it is a weakness of a time-use survey to not include these Canadians who might have notably different patterns of time-use. In the three territories, for example, the availability of metropolitan leisure activities might be very limited, as might cable TV and internet access. As such, these populations likely spend their time in unique ways compared to other Canadians. Additionally, the researchers used Computer Assisted Telephone Interviews (CATI) as their method of data collection, thus there is no data for Canadians within the targeted population that do not have home phones. This is acknowledged and accepted by the researchers, explaining their assumption that this small portion of the population is likely not significantly different from the rest of the target population (GSS-19, p.10). CATI techniques are excellent for quickly attaining a large sample size, but participation is skewed towards those that are more likely to be home (homemakers, children, retirees, unemployed). While there is the benefit of the inexpensive nature of CATI surveys, it may be less effective at garnering a representative sample.

Aside from this, the target population appears to be carefully and representatively selected for GSS-19. General demographic information was used to weight collected data across different geographic, age, sex, and months of the year in order to make the total sample more representative of the target population. Respondents were divided into 27 geographical strata to account for specific population variations and to ensure that each of these strata was represented on the national-level proportionally (GSS-19, P.10). Of course, as with all samples, there is always the risk of sampling error inherent in taking a sample from the total population, but I do think adequate steps were taken and the size of the responding sample greatly reduce the likelihood of there being a major disparity between sample data and total population data. The quantity of the data will have the implication of making almost any result I find through analysis statistically significant, although they may be such minor values that they may be discounted meaningful. The quality of the data in GSS-19 is sufficient, with some recoding and work the data becomes quite manageable and useful for the application at hand.

The data contained in GSS-19 will be useful in examining links between the amount of leisure time or levels of enjoyment and subjective wellbeing (SWB) related to the target population of Canadians 15 years or older. Recorded in the data are answers to questions about stress, health, happiness, and life satisfaction that will all serve as excellent measures of SWB. Since this survey also had time-use journals, I will be able to retrieve data on the duration of leisure activity and see if there is a relation between the duration of these and SWB. Additionally, respondents reported that their enjoyment of certain tasks such as cleaning or childcare or shopping, which will allow me to analyze if getting enjoyment out of daily activities is more or less important than leisure time as an indicator of SWB.

Research Questions

In light of the literature review above, I constructed one univariate question and nine bivariate questions to elucidate the relationships between higher levels of SWB and theoretically grounded possible predictors. My guiding question is how the uses and perceptions of time for individuals affect their SWB, thus leading to the questions below.

1. What are the levels of (SWB) for the sample?
2. To what extent does enjoyment determine levels of SWB?
3. To what extent do top-down influences determine levels of SWB?
4. What is the relationship between the experience of time crunch and levels of SWB?
5. What is the relationship between the experience of stress and levels of SWB?
6. To what extent does the importance of religion determine levels of SWB?
7. To what extent does leisure participation determine levels of SWB?
8. To what extent does non-leisure participation determine levels of SWB?
9. To what extent does child-care participation determine levels of SWB?
10. To what extent does sleep participation determine levels of SWB?

Conceptualization and Operationalization of Key terms

The definition used by Brooker and Hyman in Time Use: A Report of the Canadian Index of Wellbeing for wellbeing is informed by a definition of health used by the WHO, which reads as “a state of complete physical, mental and social wellbeing and not merely the absence of disease or infirmity” (Time Use, 2010, p.9). They go on explain that this means wellbeing can encompass bodily health, mental health, and the social relations a person has to their community. I believe this to be a good definition of wellbeing, highlighting the multifaceted nature of the concept. For my purposes, I tack on subjective to the front, bringing some of the more obvious implications with it. Subjectivity is about internal perceptions and understandings, so in the case of SWB I take it to mean a respondent’s self-assessment of the dimensions of wellbeing as it pertains to them. My outcome measure of SWB is calculated across eight different variables which are combined to form a single index evaluating participants’ total level of SWB (subjective well-being). The range of my index is from 0 to 32, with each variable contributing from 0 to 4 points. The eight variables measure their subjective opinion on their own health, mental health and life satisfaction, and connection to their community. In this way, I capture the three primary determinants of wellbeing as rated subjectively by respondents, thus arriving at SWB. This should allow for a high level of convergent, content, and face validity. This measure should also have high test-retest reliability, as a similar importance is placed on these concepts in Time Use (2010) for a comprehensive measure of SWB.

Leisure as a structural measure considers frequency and variability of activities that are defined as leisure by researchers (Kuykendall, Tay, & Ng, 2015). These standardized leisure activities will provide an important point of comparison against enjoyable non-leisure, and thus for leisure I am conceptualizing it in this structured way. Activities such as sports, games, socializing, and relaxation will be considered as “leisure activities”. However, in my research I am assuming that the type of leisure performed is unimportant; I am assuming that person x gets a similar amount of enjoyment out of activity A that they pursue as leisure as person y gets out of

activity B that they pursue as leisure. For this reason, I only examined duration of activities, and did not consider variety as an important factor. By using the available data on activity participation by the minute through time diaries, I obtained a ratio measure of time use for participants that starts from zero minutes of leisure activities and continues upwards. With this measure, concurrent validity was achieved, as determining between participants with low as opposed to high amounts of leisure participation was very easy. It also has good content validity by measuring the frequency dimension of leisure much more accurately than many other studies that employ number of occurrences rather than duration of time, with decent face validity. Again, Time Use (2010) employed a measurement system that uses duration rather than occurrences, increasing test-retest reliability.

For a measurement of non-leisure participation, I drew on the subjective definition of leisure, which adds to the structural definition an inclusion of any activities deemed to be leisure by the subject (Kuykendall, Tay, & Ng, 2015). Thus, I included in my enjoyment measure only those activities that would not standardly be deemed leisure in the structural definition. My indicator for this concept was activities such as cooking, cleaning, commuting, working and other things typically seen as labour that are rated as enjoyable by some respondents. I examined how much time these respondents used on these activities from the time-use diary to get a measure of the duration, thus operationalizing the time spent on non-leisure activities in minutes from zero and up. As above, this is a ratio measure with concurrent, content, and face validity. Test-retest reliability comes from Brooker & Hyman (2010) who also analyzed non-leisure participation as a key indicator for wellbeing.

For the measurement of enjoyment of daily activities, I used the same subjective definition of leisure, which adds to the structural definition an inclusion of any activities deemed to be leisure by the subject (Kuykendall, Tay, & Ng, 2015). My indicator for this concept was activities such as cooking, cleaning, commuting, working and other things typically seen as labour that are rated as enjoyable by some respondents. I examined how much enjoyment respondents received from a combination of these activities from “0 – Dislike a great deal” to “4 – Enjoy a great deal” from the GSS-19, thus operationalizing the level of enjoyment received from non-leisure. As above, this is a ratio measure with concurrent, content, and face validity. Test-retest reliability comes from Kuykendall, Tay, & Ng (2015), where they too employed a measure of enjoyment in their term “leisure satisfaction” (p.366) and used a combination of structural and subjective approaches in defining the amount of satisfaction, or enjoyment, respondents got from activities.

The term ‘time crunch’ is used to “describe the state in which respondents feel rushed or feel that they do not have enough time in a day to do what they need to do” (Time Use, 2010, p.26). GSS19 had a bank of questions that is meant to encapsulate this concept and measure it on a 10 point scale (0 being minimal time crunch, 10 being high time crunch). As it is a measure already developed and one that I did not see any difficulties with (conceptually or for later analysis), I chose to adopt the same scale. Obviously, the test-retest reliability is high, as my main data source (GSS19) as well as a major theoretical base for my research (Time Use, 2010) both employ the same measure. It should have the ability to differentiate those participants stressed by time constraints from those who are not, and thus has good concurrent validity. Once again, I believe that the measure also has high face validity, though some of the questions (eg. Do you consider yourself a workaholic?) do not seem necessarily related to time crunch.

Top-down influences are explained as personality traits that affect our perceptions of events (Leonardi, Spazzafumo, & Marcellini, 2005). In much of the literature, top-down influences are reported as major factors in determining SWB (Kuykendall, Tay, & Ng, 2015; Leonardi, Spazzafumo, & Marcellini, 2005). The GSS19 did not offer much in the way of personality traits, but it did have one theoretically important measure: happiness. According to Kuykendall, Tay, and Ng (2015) “happiness is a consequence of global and relatively stable features of personality” (p.367) meaning that there is some ground for my use of it as an indicator of personality traits for top-down influences. I operationalized top-down influences using happiness as a proxy, placing respondents into a 5 position scale from Negative top-down influences (Very Unhappy) to Positive top-down influences (Very Happy). Participants in the Positive top-down influences group are being interpreted as having a positive demeanour, which is often associated with higher SWB, and participants in the Negative top-down influences group are being seen as the reverse. This measure is limited in the aspect of content validity, as it takes a single report of happiness to be indicative of larger personality traits, rather than a broader scope of indicators. Due to the theoretical connection made by Kuykendall, Tay, and Ng (2015), I do think that it at least has face validity as the best available data for the concept in GSS19. I am unaware of other research that has a single variable meant to represent personality traits, so there may be limited test-retest reliability for this operationalization.

Description of Sample and Sub Sample

The following is an analysis of the general socio-demographic profile of the sample and sub-sample of my research. This serves the research in at least two ways. First, it allows us as researchers to examine how the sample varies from the Canadian population as a whole, thus ensuring that it is known about whom any discoveries may apply or be predictive for. Second, it allows for the discovery of particular differentiators that may point out key socio-demographics that exhibit particularly high SWB, potentially as a guidepost for future researchers to use in attaining a sample for research concerning individuals with high SWB.

Sample

Personal

Respondents to GSS19 (n=19597) ranged from 15 years of age to 75 years and over with the largest percentages of participants falling into the “35 to 44” category (19.7%), followed closely by the “45 to 44” category (18.6%). There is a small positive skew in the curve, which is to be expected in population graphs. The sex of respondents shows that 56.0% reported as female with the remaining 44.0% reporting as male. There were no alternative sex options for respondents that might have felt to be outside the sex dichotomy of the study, yet all respondents still answered the question.

Cultural

Respondents answering about their place of birth (n=19368) indicated that a vast majority of them were born in Canada (84.2%) with the remaining percentage (15.8%) indicating another location as their place of birth. The household language of participants (n=19380) was primarily English only (73.6%), with French only constituting the majority of the remainder (18.5%), while Other languages account for just 7.9% of household languages. I recoded the measure of religious attendance to include those that reported as non-religious, and found that on a scale of “No attendance” to “At least once a week”, the largest proportion of respondents (n=19208) reported that they do not attend religious gatherings (36.3%). Those that went a few times a year or at least once a week were the next largest groups, at 20.9% and 20.8% respectively. Thus, accounting for all attendance, more participants attend some gatherings (63.7%) than none at all.

Class

A quarter of the participants did not answer a question about their income, another frequent occurrence in surveys, leaving n=14689 for analysis. This is often assumed to be due to cultural taboos around talking about personal income in Canada. Of this number, respondents answered on a range of “no income” to “$100,000 or more”. Participants earning from $20 000 to $39 999 accounted for 30.0% of responses, with a noteworthy spike in the far-left “No income” response representing 8.0% of responses. Educationally, respondents (n=19341) were placed in categories from “No high school diploma” to “Masters/doctorate/Some grad school” after recoding. By and large, they were fairly educated, with 41.3% having completed at least a post-secondary diploma, certification, or Bachelors degree. Conversely, 21.2% of respondents did not finish their high school diploma, which may be a cause for concern.

Family

In terms of family at home, most respondents (68.6%) did not have children living in the home any more, and those that did typically had only 1 or 2 children at home (13.7% and 12.8%, respectively). After recoding for clarity, most respondents reported as married (54.5%), followed by those who have never been married (25.9%), leaving a substantial amount of people (19.7%) who have been married but have been divorced or widowed. Most respondents live in large urban centers (75.1%), leaving a minority that live in rural areas or small towns (24.9%).

Comparison to Canadian Population

Comparing the sample to the general Canadian population as reported by Statistics Canada, we see that the age of respondents is representative, while females are being overrepresented in the sample by roughly 5% more than their proportion of the Canadian population. Immigrants are slightly under-represented in the sample, as they constitute 19.8% of the population but only 15.8% of the sample. Language representation in the sample dramatically over-represents English representation which is around 58.1% in the population and underrepresents other languages which may account for up to 20.2% of languages primarily spoken at home. Considering religion, Statistics Canada provides information on the religion of participants but not their participation, making exact parallels difficult. A small percentage of the Canadian population reports having no religion (16.2%), which is significantly lower than the amount of people reporting that they do not attend religious gatherings in the survey (36.3%). It is quite possible that many who do report that they are religious do not attend religious gatherings, but the potential that there is a difference between the sample and the Canadian population on religious attitudes remains. Income and education both seem relatively well represented compared to the Canadian population, differing by only a few percentage points or less. Married and no longer married are both over-represented in the sample, as the numbers for the Canadian population are 46.2% single (25.9% in the sample), 40% married (54.5% in the sample), and 13.8% divorced and widowed (19.7% in the sample). The urban and rural divide is similar to the general Canadian population.

Sub-sample

Personal

My subsample is those that had a high measure of SWB as evaluated on a scale out of 32 that I created out of 8 scale questions from 0 to 4 in the survey. I am defining the high SWB subsample as those respondents that scored a 29 or higher on the SWB  measure. The subsample n=1982 is primarily in the 55-64 and 65-74 age groups, representing 21.6% and 19.6% respectively. This age group of the subsample is negatively skewed, with the 15-24 age group standing out by representing only 9.5%. Females represent 57.4% of the subsample with males representing the remaining 42.6%.

Cultural

The country of birth of respondents was primarily Canada at 87.1%, meaning only 12.9% of respondents are immigrants. Spoken home languages were as follows: English represents 81.9%; French represents 13.3%, and other languages represent 4.8%. I recoded the measure of religious attendance to include those that reported as non-religious, and found that on a scale of “No attendance” to “At least once a week”, the largest proportion reported that they do attend religious gatherings at least once a week (32.7%). Those that did not attend religious gatherings at all made up the next largest category (25.1%), followed by those that attended a few times a year (20.8%).

Class

Of the subsample, 31.3% did not answer a question about their income, another frequent occurrence in surveys, leaving n=1361 for analysis. This is often assumed to be due to cultural taboos around talking about personal income in Canada. Of this number, respondents answered on a range of “no income” to “$100,000 or more”. Participants earning from $20,000 to $39,999 accounted for 28.2% of responses, with a noteworthy spike in the “$60,000 to $79,999” response representing 10.3% of responses. Educationally, respondents were placed in categories from “No high school diploma” to “Masters/doctorate/Some grad school” after recoding. By and large, they were fairly educated, with 37.8% having completed at least a post-secondary diploma, certification, or Bachelors degree. Conversely, 24.4% of respondents did not finish their high school diploma, which may be a cause for concern.

Family

In terms of family at home, most respondents (76.4%) did not have children living in the home any more, and those that did typically had only 1 or 2 children at home (10.8% and 8.8%, respectively). After recoding for clarity, most respondents reported that they were married (62.6%), followed by those who had been married but had been divorced or widowed at 19.9%, leaving those who had never been married at 17.5%. About two-thirds of respondents lived in large urban centers (67.5%), with the other third (32.5%) that lived in rural areas or small towns.

Comparison to Canadian Population and Total Sample

The subsample is notably older than the sample or the general population, being most heavily represented from 55-74 instead of 35-54, a 20 year age difference. The proportion of males and females in the subsample is only slightly further from the ratio in the Canadian population than the total sample is, at 1.4% more females and 1.4% fewer males than the total sample. The subsample extends the total sample’s slight under-representation of immigrants in the Canadian population, representing only 12.9% of the subsample compared to 19.8% of the population. The subsample also contains the over-representation of English-speaking people and under-represents those that speak other languages in the Canadian population. Religious participation in the “At least once a week” category is noticeably larger in the subsample than in the total sample, having an increase of 11.9 percentage points. Income remained fairly representative of both the sample and the total Canadian population, but did have a notable negative skew in comparison, with more high-earners than the other groups. In education, however, there is a noticeable decrease in attainment. In the subsample, only 37.8% of participants have completed some degree, diploma, or certificate, opposed to the 44.2% in the general population a difference of 6.4 percentage points. There is also a slightly higher percentage of respondents that have no high school diploma, 24.4% compared to the population total of 23.4%. Married and no longer married are both over-represented in the sample, as the numbers for the Canadian population are 46.2% single (17.5% in the subsample), 40% married (62.6% in the subsample), and 13.8% divorced and widowed (19.9%% in the subsample). The urban and rural distribution remains similar in subsample as well. The subsample also has 7.8 percentage points fewer occurrences of children living at home than the total sample with 76.4% reporting no children at home in the subsample opposed to only 68.6% in the total sample.

Results and Analysis

On a scale ranging from 0 to 32, a higher SWB index score indicates a higher amount of SWB that the respondent has. With a mean of 22.7, a median of 23.0, and a mode of 24.0, the largest pool of respondents (9.1%) scored 24 on the scale, with a standard deviation of 4.97. Only 25% of cases appear below a 20 on the index, which is 4 points higher than the index’s midpoint, showcasing the negative skew of the distribution. This means that more respondents view themselves as having medium-high to high SWB than would be expected on a non-skewed distribution. See Figure 1 below for a visual representation.

Spearman correlation coefficient is a measure that is well-suited to measuring how ordered or continuous measures are related to one another. In light of my outcome and predictors all being continuous data I decided to make use of the Spearman correlation to answer my research questions. I have not used Pearsons’s r due to the non-normal distribution of my outcome and several predictors, which the Spearman correlation (ρ) is better at handling.

Figure 1 – Histogram of SWB Outcome measure

Proportion of responses to Subjective Wellbeing Index

Regarding my research question about the relation of enjoyment and SWB, I found that SWB was significant and moderately positive correlated with the enjoyment measure, indicating that those who find more enjoyment in their daily tasks experienced higher SWB (ρ (15546) = .286, p < .001). I found that there was a strong and significant positive correlation between SWB and the top-down influence measure, showing that increased happiness did play a role in higher SWB, answering my question about the effect of top-down influences on SWB levels (ρ (18666) = .467, p < .001). Concerning the question about time crunch and SWB, those who felt more time pressures had lower SWB, as the correlation between SWB and the time-crunch measure was moderate, significant, and negative (ρ (18065) = -.326, p < .001). SWB shared a similar correlation with stress, significant and moderately negative, meaning that as stress increases, feelings of SWB decrease, thus answering my question about stress and SWB (ρ (18662) = – .332, p < .001). Answering the question regarding religious importance, I found there was a weak to moderate significant correlation of religious importance with SWB, which showed a positive relation between higher SWB and higher feelings of religion’s importance in life (ρ (18487) = .161, p < .001).

The remaining bivariate questions showed very weak relations to SWB which may have been in part to the high amount of valid responses in the “0 minutes spent performing this activity”, being the modal response of leisure participation, non-leisure participation, and child-care participation. Sleep participation was actually almost normally distributed, but still appeared as a very weak correlation. At any rate, using the Spearman correlation I discovered that SWB and leisure participation were significantly correlated, and showed a weak and positive connection (ρ (18729) = .052, p < .001). To answer the question regarding leisure participation and SWB, these findings mean that higher amounts of leisure participation were weakly correlated to higher levels of SWB. Non-leisure participation, child-care participation, and sleep participation all shared a weak and negative significant correlation to SWB ((ρ (18729) = -.030, p < .001), (ρ (18729) = -.049, p < .001), (ρ (18729) = -.021, p = .002), respectively). This shared negative correlation means that my questions regarding non-leisure participation, child-care participation, and sleep participation are answered by saying that higher levels of these concepts were correlated to decreased levels of SWB. While getting a bit more leisure time and a little less working time may expectedly help increase SWB, child-care time and sleep-time surprisingly may lower SWB. Figure 2 below has all the above data in table format.

Figure 2 – Spearman Correlation (ρ) between SWB and 9 predictors

Discussion

In regards to my first research question, it appears that well over 75% of respondents view themselves as more than middle-high to high levels of SWB. Those in particular exhibiting higher levels of SWB are 55 years of age and over, non-immigrants, higher-earning and regularly attending religious gatherings at least once a week. For the high SWB population, the increase in age may be related to the decrease in children living at home and to the likelihood of retirement as child-care and non-leisure participation were both negatively correlated with SWB. Additionally, the retired also often experience less time-crunch, so it seems quite possible that the lack of that significant negative predictor would also play a role in making the high SWB population much older than the sample. Religious attendance, another predictor that was positively related to SWB, was also demonstrated in the high SWB subsample.

Research questions 2 and 3 about the relation of enjoyment and top-down influences to SWB was answered as positive and moderate-to-strong, showing that personal factors like happiness and enjoyment of daily activities are strong predictors of SWB, supporting Leonardi, Spazzafumo, & Marcellini (2005). This is important because it means that changes in how certain activities are represented may do more to increase SWB than decreasing the duration of undesirable tasks.

Both time-crunch and stress were moderate-to-strong negative factors for SWB levels, meaning that once again personal factors about daily life are shown to be good predictors of SWB. This supports the findings in Time Use (2010) that feelings of time crunch and stress would be negative factors in SWB. Episodes of non-standard hours and elderly care can indeed be stressful, and this relation shows the same adverse affects on SWB. Once again, changing personal perspectives about what you might be reasonably expected to do in a day might be more effective than actually changing what people do in a day. Additionally, changing the opinion of elderly care would be a significant change in the quality of life for senior citizens, but could also raise SWB or caretakers.

Religious importance was a moderate positive determinant of SWB, countering Baudrillard’s (1998) theory that time was yet another commodity for exchange. This use of time is standardly not for economic gain, typically a very personal use of time in which money is sometimes even given away. Adorno (1975) may be on the right track if his position is that time-use is spent on obligatory socioeconomic betterment, as religious people still typically do better politically than agnostic or atheists in North American contexts.

Specific time use for leisure was positively correlated with SWB, but only marginally so. While this positive relation was also found in Kuykendall, Tay, and Ng (2015), it is to a much smaller amount. This question whether leisure participation plays as large role in SWB as they predicted. However, my measure was quantitative rather than a frequency, which they proved lead to lower correlation values. The differences between my results and theirs are much less when comparing only the quantitative results. Additionally, that result also pushes me to consider that the other quantitative results would have shown higher correlations if measured as frequencies instead. This may be meaningful for other researchers to perform analysis in frequency measures concerning time use to better understand which factors play a larger role.

Non-leisure, child-care, and sleep participation were all minutely negatively correlated with SWB. The exceedingly low values of correlation may mean that time-use factors are not as important determinants of SWB as personal feelings are. They do uphold theoretical underpinnings in Time Use (2010) that stress and time crunch are negative factors for SWB, as non-leisure and child-care participation typically apply stress and time crunch. Sleep is a more interesting case, as the effects of a good night’s sleep are well documented. I theorize that this may have a connection to feelings of time-crunch (ie. sleeping in feels like wasting time) which may be in line with Baudrillard’s (1998) ideas about making the most out of pressured free time.

Conclusion

Implications

            The implications of this study are that any attempts to improve SWB through policy or legislature should strongly consider propaganda campaigns as a more effective means of increasing SWB than freeing up more time for leisure. My results showed that by and large, personal factors such as happiness, enjoyment of non-leisure, feelings of stress and time-crunch were all more powerful predictors of SWB than specific uses of their time.  This answers my guiding research question about which kinds of predictors are better for predicting levels of SWB. My findings support the conclusion that personal factors serve as strong predictors of SWB and that time-use predictors are only minor predictors. Leonardi, Spazzafumo, and Marcellini’s (2005) research showcases very similar results, and the results of top-down effects are mirrored in my study.

Limitations

Limitations of my study are the use of quantitative rather than frequency measures that possibly hampered the results of time-use data showing stronger correlations. Another limitation may be that as opposed to the longitudinal and meta-analysis data collected by some of the articles and books I reference, I focus on a single data set. Comparatively, their data aggregates many sources and thus may be more widely applicable than my own.

Implications for Future Research

The implication for future research is a strong encouragement to use frequency measures of time use, perhaps in conjunction with quantitative data. This will have a higher likelihood of demonstrating stronger predictors out of time-use data. More research needs to be done in the field of the relation of sleep to SWB, as the result of a negative correlation seemed very surprising given the known health benefits of proper rest.

Works Cited

Adorno, T. W. (1975). Work and pleasure. Gesammelte Schriften Band 9.2: Soziologische
Schriften II.II, 60-67.

Baudrillard, J. (1998). The Consumer Society: Myths and Structures. London, England: Sage.

Brooker, A. S., & Hyman, I. (2010). Time use: A report of the Canadian index of wellbeing. Toronto, ON, CAN: Canadian Index of Wellbeing, ProQuest ebrary. Web. 25 September 2015.

Kuykendall, L., Tay, L., Ng, V. (2015). Leisure engagement and subjective well-being: A meta-analysis. Psychological Bulletin, 141, 364-403. http://psycnet.apa.org/doi/10.1037/
a0038508

Leornardi, F., Spazzafumo, L., Marcellini, F. (2005) Subjective well-being: The constructionist point of view. A longitudinal study to verify the predictive power of top-down effects and bottom-up processes. Social Indicators Research, 70, 53-77.