Domain importance and its impact on life satisfaction
by
Jeremy Tost
A
thesis submitted to the
in partial fulfillment of the requirements
for the degree
Masters of Arts
Major Subject: Psychology
June
2005
ABSTRACT
DOMAIN IMPORTANCE AND ITS IMPACT ON LIFE SATISFACTOIN
By
Jeremy R. Tost
Master of Arts
Dr. W. Larry Gregory, Chair
The
purpose of this study was to look at the relationship between domain importance
and life satisfaction. Of particular
interest was whether or not one’s satisfaction with a particular domain changes
as a function of the importance they place on that domain. Similarly the study attempted to understand
the effects that domain importance has on global life satisfaction (i.e.
assessments of life-as-a-whole) scores.
The results of this experiment were inconclusive in that the technique
used to manipulate domain importance was unsuccessful such that no significant
changes in domain importance were observed.
TABLE OF CONTENTS
LIST OF TABLES......................................................................................... v
INTRODUTION............................................................................................ 1
Assessment
of Life Satisfaction......................................................... 1
Domain
Relevance and How to Assess Importance........................... 4
Psychometric
Issues Associated with Counting Importance.............. 6
Empirical
Issues Associated with Counting Importance.................... 7
Value Orientation................................................................................ 8
METHODS..................................................................................................... 12
Participants.......................................................................................... 12
Materials & Measures......................................................................... 12
Design & Procedures.......................................................................... 14
RESULTS....................................................................................................... 16
Manipulation
Check............................................................................ 16
Domain
Importance Ratings............................................................... 18
Domain Importance Rankings............................................................ 18
Domain Specific Satisfaction Ratings................................................ 22
Overall
Quality of Life Ratings.......................................................... 26
Reliability............................................................................................ 27
Exploratory
Correlations...................................................................... 27
DISCUSSION................................................................................................. 29
General
Discussion............................................................................. 29
Integration
of findings......................................................................... 29
Limitations
and Direction for Future Research.................................. 31
Conclusion.......................................................................................... 34
APPENDICES................................................................................................ 35
A. MANIPULATIONS.................................................................................. 35
B. RECALL EXAMS.................................................................................... 62
C. MANIPULATION CHECK...................................................................... 72
D. QUALITY OF LIFE QUESTIONNAIRE................................................ 77
E. DOMAIN IMPORTANCE RATING QUESTIONNAIRE..................... 79
F. DOMAIN IMPORTANCE RANKING QUESTIONNAIRE................... 81
G. CONSENT FORM..................................................................................... 83
H. DEBRIEFING FORM.............................................................................. 85
REFERENCES............................................................................................... 87
LIST OF TABLES
Table
1. Summary of Planned Comparisons for Manipulation Check ........................ 16
2. Means and Standard Deviations for each Set of Manipulation Questions…. 17
3. Summary of Planned Comparisons for Domain Importance Ratings............ 19
4. Means and Standard Deviations for Domain Importance Ratings................. 20
5. Summary of Planned Comparisons for Domain Importance Ranking........... 21
6. Means and Standard Deviations for Domain Importance Ranking................ 23
7. Summary of Planned Comparisons for Domain Satisfaction Ratings............ 24
8. Means and Standard Deviations for Domain Satisfaction Ratings................. 25
9. Summary of Planned Comparisons for Overall Quality of Life Rating......... 26
10. Means and Standard Deviations for Overall Quality of Life Rating........... 27
11. Alpha Levels for Quality of
Life and Importance Measures........................ 28
INTRODUCTION
In recent years
quality of life research has been forced to deal with conflicting opinions
regarding the role of how best to measure an individual’s satisfaction with
life, in particular how to account for the importance one places on various
life domains. The majority of
theoretical speculation as well as empirical evidence agree that the inclusion
of domain importance is not only unnecessary but undesirable. Yet most researchers will admit that specific
domains do not carry equal weight. The
purpose of this study is to examine how a change in the amount of importance a
respondent places on a particular domain affects (a) their s with that
particular domain as well as (b) their overall life satisfaction.
Assessment of Life Satisfaction
Satisfaction with life
is but one way researchers have attempted to examine the greater issue of
subjective well-being. Aspects of
subjective well-being associated with first-person evaluations of their own
lives has come to be known as life satisfaction. There exist two opposing schools of thought
as to how best conceptualize and measure life satisfaction; those being the
top-down and bottom-up approaches. In
the top-down approach one’s personality and/or predispositional traits are seen
as having great influence over the way in which one reacts to events and
subsequently appraises his or her current state of affairs. The bottom-up approach asserts that global
life satisfaction (an assessment looking at one’s life-as-a-whole) can be
regarded as the sum of satisfaction ratings in various domains (Diener,
1984). Discussions and research
assessing both schools of thought can be found throughout the literature with
agreement lacking as to which approach is best (Feist et al., 1995; Headey et
al., 1991; Lance et al., 1989; Scherpenaeel & Saris, 1996). Being that this research is concerned with
assessments of life satisfaction and not the role of personality traits; the
focus of this paper shall accept the bottom-up approach. By making certain assumption regarding the
bottom-up approach this paper is neither trying to defend nor reject either
view. Of particular concern is how well
a global measure can be predicted by data obtained via individual domain
satisfaction ratings. The better we are
at predicting global satisfaction the more precision we have when
intervening. Campbell et al. (1976)
points out various scenarios in which specific interventions would be deemed
more appropriate. Consider the notion that a sense of global-well being depends
on the presence of some threshold number of satisfaction, or perhaps vital
domains need to be met first before pleasures associated with lesser domains
have any impact. It is also conceivable
that there exists a ceiling in which satisfaction derived from additional
domains fail to produce corresponding gains in the sense of well-being. An understanding of the relationship between
global and domain assessments, in statistical terms, means being able to
account for the greatest amount of variability in the global assessment.
The bottom-up
approach, by assuming that the sum of domain ratings amount to one overall
global rating, suggests that there should exist a reasonable correlation
between a global index (e.g. How do you feel about your life-as-a-whole?) and
discrete domain ratings (e.g. How do you feel about your health?). However this does not seem to be the
case. Correlations between these two
ratings belie researchers’ expectations that the two assessments should have a
strong measure of association. In one instance, when examining the relationship
between satisfaction scores for individual domains and that of a global
assessment of well-being (
Andrews and Withey, while attempting to obtain
the best measure of global well-being, describe sixty-eight different measurement
techniques (Andres & Withey, 1976).
Cummins suggests that there exist at least 173 different domain names
which have been used in past literature to delineate amongst various life
aspects, arguing that the number of possible relevant domains exceeds this
number (Cummins, 1996). With such an
extensive array of measurement devices, it seems unrealistic to expect a
perfect correlation between a global assessment and summed-domain
measures. Despite a lack of agreement
between these two techniques, it has become standard practice to measure global
life satisfaction by summing the values assigned to individual life domains (Andres
& Withey, 1976; Cummins, 1994; Flanagan, 1978). In defense of this practice is the agreement
amongst researchers that domains closest and most directly associated to
people’s personal lives will ultimately carry the greatest influence on their subjective
well-being (Andres & Withey, 1976; Campbell et al., 1976).
Domain Relevance and How to Assess Importance
The notion that
life domains could be unequal in their importance ratings is nothing new. Researchers have taken to label this concept
in varying ways. Campbell et al. (1976)
use “domain importance” while Andrews and Withey (1976) employ the terminology
of “single concern measures.” Other
instances include “value priority” (Inglehart, 1978), and “psychological
centrality” (Ryff & Essex, 1992).
In
order to create the most accurate assessment tool researchers have gone to
great lengths to ensure that the domains being rated have the greatest
relevance to the greatest number of people, while at the same time
excluding those domains considered to be
unimportant (Andres & Withey, 1976; Flanagan 1978). More recent research evaluating the relevance
of domains typically found in Quality of Life (QOL) surveys seems to indicate
that the domains included tend to be considered important and therefore
applicable to any respondent. While
working on the Comprehensive Quality of Life Scale (ComQol), researchers had
participants categorize 64 variables associated with different QOL aspects into
seven domains (Cummins, 1994). With 97%
of the variables being placed under the seven domain headings it was concluded
that “this provided verification that the domain headings together encompassed
the full range of QOL variables” (Cummins, 1994, p. 373). It has been suggested that since QOL scales
have selected domains on the basis of “consensus, convergence and breadth of
convergence”, these domains should be considered “sufficiently universal and
should be regarded as already having importance built in” (Chang-Ming, 2004, p.
165).
In contrast to the
above, it has been well documented that not all researchers feel domains are
and should be weighted equally (Trauer & Mackinnon, 2001). Having importance “built in” does not imply
an equal distribution of importance amongst selected domains. As an example, an individual with failing
health may consider health to be of greater importance than an individual in
perfect health. As of recent years a
solution to this predicament has been on the forefront of quality of life
research. In order to assess the
importance a respondent places on a particular domain, some QOL instruments
have included a line of questioning in which a value associated with importance
is assigned to a domain, using a standard 5 or 7 point likert scale. In other cases the domains are ranked in
order of their importance. These methods
of assessing importance do seem valid to the extent that they accurately
quantify the construct in question. When
comparing a direct measure of importance (mean importance rating) to an
indirect measure (regression coefficient for predicting for predicting reports
of well-being), a correlation of .41 was obtained (Campbell et al., 1976). When comparing importance ratings of
populations from 1971 and 1978, it was discovered that the order of importance
for domains changed very little. As
discussed by the researcher “Specific individuals depart dramatically from this
general pattern as the circumstances of their lives differ from the average,
but the pattern for the population at large was very stable through the decade
of the 1970’s” (Campbell, 1981, p. 49).
The problem researchers are faced with is not the extent to which their
assessment techniques are accurate, but rather what is to be done with
importance ratings. As shall be
discussed, there appears to be no clear cut way of integrating satisfaction
ratings with importance ratings.
Psychometric Issues Associated with Counting Importance
The most commonly
discussed problem associated with taking into account importance ratings stems
from the practice of multiplying importance by satisfaction (Andres &
Withey, 1976; Cummins, 2002; Trauer & Mackinnon, 2001). One theoretical flaw discrediting this
approach is based on the idea that likert scale data of this kind are
quasi-interval and not ratio (Cummins, 2002).
For a multiplication product to be meaningful the terms need to be
measured in ratio data (Schmidt, 1973).
In the case of importance and QOL ratings, there seems to be
considerable evidence indicating a lack of ratio properties (Stevens,
1957). By calling into questions the
legitimacy of the multiplicative composite obtained via multiplying importance
by satisfaction, it becomes extremely difficult to defend use of this
practice.
When we are
considering the practice of weighting domain importance we need to ask whether
or not changes in domain importance proceed in a linear function or rather
follow what others have suggested to be a curvilinear pathway. The implications associated with curvilinear
vs. linear pathways become apparent when conceptualizing notions such as the
“ceiling” and “threshold” (Campbell et al., 1976). Consider the variations that different
importance levels would have on a single domain satisfaction rating: do
abysmally low importance ratings have zero effect on domain satisfaction or is
there some minimal change, positive or negative, occurring? Correspondingly, can importance ratings grow
beyond a ceiling, such that the product obtained via multiplying satisfaction
by importance is undervalued? Through
data simulations researchers demonstrated the complexities associated with
correlating the product of two variables with a third, commenting that “the
exquisite sensitivity of the multiplicative composite to the scaling of its
components may surprise many” (Trauer & Mackinnon, 2001).
Empirical Issues Associated with Counting Importance
Ultimately by
measuring importance researchers are aspiring to improve the measurement
properties of QOL assessments. What has
been empirically demonstrated is that the inclusion of importance does little
to improve our understanding between global assessments and individual domain
assessments. Andrews and Withey reported both surprise and pleasure when they
concluded that “data about the importance people assign to concerns did not
increase the accuracy with which feelings about life-as-a-whole could be
predicted” (Andrews & Withey, 1976, p. 119). Campbell, Converse and Rodgers reported
similar findings that “taking account of the direct domain importance
ratings...fails to show any detectable increase in the power of the domain
satisfaction scores to explain variations in response to the Index of
Well-Being” (Campbell et al., 1976, p. 87 ).
When performing hierarchal regression with satisfaction entered as Step
1 and importance entered as Step 2, it was found that no significant extra
variance could be accounted for (Cummins, 2002). One study examining the correlation between
global life satisfaction and domain satisfaction found that relative domain
importance did not improve ones ability to make predictions regarding their
relationship (Chang-Ming, 2003).
Although, in the same study it was noted that domain ranking (as opposed
to rating) acted as a better indicator of overall life satisfaction than the
simple sum of domain satisfaction. “Using
domain ranking as a weighting mechanism clearly improved the correlation
between the single-item satisfaction measure and domain satisfactions” (Chang-Ming,
2003, p. 237). This confirmation, via a
novel approach to assessing importance, reveals that some measurable interaction
is shaping the relationship between the two variables of global assessors and
domain specific assessors. Further
research, including a replication of Chang’s finding, should indeed work
towards establishing a greater understanding of this phenomenon. Numerous alternative weighting schemes exist
(Chang-Ming, 2004; Hickey, 1996) demonstrating further methods of
life-satisfaction research in which variations of data collection tools and
mathematical procedures are employed.
Value Orientation
In order for
domain importance to be adequately conceptualized, an understanding of the
possible influence that importance may have on life satisfaction ratings needs
to be examined. To date this has yet to
be done. Research pertaining to domain
importance has focused instead on the benefits associated with counting
importance. By shifting our focus to the
role that domain importance plays on life satisfaction we can obtain insight
regarding whether or not such ratings can provide useful information, and if so,
how best to apply this information. When
considering importance one must also reflect on its changing nature. The
literature on value orientation offers empirical evidence that what we consider
to be “most important” changes as we progress through life. In the younger years of an individual’s life
financial security, housing and standard of living are of primary concern. As we approach 65 our values shift towards
life aspects regarding our own health as well as the health of those we are
close with (Bowling, 1995). In one study
in which attitudes toward various concepts (eg. “old age,” “future,” “life”)
was assessed it was found that “over-all agreement between the age samples was
high with regard to the evaluative rank-ordering of the concepts.” Approximately
one third of the concepts, however, yielded significant age differences (Kogan,
1967, p. 279). Health related quality of
life research is constrained by this factor of shifting responses to an even
greater extent considering the subjective nature of one’s physical
well-being. Describing this response
shift as ‘reframing,’ it is argued that “patients do not assess their health
against a fixed reference point (i.e. “true” baseline) but against a frame of
reference which shifts in the light of experience” (Bernhard, 2003). By demonstrating that the significance we
place on aspects of our lives is mutable and simultaneously failing to take
into consideration any causal implication associated with a change in one’s
quality of life, we may be inaccurately measuring life satisfaction.
If it can be shown that changing the
importance levels of one or two domains alters a person’s overall satisfaction
score then a reassessment of our understanding on importance needs to be
considered. Aside from Chang-Ming’s
(2003) finding that using domain ranking improved the correlation between
single-item global life satisfaction measures and domain satisfaction, all
other evidence rejects the utility of an importance measure. Yet that singular finding, along side years
of speculation that importance should matter, gives reason to believe there is
still more to be learned. QOL
researchers should be concerned with being able to fully understand the
complexities of global as well as domain specific assessments and how the two
relate. Statistically, the focus needs
to stress the task of accounting for the greatest possible variance in global
assessments. If importance is a relevant
factor, then it, and any other variable with influence, should show up under
scrutiny.
Instead of
focusing on the psychometric aspects of assessing domain importance, it is
proposed that a classic experimental approach be taken to determine if domain
importance is related to life satisfaction.
By increasing (or decreasing) the extent to which a participant values
the importance of life domains, in specific the domains of family relations,
material comforts, community involvement and helping behaviors, it is predicted
that changes in his/her individual global QOL scores will occur.
If it can be shown
that the manipulation of domain importance affects individuals’ ratings of
quality of life, then the implications for the literature are clear: domain importance is an important variable to
take into account when assessing overall life satisfaction. Continuing to determine the particulars of
assessing domain importance and incorporating domain importance into the
computation of life satisfaction scores can be left to others, but with the
clear implication that it should be done.
METHOD
Participants
Participants
for this study consisted of 141 undergraduates from
Materials and Measures
Prior to their involvement in this study the participants attended an instructional session in which they were explained the nature of the experiment. At that time they also completed a consent form along with three quality of life questionnaires. Over the course of a week, on three separate occasions, participants were asked to read twelve short passages and answer questions regarding the content of those passages (see Appendix A). The passages as well as the participant’s responses were transmitted via email. The participants were told that the purpose of the study was to examine their ability to recall certain types of information and how that recall would be affected by their quality of life. The reason for this fallacious explanation was to prevent the participants from inferring the true purpose of the experiment. Upon their return, the participants were given a short examination assessing their recall of the passages (see Appendix B). This was done prior to completion of the three quality of life questionnaires. The experiment concluded with a debriefing session in which participants were explained the true nature of the experiment. Additionally participants completed a questionnaire assessing the effectiveness of the manipulations. This manipulation check consisted of 5 questions for each domain manipulated. (See Appendix C)
1. In order to assess participants’ quality of
life Flanagan’s Quality of Life Scale (QOLS) was utilized. This 16 item questionnaire asks participants
to rate their level of satisfaction on numerous life domains, with the scale
ranging form 1(terrible) to 7 (delighted).
Overall quality of life is calculated by adding the satisfaction ratings
for the 16 items. (See Appendix D)
2. In order to asses the amount of importance
participants place on various life domains, a modified version of Flanagan’s
Quality of Life Scale was utilized.
Keeping the 16 domains constant, each participant was asked to rate how
important each domain is in their daily life with the scale ranging from 1
(indifferent/not important at all ) to 5 (extremely important). There was no overall level of importance
calculated, as individual domain importance was the construct being
evaluated. (See Appendix E)
3. Domain importance was measured in second way
in, again using a modified version of Flanagan’s Quality of Life Scale. Rather than rating the 16 domains on a 1-5
scale, the participants were asked to rank the domains from the most important
to the least important. (See Appendix F)
Design & Procedure
The
study utilized a randomized experimental-control group design where subjects were
assigned randomly to either the control (neutral passage), experimental A
(importance increasing passage), or experimental B (importance decreasing)
group. Random assignment was conducted
employing a random number table. The null hypothesis for the study states: there will be no significant difference
between the experimental and control groups on the ratings of the three
questionnaires. The independent variable
is the neutral or priming passages with the dependant variables being various
quality of life assessments as measured by three quality of life measures.
Prior
to being randomly assigned to either of the two experimental groups or control
group, each participant filled out the three questionnaires. At this time the
participants were also explained the instructions of the experiment and given
access to the experiment website.
The
day following the instructional section mentioned above, the participants
received an email containing four short passages along with several questions
regarding the content of those passages.
The control group of 47 participants read passages pertaining to
RESULTS
Manipulation Check
The first analyses
were a series of planned comparisons (Hays, 1973; Winer, 1971) on the
manipulation checks. For each life
quality domain (family, material comforts, volunteering, community) the five
manipulation check items were summed.
Sum scores were then used to perform four contrasts (one for each
domain) of the increase condition to the control condition, and four contrasts
of the decrease condition to the control condition. For the increase to control condition
comparisons, none of the four contrasts were statistically significant (see
Table 1 for a summary).
Table 1
Summary of Planned Comparisons for Manipulation Check
_______________________________________________________________
Neutral in comparison with
________________________________
Decrease Increase
_______________________________________________________________
Domain F
df p F
df p
_______________________________________________________________
Family 2.4 (1,141) ns <1 (1,141) ns
Material comforts <1 (1,141) ns 1.2 (1,141) ns
Volunteering 4.0 (1,141) .05 <1 (1,141) ns
Community 2.0 (1,141) ns 1.1 (1,141) ns
_______________________________________________________________
For
the decrease to control condition comparisons, the manipulation checks differed
for the volunteering domain but no others.
Inspection of the means (see Table 2) reveals that the participants in
the decrease condition had lower scores than those in the control
condition. This indicates that the
manipulation resulted in participants in the decrease condition placing less
importance on the four manipulated domains than participants in the control
condition. Likewise, scores for the
increase condition are higher than those in the control condition. This finding
indicates that the manipulations, while lacking the strength desired by the
researcher, were working in the manner intended.
Table 2
Means and Standard Deviations for each set of Manipulation Questions
___________________________________________________________________
Conditions
________________________________
Decrease Control Increase
__________________________________________________________________
Domain M SD M SD M SD
__________________________________________________________________
Material comforts 11.7 3.1 11.2 3.4 11.9 2.5
Volunteering 10.8 3.2 12.2 3.1 12.5 4.0
Community 8.7
2.9 9.6
3.1 10.3 3.9
____________________________________________________________________
Note. High scores represent a greater sense of
importance felt toward domains.
Domain Importance Ratings
A series of planned comparisons were then performed contrasting domain importance difference scores for the increase condition to the control condition, and again contrasting the decrease condition scores to the control condition scores. Difference scores were calculated by subtracting scores obtained during time one (baseline) from scores obtained during time 2 (follow-up). A positive difference signifies an increase in domain importance while a negative difference signifies a decrease in domain importance. None of the difference scores for the four domains differed significantly for either set of contrasts (see Table 3). Inspection of the means (see Table 4) reveal that the participants in the decrease condition did not have lower scores than those in the control condition; likewise participants in the increase condition did not have higher scores than those in the control.
Domain Importance Rankings
Recall that
participants were required to rank order the sixteen domains at baseline and
again at follow-up. Planned comparisons
were used to contrast the difference scores for the rank orders for the four
domains. Difference scores were
calculated by subtracting scores obtained during time one (baseline) from
scores obtained during time 2 (follow-up).
A positive difference represents an increase in importance ranking while
a negative difference represents a decrease in importance ranking. None of contrasts were significant (see Table
5)
Table 3
Summary of Planned Comparisons for Domain Importance Ratings
_________________________________________________________________
Neutral in comparison with
_______________________________
Decrease Increase
_________________________________________________________________
Domain F
df p F
df p
_________________________________________________________________
Family <1
(1,141) ns <1 (1,141)
ns
Material comforts <1 (1,141) ns <1 (1,141) ns
Volunteering <1 (1,141) ns <1 (1,141) ns
Community <1 (1,141) ns <1 (1,141) ns
________________________________________________________________
Table 4
Means and Standard Deviations for Domain Importance Ratings
__________________________________________________________________ Conditions
__________________________________
Decrease Control Increase
__________________________________________________________________
Domain M SD M SD M SD
__________________________________________________________________
Family
Baseline .7 .7 .7 .9 .6 .7
Follow-up .6 .6 .5 .8 .4 .5
Difference -.1 .7 -.2 .6 -.2 .6
Material comforts
Baseline 1.0 .8 1.3 .9 1.2 .8
Follow-up .9 .8 1.1 .8 1.0 .9
Difference -.1 .6 -.2 .8 -.2 .7
Volunteering
Baseline 1.4 .7 1.4 .9 1.4 .9
Follow-up 1.6 .8 1.6 1.0 1.5 .8
Difference
.2 .8
.1 .8 .1 .6
Community
Baseline 2.2 .9 2.1 .8 2.0 .8
Follow-up 2.2 .9 2.1 .9 2.1 .8
Difference .1 .8 .0 .9 .1 .9
__________________________________________________________________
Note. High scores represent a greater sense of
importance felt toward domains.
Table 5
Summary of Planned Comparisons for Domain Importance Rankings
__________________________________________________________________
Neutral in comparison with
_______________________________
Decrease Increase
_______________________________________________________________
Domain F
df p F df p
_______________________________________________________________
Family <1 (1,121) ns <1 (1,121) ns
Material comforts <1 (1,121) ns <1 (1,121) ns
Volunteering <1 (1,121) ns <1 (1,121) ns
Community <1
(1,121) ns <1 (1,121) ns
________________________________________________________________
Inspection of the means (see Table 6) reveals that the participants in the decrease condition did not rank the domains as less important compared to those in the control condition; likewise participants in the increase condition did not rank the domains as more important compared to those in the control.
Domain Specific
Satisfaction Ratings
A series of
planned comparisons were performed contrasting domain satisfaction difference
scores for each of the four domains comparing the increase condition to the
control condition and the decrease condition to the control condition. Difference scores were calculated by
subtracting scores obtained during time one (baseline) from scores obtained
during time 2 (follow-up). A positive
difference score represents an increase in domain satisfaction while a negative
difference score represents a decrease in domain satisfaction. None of comparisons were statistically
significant (see Tables 7 and 8).
Table 6
Means and Standard Deviations
for Domain Importance Ranking
__________________________________________________________________
Conditions
__________________________________
Decrease Control Increase
__________________________________________________________________
Domain M SD M SD M SD
__________________________________________________________________
Family
Baseline 3.4 2.3 3.2 2.9 3.3 2.5
Follow-up
2.9 2.0 2.9 2.2 2.7 2.0
Difference -.5 2.1 -.3 1.9
-.6 1.8
Material comforts
Baseline 5.7
4.0 5.9 3.9 5.6 4.0
Follow-up 6.2
4.2 6.9 4.0 6.5 4.2
Difference .5
4.4 1.0 2.8 .9 3.2
Volunteering
Baseline 10.3 2.1 10.3 2.8 10.1 3.0
Follow-up 10.9
3.8 10.8 3.3
9.9 3.5
Difference .6
3.7 .5
3.1 -.2 3.7
Community
Baseline 13.7 2.3 13.3 2.5 12.8 2.5
Follow-up 13.7 2.5 13.6 2.6 13.5 2.3
Difference .0 1.9 -.3 2.9 .7 2.5
__________________________________________________________________
Note. High scores represent a greater sense of importance felt toward domains.
Table 7
Summary of Planned Comparisons
for Domain Satisfaction Ratings
__________________________________________________________________
Neutral in comparison with
_________________________________
Decrease Increase
__________________________________________________________________
Domain F
df p F df p
__________________________________________________________________
Family <1 (1,141) ns 1.1 (1,141) ns
Material comforts <1 (1,141) ns 1.6 (1,141) ns
Volunteering < 1 (1,141) ns < 1 (1,141) ns
Community 1.2 (1,141) ns 1.2 (1,141) ns
__________________________________________________________________
Table 8
Means and Standard
Deviations for Domain Satisfaction Ratings
__________________________________________________________________
Conditions
_________________________________
Decrease Control Increase
__________________________________________________________________
Domain M SD M SD M SD
__________________________________________________________________
Family
Baseline 1.1 1.0 1.4 1.3 1.0 1.0
Follow-up 1.2 .9 1.4 1.3 1.2 1.0
Difference
.1
.8 .0
.7 .2 .8
Material comforts
Baseline 1.8 1.2 1.6 1.1 1.6 1.1
Follow-up 1.6 1.2 1.4 .9 1.3 1.0
Difference -.2 1.0 -.2 .9
-.3 .7
Volunteering
Baseline 1.5 .8 1.4 1.2 1.4 .9
Follow-up 1.9 .9 1.7 1.1 1.8 .9
Difference .4 .8 .3
1.1 .4
.8
Community
Baseline 2.1 1.1 2.4 1.3 2.2 1.1
Follow-up 2.3 .9 2.4 1.2 2.4 .9
Difference .2 .9 .0 .8 .2 1.3
___________________________________________________________________
Note. High scores represent a greater satisfaction with respective domains
Overall Quality of
Life Ratings
The last set of
analyses was two planned comparisons on the overall quality of life rating
scores. Difference scores were
calculated by first summing the quality of life ratings for each of the sixteen
domains for both baseline and follow-up, and then subtracting the total score
for time one (baseline) from total scores obtained during time 2
(follow-up). A positive difference score
represents an increase in overall satisfaction while a negative difference
score represents a decrease in overall satisfaction. Neither the contrast of the increase
condition to the control condition nor the contrast of the decrease condition
to the control condition was statistically significant (see Table 9 and 10).
Table 9
Summary of Planned
Comparisons for Overall Quality of Life Rating
__________________________________________________________________
Neutral in comparison with
_______________________________
Decrease Increase
__________________________________________________________________
Domain F
df p F df p
__________________________________________________________________
Overall QOL 1.5 (1,141) ns <1 (1,141) ns
__________________________________________________________________
Table 10
Means and Standard
Deviations for overall Quality of Life Ratings
__________________________________________________________________
Conditions
_________________________________
Decrease Control Increase
__________________________________________________________________
Domain M SD M SD M SD
__________________________________________________________________
Baseline 25.7 9.7 25.0 9.7 26.0 9.5
Follow-up 26.5 10.1 24.6 9.5 25.6 7.2
Change in QOL .8
6.6 -.6 3.9
- .4 5.5
_____________________________________________________________________
Note. High scores represent a greater satisfaction with life.
Reliability
A reliability analysis was conducted on both the domain
satisfaction scale and domain importance scale.
In both cases reliability was considerably high. (See Table 11)
Exploratory
Correlations
Correlations were conducted to examine the strength of
association between domain importance and satisfaction ratings. The importance rating and satisfaction rating
for each of the 16 domains correlated significantly at the .05 level with an
average correlation of .38. The
strongest correlation was found for the domain of close friend; .57, while the
weakest was for the domain of material comforts; .20.
Table 11
Alpha levels for quality of life and
importance measures
____________________________________________________________________
Measure Alpha Number of cases Number of Items
____________________________________________________________________
Satisfaction .80 125 16
Importance .74 125 16
____________________________________________________________________
DISCUSSION
General Discussion
The
results of this study did not support the prediction that changes in importance
would lead to changes in satisfaction. There were no significant differences
between the experimental and control groups on any of the three dependant
measures (domain satisfaction ratings, domain importance ratings and domain
ranking). There was also no significant difference in overall quality of life
between the increase condition to control or between the decrease condition and
control. The manipulation checks
revealed that of the four domains manipulated only volunteering (decrease
condition) differed significantly from the control. The remaining domains, when comparing either
increase or decrease to control, showed little difference.
Integration of findings
Due
to the manipulations inability to increase or decrease domain importance it is
quite difficult to infer any systematic relationship between importance and
satisfaction. This lack of difference
between the experimental and control conditions mutes any and all predictions
made prior. The importance ratings
obtained from the participants vary only slightly, not only across conditions
but over time also. Perhaps values and
beliefs associated with the domains examined are less malleable than previously
assumed, or at the very least, require a manipulation more powerful and
inspiring than brief readings. Similar
results were found concerning quality of life scores; where participants
reported consistent levels of satisfaction across conditions and time as well.
Despite
the manipulation’s inabilities to foster a change in importance, there exist
several findings of interest. An eyeball
comparison of the relative rank of domain satisfaction ratings to domain
importance ratings find that the absolute difference between the two is quite
small, such that domains that are low in satisfaction are similarly low in
importance and vice versa. An analysis
of the correlations between the satisfaction rating and importance rating for
each domain found significant results in each of the 16 cases, with an average
importance to satisfaction correlation of .38.
Intuitively, it would seem rather beneficial for one to place greater
value on those domains they are already quite satisfied with. If importance
does have a bearing on overall quality of life, then this tendency to value
that which is good in our lives should act as a morale booster. This idea fits well with past findings that
most people report a positive level of well being (Diener, 1996.)
The
two most noticeable exceptions to the finding that we value what we are
satisfied with occur for the domains of health and passive entertainment. The domain of passive entertainment (reading,
listening to music, observing entertainment) scored 2nd highest on
satisfaction yet 10th on importance.
Conversely, participants rated their health as something moderately
important (6th most important) but with a satisfaction rating of
11. Not surprising, learning and
attending school was ranked as the most important domain followed by family
relations, independence, understanding one’s self, and relationship with
significant other.
Limitations and Direction for Future Research
The
greatest limitation of the current study would certainly be the failure of the
manipulations to elicit any change in domain importance. When informally asked about the
manipulations, the participants responded that the passages did little towards
influencing their attitude or position towards the respective domains. While the passages selected for the study
appear to have been ineffective in swaying opinion, several respondents voiced
that written statements in general seem like an unlikely catalyst for attitude
change of this sort. Since the
post-manipulation assessment took place the day following the 3rd
email, the expectation of the passages to maintain their effect a full 24 hours
may have been asking too much. A more practical approach may have sought a
shorter term change with the assessment immediately following exposure to a
manipulation. By condensing the time
frame, the circumstances under which the participants read the passages and
complete quality of life questionnaires would have been kept more stable. In the current study participants read the
passages at their discretion under varying conditions.
As
discussed earlier, the difficulty associated with altering importance needs to
be critically examined. Past research has had little trouble in assessing
importance (Bowling, 1995; Kogan, 1967; Cummins, 1996) yet little, if any
research exists attempting to sway domain importance. Pilot research examining
possible methods of importance manipulation would be prudent first step. When
attempting to manipulate domain importance it is vital that domain satisfaction
be left untouched. This can become difficult
depending on the nature of the importance manipulation. A manipulation altering behavior may be
effective in changing attitudes associated with domain importance but may have
the undesired consequence of changing attitudes associated with satisfaction. A clear understanding of the relationship
between importance and satisfaction needs to come from research in which
satisfaction is sheltered from the stimulus manipulating importance.
An
approach that has been demonstrated in past research to have influence over
future judgments involves the active imagination of scenarios. Prior to the 1976 presidential election,
individuals were asked to imagine either Jimmy Carter or Gerald Ford winning
the election. Those individuals who
imagined Carter winning believed that indeed Carter would win, while those who
had imagined Ford winning believed that Ford would win (Carroll, 1978). The results of such research has been
explained by the availability heuristic; “A person is said to employ the
availability heuristic whenever he estimates frequency or probability by the
ease with which instances or associations could be brought to mind”( Tversky
& Khaneman, 1973, p. 208). The same
technique could be used to manipulate domain importance. Under such conditions participants would be
instructed to imagine scenarios in which their family is very important to them
or perhaps not important at all. The
hope in doing so would be to observe expected changes in importance. As
mentioned earlier, domain selection would need to be carefully thought through.
Pilot
research assessing which domains would be most susceptible to change would
provide researchers with an idea of which domains to manipulate. It is quite
likely that the domain of family and public service would differ in their
response to a manipulation. A reversal
of the methodology used in the current experiment in which satisfaction would
be manipulated while changes in importance observed could also potentially be
revealing. A longitudinal study in which
changes in importance and satisfaction are simply observed over a span of time
could be useful in understanding quality of life trends. Similarly, changes in importance and
satisfaction by culture or socio-economic background could be informative.
Although
the sample size in the current study was rather large, the observed power for
the manipulation checks was quite low.
The average observed power for the manipulation checks was .24, with a
range of .07 - .51. Observed power for
domain importance ratings were even lower, averaging .08, with a range of .05 -
.12. A power of .5 stands a 50/50 chance
of obtaining a significant F when
true significance exists. An agreed up
power of .8 has become a reasonable and realistic value for researchers in the
behavioral sciences (Keppel, 1991).
With such low power values, the chances of finding significant findings,
should they exist, becomes slim.
Conclusion
In conclusion, the present study failed to establish whether or not a change in domain importance is related to changes in domain specific satisfaction and/or global life satisfaction. Although the question of how to best incorporate domain importance into the computation of overall life satisfaction remains unanswered, the potential exists for future research that could further our understanding of quality of life assessment.