Delay and Death-Thought Accessibility

Abstract

The dual-process component of Terror Management Theory (TMT) proposes that different types of threats lead to increases in death-thought accessibility (DTA) after different delay intervals. Experimental studies of terror management threats’ effect on DTA were collected and coded for their use of explicitly death-related (vs. not explicitly death-related) threats, and for their use of delay and task-switching during the delay. Results reveal that studies using death-related threats achieved larger DTA effect-sizes when they included more task-switching or a longer delay between the threat and the DTA measurement. In contrast, studies using threats that were not explicitly death-related achieved smaller DTA effect-sizes when they included more task-switching between the threat and the DTA measurement. These findings provide partial support for the dual-process component’s predictions regarding delay and DTA. Limitations and future directions are discussed.

Keywords

terror management death-thought accessibility delay dual process meta-analysis

Under what conditions do individuals most readily think of death? The question of death-thought accessibility (DTA), or when individuals most readily think of death, has been a growing subject of research, most notably within Terror Management Theory (TMT; Greenberg, Pyszczynski, & Solomon, 1986; Solomon, Greenberg, & Pyszczynski, 1991). TMT proposes that humans are faced with a unique psychological conflict: We have an instinctual desire to survive, yet also the self-awareness to recognize that we will inevitably cease to exist. This conflict is thought to produce a profound sense of terror. Individuals are compelled to manage this terror by embracing cultural worldviews, social value systems that promote perceptions of life as being coherent and meaningful, and that provide individuals with a sense of symbolic immortality (Dechesne et al., 2003). These cultural values thus function to thwart concerns regarding mortality. Consequently, TMT implies a social and cognitive link between cultural values and concerns regarding mortality or death.

In support of this perspective, early TMT studies showed that invoking explicit thoughts of death leads individuals to more strongly defend their cultural values (Greenberg et al., 1990; Rosenblatt, Greenberg, Solomon, Pyszczynski, & Lyon, 1989). In these studies, participants were assigned to either a Mortality Salience (MS) condition, where they were asked to reflect on their own death, or a control condition. Following exposure to MS, participants were more willing to punish and derogate “worldview violators,” those who expressed beliefs or behaviors that were inconsistent with the participants’ cultural values. Inversely, MS also increased participants’ willingness to reward and positively evaluate others who confirmed or upheld participants’ cultural values. These first studies were critical in establishing a link between explicit thoughts of death and desire to uphold cultural values, a finding that has been confirmed by later studies (e.g., Greenberg, Simon, Pyszczynski, Solomon, & Chatel, 1992; McGregor et al., 1998).

However, some early studies also presented conflicting results (Greenberg, Pyszczynski, Solomon, Simon, & Breus, 1994). Individuals seemed less willing to defend cultural values while they were still experiencing strongly explicit thoughts of death. Rather, it seemed individuals defended cultural values after explicit thoughts of death had subsided or weakened, becoming implicit: that is, cognitions related to death that are accessible, but reside outside full conscious awareness. The implications of this are twofold. First, stronger explicit reminders of death should require more time to pass or more distraction before affecting cultural value defense. Second, cultural value defense should only be expected to occur when thoughts of death have become implicit. Exploring this second idea required a means to quantify implicit thoughts of death. Accordingly, Greenberg et al. (1994) developed a measure of DTA, which is now widely used.

The classic measure of DTA takes the form of a word fragment completion task (Greenberg et al., 1994). Participants are asked to complete a series of word fragments (e.g., C O F F _ _) with the first word that comes to mind. Unbeknownst to the participants, several of these fragments can be completed with either a death-related or neutral word (e.g., coffin or coffee, respectively). Thus, completing a greater number of fragments with death-related words indicates heightened DTA, or activation of implicit thoughts of death. Alternative methods for measuring DTA have also been used, including the measurement of reaction times to death-related stimuli (e.g., Fritsche, Jonas, & Fankhänel, 2008; Vail, Arndt, Motyl, & Pyszczynski, 2012), and the number of references to death in free-writing tasks (e.g., Gailliot, Schmeichel, & Maner, 2007; Trafimow & Hughes, 2012).

Greenberg and colleagues’ (1994) first study of DTA found that immediately following explicit thoughts of death, DTA was suppressed; however, following explicit thoughts of death and a delay, DTA increased. This effect has been replicated (e.g., Harmon-Jones et al., 1997), although some researchers have recently found the opposite effect of delay (Trafimow & Hughes, 2012). In addition, Greenberg and colleagues’ (1994) findings suggested that individuals tend to defend cultural values only when DTA is high, that is, when thoughts of death are implicitly rather than explicitly active. This provided the first direct evidence that implicit thoughts of death play a role in predicting defense of cultural values, and laid the foundation for future use of DTA in investigations of TMT processes.

This finding, combined with other evidence regarding the effects of cognitive load and subliminal death primes on DTA (Arndt, Greenberg, Pyszczynski, & Solomon, 1997; Arndt, Greenberg, Solomon, Pyszczynski, & Simon, 1997), ultimately led to the articulation of the dual-process component of TMT (Greenberg, Arndt, Simon, Pyszczynski, & Solomon, 2000; Pyszczynski, Greenberg, & Solomon, 1999). The dual-process component of TMT theorizes that there are distinct psychological defenses in response to implicit and explicit processing of death-related information, and also indicates when implicit thoughts of death (heightened DTA) should be expected to occur. Specifically, the dual-process component argues that explicit processing of death-related information, such as that produced by MS manipulations, first leads to a set of proximal defenses (Pyszczynski, Greenberg, & Solomon, 1999). These proximal defenses include pseudo-rational distortions of one’s vulnerability to death (e.g., underestimating one’s risk of death), and also active suppression of death-related thoughts, including suppression of DTA. However, once an individual has become distracted from explicit thoughts of death, such as after completing neutral delay tasks, DTA resurges and proximal defenses subside. This resurgence of DTA following suppression is thought to be the result of cognitive mechanisms similar to those that drive the classic paradoxical effect of thought suppression (i.e., attempting “not to think of a white bear” actually makes it more likely that one will think of a white bear; Wegner, Schneider, Carter, & White, 1987). Along with resurgence in DTA, individuals begin to engage in distal defenses, which include defending and affirming cultural values, and attempts to enhance one’s self-esteem (Pyszczynski, Solomon, & Greenberg, 1999).

In addition, the dual-process component describes what is expected to happen in response to threats that do not invoke explicit processing of death-related information, such as subliminal priming of death-related constructs (Pyszczynski, Solomon, & Greenberg, 1999). In this case, proximal defenses are “skipped.” Instead, individuals begin to engage in distal defenses, such as affirmation of values, immediately. Importantly, because proximal defenses are skipped, DTA is also not suppressed and therefore becomes heightened immediately following these threats. Thus, the dual-process component predicts that threats involving explicit thoughts of death lead only to delayed increases in DTA, whereas threats that do not involve explicit thoughts of death lead to immediate increases in DTA. Some additional evidence implies that, in the case of threats that are not explicitly death-related, the immediate increase in DTA might taper off following a delay (see Figure 1 for a hypothetical model of DTA’s associations with delay; Martens, Burke, Schimel, & Faucher, 2011; Schimel, Hayes, Williams, & Jahrig, 2007). Thus, the dual-process component carries important implications for terror management studies’ procedural designs, as inclusion of a delay may either enhance or diminish DTA or particular defenses (i.e., proximal or distal) in response to terror management threats.

Figure 1.

Hypothetical model of threat type and delay predicting DTA.

Providing insight into the dual-process component of TMT is not the only major contribution of DTA to TMT, however. Although early studies revealed that thoughts of death predict concerns related to cultural values, DTA enabled researchers to test the opposite hypothesis: that concerns regarding cultural values affect thoughts of death. If cultural values buffer thoughts of death, as TMT suggests, then challenging cultural values may instigate thoughts of death or increases in DTA. Indeed, several studies have supported this. Among others, threats to relationship solidarity (e.g., Florian, Mikulincer, & Hirschberger, 2002; Mikulincer, Florian, Birnbaum, & Malishkevich, 2002; Taubman-Ben-Ari & Katz-Ben-Ami, 2008), self-esteem (e.g., Hayes, Schimel, Faucher, & Williams, 2008), and cultural identity or worldview (e.g., Schimel et al., 2007) have all been shown to increase DTA. Critically, many of these studies suggest that cultural value threats have an immediate effect on DTA. This is consistent with the dual-process component’s predictions; because these threats do not produce explicit thoughts of death, they should be expected to result in immediate increases in DTA.

Testing the Dual-Process Components Predictions Regarding Delay and DTA

Three prior reviews have addressed the dual-process component of TMT, but none has quantitatively synthesized the role of delay on DTA. The closest review to address this question was a qualitative review of the literature from 1997 to 2009, in which Hayes, Schimel, Arndt, and Faucher (2010) concluded that DTA is suppressed immediately following MS, but resurges following a delay or distraction. Martens et al. (2011) meta-analytically showed that relative to meaning threats, MS manipulations are associated with larger effect-sizes following a delay, but this review only compared MS manipulations with meaning threats so that delay’s unique association with each particular type of threat could not be isolated. Burke, Martens, and Faucher’s (2010) meta-analysis primarily examined the role of death-related threats on worldview- and esteem-related outcomes rather than DTA, and found stronger effects of mortality salience on worldview and esteem outcomes after a delay compared with after no delay. Thus, these prior reviews provide indirect support for the notion that DTA is suppressed immediately following MS but resurges afterward.

Direct support for the role of delay on DTA, however, is mixed. Although some studies experimentally test the dual-process component’s predictions regarding delay and DTA, these studies are few and most of them were conducted early in the history of the development of TMT (e.g., Arndt, Greenberg, Solomon, et al., 1997; Harmon-Jones et al., 1997; Schimel et al., 2007). More recently, a series of six studies failed to find evidence of delayed increases in DTA following mortality salience (Trafimow & Hughes, 2012), thus casting some doubt on this element of the dual-process component of TMT.

Present Study

The literature on DTA is now sufficiently large to meta-analytically evaluate the influence of delay on DTA. Across this literature, there exists significant variability in studies’ use of delay when measuring DTA. This presents a unique opportunity to meta-analyze whether studies’ use of delay is associated with differences in DTA, and whether this influence of delay differs according to type of threat. Thus, the primary aim of the present study is to test the dual-process component’s predictions regarding delay and DTA. Specifically, we tested the following hypotheses:

Hypothesis 1: There will be no effect of death-reminders or MS on DTA in studies that do not use a delay, but DTA effect-sizes will be larger in studies that use longer delays.

Hypothesis 2: There will be a moderate or strong effect of threats that are not explicitly death-related on DTA in studies that do not include a delay, but the effect will be diminished in studies that use longer delays.

In addition, we examined whether MS—reflection on one’s personal mortality—results in distinct effects as compared with less direct reminders of death (e.g., thinking about war; that is, death-reminders). MS may result in unique effects because it is a stronger, more personal reminder of death (Greenberg et al., 1994; Hayes et al., 2010).

Method

To analyze these hypotheses, we computed a mean difference indicator of effect-size, Hedges’s g, from experimental studies of threats’ effect on DTA. Hedges’s g is similar to Cohen’s d in that it reflects a standardized difference in means, although it also corrects for inflated effect-sizes among studies with small samples. Thus, a value of g = 1 reflects a mean difference that is approximate to the value of one standard deviation. Hedges’s g can also be interpreted following J. Cohen’s (1988) suggestion for Cohen’s d, that .2 indicates a small effect-size, .5 indicates a medium effect-size, and .8 indicates a large effect-size. All effect-sizes were computed using the standard formulas contained in Comprehensive Meta-Analysis (CMA) software (Borenstein, Hedges, Higgins, & Rothstein, 2005), and all analyses were conducted using meta-analysis packages in STATA 12.

Study Search and Criteria for Inclusion

Studies were identified from Google Scholar and PsycINFO using combinations of the following search terms: death-thought, accessibility, death-related, thoughts, death-construct, and constructs. Studies were also identified using citations from a previous review of DTA (Hayes et al., 2010). This search returned a total of 556 unique works. From these, 121 empirical papers that measured DTA were identified. Several papers, studies, or effects were excluded from this analysis because their methods conflicted with the present study’s primary aim and analytic strategy, or because there were insufficient data for estimating an effect-size. This resulted in a total of 78 papers and 99 effect-sizes included in the present analysis. Explanations for other exclusions are as follows.

To best capture the causal effects of threat manipulations on DTA, and to regress effects on delay, we used only between-subjects experimental studies in our analysis. Thus, some studies were excluded for being non-experimental (e.g., Arndt, Cook, Goldenberg, & Cox, 2007, Study 4; Dunkel, 2009, Study 3; Scharf & Cohen, 2013, Study 1; Vess, Arndt, & Cox, 2012, Study 1). In a similar vein, studies were excluded for their use of the DTA measure as an independent variable (D. P. Cooper, Goldenberg, & Arndt, 2011, Study 1; Zhou, Liu, Chen, & Yu, 2008, Study 1), or for use of within-subjects manipulations of delay following threats (Mikulincer & Florian, 2000, Study 1; Simon et al., 1997, Study 4; Trafimow & Hughes, 2012, Studies 1, 2, and 6; Ullrich & Cohrs, 2008, Study 3). Within-subjects manipulations were excluded because it is possible that completing one DTA measure could affect subsequent DTA measures, whether in terms of activation or suppression of DTA, and therefore, we reasoned that between-subjects manipulations would more clearly capture the effects of a single experimental manipulation of threat. Studies were also excluded if they did not manipulate threat in a clear fashion or for not including a theoretically neutral control condition regarding the question at hand (Arndt, Greenberg, Simon, Pyszczynski, & Solomon, 1998, Study 2; Arndt, Greenberg, Solomon, et al., 1997, Study 3; Chatard & Selimbegović, 2011, Studies 1 and 2; King, Hicks, & Abdelkhalik, 2009, Studies 1, 2, and 3; Martens, Greenberg, Schimel, & Landau, 2004, Study 1; Mikulincer & Florian, 2002, Study 3; O’Connor & McFadden, 2012, Study 1; Schmeichel et al., 2009, Study 3; Schmeichel & Martens, 2005; Silvia, 2001, Study 1).

Studies were excluded if they allowed for the possibility of defense of values between the threat exposure and the measurement of DTA (F. Cohen, Soenke, Solomon, & Greenberg, 2013, Study 4; F. Cohen, Sullivan, Solomon, Greenberg, & Ogilvie, 2011, Study 5; Cox et al., 2008, Study 1; Davis, Juhl, & Routledge, 2011, Study 1; Edmondson et al., 2011, Study 2; Goldenberg, Cox, Pyszczynski, Greenberg, & Solomon, 2002, Study 1; Goldenberg, Pyszczynski, McCoy, Greenberg, & Solomon, 1999, Study 3; Harmon-Jones et al., 1997, Study 3; Hayes, Schimel, Faucher, & Williams, 2008, Study 3; Hayes, Schimel, & Williams, 2008, Study 1; Mikulincer & Florian, 2000, Study 1; Mikulincer & Florian, 2002, Study 3; Norenzayan, Dar-Nimrod, Hansen, & Proulx, 2009, Study 3; Schimel et al., 2007, Study 2; Vaes, Heflick, & Goldenberg, 2010, Study 3; Williams, Schimel, Hayes, & Martens, 2010, Study 3; Wojtkowiak & Rutjens, 2011, Study 1). This exclusion criterion was used because defense of values and self-affirmation have been shown to prevent increases in DTA (Davis et al., 2011; Greenberg et al., 1990), and so reflect theoretically predicted reduced DTA effect-sizes. Indeed, supporting this rationale, random-effects meta-regression revealed that studies permitting defense between threat and DTA measurement (k = 17) achieved a significantly smaller DTA effect-size (g = .27, 95% confidence interval [CI] = [0.05, 0.5]; p < .001) than those in the primary sample, t(119) = −2.41, p = .02.

Because a major aim of this study was to investigate the impact of delay on DTA, studies were excluded for use of cognitive load instead of delay for two reasons (Arndt et al., 2007, Study 2). First, it is difficult to map cognitive load onto the same dimension as delay. Second, there is evidence that ego-depletion can have a causal effect on TMT processes (Gailliot, Schmeichel, & Baumeister, 2006), which may suggest that cognitive load directly affects rather than moderates DTA. Thus, because we have reason to believe that cognitive load may function differently than delay, we focused only on delay in the present study.

Studies were excluded if they used explicit measures of DTA, such as the Death-Anxiety Scale (Gailliot et al., 2006, Study 3; Trafimow & Hughes, 2012, Study 5). These were excluded because TMT makes a critical distinction between explicit and implicit thoughts of death, and the former is opposed to theoretical perspectives on DTA (Pyszczynski, Greenberg, & Solomon, 1999). Studies were also excluded if they involved interactions that could not be broken down into a pure main effect (i.e., the manipulation was unavoidably either a double-threat or a threat + affirmation; for example, Agroskin & Jonas, 2013, Study 3; Cox et al., 2008, Study 1; Goldenberg et al., 2002, Study 1; Hirschberger, 2006, Study 4; Koole & Van den Berg, 2005, Study 4; Martin & Kamins, 2010, Study 1 pretest), as well as for use of unusually long-term delay (Cox, Reid-Arndt, Arndt, & Moser, 2012, Study 1).

Finally, in the case of studies that met criteria but did not include sufficient information for calculating effect-sizes, authors were emailed. In some cases, the needed data were obtained. If sufficient data were not obtained, the study was excluded. The only exception to this rule was for studies that did not provide condition-level ns, but reported a study-level n and also the total number of conditions in the study. In this case, condition-level ns were estimated by dividing the study-level n by the number of conditions.

Coding of Studies

Studies were coded on a variety of factors. First, independent variables were coded as being either MS, death-reminders, or not explicitly related to death (e.g., subliminal primes, cultural value threats). Studies using a manipulation that focused on personal mortality, such as the classic MS manipulation, were coded in the MS category (e.g., Arndt et al., 2007; Florian, Mikulincer, & Hirschberger, 2001). Studies that made a simple reference to death or death-related violence were coded death-reminders. Importantly, some of these may have been intended to threaten cultural values, but did so by means of exposure to some death-related concept (e.g., terrorism in one’s own country; Das, Bushman, Bezemer, Kerkhof, & Vermeulen, 2009). Finally, all other threats were classified as not explicitly death-related (e.g., Schimel et al., 2007).

If studies included more than one of these types of threats, an effect-size was calculated for each threat type.¹ If studies included multiple of the same type of threat, the threat condition that was most consistent with theory and/or other research was selected for use (e.g., standard MS instead of suicide-related MS; Fritsche et al., 2008). The exception to this was when a study used the same threat type twice, but with different delays and control conditions with respective delays. Our reason for selecting between similar threats rather than pooling them was to avoid ambiguity and to maintain theoretical consistency.

Similarly, when studies utilized multiple control conditions, the most conservative control condition was selected for use in two steps. First, when possible, we avoided control conditions that were likely to affirm personal values, which could reduce DTA in the control condition and thus inflate the DTA effect-size. For example, in a sample of Americans, we assumed that reading about a local synagogue was less affirming than reading about a church, given that Americans are predominantly Christian (F. Cohen et al., 2013), and that pictures of intact buildings were less affirming than pictures of buildings being constructed (as construction denotes human progress; Vail et al., 2012). There were few studies that required this decision, and they generally tested the effect of value threats on DTA. Second, when possible, we selected the control condition that was most likely to be related to pain (e.g., dental pain salience rather than television salience) or negative emotion (e.g., exam anxiety over watching television; anger, Chatard & Selimbegović, 2011), except in the case of body-related disgust (e.g., dust mites cause less body-related disgust than bed bugs; Burris & Rempel, 2004). This is because general negative emotion has been argued to be an important control, although evidence suggests it does not affect TMT processes (Greenberg et al., 1995), except in the case of fear of death (e.g., Mikulincer & Florian, 2000), and sometimes disgust (see Goldenberg et al., 2001; but cf. Fessler & Navarrete, 2005).

Second, studies were also coded on the type of DTA operationalization they had used (word fragment, reaction time, or free writing). Word fragment studies were additionally coded for the number of death-related word fragments that they included, and the total number of neutral word fragments they included.

Finally, studies were also coded on their use of delay. Previous TMT meta-analyses have used two different methods for coding delay, each involving unique limitations. First, the number of tasks that participants completed during the delay has been operationalized as delay length (Burke et al., 2010). The potential problem with this method is that it is impossible to be certain that all delay tasks are equivalent in length. As a result, it is perhaps more accurate to describe this coding method as task-switching than delay length. Alternatively, previous studies have used dichotomous coding of whether the study included some delay or no delay (Martens et al., 2011). This does address the issue of potentially non-ordinal data. However, dichotomizing a continuous variable naturally results in a loss of statistical power, which might mask effects. Because of the unique limitations of each of these methods, both were used in the present study.

In addition, the present study introduces a third method for coding delay: estimated delay length (seconds). To estimate delay length, studies were coded on the types of delay tasks they used, and authors’ estimations of total delay length. If any of the delay tasks were Likert-style questionnaires, the studies were also coded on the number of questionnaire items included during the delay. We then estimated length data for each delay task type using data from an independent sample (n = 43), in which we timed participants on typical TMT delay tasks (i.e., the Positive and Negative Affect Schedule [PANAS] and reading an excerpt from “The Growing Stone” by Albert Camus), as well as author’s reports of delay length (see Table 1 for descriptive statistics regarding delay task length estimates). We used this information to impute total delay length estimates for each study. Specifically, authors’ reports were used to impute delay estimates for use of puzzles as delay (3.5 min; Carey & Sarma, 2011; Maxfield et al., 2007). Data from the independent sample were used to impute values for the average questionnaire item length (to be multiplied by the number of questionnaire items that were used during the study’s delay), and also for the use of written passages as a delay task. Any other type of delay task (k = 5 of total k = 99) was given the average value of all “other” delay task types (141.05 s). For studies that used multiple delay tasks, the length estimates for each delay task were summed.

Table 1.

Delay Length Estimates by Task Type.

Delay task type	Derived from	M/estimated seconds	SD
Passage	Independent sample (n = 43)	162.32	57.95
One questionnaire item	Independent sample (n = 43)	3.76	0.35
PANAS (20 item)	Independent sample (n = 43)	75.2	20.71
PANAS-X (60 item)	Length of Item × Number of Items	225.6	—
Puzzle	Authors’ reports (3.5 min)	210	—
Other	Average task length	141.05	—

Note. PANAS = Positive and Negative Affect Schedule.

One limitation to operationalizing delay length in this manner is that imputing mean delay length by task type cannot account for variability within each delay task type (as indicated by the standard deviations in Table 1; H. Cooper, Hedges, & Valentine, 2009). However, because of studies’ varied use of delay task types and combinations of different delay tasks, this method still results in noticeable variability in delay length estimates across studies (m = 120.57 s, SD = 147.32). Thus, this technique should significantly increase both the likelihood that our delay variable is ordinal and statistical power to detect associations with delay. Because each of the three methods of operationalizing delay is associated with its own limitations, using all three methods (delay vs. no delay, task-switching, and estimated delay length) should provide the clearest picture of how delay relates to effect-sizes across studies.

A summary of studies and their codes can be found in Table 2.

Table 2.

DTA Effect-Size, Threat Type, and Delay Codes for DTA Studies.

Authorship	Year	Study	Threat code	Delay tasks	Delay length (s)	Sample size	g	SE
Arndt, Cook, Goldenberg, and Cox	2007	1	MS	2	420	33	0.88	.36
Arndt, Cook, Goldenberg, and Cox	2007	1	DR	2	420	32	0.01	.34
Arndt, Cook, Goldenberg, and Cox	2007	2	DR	2	420	20	−0.4	.43
Arndt, Greenberg, Pyszczynski, and Solomon	1997	1	NEDR	0	0	22	1.07	.44
Arndt, Greenberg, Pyszczynski, and Solomon	1997	3	NEDR	0	0	25	0.94	.41
Burris and Rempel	2004	6	DR	0	0	67	0.56	.25
Burris and Rempel	2004	6 F	DR	0	0	29	0.69	.37
Burris and Rempel	2004	6 F	DR	1	214.82	28	0.61	.38
Carey and Sarma	2011	1	DR	1	210	80	1.16	.24
Chatard and Selimbegović	2011	5	NEDR	0	0	28	−0.07	.37
Chatard et al.	2012	1	MS	0	0	105	0.01	.19
Cohen, Soenke, Solomon, and Greenberg	2013	2	MS	2	387.92	67	0.167	.28
Cohen, Soenke, Solomon, and Greenberg	2013	2	NEDR	2	387.92	67	1.92	.29
Cohen, Soenke, Solomon, and Greenberg	2013	3	MS	2	387.92	28	1.02	.39
Cohen, Soenke, Solomon, and Greenberg	2013	3	NEDR	2	387.92	26	1.57	.44
Cohen, Soenke, Solomon, and Greenberg	2013	4	MS	2	387.92	41	1.07	.33
Cox et al.	2008	1	MS	2	285.2	114	0.87	.34
Das, Bushman, Bezemer, Kerkhof, and Vermeulen	2009	1	DR	0	0	44	0.35	.28
Das, Bushman, Bezemer, Kerkhof, and Vermeulen	2009	1	DR	0	0	56	0.22	.28
Das, Bushman, Bezemer, Kerkhof, and Vermeulen	2009	2	DR	0	0	101	0.5	.2
Echebarria-Echabe	2013	1	MS	2	365.22	45	1	.31
Echebarria-Echabe	2013	1	NEDR	2	365.22	45	0.23	.3
Echebarria-Echabe	2013	2	MS	2	365.22	45	1.3	.32
Echebarria-Echabe	2013	2	NEDR	2	365.22	45	0.15	.29
Edmondson et al.	2011	1	MS	0	0	125	0.76	.18
Edmondson et al.	2011	1	MS	2	420	179	0.68	.15
Edmondson et al.	2011	1	NEDR	0	0	127	0.18	.18
Edmondson et al.	2011	1	NEDR	2	420	174	0.21	.15
Edmondson et al.	2011	2	MS	0	0	142	0.42	.17
Florian, Mikulincer, and Hirschberger	2001	2	MS	1	71.44	120	0.52	.18
Florian, Mikulincer, and Hirschberger	2002	3	NEDR	0	0	66	0.71	.22
Fritsche, Jonas, and Fankhänel	2008	4	MS	2	116.56	57	0.43	.26
Fritsche et al.	2007	2	MS	1	75.2	140	0.28	.17
Gailliot, Schmeichel, and Baumeister	2006	2	NEDR	1	86.48	19	0.94	.46
Gailliot, Schmeichel, and Maner	2007	1	NEDR	1	60.16	46	1.06	.48
Goldenberg, Arndt, Hart, and Routledge	2008	2 P	NEDR	0	0	34	0.65	.34
Goldenberg, Pyszczynski, McCoy, Greenberg, and Solomon	1999	2	NEDR	0	0	38	0.32	.23
Goldenberg, Pyszczynski, McCoy, Greenberg, and Solomon	1999	3	NEDR	0	0	26	0.57	.28
Golec de Zavala, Cichocka, Orehek, and Abdollahi	2012	1	MS	0	0	158	0.32	.16
Grover, Miller, Solomon, Webster, and Saucier	2010	1	MS	2	169.8	103	0.43	.2
Grover, Miller, Solomon, Webster, and Saucier	2010	1	DR	2	169.8	103	0.43	.2
Harmon-Jones et al.	1997	3	MS	1	225.6	26	0.39	.38
Harmon-Jones et al.	1997	3	MS	2	387.9	25	2.35	.5
Hayes, Schimel, and Williams	2008	1	NEDR	0	0	50	0.62	.29
Hayes, Schimel, Faucher, and Williams	2008	1	NEDR	0	0	38	0.46	.32
Hayes, Schimel, Faucher, and Williams	2008	2	NEDR	0	0	30	0.88	.37
Hayes, Schimel, Faucher, and Williams	2008	3	NEDR	0	0	65	0.36	.25
Hirschberger, Ein-Dor, and Almakias	2008	1 P	NEDR	1	75.2	27	0.96	.4
Hirschberger, Ein-Dor, and Almakias	2008	2 P	DR	1	75.2	40	0.72	.32
Hirschberger, Florian, and Mikulincer	2005	3	NEDR	0	0	31	0.15	.18
Jessop and Wade	2008	1	MS	1	210	39	0.67	.32
Jessop and Wade	2008	1	DR	1	210	39	0.92	.33
Jessop, Albery, Rutter, and Garrod	2008	4	DR	1	210	61	0.46	.26
Jonas and Fischer	2006	3	MS	1	75.2	50	0.1	.28
Landau et al.	2004	2	NEDR	0	0	31	0.01	.28
Landau et al.	2004	2	NEDR	0	0	31	0.96	.37
Ma-Kellams and Blascovich	2012	1	MS	1	75.2	63	0.35	.25
Maxfield et al.	2007	1	MS	1	210	116	−0.21	.26
Mikulincer and Florian	2000	2	MS	0	0	87	0.15	.17
Mikulincer and Florian	2000	3	MS	1	82.7	83	1.38	.19
Mikulincer, Florian, Birnbaum, and Malishkevich	2002	1	NEDR	0	0	72	0.65	.24
Mikulincer, Florian, Birnbaum, and Malishkevich	2002	1	DR	0	0	72	0.8	.24
Mikulincer, Florian, Birnbaum, and Malishkevich	2002	2	NEDR	0	0	60	0.85	.27
Mikulincer, Florian, Birnbaum, and Malishkevich	2002	3	NEDR	0	0	58	0.84	.23
Motyl et al.	2012	1	NEDR	1	90	36	0.16	.24
Motyl et al.	2012	3	NEDR	1	75.2	62	−0.4	.33
Navarrete, Kurzban, Fessler, and Kirkpatrick	2004	2	MS	1	225.6	50	0.62	.29
Navarrete, Kurzban, Fessler, and Kirkpatrick	2004	2	NEDR	1	225.6	51	0.27	.28
Proulx and Heine	2008	1b	NEDR	1	75.2	40	−0.15	.31
Proulx and Heine	2009	1	NEDR	1	75.2	40	−0.37	.31
Proulx and Heine	2009	2	NEDR	1	75.2	53	−0.11	.27
Routledge, Arndt, Sedikides, and Wildschut	2008	3	MS	1	162.32	38	0.69	.33
Rutjens and Loseman	2010	1	MS	1	75.2	52	1.21	.3
Rutjens and Loseman	2010	1	NEDR	1	75.2	52	0.3	.27
Rutjens, van der Pligt, and Harreveld	2009	1	MS	2	285.2	53	0.29	.28
Rutjens, van der Pligt, and Harreveld	2009	2	NEDR	0	0	43	0.8	.31
Rutjens, van der Pligt, and Harreveld	2009	3	MS	1	75.2	45	0.72	.3
Schimel, Hayes, Williams, and Jahrig	2007	1	NEDR	2	435.6	31	0.22	.35
Schimel, Hayes, Williams, and Jahrig	2007	1	NEDR	0	0	30	1.08	.38
Schimel, Hayes, Williams, and Jahrig	2007	3	NEDR	0	0	38	1.03	.34
Schimel, Hayes, Williams, and Jahrig	2007	4	NEDR	0	0	35	1.07	.35
Schimel, Hayes, Williams, and Jahrig	2007	5	NEDR	0	0	40	0.88	.33
Shehryar and Hunt	2005	1 P	DR	0	0	49	0.83	.29
Silvia	2001	1	NEDR	1	3.76	20	0.38	.43
Taubman-Ben-Ari	2004	2	NEDR	0	0	68	0.98	.25
Taubman-Ben-Ari	2004	3	NEDR	0	0	74	0.74	.24
Taubman-Ben-Ari	2011	2	MS	0	0	46	0.89	.3
Taubman-Ben-Ari	2011	2	NEDR	0	0	45	0.97	.31
Taubman-Ben-Ari	2011	3	NEDR	0	0	51	0.82	.29
Taubman-Ben-Ari and Katz-Ben-Ami	2008	1	NEDR	0	0	60	0.62	.26
Trafimow and Hughes	2012	3	MS	2	162.32	60	0	.25
Trafimow and Hughes	2012	3	MS	0	0	60	0.31	.26
Ullrich and Cohrs	2008	3	DR	0	0	26	−0.1	.26
Vail, Arndt, Motyl, and Pyszczynski	2012	1	NEDR	2	285.2	25	0.86	.41
Vail, Arndt, Motyl, and Pyszczynski	2012	4	NEDR	0	0	26	1.15	.41
Van Beest, Williams, and Van Dijk	2011	2 P	DR	0	0	50	0.63	.29
Van Tongeren and Green	2010	1	NEDR	3	225.6	101	−0.13	.2
Van Tongeren and Green	2010	2	NEDR	2	319.6	122	−0.22	.18
Wojtkowiak and Rutjens	2011	1	MS	1	210	36	0.93	.35

Note. DTA = death-thought accessibility; MS = mortality salience; DR = death-reminder; NEDR = not explicitly death-related threat; F = follow-up study; P = pretest/pilot study.

Publication Bias and Small-Study Effects

The present analysis only included published studies, partly as a means of quality control (e.g., peer review) and partly because our coding of delay required specific procedural information that is relatively incidental, and so unlikely to be included with unpublished data. Publication bias and small-study effects were explored using funnel plots (Sterne & Harbord, 2004) and p-curve analysis (Simonsohn, Nelson, & Simmons, 2014). These analyses were not only conducted on the full sample of effect-sizes but also broken down by threat type, and by delay within each threat type. The funnel plots revealed some small-study effects. However, the p-curve analyses suggest that these small-study effects were not driven by a preferential publishing of findings near the cutoff for statistical significance. Moreover, approximately 43% of the effects in the present study were published although they were not statistically significant, which can perhaps be explained by the fact that DTA was often of secondary rather than primary interest in the publications. Importantly, small-study effects did not appear to be meaningfully associated with studies’ use of delay.²

Results

Overall Effect-Size and Comparison of Threats

Because the studies to be analyzed involve diversity in sample characteristics and method, we used a random-effects model for effect-size estimates (H. Cooper et al., 2009; DerSimonian & Laird, 1986). This decision was supported by a test revealing that effect-sizes were not homogeneous across all studies, k = 99; Q_T(98) = 274.1, p < .001, with other statistics indicating a notable degree of heterogeneity in the effect-sizes (T² = .13; I² = 64.2%). Besides justifying the use of a random-effects model, the existence of heterogeneity in effect-sizes also confirms an important auxiliary hypothesis—that variance in effect-sizes exists—to test for potential moderation across studies. Overall, there was a moderate effect of all experimental manipulations on DTA effect-size (g = .57, 95% CI = [.48, .66], p < .001).

Next, we analyzed studies by type of threat and compared them using the Analog with the ANOVA (Q test), which revealed that MS (MS; k = 31), death-reminders (k = 20), and other types of threats (k = 48) significantly differed in their effect on DTA, Q_B(2) = 10.55, p < .05, (see Table 3 for effect-size estimates). Direct group comparisons using random-effects meta-regression revealed that MS studies resulted in marginally larger effect-sizes on DTA than death-reminder studies (β = .25, SE = .13), t(50) = 1.97, p = .054. However, MS studies did not display larger effect-sizes than studies of threats that were not explicitly related to death (β = .16, SE = .11), t(78) = 1.51, p = .14, and effect-sizes did not significantly differ between death-reminders and studies without explicitly death-related threats (β = −.09, SE = .12) t(67) = −0.71, p = .48.

Table 3.

DTA Effect-Size Estimates by Threat Type.

Threat	k	g	95% CI	T ²
All threat types	99	.57	[.48, .66]	.12
Death-related	51	.6	[.48, .72]	.11
Mortality salience	31	.7	[.54,.85]	.12
Death-reminders	20	.44	[.28, .61]	.07
Not explicitly death-related	48	.54	[.4, .67]	.14

Note. T² represents tau-squared, with the square root of this number reflecting the estimated standard deviation of underlying effects across studies. DTA = death-thought accessibility; CI = confidence interval.

Effect-Size by Study Characteristics

Analyses were conducted to determine whether differences in studies’ methodological or procedural characteristics were associated with the effect-sizes achieved, including differences in operationalization of DTA, delay length, and in a subsample of MS studies, the control condition used.

DTA measurement

Studies operationalized DTA in three different ways: word fragment completion tasks (k = 91), reaction time to recognize death-related words (k = 5), and the number of death-related references in free writing–type tasks (k = 3). Because the sample of studies was small for the free-writing group, we excluded it from comparative analysis. Analysis revealed that studies using the word fragment task did not differ in effect-size from those that used a reaction-time paradigm (β = .19, SE = .23), t(95) = 0.81, p = .42.

Specifically regarding the word fragment completion task, studies tended to include different numbers of death-related word fragments in the measure. Measures with more death-related word fragments may be more or less likely to capture DTA effects, especially if the early death-related items prime or suppress identification of later ones. As well, having more neutral word fragments—space between the death-related fragments—could enhance or diminish this effect. Thus, hierarchical meta-regression was used to, first, regress effect-sizes on the number of death-related word fragments, and second, regress effect-sizes on death-related fragments while controlling for the number of neutral word fragments. The number of death-related word fragments did not predict DTA effect-size either by itself (β = −.005, SE = .02), t(83) = −0.29, p = .77, or when the number of neutral fragments was entered as a covariate (b = −.03, SE = .03), t(77) = −0.97, p = .33.

Delay between threats and DTA measurement

Having shown that DTA was not influenced by the type of DTA measure used, we collapsed across DTA measure type to investigate the role of delay on DTA. Three different operationalizations of delay were used, as described earlier: whether studies included a delay task or not (Martens et al., 2011), the number of delay tasks that were included in the study, representing task-switching (Burke et al., 2010), and estimated delay length (seconds). Importantly, only a few studies included three delay tasks (the maximum), which might serve as outliers. However, forcing a coding cap of two delay tasks did not change any results in terms of statistical significance, and so we present all findings using the original coding of up to three delay tasks.

As a preliminary step, analyses of the general effects of delay on DTA, regardless of threat type, were conducted (k = 99). There was no significant effect of delay on DTA, whether delay was assessed as inclusion of any delay (β = −.04, SE = .09; t = −0.43, p = .67), task-switching (β = −.007, SE = .06; t = 0.12, p = .9), or estimated delay length (β = .0003, SE = .0003; t = 0.91, p = .37). This finding is consistent with TMT, which suggests that delay should not have a main effect on DTA independent of the type of threat.

Next, analyses were conducted to examine whether delay predicts DTA following specific types of threats. Because in the present study we found effect-size differences between MS and death-reminders, we ran analyses at three levels to examine this relationship. Specifically, we sought to determine whether delay was positively associated with DTA following any type of death-related threat, following MS threats only, and following death-reminders only. Finally, we tested for the influence of delay on DTA following threats that were not explicitly death-related.

All death-related threats

Regressing DTA effect-sizes on delay for all types of death-related threats (both MS and death-reminders; k = 51) revealed marginally significant, positive effects of delay for studies with no delay tasks versus studies with delay tasks (β = .25, SE = .13; t = 1.92, p = .06). Analysis also revealed a significant positive effect of task-switching (β = .17, SE = .07; t = 2.41, p = .02) and estimated delay length (β = .001, SE = .0004; t = 2.48, p = .007; see Figure 2) in predicting the effect of all death-related threats on DTA. A summary of these findings can be found in Table 4. These findings are generally consistent with the dual-process component of TMT. In contrast, the effect-size at the intercept (no delay) was significantly larger than g = 0 (k = 16, g = .43, 95% CI = [.29, .57], p < .001), suggesting that immediate DTA suppression may not be occurring, which is inconsistent with the dual-process component.

Figure 2.

Effect-size by estimated delay length (seconds) for all death-related threats.

Table 4.

Independent Analyses of Delay as a Moderator of DTA Effect-Size, by Threat Type.

Type of threat	k	β	SE	t	p
All death-related threats
No delay vs. delay	51	.25	.13	1.92	.06
Task-switching	51	.17	.07	2.41	.02
Estimated delay length (s)	51	.001	.0004	2.84	.007
Mortality salience only
No delay vs. delay	31	.34	.19	1.79	.08
Task-switching	31	.25	.1	2.65	.01
Estimated delay length (s)	31	.002	.0005	3.92	<.001
Death-reminders only
No delay vs. delay	20	.002	.18	0.01	.99
Task-switching	20	−.05	.12	−0.44	.67
Estimated delay length (s)	20	−.0006	.0007	−0.83	.42
Not explicitly death-related threats
No delay vs. delay	48	−.4	.13	−2.99	.004
Task-switching	48	−.19	.08	−2.45	.02
Estimated delay length (s)	48	−.0007	.0005	−1.46	.15

Note. DTA = death-thought accessibility.

Mortality salience

Conducting similar analyses on only MS studies (k = 31) provided similar results. MS studies’ inclusion of a delay versus no delay did not significantly predict DTA effect-size (β = .34, SE = .19; t = 1.79, p = .08). However, DTA effect-size was significantly predicted by both task-switching (β = .25, SE = .1; t = 2.65, p = .01) and estimated delay length (β = .002, SE = .0004; t = 3.92, p < .001; see Figure 3). Although the sample was small (k = 6), observing effect-size of MS on DTA at the intercept (no delay) revealed an effect that was significantly larger than g = 0 (g = .42, 95% CI = [.18, .66], p = .001).

Figure 3.

Effect-size by estimated delay length for mortality salience threats.

Death-reminders

Finally, regressing effect-sizes on delay for only death-reminders (k = 20) did not reveal any significant effects (all ps > .4; see Table 4 for estimates). This may imply that there is something unique about personal reflections on MS as compared with other reminders of death, but the limitations of this interpretation will be highlighted in the discussion. Again, the estimated effect-size at the intercept (no delay) was significantly larger than g = 0 (k = 10, g = .43, 95% CI = [.24, .62], p < .001).

Not explicitly death-related threats

Besides death-related threats, TMT also proposes that threats that are not explicit related to death, such as cultural value threats, can increase DTA. In addition, some research indicates that the effect of these threats on DTA decreases following a delay (Martens et al., 2011; Schimel et al., 2007). Thus, DTA effect-sizes were regressed on studies’ use of delay to test for the effect of delay following threats that are not explicitly related to death (k = 48). Results indicate that studies including a delay versus no delay (β = −.4, SE = .13; t = −2.99, p = .004) and more task-switching (β = −.19, SE = .08; t = −2.45, p = .02; see Figure 4) tended to achieve a smaller DTA effect-size following threats that are not explicitly related to death (see Table 4 for estimates). However, estimated length of delay did not predict significant differences in DTA (β = −.0007, SE = .0005; t = −1.46, p = .15).

Figure 4.

Effect-size by task-switching for not explicitly death-related threats.

Control conditions

In many cases, studies used unique control conditions, making meta-analytic comparison of control conditions difficult. The exception to this is MS studies, which often used either dental pain salience (k = 21) or television salience (k = 5) as a control (i.e., think about dental pain or watching television). Traditionally, the distinction between these two control conditions is important, as it is expected that pain can control for general negative experience but television watching cannot (Greenberg et al., 1995). Comparison of MS studies with these two types of control conditions revealed that the control conditions did not predict differences in DTA (β = −.29, SE = .25; t = −1.14, p = .26), even when estimated delay length was entered as a covariate (b = −.32, SE = .2; t = −1.59, p = .13). This suggests that, at least in the case of studies of MS’s effect on DTA, general negative emotional experience may not be a crucial control.

Discussion

The present analysis sought to test two aspects of the dual-process component of TMT, and a third related hypothesis. First, the dual-process component of TMT argues that following explicitly death-related threats: (a) DTA should be immediately suppressed, but that (b) after a delay, this suppression should subside and DTA differences should emerge. In contrast, we additionally hypothesized that delay would be negatively associated with DTA effect-size in the case of threats that are not explicitly related to death.

Before accounting for delay, studies of the effect of TMT threats on DTA seem to have resulted in a moderately strong effect of TMT threats on DTA (g = .57). Analysis also revealed that MS manipulations seem to lead to greater DTA than death-reminders. This seems to suggest that reflecting on one’s personal mortality leads to particularly high levels of DTA. One critical limitation to this interpretation of the present data, however, is that these studies are independent and used a variety of control conditions. It is possible that different types of control conditions in each class of threat may drive the difference in effect-size. This perspective is supported by the fact that MS studies tend to use relatively similar control conditions (e.g., two questions about dental pain; two questions about watching television), whereas death-reminder and other threat studies tend to use control conditions that are unique to the study. Although we present some evidence that neutral versus negative controls do not influence effects in the case of MS and DTA, it is unclear that this should be extrapolated to, for example, death-reminders. We do acknowledge that we computed effect-sizes using the most conservative control conditions available, and so attempted to minimize the likelihood that the effects of threat would be confounded with the nature of the control conditions.

The present study also observed whether there were differences in DTA effect-size as a function of DTA’s operationalization. Interestingly, the word fragment completion task and reaction-time measures of DTA did not predict differences in DTA effect-size. As well, the number of word fragments in the word fragment completion task was not associated with different effect-sizes. This latter finding in particular may be conflated with culture, lab, or language, however. For example, the frequency at which certain death-related words or concepts are commonly used may vary by language, which might affect priming sensitivity.

In terms of the immediate suppression of DTA following death-related threats, there seem to be several instances where DTA was not immediately suppressed. Although the dual-process component does not articulate that there will be no effect of DTA immediately following death-related threats, DTA suppression does imply this. One possible explanation for immediate, but smaller DTA effect-size following death-related threats is that only strongly death-related concepts are immediately suppressed, whereas weakly death-related ones are not. Alternatively, it may be that only strongly death-related concepts are immediately primed, but that a spreading activation to weakly death-related concepts is temporarily stifled.

As for delay effects, the present study found that studies including more task-switching or a longer delay following MS tended to achieve larger DTA effects. In the case of threats that were not explicitly related to death, including any delay or more task-switching during the delay was associated with smaller DTA effects. These findings are fairly consistent with the dual-process component’s predictions regarding the effect of delay (Pyszczynski, Greenberg, & Solomon, 1999), as well as with studies that imply a negative association between delay and DTA following threats that are not explicitly death-related (Martens et al., 2011).

Limitations

There are several limitations to the present study. First, the present analysis only included a subset of all DTA studies. Specifically, only studies that were both experimental and met strict theoretical design criteria (aside from delay) for increasing DTA were included. Because we used strict criteria for excluding studies based on theoretical design, the present study offers a rigorous test of TMT’s predictions regarding DTA, but does not account for other manipulations that could influence DTA. In addition, because some DTA studies were excluded for having designs that should lead to theoretically predicted reductions in DTA effect-size, it is unclear whether these studies might have detected differences in DTA, or the extent to which they reduced DTA effects. (Notably, reductions in DTA were the objective of some of these studies; for example, F. Cohen et al., 2011.)

All comparative analyses in the present study are strictly correlational. These tests may suggest conditions under which threats are most likely to lead to causal increases in DTA. However, the present tests comparing different types of threats or studies’ methodological and procedural characteristics do not imply causal influence of these factors on DTA. Similarly, because these comparisons are correlational, neither do these results imply that causal relationships are not underlying these relationships, even when the correlation appears null. Last, the analyses were based only on published studies. Although it is unclear whether biases may have favored the publication of results consistent with the TMT perspective of delay on DTA, it is important to recognize that nearly all of these studies were not explicitly testing the role of delay on DTA (cf. Schimel et al., 2007; Trafimow & Hughes, 2012). Rather the majority of studies were simply testing DTA responses to threats, but happened to contain information useful for testing the overall relationship between delay length and the magnitude of DTA. As well, we did not find any evidence that studies’ use of delay was associated with small-study effects.

Finally, it is important to note that the present data cannot control for the possibility that particular types of delay tasks (e.g., PANAS) influence DTA independently of delay length. However, we emphasize that studies used a variety of different delay tasks (different questionnaires, passages, puzzles, and other tasks), and that many studies included no delay task or multiple types of delay tasks, which should minimize the influence of any particular type of delay task.

Future Directions

The present study provides evidence that not all threats are the same in terms of DTA. They may lead to different effect-sizes and also may interact with delay differently (positive vs. negative effect on DTA). Especially interesting is that there might be a distinction between the effects of MS and that of death-reminders, both in terms of effect-size and delay. This could suggest that participants who are less engaged in MS express DTA in a way that is more similar to the effects of death-reminders (smaller effect, no association with delay). It would be worthwhile for future research to explore how impersonal death-reminders are different than moderate to deep elaboration on personal mortality. This may provide the opportunity to expand TMT’s predictive power, but is also practically useful because death-reminders are quite common in the context of daily living.

Second, the present study presents strong evidence that, across a wide range of studies, delay seems to share a positive relationship with DTA following MS. Thus, it is striking that researchers have sometimes consistently found the opposite effect of delay following MS (Trafimow & Hughes, 2012). Most of the studies conducted by Trafimow and Hughes (2012) did not meet the design and methodological specifications of the current study (e.g., some were not between-subjects), and thus could be considered anomalies within the larger context of DTA research.

However, the current results indicate that DTA is not truly or fully suppressed immediately following MS. Although this cannot explain the delayed reductions in DTA observed by Trafimow and Hughes (2012), it does agree with their finding immediate effects of MS on DTA. The present finding that DTA is not suppressed may reflect chance, but highlights the need for closer inspection of DTA activation and suppression immediately following MS. One possible explanation for these findings is that DTA is only partially suppressed. For example, perhaps only the most strongly or weakly death-related words are initially suppressed. If this is the case, it could lead to unique predictions and warrant some re-specifications of TMT. For example, partial DTA suppression could imply that proximal defenses are actually associated with activation of strongly death-related concepts in the absence of weakly death-related concepts, whereas distal defenses are associated with a spreading activation of weakly death-related concepts. Thus, TMT may benefit from greater specification and more precise measurement of DTA.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Supplemental Material

The online supplemental material is available at

Notes

References

Agroskin

Jonas

(2013). Controlling death by defending ingroups—Mediational insights into terror management and control restoration. Journal of Experimental Social Psychology, 49, 1144-1158.

*Arndt

Cook

Goldenberg

J. L.

Cox

C. R.

(2007). Cancer and the threat of death: The cognitive dynamics of death-thought suppression and its impact on behavioral health intentions. Journal of Personality and Social Psychology, 92, 12-29.

*Arndt

Greenberg

Pyszczynski

Solomon

(1997). Subliminal exposure to death-related stimuli increases defense of the cultural worldview. Psychological Science, 8, 379-385.

Arndt

Greenberg

Simon

Pyszczynski

Solomon

(1998). Terror management and self-awareness: Evidence that mortality salience provokes avoidance of the self-focused state. Personality and Social Psychology Bulletin, 24, 1216-1227.

Arndt

Greenberg

Solomon

Pyszczynski

Simon

(1997). Suppression, accessibility of death-related thoughts, and cultural worldview defense: Exploring the psychodynamics of terror management. Journal of Personality and Social Psychology, 73, 5-18.

Borenstein

Hedges

Higgins

Rothstein

(2005). Comprehensive meta-analysis (Version 2). Englewood, NJ: Biostat.

Burke

B. L.

Martens

Faucher

E. H.

(2010). Two decades of terror management theory: A meta-analysis of mortality salience research. Personality and Social Psychology Review, 14, 155-195.

*Burris

C. T.

Rempel

J. K.

(2004). “It’s the end of the world as we know it”: Threat and the spatial-symbolic self. Journal of Personality and Social Psychology, 86, 19-42.

*Carey

R. N.

Sarma

K. M.

(2011). The impact of threat appeal messages on risky driving intentions: A terror management theory perspective. Journal of the Australasian College of Road Safety, 22(4), 51-56.

10.

*Chatard

Pyszczynski

Arndt

Selimbegović

Konan

P. N.

Van der Linden

(2012). Extent of trauma exposure and PTSD symptom severity as predictors of anxiety-buffer functioning. Psychological Trauma: Theory, Research, Practice, and Policy, 4, 47-55.

11.

*Chatard

Selimbegović

(2011). When self-destructive thoughts flash through the mind: Failure to meet standards affects the accessibility of suicide-related thoughts. Journal of Personality and Social Psychology, 100, 587-605.

12.

*Cohen

Soenke

Solomon

Greenberg

(2013). Evidence for a role of death thought in American attitudes toward symbols of Islam. Journal of Experimental Social Psychology, 49, 189-194.

13.

Cohen

Sullivan

Solomon

Greenberg

Ogilvie

D. M.

(2011). Finding Everland: Flight fantasies and the desire to transcend mortality. Journal of Experimental Social Psychology, 47, 88-102.

14.

Cohen

(1988). Statistical power analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Lawrence Erlbaum.

15.

Cooper

D. P.

Goldenberg

J. L.

Arndt

(2011). Empowering the self: Using the terror management health model to promote breast self-examination. Self and Identity, 10, 315-325.

16.

Cooper

Hedges

L. V.

Valentine

J. C.

(2009). Handbook of research synthesis and meta-analysis (2nd ed.). New York, NY: Russell Sage Foundation.

17.

*Cox

C. R.

Arndt

Pyszczynski

Greenberg

Abdollahi

Solomon

(2008). Terror management and adults’ attachment to their parents: The safe haven remains. Journal of Personality and Social Psychology, 94, 696-717.

18.

Cox

C. R.

Reid-Arndt

S. A.

Arndt

Moser

R. P.

(2012). Considering the unspoken: The role of death cognition in quality of life among women with and without breast cancer. Journal of Psychosocial Oncology, 30, 128-139.

19.

*Das

Bushman

B. J.

Bezemer

M. D.

Kerkhof

Vermeulen

I. E.

(2009). How terrorism news reports increase prejudice against outgroups: A terror management account. Journal of Experimental Social Psychology, 45, 453-459.

20.

Davis

W. E.

Juhl

Routledge

(2011). Death and design: The terror management function of teleological beliefs. Motivation and Emotion, 35, 98-104.

21.

Dechesne

Pyszczynski

Arndt

Ransom

Sheldon

K. M.

Van Knippenberg

Janssen

(2003). Literal and symbolic immortality: The effect of evidence of literal immortality on self-esteem striving in response to mortality salience. Journal of Personality and Social Psychology, 84, 722-737.

22.

DerSimonian

Laird

(1986). Meta-analysis in clinical trials. Controlled Clinical Trials, 7, 177-188.

23.

Dunkel

C. S.

(2009). The association between thoughts of defecation and thoughts of death. Death Studies, 33, 356-371.

24.

*Echebarria-Echabe

(2013). Mortality salience and uncertainty: Similar effects but different processes? European Journal of Social Psychology, 43, 185-191.

25.

*Edmondson

Chaudoir

S. R.

Mills

M. A.

Park

C. L.

Holub

Bartkowiak

J. M.

(2011). From shattered assumptions to weakened worldviews: Trauma symptoms signal anxiety buffer disruption. Journal of Loss and Trauma, 16, 358-385.

26.

Fessler

D. M.

Navarrete

C. D.

(2005). The effect of age on death disgust: Challenges to Terror management perspectives. Evolutionary Psychology, 3, 279-296.

27.

*Florian

Mikulincer

Hirschberger

(2001). An existentialist view on mortality salience effects: Personal hardiness, death-thought accessibility, and cultural worldview defense. British Journal of Social Psychology, 40, 437-453.

28.

*Florian

Mikulincer

Hirschberger

(2002). The anxiety-buffering function of close relationships: Evidence that relationship commitment acts as a terror management mechanism. Journal of Personality and Social Psychology, 82, 527-542.

29.

*Fritsche

Jonas

Fankhänel

(2008). The role of control motivation in mortality salience effects on ingroup support and defense. Journal of Personality and Social Psychology, 95, 524-541.

30.

*Fritsche

Jonas

Fischer

Koranyi

Berger

Fleischmann

(2007). Mortality salience and the desire for offspring. Journal of Experimental Social Psychology, 43(5), 753-762.

31.

*Gailliot

M. T.

Schmeichel

B. J.

Baumeister

R. F.

(2006). Self-regulatory processes defend against the threat of death: Effects of self-control depletion and trait self-control on thoughts and fears of dying. Journal of Personality and Social Psychology, 91, 49-62.

32.

*Gailliot

M. T.

Schmeichel

B. J.

Maner

J. K.

(2007). Differentiating the effects of self- control and self-esteem on reactions to mortality salience. Journal of Experimental Social Psychology, 43, 894-901.

33.

*Goldenberg

J. L.

Arndt

Hart

Routledge

(2008). Uncovering an existential barrier to breast self-exam behavior. Journal of Experimental Social Psychology, 44, 260-274.

34.

Goldenberg

J. L.

Cox

C. R.

Pyszczynski

Greenberg

Solomon

(2002). Understanding human ambivalence about sex: The effects of stripping sex of meaning. Journal of Sex Research, 39, 310-320.

35.

Goldenberg

J. L.

Pyszczynski

Greenberg

Solomon

Kluck

Cornwell

(2001). I am not an animal: Mortality salience, disgust, and the denial of human creatureliness. Journal of Experimental Psychology: General, 130, 427-435.

36.

*Goldenberg

J. L.

Pyszczynski

McCoy

S. K.

Greenberg

Solomon

(1999). Death, sex, love, and neuroticism: Why is sex such a problem? Journal of Personality and Social Psychology, 77, 1173-1187.

37.

*Golec

Zavala

Cichocka

Orehek

Abdollahi

(2012). Intrinsic religiosity reduces intergroup hostility under mortality salience. European Journal of Social Psychology, 42, 451-461.

38.

Greenberg

Arndt

Simon

Pyszczynski

Solomon

(2000). Proximal and distal defenses in response to reminders of one’s mortality: Evidence of a temporal sequence. Personality and Social Psychology Bulletin, 26, 91-99.

39.

Greenberg

Pyszczynski

Solomon

(1986). The causes and consequences of the need for self-esteem: A terror management theory. In Baumeister

R. F.

(Ed.), Public self and private self (pp. 189-212). New York, NY: Springer-Verlag.

40.

Greenberg

Pyszczynski

Solomon

Rosenblatt

Veeder

Kirkland

Lyon

(1990). Evidence for terror management theory II: The effects of mortality salience on reactions to those who threaten or bolster the cultural worldview. Journal of Personality and Social Psychology, 58, 308-318.

41.

Greenberg

Pyszczynski

Solomon

Simon

Breus

(1994). Role of consciousness and accessibility of death-related thoughts in mortality salience effects. Journal of Personality and Social Psychology, 67, 627-637.

42.

Greenberg

Simon

Harmon-Jones

Solomon

Pyszczynski

Lyon

(1995). Testing alternative explanations for mortality salience effects: Terror management, value accessibility, or worrisome thoughts. European Journal of Social Psychology, 25, 417-433.

43.

Greenberg

Simon

Pyszczynski

Solomon

Chatel

(1992). Terror management and tolerance: Does mortality salience always intensify negative reactions to others who threaten one’s worldview? Journal of Personality and Social Psychology, 63, 212-220.

44.

*Grover

K. W.

Miller

C. T.

Solomon

Webster

R. J.

Saucier

D. A.

(2010). Mortality salience and perceptions of people with AIDS: Understanding the role of prejudice. Basic and Applied Social Psychology, 32, 315-327.

45.

*Harmon-Jones

Simon

Greenberg

Pyszczynski

Solomon

McGregor

(1997). Terror management theory and self-esteem: Evidence that increased self-esteem reduced mortality salience effects. Journal of Personality and Social Psychology, 72, 24-36.

46.

Hayes

Schimel

Arndt

Faucher

E. H.

(2010). A theoretical and empirical review of the death-thought accessibility concept in terror management research. Psychological Bulletin, 136, 699-739.

47.

*Hayes

Schimel

Faucher

E. H.

Williams

T. J.

(2008). Evidence for the DTA hypothesis II: Threatening self-esteem increases death-thought accessibility. Journal of Experimental Social Psychology, 44, 600-613.

48.

*Hayes

Schimel

Williams

T. J.

(2008). Fighting death with death: The buffering effects of learning that worldview violators have died. Psychological Science, 19, 501-507.

49.

Hirschberger

(2006). Terror management and attributions of blame to innocent victims: Reconciling compassionate and defensive responses. Journal of Personality and Social Psychology, 91, 832-844.

50.

*Hirschberger

Ein-Dor

Almakias

(2008). The self-protective altruist: Terror management and the ambivalent nature of prosocial behavior. Personality and Social Psychology Bulletin, 34, 666-678.

51.

*Hirschberger

Florian

Mikulincer

(2005). Fear and compassion: A terror management analysis of emotional reactions to physical disability. Rehabilitation Psychology, 50, 246-257.

52.

*Jessop

D. C.

Albery

I. P.

Rutter

Garrod

(2008). Understanding the impact of mortality-related health-risk information: A terror management theory perspective. Personality and Social Psychology Bulletin, 34, 951-964.

53.

*Jessop

D. C.

Wade

(2008). Fear appeals and binge drinking: A terror management theory perspective. British Journal of Health Psychology, 13, 773-788.

54.

*Jonas

Fischer

(2006). Terror management and religion: Evidence that intrinsic religiousness mitigates worldview defense following mortality salience. Journal of Personality and Social Psychology, 91, 553-567.

55.

King

L. A.

Hicks

J. A.

Abdelkhalik

(2009). Death, life, scarcity, and value: An alternative perspective on the meaning of death. Psychological Science, 20, 1459-1462.

56.

Koole

S. L.

Van den Berg

A. E.

(2005). Lost in the wilderness: Terror management, action orientation, and nature evaluation. Journal of Personality and Social Psychology, 88, 1014-1028.

57.

*Landau

M. J.

Solomon

Greenberg

Cohen

Pyszczynski

Arndt

. . .Cook

(2004). Deliver us from evil: The effects of mortality salience and reminders of 9/11 on support for President George W. Bush. Personality and Social Psychology Bulletin, 30, 1136-1150.

58.

*Ma-Kellams

Blascovich

(2012). Enjoying life in the face of death: East–West differences in responses to mortality salience. Journal of Personality and Social Psychology, 103, 773-786.

59.

Martens

Burke

B. L.

Schimel

Faucher

E. H.

(2011). Same but different: Meta-analytically examining the uniqueness of mortality salience effects. European Journal of Social Psychology, 41, 6-10.

60.

Martens

Greenberg

Schimel

Landau

M. J.

(2004). Ageism and death: Effects of mortality salience and perceived similarity to elders on reactions to elderly people. Personality and Social Psychology Bulletin, 30, 1524-1536.

61.

Martin

I. M.

Kamins

M. A.

(2010). An application of terror management theory in the design of social and health-related anti-smoking appeals. Journal of Consumer Behaviour, 9, 172-190.

62.

*Maxfield

Pyszczynski

Kluck

Cox

C. R.

Greenberg

Solomon

Weise

(2007). Age-related differences in responses to thoughts of one’s own death: Mortality salience and judgments of moral transgressions. Psychology and Aging, 22, 341-353.

63.

McGregor

H. A.

Lieberman

J. D.

Greenberg

Solomon

Arndt

Simon

Pyszczynski

(1998). Terror management and aggression: Evidence that mortality salience motivates aggression against worldview-threatening others. Journal of Personality and Social Psychology, 74, 590-605.

64.

*Mikulincer

Florian

(2000). Exploring individual differences in reactions to mortality salience: Does attachment style regulate terror management mechanisms? Journal of Personality and Social Psychology, 79, 260-273.

65.

Mikulincer

Florian

(2002). The effects of mortality salience on self-serving attributions—Evidence for the function of self-esteem as a terror management mechanism. Basic and Applied Social Psychology, 24, 261-271.

66.

*Mikulincer

Florian

Birnbaum

Malishkevich

(2002). The death-anxiety buffering function of close relationships: Exploring the effects of separation reminders on death-thought accessibility. Personality and Social Psychology Bulletin, 28, 287-299.

67.

*Motyl

Hart

Cooper

D. P.

Heflick

Goldenberg

Pyszczynski

(2013). Creatureliness priming reduces aggression and support for war. British Journal of Social Psychology, 52, 648-666.

68.

*Navarrete

C. D.

Kurzban

Fessler

D. M.

Kirkpatrick

L. A.

(2004). Anxiety and intergroup bias: Terror management or coalitional psychology? Group Processes & Intergroup Relations, 7, 370-397.

69.

Norenzayan

Dar-Nimrod

Hansen

I. G.

Proulx

(2009). Mortality salience and religion: Divergent effects on the defense of cultural worldviews for the religious and the non-religious. European Journal of Social Psychology, 39, 101-113.

70.

O’Connor

M. L.

McFadden

S. H.

(2012). A terror management perspective on young adults’ ageism and attitudes toward dementia. Educational Gerontology, 38, 627-643.

71.

*Proulx

Heine

S. J.

(2008). The case of the transmogrifying experimenter affirmation of a moral schema following implicit change detection. Psychological Science, 19, 1294-1300.

72.

*Proulx

Heine

S. J.

(2009). Connections from Kafka: Exposure to meaning threats improves implicit learning of an artificial grammar. Psychological Science, 20, 1125-1131.

73.

Pyszczynski

Greenberg

Solomon

(1999). A dual-process model of defense against conscious and unconscious death-related thoughts: An extension of terror management theory. Psychological Review, 106, 835-845.

74.

Rosenblatt

Greenberg

Solomon

Pyszczynski

Lyon

(1989). Evidence for terror management theory I: The effects of mortality salience on reactions to those who violate or uphold cultural values. Journal of Personality and Social Psychology, 57, 681-690.

75.

*Routledge

C. D.

Arndt

Sedikides

Wildschut

(2008). A blast from the past: The terror management function of nostalgia. Journal of Experimental Social Psychology, 44, 132-140.

76.

*Rutjens

B. T.

Loseman

(2010). The society-supporting self: System justification and cultural worldview defense as different forms of self-regulation. Group Processes & Intergroup Relations, 13, 241-250.

77.

*Rutjens

B. T.

van der Pligt

van Harreveld

(2009). Things will get better: The anxiety- buffering qualities of progressive hope. Personality and Social Psychology Bulletin, 35, 535-543.

78.

Scharf

Cohen

(2013). Relatedness and individuation among young adults born preterm: The role of relationships with parents and death anxiety. Journal of Adult Development, 20, 212-221.

79.

*Schimel

Hayes

Williams

Jahrig

(2007). Is death really the worm at the core? Converging evidence that worldview threat increases death-thought accessibility. Journal of Personality and Social Psychology, 92, 789-803.

80.

Schmeichel

B. J.

Gailliot

M. T.

Filardo

E. A.

McGregor

Gitter

Baumeister

R. F.

(2009). Terror management theory and self-esteem revisited: The roles of implicit and explicit self-esteem in mortality salience effects. Journal of Personality and Social Psychology, 96, 1077-1087.

81.

Schmeichel

B. J.

Martens

(2005). Self-affirmation and mortality salience: Affirming values reduces worldview defense and death-thought accessibility. Personality and Social Psychology Bulletin, 31, 658-667.

82.

*Shehryar

Hunt

D. M.

(2005). A terror management perspective on the persuasiveness of fear appeals. Journal of Consumer Psychology, 15, 275-287.

83.

Silvia

P. J.

(2001). Nothing or the opposite: Intersecting terror management and objective self-awareness. European Journal of Personality, 15, 73-82.

84.

Simon

Greenberg

Harmon-Jones

Solomon

Pyszczynski

Arndt

Abend

(1997). Terror management and cognitive-experiential self-theory: Evidence that terror management occurs in the experiential system. Journal of Personality and Social Psychology, 72, 1132-1146.

85.

Simonsohn

Nelson

L. D.

Simmons

J. P.

(2014). P-curve: A key to the file-drawer. Journal of Experimental Psychology: General, 143, 534-547.

86.

Solomon

Greenberg

Pyszczynski

(1991). A terror management theory of social behavior: The psychological functions of self-esteem and cultural worldviews. In Zanna

M. P.

(Ed.), Advances in experimental social psychology (Vol. 24, pp. 91-159). San Diego, CA: Academic Press.

87.

Sterne

J. A.

Harbord

R. M.

(2004). Funnel plots in meta-analysis. Stata Journal, 4, 127-141.

88.

*Taubman-Ben-Ari

(2004). Intimacy and risky sexual behavior—What does it have to do with death? Death Studies, 28, 865-887.

89.

*Taubman-Ben-Ari

(2011). Is the meaning of life also the meaning of death? A terror management perspective reply. Journal of Happiness Studies, 12, 385-399.

90.

*Taubman-Ben-Ari

Katz-Ben-Ami

(2008). Death awareness, maternal separation anxiety, and attachment style among first-time mothers—A terror management perspective. Death Studies, 32, 737-756.

91.

*Trafimow

Hughes

J. S.

(2012). Testing the death thought suppression and rebound hypothesis: Death thought accessibility following mortality salience decreases during a delay. Social Psychological & Personality Science, 3, 622-629.

92.

*Ullrich

Cohrs

J. C.

(2007). Terrorism salience increases system justification: Experimental evidence. Social Justice Research, 20, 117-139.

93.

Vaes

Heflick

N. A.

Goldenberg

J. L.

(2010). “We are people”: Ingroup humanization as an existential defense. Journal of Personality and Social Psychology, 98, 750-760.

94.

*Vail

K. E.

III Arndt

Motyl

Pyszczynski

(2012). The aftermath of destruction: Images of destroyed buildings increase support for war, dogmatism, and death thought accessibility. Journal of Experimental Social Psychology, 48, 1069-1081.

95.

*Van Beest

Williams

K. D.

Van Dijk

(2011). Cyberbomb: Effects of being ostracized from a death game. Group Processes & Intergroup Relations, 14, 581-596.

96.

*Van Tongeren

D. R.

Green

J. D.

(2010). Combating meaninglessness: On the automatic defense of meaning. Personality and Social Psychology Bulletin, 36, 1372-1384.

97.

Vess

Arndt

Cox

C. R.

(2012). Faith and nature: The effect of death-relevant cognitions on the relationship between religious fundamentalism and connectedness to nature. Social Psychological & Personality Science, 3, 333-340.

98.

Wegner

D. M.

Schneider

D. J.

Carter

S. R.

White

T. L.

(1987). Paradoxical effects of thought suppression. Journal of Personality and Social Psychology, 53, 5-13.

99.

Williams

Schimel

Hayes

Martens

(2010). The moderating role of extrinsic contingency focus on reactions to threat. European Journal of Social Psychology, 40, 300-320.

100.

*Wojtkowiak

Rutjens

B. T.

(2011). The postself and terror management theory: Reflecting on after death identity buffers existential threat. The International Journal for the Psychology of Religion, 21, 137-144.

101.

Zhou

Liu

Chen

(2008). Do children transcend death? An examination of the terror management function of offspring. Scandinavian Journal of Psychology, 49, 413-418.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.17 MB