OMEGA, Vol. 61(2) 103-120, 2010
AN EXAMINATION OF STAGE THEORY OF
GRIEF AMONG INDIVIDUALS BEREAVED
BY NATURAL AND VIOLENT CAUSES:
A MEANING-ORIENTED CONTRIBUTION*
JASON M. HOLLAND
VA Palo Alto Health Care System and
Stanford University Medical Center
ROBERT A. NEIMEYER
University of Memphis
Despite its popularity, few attempts have been made to empirically test
the stage theory of grief. The most prominent of these attempts was con-
ducted by Maciejewski, Zhang, Block, and Prigerson (2007), who found that
different states of grieving may peak in a sequence that is consistent with
stage theory. The present study aimed to provide a conceptual replication
and extension of these findings by examining the association between time
since loss and five grief Indicators (focusing on disbelief, anger, yearning,
depression, and acceptance), among an ethnically diverse sample of young
adults who had been bereaved by natural (n = 441) and violent (n = 173)
causes. We also examined the potential salience of meaning-making and
assessed the extent to which participants had made sense of their losses.
In general, limited support was found for stage theory, alongside some
evidence of an “anniversary reaction” marked by heightened distress and
*The views expressed here are the authors’ own and do not necessarily represent the views
of the Department of Veterans Affairs or the United States Government.
? 2010, Baywood Publishing Co., Inc.
reduced acceptance for participants approaching the second anniversary of
the death. Overall, sense-making emerged as a much stronger predictor
of grief Indicators than time since loss, highlighting the relevance of a
Several grief theorists have postulated that grief tends to proceed along a series
of predictable stages (Bowlby, 1980; Jacobs, 1993; Kübler-Ross, 1969; Kübler-
Ross & Kessler, 2005; Parkes & Weiss, 1983). Though these stage models have
included subtle variations, each articulates a pathway through bereavement that
typically begins with a sense of disbelief or numbness and—after proceeding
through intermediate stages that commonly include some form of protest and
depression—ultimately ends with acceptance of a loss. Since the birth of these
theories, thenotion of stagesof grief has, toasignificantextent,becomeingrained
in our cultural beliefs about loss, and these models of grieving have been rou-
tinely taught as part of the curriculum in medical schools and nursing programs
(Downe-Wamboldt & Tamlyn, 1997).
Despite the popularity and intuitive appeal of such models, only a few attempts
have been made to empiricallytest the merits of stage theory. Generally speaking,
these studies have provided mixed support, with one study failing to find sup-
porting evidence (Barrett & Schneweis, 1981) and two others finding some
support for a stage-like model (Maciejewski, Zhang, Block, & Prigerson, 2007;
Meuser & Marwit, 2001). The most prominent of these tests was conducted by
Maciejewski and colleagues (2007) who tested a phasic model of grief based on
the work of Jacobs (1993), whereby normal grief is initially characterized by
disbelief (which gradually decreases over time), followed by yearning, anger, and
depression (which show distinct peaks in the order presented), and concludes
with acceptance (which gradually increases over time).
These researchers examined patterns of change in these grief experiences
across three time periods (i.e., 1-6, 6-12, and 12-24 months post-loss) among a
sample of bereaved individuals, mostly made up of older adults grieving the
loss of a spouse by natural causes. In order to examine the smooth peaks and
valleys of these grief experiences across a multitude of time points, another
analysis was also conducted, which involved selecting one wave of data (at
random) for each participant and then modeling the trajectories of different grief
Indicators as a function of time since loss. These analyses revealed that par-
ticipants’ predominant response, regardless of the amount of time since loss,
was acceptance—a finding that was inconsistent with stage theory. However,
Maciejewski and colleagues (2007) did find that each of these grief Indicators
reached their respective maximum values in a sequence that was in line with
stagetheory(i.e.,disbelief? yearning? anger? depression? acceptance),
104/ HOLLAND AND NEIMEYER
The present study aims to provide a conceptual replication and extension of
these findings by examining the relation between time since loss and several
grief Indicators, similar to the ones tested previously (Maciejewski et al., 2007),
among a large, ethnically diverse sample of bereaved young adults who experi-
enced losses by natural causes (both sudden and anticipated) as well as by violent
causes (including homicide, suicide and fatal accidents). Though our cross-
sectionaldesign prohibits replication of longitudinal analyses, we plan to examine
the relative peaks and valleys of grief Indicators among participants with varying
durations of time since loss—similar to Maciejewski and colleagues’ (2007)
analysis involving one randomly selected wave of data from each participant.
This investigation also attempts to shed some light on how a meaning-oriented
perspective, which stresses the salience of finding meaning and making sense of
loss, might enhance our understanding of the grief experiences associated with
stage theory. In contrast to stage theories of grief, a meaning-oriented perspective
conceptualizes grief as a highly individualized process that is largely influenced
by the personal meanings people ascribe to a loss (Neimeyer, 2001, 2006).
Notably, this model of grieving has received empirical support in a number of
studies (e.g., Currier, Holland, & Neimeyer, 2006; Davis, Nolen-Hoeksema, &
Larson, 1998; Holland, Currier, &Neimeyer,2006; Keesee,Currier, &Neimeyer,
2008). In this investigation, the extent to which a participant has made sense of
a loss is examined side-by-side with other grief Indicators, to allow for a com-
parison of patterns across different durations of time since loss. We also examine
sense-making as a predictor of these grief Indicators, and it is hypothesized that
sense-making will outperform time since loss in predicting these grief outcomes.
Participants and Procedure
Participants in this study were drawn from a larger data set of bereaved college
students who, following institutional review, were recruited in their introductory
psychology courses across four waves of data collection at the University of
Memphis, a large state university serving an ethnically and economically diverse
student body (see Currier, Holland, Coleman, & Neimeyer, 2008, for a complete
description of this larger sample). Each participant was at least 18 years of age
and also reported the death of a friend or loved one within the past 2 years.
For each wave, eligible participants completed a single-session questionnaire that
included measures of meaning-making and grief symptoms as well as questions
concerning their background and the circumstances surrounding their loss
(e.g., How did the death occur? How long ago did the death occur?).
Some waves of the data collection used different measures and omitted
others. Therefore, these analyses were restricted to only that subset of partici-
pants (n = 717) who completed the Inventory of Complicated Grief-Revised
AN EXAMINATION OF STAGE THEORY/ 105
(ICG-R; Prigerson & Jacobs, 2001), the Core Bereavement Items (CBI; Burnett,
Middleton, Raphael, & Martinek, 1997), a one-item measure of sense-making,
and information about the cause of death and number of months since the loss—
all of which were necessary for the present analysis. Because we were primarily
interested in individuals who had lost a relationship of some significance, the
sample was further restricted by excluding 74 participants who reported that they
had lost an “acquaintance” or someone they “barely knew.” In order to maintain
consistency with Maciejewski and colleagues’ (2007) work, we also excluded 29
individuals who, based on their responses on the ICG-R, appeared to meet the
proposed criteria for Prolonged Grief Disorder (PGD; Prigerson & Maciejewski,
2006; Prigerson, Vanderwerker, & Maciejewski, 2008). Thus, 614 bereaved
individuals who lost a person of significance and were exhibiting grief symptoms
within the normal range, made up the sample for the present study.
The average age for this group of participants was 21.3 years (SD = 5.3), and
the average time since the loss was 11.1 months (SD = 7.5). The majority of
participants were women (76.4%, n = 469) and most had lost a loved one to
natural causes (71.8%, n = 441), with a sizable minority (28.2%, n = 173) losing a
loved one by violent means (i.e., homicide, suicide, or accident). The sample was
also ethnically diverse with 55.4% of the sample being Caucasian, 39.6% African
American, 1.1% Native American, 0.7% Asian American, 0.2% Hispanic/Latino,
and the remaining 3.0% of the sample identifying themselves as some “other”
race/ethnicity. The majority of participants (66.1%) lost an extended family
member (e.g., grandparent, cousin, aunt/uncle), and 8.1% and 25.7% lost an
immediate family member (e.g., spouse/partner, child, sibling) or friend, respec-
tively. The average age of the deceased was 53.3 years (SD = 25.9).
Individuals bereaved by violent causes were similar to those who had lost
a loved one to natural causes in terms of time since loss (t(612) = .54, p = .59)
and sex (?(1) = .46, p = .50). However, those who lost a loved one to violent
causes (compared to those who lost someone to natural causes) were some-
what more likely to be younger in age (t(611) = 3.43, p = .001) and to have
lost a friend (?(2) = 204.22, p < .001) who was younger in age (t(598) = 28.29,
p < .001).
Similar to Maciejewski and colleagues’ (2007) study, one-item Indicators
of Disbelief, Yearning, Anger, Depression, and Acceptance were created using
a pool of items from existing measures, as there are currently no established,
bereavement-specific instruments that tap into these constructs. In particular,
items from the ICG-R (Prigerson & Jacobs, 2001) were used to create Indicators
of Disbelief (Item 8; I feel disbelief over [the deceased]’s death), Yearning
(Item 5; I feel myself longing and yearning for [the deceased]), Anger (Item 7;
I can’t help feeling angry about [the deceased]’s death), and Acceptance (Item 4,
106/ HOLLAND AND NEIMEYER
reverse-coded; I feelthat I have trouble accepting the death)1. These ICG-R items
were drawn from a larger set of 30 items, each of which consists of a declarative
statement to which a response is made on a 5-point Likert-type scale describing
the frequency of symptoms ranging from 1 (Never) to 5 (Always). The ICG-R
and the original 19-item version of this measure have evidenced strong psycho-
metric properties in a number of studies (Boelen, van den Bout, de Keijser, &
Hoijtink, 2003; Neimeyer, Hogan, & Laurie, 2008). For example, this measure
has been tested in the Netherlands where the 29-item Dutch version displayed
high internal consistency (? = .94), concurrent validity with the Texas Revised
Inventory of Grief (r = .71; Faschingbauer, 1981), and good test-retest reliability
(r = .92) over a period ranging from 9 to 28 days (Boelen et al., 2003).
Because the ICG-R does not include an item that taps into feelings of sadness/
depression related to the loss, Item 16 from the CBI (Burnett et al., 1997; Do
reminders of [the deceased] such as photos, situations, music, places etc. cause
you tofeelsadness?) was used asan Indicator of Depression. Thisitemwas drawn
from a larger set of 17 items that make up the CBI. Each item from the CBI
is presented as a question, and responses are given on a 4-point Likert-type scale
based on how often the respondent experiences a particular symptom, ranging
from 0 (Never) to 3 (Always). To ensure that all Indicators were all on the
same scale, Item 16 from the CBI was rescaled in such a way that the lowest
possible score was 1 and the highest possible score was 5. Among a sample of
158 bereaved adults under the age of 70, the CBI has been shown to have high
internal consistency (? = .92) and successfully discriminate between different
groups of grievers in expected ways (e.g., higher scores for those who experi-
enced an unexpected versus an expected loss; Burnett et al., 1997).
Finally, in addition to the grief Indicators used in Maciejewski and colleagues’
(2007) study, an additional one-item Indicator of how much sense had been
made of the loss was used to examine meaning-reconstruction alongside the
other grief Indicators. Sense-making was assessed by having participants respond
to the question, How much sense would you say you have made of the loss?, using
a 4-point Likert-type scale from no sense of my loss to a good deal of sense.
Again, to maintain consistency with the other Indicators, this item was rescaled
so that the lowest possible degree of sense-making was represented by a 1, and
the highest degree of sense-making was represented by a 5. This single-item
sense-making item corresponds closely to the single-item questions that other
researchers have used to quantitatively measure this construct (e.g., Davis
et al., 1998; McIntosh, Silver, & Wortman, 1993). Significantly, Davis and
colleagues (1998) found that their sense-making item showed adequate test-retest
AN EXAMINATION OF STAGE THEORY/ 107
1Because Item 4 on the ICG-R is structured in such a way that higher scores indicate more
“trouble accepting the death,” this item was recoded when creating an Indicator of Acceptance,
so higher scores corresponded with greater acceptance of the loss.
reliability with Pearson correlations greater than .5 from 6 months to 13 months
post-loss. In our own work, this one-item sense-making measure has been
shown to have impressive utility as a strong predictor of overall grief symp-
tomatology in diverse populations and as a factor that can reliably distinguish
between those bereaved by violent and natural causes (Currier et al., 2006;
Holland et al., 2006; Keesee et al., 2008).
The analyses in this study were performed separately for those who were
bereaved by natural and violent means. Participants were divided in this way
so that we could run analyses on a sample that was as similar as possible to
Maciejewski and colleagues’ (2007) sample, which included only those bereaved
by natural causes, while at the same time examining this research question
with a previously uninvestigated population of grievers (i.e., those bereaved by
violent means). In each of these two subsamples, a series of four linear regres-
sion models was tested with time since loss as the independent variable, which
was expressed as a
3. cubic; and
function in a hierarchical fashion. These regression models were tested separ-
ately for each of the six Indicators, which served as the dependent variables.2
This analytic approach was adopted as a means of examining whether or not the
variability of these Indicators could be modeled as a function of time since
loss, either linearly or curvilinearly. Distinctions were made between signifi-
cant and non-significant models, and in each case, models with the lowest value
for Akaike’s Information Criterion (AIC) were selected, as lower values are
indicative of better fit. The resulting curves produced by these best-fitting
models were then graphed to examine the relative position of the peaks and
valleys among the six Indicators.
108/ HOLLAND AND NEIMEYER
2It should be noted that all of the Indicators being used as dependent variables are on
an ordinal scale. Therefore, assumptions of linear regression (e.g., normality) are violated to
some extent. Given this potential problem, all of the analyses were duplicated using ordinal
regression, which allows for a rank-ordered dependent variable. Interestingly, the ordinal
regression analyses yielded an identical pattern of results as when linear regression was
used. Because of the identical pattern of results and the practical advantages of linear regres-
sion (e.g., predicted values can be plotted as a smooth curve), linear regression was used to
analyze these data.
Sample Bereaved by Natural Causes
As can be seen in Table 1, among those bereaved by natural causes (n = 441),
the model that included time since loss as a quartic term was the best at predicting
scores on the Disbelief Indicator. Notably, this model accounted for 3.3% of the
variance overall and was statisticallysignificant, R2= .033, p = .005. In contrast, a
statistically significant quadratic model was the best at predicting scores on the
Yearning Indicator, and this model explained 1.5% of the variance, R2= .015,
p = .04. In this same group of grievers, the cubic model accounted for 2.2%
of the variability in Anger Indicator scores and was statistically significant,
R2= .022, p = .02. It should be noted that none of the models significantly
predicted scores on the Depression Indicator; however, the quadratic model
best fitting model for the Acceptance Indicator was the cubic model, which
explained 2.7% of the variability and was statistically significant, R2= .027,
p = .007. Finally, none of the models significantly predicted scores for the
Sense-Making Indicator. In fact, for this Indicator a linear model with a very
shallow slope appeared to fit the data best, R2= .000, p = .69.
The predicted curves for these best fitting models are presented in Figure 1
alongside group means from the raw data.3As can be seen in this figure, the rela-
tive positions of the graphed curves for the models that were found to be statis-
tically significant (i.e., those for Disbelief, Yearning, Anger, and Acceptance)
provide mixed support for stage theory. First, it is worth noting that the curves
for the Disbelief, Yearning, and Anger Indicators appear to have two distinct
peaked areas—one occurring within the first 12 months after a loss and the other
peaking at around 24 months. Focusing solely on the initial peaked area within
the first 12 months, it appears that Yearning is at its highest for those in the
very early stages of grief (peaking around 0-2 months), which is followed
by Disbelief (peaking around 3-4 months) then Anger (peaking around 3½-5
months). This ordering is only partially consistent with stage theory, which would
predict the relative position of the peaks to follow a somewhat different pattern
(i.e., Disbelief, Yearning, and then Anger). Looking beyond the first year of
AN EXAMINATION OF STAGE THEORY/ 109
3Group means from the raw data were generated by dividing each of the samples into
10 groups based on the number of months since the loss. Groups were divided as evenly
as possibly without making arbitrary divisions between individuals who reported the same
number of months since the loss. Using this strategy, the sample of bereaved individuals who
experiencedaloss bynatural causeswasdivided into 10 groups ranging from39-49 individuals
per group, and those who lost a loved one by violent means were divided into 10 groups
ranging from 13-20 individuals per group. For each group, the means for the independent
(i.e., months since loss) and dependent (i.e., the grief Indicators) variables were plotted on
the x and y axes, respectively.
110/ HOLLAND AND NEIMEYER
Table 1. Summary of Regression Analyses Predicting Grief Indicators
as a Function of Time Since Loss
Participants bereaved by
natural causes (n = 441)
Participants bereaved by
violent causes (n = 173)
*p ? .05. **p < .01. ***p < .001.
bereavement, however, it appears that the Acceptance Indicator is largely in line
with stage theory, exhibiting its first substantial peak after those of Disbelief,
Yearning, and Anger among participants who had lost a loved one approximately
15 to 17 months ago. However, for those persons approaching the second anni-
versary of the death, this Indicator seemed to take a downturn, approximating
levels shown in the near aftermath of loss.
Though derived from a non-significant model, the best fitting curve for the
Depression Indicator yielded two peaked areas—one around 0-2 months and
another around 22-24 months after a loss—neither of which would be consistent
with stage theory. Additionally, the non-significant linear model that provided the
AN EXAMINATION OF STAGE THEORY/ 111
Figure 1. This graph depicts scores on the various Indicators (y-axis) as a
function of months since loss (x-axis) among participants bereaved by natural
causes. Predicted curves derived from the linear regression models are labeled
with arrows. The initials “n.s.” are used to identify statistically non-significant
models. As a means of assessing how well each predicted curve maps
onto actual scores, group means from the raw data are presented as
single points of varying shapes, each of which represents scores on
a particular Indicator as indicated by the key that appears below.
best fit for the Sense-Making Indicator appeared as a relatively straight line. It
should also be noted that this Indicator along with the Acceptance Indicator
appeared to be at consistently higher levels overall compared to the other
Indicators, suggesting that the predominant response for many of these
participants was to accept and make sense of their losses. In contrast, across the
entire2 years,theso-called“negative”grief Indicators of depression and yearning
were much less evident, though each appeared to rise visibly as the second
anniversary of the death drew closer. In terms of absolute levels, these Indicators
were followed in intensity by denial and anger, which showed a low, oscillating
pattern across this same period, cresting both early and late in the two-year
window of bereavement represented by our sample.
Sample Bereaved by Violent Causes
Among those bereaved by violent causes (n = 173), a statistically significant
linear association between time since loss and scores on the Disbelief Indicator
provided the best fit, accounting for 11.1% of the variance, R2= .111, p < .001.
None of the tested models was statistically significant for the Yearning Indicator,
although the quadratic model provided the best fit, R2= .021, p = .17. Similarly,
theAnger Indicator wasalsonot significantlypredicted byanyof thefour models.
For this Indicator, the linear model provided the best fit, explaining 0.8% of the
variability in Anger Indicator scores, R2= .008, p = .25. Scores on the Depression
Indicator were significantly predicted by the quadratic model, which provided the
best fit and accounted for 4.1% of the variability, R2= .041, p = .03. The best
fitting model for the Acceptance Indicator was also the quadratic model, which
explained 5.1% of the variability and was statistically significant, R2= .051, p =
.01. Likewise, the quadratic model provided the best fit for the Sense-Making
Indicator, although this model was non-significant, R2= .016, p = .24.
participants bereaved by violent causes are presented in Figure 2. For those
Indicators thatyieldedastatisticallysignificantmodel(i.e., Disbelief,Depression,
and Acceptance), it is noteworthy that the Disbelief and Depression Indicators
peaked first (around 0-2 months), with Disbelief steadily decreasing across time
since loss and Depression peaking again for participants bereaved around 22-24
months ago. Consistent with stage theory, the Acceptance Indicator peaked after
theDisbeliefand Depression Indicators amongparticipantsbereaved around 16 to
19 months ago.
The graphed curve of the non-significant, best fitting model for the Yearning
Indicator yielded peaks for those bereaved early on (peaking around 0-2 months)
and those who had been bereaved for nearly 2 years (peaking slightly around
22-24 months). Conversely, the best fitting curve for the Anger Indicator, which
was also derived from a non-significant model, was linear with a slight negative
slope. Thus, the peak for this Indicator occurred among those who were most
112/ HOLLAND AND NEIMEYER
recently bereaved (i.e., around 0-2 months after a loss). Finally, the best-fitting,
yet non-significant, model for the Sense-Making Indicator yielded a peak around
11-13 months, with a visible reduction for those respondents approaching the
second anniversary of the loss. It is also interesting to note that among those
bereaved by violent causes, sense-making and acceptance were not necessarily
the dominant responses, particularly among participants in the early months of
bereavement, who tended to score highest on the Disbelief Indicator. Generally
speaking, it also appeared that values for the negative grief Indicators were
AN EXAMINATION OF STAGE THEORY/ 113
Figure 2. This graph depicts scores on the various Indicators (y-axis) as a
function of months since loss (x-axis) among participants bereaved by
violent means. Predicted curves derived from the linear regression models
are labeled with arrows. The initials “n.s.” are used to identify statistically
non-significant models. As a means of assessing how well each predicted
curve maps onto actual scores, group means from the raw data are presented
as single points of varying shapes, each of which represents scores on a
particular Indicator as indicated by the key that appears below.
to those bereaved by natural causes. A notable exception was Yearning, which
displayed an elevation and general pattern that closely approximated that evi-
denced by persons bereaved by natural death. Relative to Depression, Disbelief
and Anger, however, it was less prominent in the grief response of violently
As a point of comparison for these findings, several regression analyses
were performed using sense-making scores as the independent variable and
the grief Indicators proposed by stage theory (i.e., Disbelief, Yearning, Anger,
Depression, and Acceptance) as the dependent variables. The full sample of
614 bereaved participants was used for these analyses, as the purpose was simply
to examinesense-making’s abilityto predict scores on the grief Indicators overall.
These analyses revealed that the degree of sense-making a participant reported
was negatively associated with scores on the Disbelief Indicator (R2= .108,
p < .001), Yearning Indicator (R2= .042, p < .001), Anger Indicator (R2= .174,
p < .001), and Depression Indicator (R2= .071, p < .001). Conversely, higher
degrees of sense-making were significantly associated with higher levels of
Acceptance (R2= .149, p < .001). Overall, these analyses revealed that sense-
making significantly predicted each of the five Indicators, accounting for 4.2%
to 17.4% of the variability in scores.
As a whole, these results do not closely corroborate a stage theory of grief;
however, neither are they entirely inconsistent with it. Though the grief Indi-
cators examined in this investigation did not strictly follow the sequential pattern
found in Maciejewski and colleagues’ (2007) study, there was some evidence
suggesting that experiences of grief distress, such as disbelief, yearning, anger,
and depression, weregenerallymostsalientfor thosebereaved for ashorter period
of time (compared to those bereaved for longer periods). Likewise, acceptance
was typically most evident among those bereaved for longer periods of time.
This general pattern, with grief distress and acceptance rising and falling
asynchronously, fits with Prigerson and Maciejewski’s (2008) model, whereby
grief and acceptance are conceptualized as “opposite sides of the same coin”
(p. 435). In this model, specific manifestations of grief distress (e.g., disbelief,
yearning, anger) are regarded as different “grief states” (all of which represent
a single underlying construct of grief), rather than distinct psychological phases
or stages (Prigerson & Maciejewski, 2008). Therefore, such a model would
simply hypothesize that grief distress and acceptance would be inversely related,
but it would not necessarily predict that different grief states would adhere to
a strict sequential pattern.
114/ HOLLAND AND NEIMEYER
Looking beyond the sequence of these grief experiences, it is worth noting
that participants in the present study who were bereaved by natural causes
were primarily characterized by acceptance regardless of time since loss—
a finding that is consistent with Maciejewski and colleagues’ (2007) study.
These participants were also found to be characterized by a high degree of
sense-making, which was essentially unrelated to time since loss. Interestingly,
these trends were less pronounced for those bereaved by violent causes, who
generally had higher scores on the more negative grief Indicators. In fact, among
those bereaved by violent loss for 3 months or less, disbelief showed up as the
Also of concern is the predominance of possible Indicators of more traumatic
adaptation (e.g., anger, disbelief, and depression), which could interfere with
processing of attachment-related emotions such as yearning in this tragically
bereaved population. Although grief-related distress was much more marked
and persistent for these violently bereaved individuals, acceptance still emerged
as the dominant response among individuals bereaved for a longer period of
time, possibly indicating that most grievers are fairly resilient even in the face
of objectively traumatic loss.
Though we found some evidence that might suggest that grief distress and
acceptance wax and wane in a predictable fashion, these results also indicate
that other factors, even with regard to time, may be just as (or more) important.
For example, in several instances, grief Indicators exhibited peaks beyond the
first year that appeared to reach a maximum point among participants who had
been bereaved for roughly 24-months. These results could suggest the presence
of an anniversary effect (Musaph, 1990), whereby grief is aggravated as the
bereaved approach significant dates that call to mind their losses. However, peaks
were not generally observed among bereaved individuals at the 1-year mark.
Though subject to a variety of interpretations—including the possibility that
this finding is an artifact of our cross-sectional design—it could be that anni-
versary effects may be more salient in the second year of bereavement (and
conceivably beyond), perhaps once the protective coping responses or social
support of the first year have waned.
This study also indicates that sense-making (which appears to be somewhat
independent of length of bereavement) is a more robust predictor of these grief
Indicators, compared to time since loss—even when time is modeled in complex
ways. Specifically, the one-item sense-making measure consistently accounted
for more variability in the five grief Indicators than the “best-fitting” regression
models that used time since loss as the independent variable, illustrating the
predictive advantages of a meaning-oriented perspective. The only exception to
this general trend was for the disbelief Indicator within the subsample of partici-
pants bereaved by violent causes. In this case, time since loss and sense-making
performed similarly, possibly indicating the particular importance of time in
helping these individuals come to grips with the painful reality of violent loss.
AN EXAMINATION OF STAGE THEORY/ 115
Given the uneven support for a stage-like conception of grieving offered by
these findings as well as those of Maciejewski and associates (2007), it is worth
considering why a phasic model of grief continues to hold such appeal for the
general public, and, to a lesser extent, bereavement professionals. One answer
might be that human beings are inveterate seekers of patterns to organize the
flux of experience in a fashion that enhances their sense of prediction and
control (Kelly, 1955). Certainly such motivation could be greatly enhanced by
the cognitive, emotional, and social turmoil occasioned by the death of a loved
one, when the promise of a modicum of predictability conferred by even a
provisional “roadmap” through the terrain of loss would be welcomed by many
of the bereaved.
A second and perhaps still more basic reason that a sequential plotting of grief
responses is difficult to resist is that it may correspond to the fundamentally
narrative structure of much of human thought (Fireman, McVay, & Flanagan,
2003; Neimeyer, van Dyke, & Pennebaker, 2009), which seeks to order temporal
events so as to “tell a good story” by arranging them in terms of a meaningful
beginning, middle, and end, moving from an early perturbation of the plot
through various obstacles to the achievement of valued goals (Bruner, 1990;
Neimeyer & Levitt, 2000).
A further and related factor could be that stage theory suggests a particular
kind of narrative structure to grieving, one in which the protagonist is thrust into
what Joseph Campbell (1949) described as the hero’s journey, an archetypal
narrative discernible across widely varying cultures in which the hero is called
forth from the normal world to face and overcome a series of great trials,
ultimately to return triumphant with special knowledge or a special gift to bestow
on others. This epic narrative structure is easily enough seen in popular depic-
tions of “the griever’s journey,” which like Campbell’s “monomyth” commonly
entail a shift in the protagonist’s spiritual center of gravity as he or she crosses
a liminal threshold into an unknown and dangerous world, typically under-
going a personal metamorphosis as the journey proceeds, before reentering the
known world transformed and bearing a special boon to confer on his or her
fellows. This implicit description of grieving as a heroic quest for recovery
of a lost treasure or conquest of a series of obstacles could ennoble the bereaved
in their own eyes and that of their community, and provide a canonical cultural
script for a positive “ritual re-inscription of identity” that is part of the rite of
passage for survivors of a loved one’s death (Romanoff & Terenzio, 1998). In
this view, the seemingly magnetic draw of a stage-like depiction of grieving that
begins with a disorienting separation from the “normal,” pre-bereavement world,
and that progresses heroically through a series of clearly marked emotional trials
before eventuating in a triumphant stage of acceptance, recovery, or symbolic
return, may owe more to its compelling coherence with a seemingly universal
narrative structure than to its objective accuracy. Simply stated, stage theory
may have functioned as the bereavement field’s own cultural “monomyth.”
116/ HOLLAND AND NEIMEYER
Finally, although stage theorists typically have focused their attention on the
sequential nature of grief experiences, it should be noted that the conclusions
drawn here are not necessarily diametricallyopposed to more flexible accounts of
stage theory, as most also acknowledge the unpredictability of individual responses
Although it is sometimes instructive to conceptualize the manifestations
of grief in this manner, it is important to emphasize that the idea that grief
unfolds inexorably in regular phases is an oversimplification of the highly
complex, personal waxing and waning of the emotional process. (p. 18)
Thus, rather than refuting these models of grief, the findings of the present
study would suggest that more attention should probably be placed on the qualifi-
cations and caveats associated with stage theory than the theory itself.
Limitations and Implications
The conclusions drawn from this study are limited by the fact that a cross-
sectional data set was used to address what may be considered an inherently
longitudinal question. Therefore, variability among the grief Indicators accounted
for by timesince loss mightbe due to the passage of timeor individual differences
among participants. Though doing so would pose obvious practical chal-
lenges, future studies would do well to examine the merits of stage theory
using a prospective, longitudinal design with enough time points (perhaps using
monthly assessments) to fully capture the waxing and waning of different grief
experiences. In addition, even though the items used to construct our grief
Indicators were drawn from psychometrically-validated measures of grief, these
Indicators, by themselves, have not been rigorously evaluated as measures of
the constructs of interest. Because no grief-specific measures presently exist
for constructs like anger or yearning, future examination of stage theory would
benefit from the development and validation of measures that reliably tap into
these grief experiences.
It is also worth noting that this sample was made up primarily of young
adults who were grieving the loss of an extended family member, which limits
the generalizability of the results and could account for differences between
our findings and those of Maciejewski and colleagues (2007). Within our own
sample, it also appeared that those who had lost a loved one to violent causes
tended to be younger and more likely to have lost a young-age friend, compared
to those bereaved by natural causes. Thus, it is possible that differences in
patterns of grief observed between those bereaved by violent versus natural
causes could be influenced by these confounding factors.
Despite these limitations, the present findings suggest that, even though some
predictable patterns in grieving maybe observed as a function of timesince loss, a
meaning-oriented approach may prove to be more relevant. Thus, clinicians
AN EXAMINATION OF STAGE THEORY/ 117
working with the bereaved would do well to focus their assessments on how
a client has come to understand or make sense of a loss, rather than trying to
determine one’s stage of grieving. This information about the personal meanings
a client has ascribed to a loss may also serve as a focal point for treatment, as
problematic grief reactions are often conceptualized as arising from difficulties
in finding meaning in the aftermath of these painful experiences (Neimeyer,
2001; Park, 2008). By drawing upon interventions with a strong narrative com-
ponent (e.g., revisiting/ re-telling the story of the loss, journaling about any sense
made of the loss or its unsought benefits, enacting imaginal dialogues with
the deceased; Lichtenthal & Cruess, in press; Shear, Frank, Houck, & Reynolds,
problematic reactions to bereavement through reconstructing a world of meaning
that allows them to make sense of loss and envision a more hopeful future.
Barrett, C. J., & Schneweis, K. M. (1981). An empirical search for stages of widowhood.
Omega, 11, 97-104.
Boelen, P. A., van den Bout, J., de Keijser, J., & Hoijtink, H. (2003). Reliability and
validity oftheDutchversionoftheInventory ofTraumaticGrief(ITG).DeathStudies,
Bowlby, J. (1980) Attachment and loss. New York: Basic Books.
Bruner, J. (1990). Acts of meaning. Cambridge, MA: Harvard University Press.
Burnett, P., Middleton, W., Raphael, B., & Martinek, N. (1997). Measuring core bereave-
ment phenomena. Psychological Medicine, 27, 49-57.
Campbell, J. (1949). The hero with a thousand faces. Princeton, NJ: Princeton University
Currier, J. M., Holland, J. M., Coleman, R. A., & Neimeyer, R. A. (2008). Bereavement
following violent death: An assault on life and meaning. In R. Stevenson & G. Cox
(Eds.), Perspectives on violence and violent death (pp. 177-202). Amityville, NY:
Currier, J. M., Holland, J. M., & Neimeyer, R. A. (2006). Sense-making, grief, and the
Davis, C. G., Nolen-Hoeksema, S., & Larson, J. (1998). Making sense of loss and
benefiting from experience: Two construals of meaning. Journal of Personality and
Social Psychology, 75, 561-574.
Downe-Wamboldt, B., & Tamlyn, D. (1997). An international survey of death educa-
tion trends in faculties of nursing and medicine. Death Studies, 21, 177-188.
Faschingbauer, T. R. (1981). Texas Revised Inventory of Grief manual. Houston, TX:
Fireman, G. D., McVay, T. E., & Flanagan, O. J. (Eds.). (2003). Narrative and conscious-
ness. New York: Oxford.
Holland, J. M., Currier, J. M., & Neimeyer, R. A. (2006). Meaning reconstruction in
the first two years of bereavement: The role of sense-making and benefit-finding.
Omega, 53, 175-191.
118/ HOLLAND AND NEIMEYER
Jacobs, S. (1993). Pathologic grief: Maladaptations to loss. Washington, DC: American
Kelly, G. A. (1955). The psychology of personal constructs. New York: Norton.
Keesee, N. J., Currier, J. M., & Neimeyer, R. A. (2008). Predictors of grief following
the death of one’s child: The contribution of finding meaning. Journal of Clinical
Psychology, 64, 1145-1163.
Kübler-Ross, E. (1969). On death and dying. New York: Macmillan.
Kübler-Ross, E., & Kessler, D. (2005). On grief and grieving: Finding the meaning of
grief through the five stages of loss. New York: Scribner.
Lichtenthal, W. G., & Cruess, D. G. (in press). Effects of directed written disclosure
on grief and distress symptoms among bereaved individuals. Death Studies.
Maciejewski, P. K., Zhang, B., Block, S. D., & Prigerson, H. G. (2007). An empirical
examination ofthe stage theory ofgrief. Journalofthe American Medical Association,
McIntosh, D. N., Silver, R. C., & Wortman, C. B. (1993). Religion’s role in adjust-
ment to a negative life event: Coping with the loss of a child. Journal of Personality
and Social Psychology, 65, 812-821.
Meuser, T. M., & Marwit, S. J. (2001). A comprehensive, stage-sensitive model of grief
in dementia caregiving. The Gerontologist, 41, 658-670.
Musaph, H. (1990). Anniversary reaction as a symptom of grief in traumatized persons.
Israel Journal of Psychiatry and Related Sciences, 27, 175-179.
Neimeyer, R. A. (2000). Narrative disruptions in the construction of the self. In R. A.
Neimeyer & J. D. Raskin (Eds.), Constructions of disorder: Meaning-making frame-
works for psychotherapy (pp. 207-242). Washington, DC: American Psychological
Neimeyer, R. A. (2001). Meaning reconstruction & the experience of loss. Washington,
DC: American Psychological Association.
Neimeyer, R. A. (2006). Restorying loss: Fostering growth in the posttraumatic narrative.
practice (pp. 68-80). Mahwah, NJ: Lawrence Erlbaum.
Neimeyer, R. A., Hogan, N., & Laurie, A. (2008). The measurement of grief: Psycho-
metric considerations in the assessment of reactions to bereavement. In M. Stroebe,
R. O. Hansson, H. Schut, & W. Stroebe (Eds.), Handbook of bereavement research:
21st century perspectives (pp. 133-186). Washington, DC: American Psychological
Neimeyer, R. A., & Levitt, H. (2000). What’s narrative got to do with it? Construction and
coherence in accounts of loss. In J. Harvey (Ed.), Loss and trauma (pp. 401-412).
Philadelphia: Brunner Routledge.
Neimeyer, R. A., van Dyke, J. G., & Pennebaker, J. W. (2009). Narrative medicine:
Writing through bereavement. In H. Chochinov & W. Breitbart (Eds.), Handbook of
psychiatry in palliative medicine (pp. 454-469). New York: Oxford.
Park, C. L. (2008). Testing the meaning-making model of coping with loss. Journal of
Social and Clinical Psychology, 27, 970-994.
Parkes, C. M., & Weiss, R. S. (1983). Recovery from bereavement. New York: Basic
Prigerson, H. G., & Jacobs, S. C. (2001). Diagnostic criteria for traumatic grief: A
rationale, consensus criteria, and a preliminary empirical test. In M. S. Stroebe,
AN EXAMINATION OF STAGE THEORY/ 119
R. O. Hansson, W. Stroebe, & H. Schut (Eds.), Handbook of bereavement research: Download full-text
Consequences, coping, and care (pp. 613-645). Washington, DC: American Psycho-
Prigerson, H. G., & Maciejewski, P. K. (2006). A call for sound empirical testing and
evaluation of criteria for complicated grief proposed for DSM-V. Omega, 52, 9-19.
Prigerson, H. G., & Maciejewski, P. K. (2008). Grief and acceptance as opposite sides
of the same coin: Setting a research agenda to study peaceful acceptance of loss.
The British Journal of Psychiatry, 193, 435-437.
Prigerson, H. G., Vanderwerker, L. C., & Maciejewski, P. K. (2008). A case for inclu-
sion of prolonged grief disorder in DSM-V. In M. Stroebe, R. Hansson, H. Schut,
& W. Stroebe (Eds.), Handbook of bereavement research and practice: Advances
in theory and intervention (pp. 165-186). Washington, DC: American Psychological
Romanoff, B. D., & Terenzio, M. (1998). Rituals and the grieving process. Death Studies,
Shear, K., Frank, E., Houck, P. R., & Reynolds, C. F. (2005). Treatment of complicated
grief: A randomized controlled trial. Journal of the American Medical Association,
Wagner, B., Knaevelsrud, C., & Maercker, A. (2006). Internet-based cognitive-behavioral
therapy trial. Death Studies, 30, 429-453.
Direct reprint requests to:
Jason M. Holland
Center for Health Care Evaluation (152-MPD)
VA Palo Alto Health Care System
795 Willow Rd.
Menlo Park, CA 94025
120/ HOLLAND AND NEIMEYER