A Prescriptive, Intergenerational-Tension Ageism Scale: Succession, Identity, and Consumption (SIC)

Abstract

We introduce a novel ageism scale, focusing on prescriptive beliefs concerning potential intergenerational tensions: active, envied resource Succession, symbolic Identity avoidance, and passive, shared-resource Consumption (SIC). Four studies (2,010 total participants) developed the scale. EFA formed an initial 20-item, three-factor solution (Study 1). The scale converges appropriately with other prejudice measures and diverges from other social control measures (Study 2). It diverges from anti-youth ageism (Study 3). Study 4’s experiment yielded both predictive and divergent validity apropos another ageism measure. Structural equation modeling confirmed model fit across all studies. Per an intergenerational-tension focus, younger people consistently scored the highest. As generational equity issues intensify, the scale provides a contemporary tool for current and future ageism research.

Ageism is a peculiar prejudice. Despite the reality that every living person potentially joins every age group, ageism remains relatively under-researched (comparatively rare in prejudice literature), underappreciated (overlooked as a prejudice), and under-the-radar (subtle in nature; North & Fiske, 2012). Nevertheless, a rapidly growing older population necessitates its increased understanding—for both social psychology and society at large.

The current paper analyzes potential intergenerational tensions over practical and symbolic resources and introduces a measure of ageism with contemporary relevance. Althoughprior scales focus mainly on what older people allegedly “are” (descriptive stereotypes), the current analysis centers on the role of more controlling, “should”-based, prescriptive beliefs. This approach proposes three prescriptive dimensions that younger generations are particularly likely to endorse: (1) active Succession of enviable positions and influence, (2) age-appropriate, symbolic Identity maintenance, and (3) minimizing passive shared-resource Consumption (SIC). We argue that a rapidly growing older population—intensifying potential intergenerational tensions—necessitates the new, prescriptive ageism scale presented here.

The Potential Rise of Prescriptive (Hostile) Ageism

Demographic shifts render ageism a particularly ripe research topic. Already the largest proportion in history—currently 13 percent of the U.S. populace—the older population is expected to compose almost 20 percent by 2030 (U.S. Census, 2012). Though prevailing stereotypes place elders outside mainstream consciousness—spurring negative (or at best, mixed) descriptive elder stereotypes of ineptness, illness, and irrelevance—an era of more conspicuous older age is forthcoming.¹

How increased visibility will change elder images is an empirical question (North & Fiske, 2012). An optimistic standpoint posits a more visible older age debunking negative elder stereotypes. The pessimistic counterpoint cites the potential for hostile ageism to brew among younger generations, if elders do not step aside and cede resources in the traditional manner (e.g., if they postpone retirement or reap disproportionate government benefits).

Theoretically, backlash for overstepping societal boundaries is particularly likely when group outcomes are interdependent (as younger and older age groups are). For instance, because the genders are intimately interconnected in everyday outcomes, women face negative repercussions for violating expectations (e.g., by being too agentic; Glick & Fiske, 1996). Such controlling, prescriptive (“should”-based) stereotypes aim to dictate other groups’ behavior so as to benefit ingroup outcomes (Rudman & Glick, 2001; Prentice & Carranza, 2002). Thus, prescriptive expectations yield far greater between-group differences in endorsement than do descriptive stereotypes.

But despite the reality that age groups co-exist within society, and the fact that everyone (with luck) eventually joins each age group, different ages’ inherent interdependence has not been considered as an integral factor in driving ageism. Therefore, even though age groups largely agree about descriptive elder-stereotype content (Greenberg et al., 2004; Nosek et al., 2002), because of resource interdependence, younger people theoretically should endorse prescriptive stereotypes more than older people do. Ageism measures have largely overlooked these prescription-based possibilities, as discussed next.

Extant (Descriptive) Ageism Measures

Although useful, extant prejudice measures focus primarily on content (“are”)-based notions about older people. Various scales gauge ageist sentiment indirectly, focusing on descriptive aging-process knowledge. A prominent example is the Facts on Aging Quiz (Palmore, 1998).

Other instruments focus more directly on prejudicial attitudes. Two early scales— Tuckman and Lorge’s (1953) Attitudes toward Old People measure and The Negative Attitudes Toward Old People Scale (KOPS; Kogan, 1961)—both gauge agreement with descriptive elder statements. The Aging Semantic Differential (Rosencranz & McNevin, 1969) organizes descriptive statements into three overarching constructs.

The three-factor Fraboni Scale of Ageism (FSA; Fraboni et al., 1990) aims to “measure the affective component of [ageist] attitude, to supplement the cognitive aspect measured by other instruments” (p. 56). It includes both attitudinal beliefs—via descriptive, antilocution items (e.g., “Many old people just live in the past”)—and discriminatory behavior (avoidance; e.g., “I sometimes avoid eye contact with old people when I see them). Though the word “should” does appear in a third subscale, discrimination (e.g., “Old people should be encouraged to speak out politically”), these prescriptions tend to focus more on what society as a whole should do, rather than expectations targeting elders themselves. Moreover, prescription is not a conceptual focus of the largely descriptive measure. Nevertheless, to demonstrate the current (SIC) measure’s divergence from the FSA, we conduct a study with circumstances in which we expect the SIC measure to have greater predictive power than the FSA.

Prescriptive Domains: Succession, Identity, and Consumption

Age differs from any other social category in its permeability: provided they live long enough, all people eventually join each group. Another way to think of this is that age groups take turns along a hypothetical age queue—with younger people entering, middle-agers enjoying, and elders exiting (e.g., retiring). Although societal allocation of practical and figurative resources tends to favor the middle-aged (North & Fiske, 2012), as long as the line keeps moving, everyone generally gets his/her privileged turn.

However, those at the back of the line are dependent on those at the front transitioning away in order to keep the line moving. Thus, we posit three key ways that older people particularly are expected to relinquish resources (North & Fiske, in press), each pertinent to blocking a different aspect of the theoretical queue. Although not the only possible prescriptive dimensions, we focus on these three as central ones.

Succession-based prescriptions derive from expectations surrounding enviable resources and societal positions. Although middle-agers predominately hold the greatest societal influence, younger people’s opportunities more realistically depend on the old stepping aside—primarily in employment (where retirement opens up jobs for the young) and political influence (where older voters form a powerful bloc, while minors face age restrictions). In other words, acceding to Succession means allowing those waiting to move predictably toward their turn at the front.

Consumption-based prescriptive stereotypes center on passive depletion of currently shared resources. Elder violations derive from apparent exploitativeness, reaping more than a fair share of allotted societal resources—characterized by dilemmas involving government money (e.g., healthcare) and shared public space (e.g., the highway). Put another way, sharing Consumption of societal resources means not using up everything before others get there.

Identity involves resources more symbolic than Succession or Consumption, limiting elder participation in activities usually reserved for younger people. Although in this case who is considered “old” often depends on context, elders are particularly barred from youth culture (Greenberg et al., 2004). Unwanted intrusions into young ingroup territory can be both direct (e.g., frequenting youth-centered hangouts) and indirect (e.g., attempting to act “cool”). Thus, avoiding Identity invasion means not trying to go back through the line again by adopting youth’s territory.

Research Overview and Hypotheses

A priori, we aimed to test empirically a three-factor model of prescriptive ageism. The methodological foundation for the proposed scale utilized four samples, totaling 2,010 participants.

Study 1 (scale development) had participants rate their agreement with pilot-generated SIC-based statements; exploratory factor analysis (EFA) ascertained the number of latent factors underlying these items. Study 2 (initial scale validation) tested the scale’s convergent and divergent validity, comparing it to other measures of prejudice and potentially related factors. Study 3 tested elder-focused SIC’s divergent validity from anti-young ageism. Finally, Study 4 tested the scale’s predictive validity, exploring whether high scorers—controlling for another ageism measure (the FSA)—would exhibit the most bias toward older prescription-violators. This also served as a test of divergent validity from the descriptive-focused FSA, exploring which scale best predicts reactions to older people who do and do not “know their place” (i.e., adhere to versus violate expectations).

Using structural equation modeling, each study conducted a confirmatory factor analysis (CFA), testing the proposed model’s fit to the data. We hypothesized a three-factor solution to be the best fit via both EFA and CFA. We also expected younger people to score the highest on the scale, given SIC’s emphasis on prescriptive tensions arising from generational interdependence.

Study 1

Method

Item generation

Forty-one potential scale items derived from lab and participant samples’ open-ended reports. Responses generally answered the question, “What are things older people should or shouldn’t do?”

Participants

Participants (N = 427; 264 female, mean age = 32.9, median = 32, range 16-81) included 397 online participants from Amazon Mechanical Turk (mTurk) and 30 undergraduates. Participants were primarily White (74.5%), followed by 6.3% or fewer East Asian, African American, Latino, South Asian, and “Other”/mixed.

Procedure

As part of a “social statements survey,” participants rated the SIC-based items from 1 (strongly disagree) to 6 (strongly agree).² Online participants received a nominal payment; undergraduate lab participants received appropriate course credit.

Results

EFA examined the intercorrelation pattern among the 41 preliminary items, utilizing principal components extraction and a Varimax rotation.³ All items with loadings below .40 on their respective factors were discarded, as were strongly double-loading items. Based on the scree plot of variance explained, three overall factors were specified for subsequent extraction, (explaining 46.51% of the variance). Factor 1 apparently represented Consumption, Factor 2 reflected Succession, and Factor 3 comprised Identity items.

The final total number of items across these three factors was 20 (Table 1 shows item factor loadings across studies), with an overall alpha reliability of .90, and substantial subscale reliabilities (Table 2). These three subscales correlated moderately with each other, Pearson r’s ranging from .46 to .61. Descriptive statistics for this and all subsequent studies appear in Table 3.

Table 1. Item factor loadings across studies.

Factor 1: Consumption	S1	S2	S3	S4
Doctors spend too much time treating sickly older people.*	.69	.74	.71	.76
Older people are too big a burden on the healthcare system.*	.64	.73	.67	.72
Older people are often too much of a burden on families.	.62	.67	.56	.72
At a certain point, older people’s maximum benefit to society is passing along their resources.	.60	.59	.51	.65
Older people shouldn’t be so miserly with their money if younger relatives need it.	.57	.43	.42	.55
Older people don’t really need to get the best seats on buses and trains.	.53	.45	.68	.63
AARP (American Association of Retired Persons) wastes charity money.	.48	.30	.38	.38

Factor 2: Succession
If it weren’t for older people opposed to changing the way things are,  we could probably progress much more rapidly as a society.	.66	.65	.69	.77
The older generation has an unfair amount of political power compared to younger people.	.65	.80	.65	.74
Most older people don’t know when to make way for younger people.+	.58	.60	.75	.65
Most older workers don’t know when it’s time to make way for the younger generation.+	.55	.53	.74	.67
Older people are often too stubborn to realize they don’t function like they used to.	.53	.33	.62	.45
Younger people are usually more productive than older people at their jobs.	.50	.44	.59	.54
Job promotions shouldn’t be based on older workers’ experience rather than their productivity.	.50	.28	.47	.54
It is unfair that older people get to vote on issues that will impact younger people much more.	.50	.58	.65	.71

Factor 3: Identity
Older people typically shouldn’t go to places where younger people hang out.x	.81	.80	.81	.83
In general, older people shouldn’t hang out at places for younger people. x	.78	.80	.84	.79
Generally older people shouldn’t go clubbing.	.70	.74	.72	.82
Older people probably shouldn’t use Facebook.	.66	.74	.47	.57
Older people shouldn’t even try to act cool.	.62	.82	.63	.60

Note.

^*+xSimilar items denoted as co-varying in the structural equation model.

Table 2. SIC Scale alpha reliabilities across studies.

Scale	SI	S2	S3	S4
Succession	.84	.84	.85	.85
Identity	.85	.87	.83	.84
Consumption	.83	.75	.83	.86
Total Scale	.90	.90	.91	.91
N	427	92	1283	207

Table 3. SIC Scale M, SD, and range across studies.

SIC Scale	S1	S2	S3	S4
Succession
M	2.97	3.01	2.96	3.24
SD	0.92	0.82	0.96	0.93
Min	1.00	1.00	1.00	1.38
Max	5.88	5.75	6.00	6.00
Identity
M	2.62	2.84	2.65	2.63
SD	1.09	1.20	1.05	1.02
Min	1.00	1.00	1.00	1.00
Max	6.00	6.00	6.00	5.80
Consumption
M	2.30	2.31	2.34	2.44
SD	0.88	0.72	0.91	0.90
Min	1.00	1.00	1.00	1.00
Max	5.86	5.00	6.00	5.71
Total Scale
M	2.65	2.72	2.66	2.81
SD	0.79	0.74	0.83	0.79
Min	1.00	1.00	1.00	1.25
Max	5.65	5.25	5.50	5.85

Note. Each scale ranged from 1 (strongly disagree) to 6 (strongly agree).

An initial CFA examined whether the proposed model (Figure 1)⁴ fit well with the current dataset. A structural equation modeling technique used AMOS 7.0 (Arbuckle, 2006). Given that the model was created from this dataset, this was not an independent test of fit (Studies 2-4 conduct this analysis). Nevertheless, across various standard indices, initial evidence emerged for good 3-factor model fit (Table 4).⁵ Moreover, in line with comparative practices utilized by other SEM scale-development studies (e.g. Glick & Fiske, 1996; Luhtanen & Crocker, 1992), these same fit indices were considerably worse for a comparative one-factor model (Table 4).

Figure 1.

Three-factor structural equation model.

Table 4. Full (3-Factor) and restricted (1-Factor) model fit indices across studies.

Fit index	S1	S2	S3	S4
Three-factor model
X2/ df	1.94	1.38	4.36	1.87
IFI	.96	.92	.95	.93
CFI	.96	.92	.95	.93
TLI	.95	.91	.94	.92
RMSEA	.047	.065	.05	.065
One-factor model
X2/ df	4.19	1.90	9.14	3.05
IFI	.85	.81	.87	.83
CFI	.85	.81	.87	.82
TLI	.85	.78	.85	.80
RMSEA	.087	.10	.08	.10

Note. X²/ df = chi-square-to-degrees-of-freedom ratio, IFI = Incremental Fit Index, CFI = Comparative Fit Index, TLI = Tucker-Lewis coefficient, RMSEA = Root Mean Square Error of Approximation (see Kenny, 2011, for explanations).

Study 2

Measuring potential convergence and divergence, Study 2’s participants completed the SIC scale, in addition to measures of prejudice, social control orientation, and political ideology. We expected the SIC dimensions to correlate with each other (as in Study 1). We also hypothesized SIC to correlate relatively highly with another ageism measure (due to measuring the same general construct), moderately with other types of prejudice (because biases tend to correlate), and slightly with general social control measures (given SIC’s emphasis on control-oriented stereotypes). Finally, we expected political ideology to be uncorrelated with SIC-based bias because ageism has not provoked partisan debate, not yet having featured a salient civil rights movement; thus (unlike sexism and racism) strong political correctness norms have not developed regarding ageism.

Method

Participants

Participants (N = 93; 69 female, M age = 25.11, median = 21, range 18-60) were online paid university participants in a “social statements survey.” Participants were 60.9% White, followed by 12.0% or less of the noted minority or mixed groups.

Procedure

Participants completed the SIC measure using the same 6-point scale as Study 1. Additional prejudice measures included the 29-item FSA (Fraboni et al., 1990), the 22-item Ambivalent Sexism Inventory (ASI; Glick & Fiske, 1996), and an 8-item version of the Symbolic Racism Scale (Henry & Sears, 2000). Intergroup threat/control scales included the 16-item Social Dominance Orientation scale (SDO; Pratto, Sidanius, Stallworth, & Malle, 1994) and the 20-item Right-Wing Authoritarianism scale (RWA; Altemeyer, 1998). Finally, participants reported their political orientation (1 = definitely liberal; 7 = definitely conservative) and party affiliation (1 = strongly Democrat; 7 = strongly Republican). After the SIC scale, the sequence of the other scales was counter-balanced to avert potential order effects. Participants were entered into a monetary-prize lottery.

Results

Inter-item reliability was once again high for the 20-item SIC scale (.91) and each subscale (Table 2).

Model fit

Like Study 1, CFA found the three-factor model to provide good structural fit for the new dataset, and a better fit across all indices than a one-factor model comprising all 20 items (see Table 4).

Divergent and convergent validity

Alpha reliabilities were high for all comparative scales: FSA (.86), ASI-Hostile (.93), ASI-Benevolent (.88), SDO (.93), RWA (.94), and Symbolic Racism (.81).

As expected, the SIC subscales correlated with each other moderately to strongly (r’s .48 to .65). Also anticipated, SIC correlated most strongly with the FSA (r = .70, p < .001), then more moderately with Symbolic Racism, ASI-Hostile, and ASI-Benevolent (rs .32 to .40, ps < .003). Though total SIC correlated significantly with intergroup-hierarchy-focused SDO (r = .31, p = .003), it did not so with intergroup-value-conflict-based RWA (r = .15, p = .14). As predicted, correlations with political and party affiliations were nonsignificant (both rs < .02).

Notably, with the exception of the FSA (β = −.24, p = .02) and its avoidance subscale (β = −.27, p = .009), rater age did not predict scores on any comparative scale, all ps > .05. This contrasts with consistent age trends for the SIC scale (see “Demographic Analyses”).

Study 3

Method

Participants

Participants (N = 1283; 808 female) were recruited from Princeton University’s paid participant pool (N = 97) and mTurk (N = 1186; M age across samples = 33.23, median = 30, range 18-81). Participants were 75.4% White, followed by 6.0% or less of the noted minority groups.

Measures and procedure

Participants completed the 20 validated SIC items using the same 6-point scale; in addition, a subset of the sample completed 21 items reflecting prescriptive stereotypes of younger people taken from a separate project by the current authors (e.g., “Young people shouldn’t use so much foul language”; “Today’s youth are too idealistic”).

Results

Substantial reliability again emerged among all 20 SIC scale items (α =.91), as well as each subscale (Table 2), and the 21 anti-young items (α =.90).

As with the prior two studies, the Succession, Identity, and Consumption subscales correlated with one another, rs ranging from .53 to .66, all ps < .001.

Model fit

The three-factor model again provided good structural fit for this dataset, despite a large chi-square/df ratio (as noted by Kenny [2011] and others, this is a poor test of fit for data sets with more than 400 cases); see Table 4. A three-factor model again bested a one-factor version on all noted indices.

Divergence from anti-young ageism

Though the total SIC scale did correlate significantly with the anti-young items, the effect size was small (r = .14, p = .01). Moreover, only the Identity subscale significantly correlated with the anti-young scale (r = .19, p = .001); Consumption was a marginal correlate (r = .10, p = .08), and Succession was nonsignificant (r = .07, p = .22).

Demographic analysis also underscored divergence between SIC and anti-young items. Participant age was not a significant predictor of anti-young ageism (β = .02, p = .68). Moreover, the two genders did not differ in their endorsement of the anti-young items (t(332) < 1), nor did Whites and non-Whites differ, t(332) = 1.17, p = .24. These null results contrast with predicted, consistent, significant trends for the SIC scale (see “Demographic Analyses”).

Study 4

Method

Participants

These mTurk participants (N = 207, 88 female, mean age = 26.87, median age = 25, age range 18-59) were 70.0% White, followed by 9.2% or fewer of the noted minority groups.

Procedure

Participants completed a study ostensibly connecting mTurk participants with professionals from an “online career network.” This involved reading a randomly assigned profile of another person’s brief self-description.

Participants read about a 74-year-old with the name “Max [last name withheld]” who either violated or adhered to SIC prescriptions. Six between-subjects versions were possible: Three different targets violated, respectively, Succession (refusing to retire despite blocking younger potential hires), Identity (displaying affinity for the latest pop music), and Consumption (undergoing a resource-intensive medical procedure). Three adhering counterparts, respectively, stated that “it’s probably time to step aside,” enjoyed oldies music, and decided against the burdensome treatment.

After viewing one of the six profiles, participants provided two types of target ratings: Three separate items composed an overall measure of perceived warmth (α = .88); three other items formed a competence measure (α = .69). These two dimensions, considered by many to be fundamental dimensions of person perception, serve as useful valence snapshots of target judgments, at both the individual and group level (Fiske et al., 2007). The dimensions also hold particular relevance in perceptions of and selection bias toward older workers (Krings, Sczesny, & Kluge, 2011). Participants also responded to six behavioroid items (α = .80) indicating their desire to interact with Max (Table 5).

Table 5. Study 4 elder target ratings as a function of SIC Scale and FSA.

	SIC (cont. for FSA)	FSA (cont. for SIC)
	Standardized β	Standardized β
Warmth1
Violators	−.32+	.17
Adherers	−.03	−.27*
Competence2
Violators	−.33+	.07
Adherers	.12	−.45***
Interaction Desire3
Violators	−.46**	.03
Adherers	−.15	−.21+
Participant Age	−.22*	−.05

Notes. SIC = Succession, Identity, and Consumption scale; FSA = Fraboni Scale of Ageism.

¹“Warmth” comprises 3 items: “Overall, how warm might people think Max is?”, “Overall, how kind might people think Max is?”, “Overall, how good-natured might people think Max is?”

²“Competence” comprises 3 items: “Overall, how competent might people think Max is?”; “Overall, how capable might people think Max is?”; “Overall, how confident might people think Max is?”

³“Interaction Desire” represents 6 items: “Would you be willing to interact further with Max after the study is over?”; “Would you recommend other participants in this survey to interact with Max?”; “Would you be willing to write and send Max a supportive message?”; Reverse-scored: “Would you prefer to ignore Max altogether?”; “If you were to interact further, how likely would you be to say mean things to Max?”; “Would you suggest to other participants in this survey that they ignore Max.”

Afterward, participants completed both the SIC scale and the FSA. Participants were thanked, debriefed, and compensated.

Results

The total SIC scale exhibited strong reliability (α = .91), as did each subscale (Table 2). The FSA also had high reliability (α = .91).

Model fit

The three-factor model again fit the data, and outperformed a one-factor counterpart per all noted indices (Table 4).

Main effects

As the prescriptive framework predicts, SIC-violating targets were perceived as less warm (M = 3.16, SD = 0.93) than adherers (M = 3.96, SD = 0.70), and less worthy of interaction (violator M = 3.23, SD = 0.82; adherer M = 3.58, SD = 0.73), both ts > 3.20, both ps < .003. No differences in competence emerged (violator M = 3.54, SD = 0.72; adherer M = 3.57, SD =0.80), t < 1.

Predictive ability of SIC versus FSA

The framework also predicts that SIC scores will predict reactions to SIC-violating older people; thus, multiple regressions tested the ability of each scale to predict prescriptive reactions toward targets. Indeed, SIC score (controlling for FSA) marginally predicted violating targets’ perceived warmth, (β = −.32, p = .08), competence (β = −.33, p = .06), and significantly predicted interaction desire (β = −.46, p = .007). However, FSA (controlling for SIC) did not significantly predict any of these dependent variables, all ps > .34 (Table 5).

As expected, for adhering targets, SIC score (controlling for FSA) did not significantly predict target ratings (Table 5). Controlling for SIC, FSA did significantly predict warmth, competence, and (marginally) interaction desire, but in the opposite direction: The more ageist participants were, the more negatively they viewed the older, adhering targets.⁶

Rater age significantly predicted FSA score, β = −.21, p = .002. However, controlling for FSA, rater age still significantly predicted SIC (β = −.22, p = .04), whereas the reverse was nonsignificant (FSA controlling for SIC, β = −.05, p = .67). A cross-study demographic analysis on SIC, including rater age, appears next.

Demographic Analyses (All Studies)

Pooling all 2,010 responses, consistent and predicted demographic trends emerge for age, gender, and race/ethnicity. Younger people and men should show more prescriptive ageism; the ethnicity results are unexpected.

Age

As expected, participant age was a significant inverse predictor of total SIC score (β = −.31, p < .001) and each subscale: Succession (β = −.37, p < .001), Identity (β = −.22, p < .001), and Consumption (β = −.17, p < .001).

Gender

Men (M = 2.94, SD = 0.82) scored significantly higher overall than women (M = 2.51, SD = 0.76), t(1558.99) = 11.84, p < .001). This held for sub-scores of Succession (male M = 3.24, SD = 0.99; female M = 2.83, SD = 0.89, t(1530.72) = 9.41, p < .001), Identity (male M = 2.89, SD = 1.09; female M = 2.50, SD = 1.02, t(1587.79) = 8.03, p < .001), and Consumption (male M = 2.64, SD = 0.92; female M = 2.15, SD = 0.82, t(1519.42) = 12.20, p < .001).

Race/Ethnicity

A one-way ANOVA found significant ethnic differences on the total SIC scale [F(4, 1883) = 13.84, p < .001, partial η² = .03] and each subscale [all Fs > 6.90, all ps < .001].

On the total scale, East Asians (M = 3.07, SD = 0.71) and South Asians (M = 3.02, SD = 0.84) scored the highest. Post-hoc Tukey comparisons found both scores to be significantly higher than Whites’ (M = 2.63, SD = 0.80) and Blacks’ (M = 2.56, SD = 0.86), all ps ≤ .001. Latinos (M = 2.87, SD = 0.93) scored marginally higher than both Whites and Blacks (both ps = .06), but not significantly different from East or South Asians (ps > .38).

Subscales showed similar trends, with East Asians scoring significantly higher than Whites and Blacks on all three subscales, South Asians outpacing Whites and Blacks on the Succession and Consumption subscales, and Latinos similarly scoring higher than Blacks and Whites on the Succession subscale and lower than East Asians on the Consumption subscale (all ps < .05). All other comparisons were nonsignificant.

General Discussion

Four studies created a three-factor (SIC) scale of prescriptive ageism, affirming its convergent validity (with other ageism and prejudice measures), divergent validity (from other types of prejudice, group hierarchy endorsement, political orientation and anti-young ageism) and predictive validity (in a SIC-based experiment, doubling as evidence for divergence from FSA).

Though overall scale means emerged toward the lower portion of the scale, demographic differences contribute to the larger body of ageism research, helping answer unresolved questions. The younger participants were, the more they endorsed prescriptive, ageist stereotypes across all three SIC domains. Similar young-as-ageist results have emerged with the FSA (Rupp et al., 2005); however, the opposite trend appears on other descriptive measures (e.g. the KOPS; Hellbusch et al., 1994). Such conflicting findings reflect the ageism literature as whole, which has left largely unresolved the question of whether the young are the most ageist (North & Fiske, 2012). The current prescription-based findings implicate the young as focal proponents of this explicitly controlling prejudice.

Moreover, men scored consistently higher than women. Men do tend to exhibit higher levels of prejudice than women in various other domains (e.g. racism; Ekehammar & Sidanius, 1982; Nosek et al., 2002), and the FSA yields the same pattern (Fraboni et al., 1990).

That Asians scored consistently higher than other groups on the SIC scale contrasts with lay beliefs that Eastern filial piety traditions reduce ageism (Ng, 1998). Admittedly, this paper’s U.S.-only samples do not necessarily speak directly to this issue; to clarify this picture, future studies might employ the SIC scale using non-American Asian samples. Meanwhile, Whites and Blacks tended to score at or among the lowest (with Latinos falling somewhere in between). This latter result echoes prior findings of young African Americans’ respect for their elders (Fiske et al., 2009).

One potential limitation is that the current measure’s development relied heavily on mTurk samples. Though initial testimony touted the website as an inexpensive, reliable source of data (Buhrmester et al., 2011; Paolacci et al., 2010), others have expressed concerns about reduced quality (Reips et al., 2012). Less debatable is mTurk’s wider age range of participants than undergraduate samples offer—particularly relevant for the current research’s focus on age-based bias and the moderating impact of perceiver age on such perceptions.

Another potential limitation is the current research’s focus on a Western population. This is an intentional starting point; many of the issues concerning intergenerational resource distribution seem particularly salient in industrialized societies, where modernization perhaps has made elders’ utility less obvious than in the past (Nelson, 2005). Nevertheless, it remains to be seen whether prescriptive ageism exists at equally high levels in other cultures that are at least believed to be more reverent of their elders.

Despite limitations and unanswered questions, the current ageism scale’s utility lies in its departure from existing scales—which tend to focus on infirmed, irrelevant “elderly.” Given an increasingly visible and relevant older population, the measure’s divergence presents greater contemporary relevance for research in a number of key areas. In gauging attitudes toward Succession of envied resources, the measure can be used to investigate shifting workplace age dynamics and attitudes toward mandatory retirement. The scale’s shared resource-Consumption subscale factors in philosophical and psychological work (e.g., “trolley problem” studies) on people’s valuing of elder lives—particularly important in current hot-button debates over Social Security and healthcare. The symbolic Identity subscale can be included to both buttress psychological theory on ageism (e.g., Social Identity Theory perspectives, which characterize ageism as deriving from young ingroup identity motivation) and in more applied domains (e.g., firms marketing products to the fast-growing older demographic should bear in mind limitations for the amount of “cool” older people are allotted). Taken together, the current measure’s focus on expectations concerning proper roles, behaviors, and resources seems vital, as the population ages and notions of what signifies “old” changes with it (North & Fiske, 2012). It is possible that such prescriptions will shift over time, and the scale can be used to document that change.

Conclusion

Departing from extant ageism scales, which focus primarily on descriptive elder stereotypes, the SIC scale introduces a novel, prescriptive ageism measure. In incorporating intergenerational tensions over practical and symbolic resources—and finding the young to most endorse ageist prescriptions—the SIC measure diverges from prior measures in both focus and findings. Give its relevance to current real-world resource concerns (such as healthcare and employment), the scale is a promising tool for cutting-edge ageism research, as the population grays and generational equity concerns grow more salient.