Socially Desirable Response Set (SDRS-5)

Abstract

The Socially Desirable Response Set (SDRS-5) is a brief, psychometrically sound, five-item self-report instrument developed by Ron D. Hays, Toshi Hayashi, and Anita L. Stewart in 1989. Engineered under the auspices of the landmark RAND Medical Outcomes Study (MOS), the SDRS-5 was constructed to detect and statistically control for social desirability bias—the pervasive tendency of survey respondents to present themselves in an overly favorable light by endorsing culturally sanctioned, virtuous behaviors that are statistically improbable. Derived from the classic 33-item Marlowe-Crowne Social Desirability Scale (MCSDS), the SDRS-5 distills the construct into an ultra-short screening tool optimized for large-scale epidemiological, clinical, and health services research where respondent burden must be minimized. The instrument utilizes a 5-point response scale ranging from Definitely True (1) to Definitely False (5), featuring two reverse-scored items. Psychometric evaluations across diverse community and clinical samples demonstrate respectable internal consistency reliability for a five-item measure (Cronbach’s alpha typically ranging between .66 and .68) and exceptional convergent validity with the full Marlowe-Crowne inventory ($r > .85$). Confirmatory factor analyses support an essentially unidimensional latent structure reflecting impression management and defensive self-presentation. By providing a low-burden, reliable proxy for socially desirable responding, the SDRS-5 enables researchers to evaluate the substantive validity of self-reported psychological states, health behaviors, treatment adherence, and subjective quality-of-life outcomes.

Keywords

Social Desirable Response Set, SDRS-5, Social Desirability Bias, Marlowe-Crowne, Response Bias, Impression Management, Self-Deceptive Enhancement, Survey Methodology, Psychometrics, Medical Outcomes Study, Validity Screener.

Authors

The SDRS-5 was authored by a team of prominent health services researchers and psychometricians associated with the RAND Corporation:

• Ron D. Hays, Ph.D. – Distinguished Professor of Medicine at the David Geffen School of Medicine at UCLA, Division of General Internal Medicine and Health Services Research; Senior Adjunct Behavioral/Social Scientist at the RAND Corporation, Santa Monica, California.
• Toshi Hayashi, M.A. – Former Statistical and Psychometric Analyst at the RAND Corporation, Santa Monica, California.
• Anita L. Stewart, Ph.D. – Professor Emerita at the Institute for Health & Aging, School of Nursing, University of California, San Francisco (UCSF); former behavioral scientist at the RAND Corporation.

Purpose

Self-report methodologies constitute the cornerstone of empirical inquiry within clinical psychology, behavioral medicine, and health services evaluation. However, the integrity of self-reported data is continuously threatened by systematic measurement error, most notably response styles. Among these, socially desirable responding (SDR)—the tendency of individuals to project favorable images of themselves across evaluative dimensions—represents one of the most pervasive threats to both internal and construct validity. Individuals exhibiting high levels of SDR systematically inflate desirable attributes (such as medical adherence, positive affect, altruism, and self-control) while attenuating or denying undesirable attributes (such as emotional distress, psychiatric symptoms, substance abuse, and interpersonal hostility).

Historically, investigators relied on extensive batteries, such as the original 33-item Marlowe-Crowne Social Desirability Scale (Crowne & Marlowe, 1960) or the 40-item Balanced Inventory of Desirable Responding (Paulhus, 1991). While psychometrically robust, the inclusion of 30 to 40 supplementary items designed exclusively to assess response bias imposes substantial cognitive and temporal burdens on participants. In multi-instrument batteries, clinical trials, or longitudinal health surveys—such as the Medical Outcomes Study—lengthy bias scales are routinely abandoned due to participant fatigue, missing data, and administrative expense.

The primary purpose of the SDRS-5 was to resolve this methodological impasse. Hays, Hayashi, and Stewart (1989) sought to construct a brief, five-item proxy measure that preserved the construct coverage and diagnostic sensitivity of the full MCSDS while requiring less than sixty seconds to complete. The instrument serves three distinct functions across empirical and clinical domains:

• Statistical Control and Covariate Adjustment: The SDRS-5 can be entered as a nuisance covariate in multivariate regression models, structural equation models (SEM), and general linear models to statistically partial out the variance attributable to defensive self-presentation, thereby uncovering the “uncontaminated” associations between substantive predictors and outcomes.
• Screening and Data Cleansing: Researchers can deploy the SDRS-5 to screen for respondents demonstrating extreme defensive posturing, establishing a priori cutoffs to flag or exclude uninterpretable protocols in forensic, occupational, or high-stakes diagnostic contexts.
• Construct Validation and Disentanglement: In the development of novel psychological scales, correlating proposed target measures with the SDRS-5 demonstrates whether a new inventory measures a distinct psychological phenomenon or primarily taps into shared socially desirable response tendencies.

Psychological Construct

The psychological construct assessed by the SDRS-5 is Socially Desirable Responding (SDR), conceptualized within the psychometric tradition of Douglas P. Crowne and David Marlowe. Unlike early conceptualizations that linked response bias directly to psychopathology (such as the Lie and Correction scales of the MMPI), Crowne and Marlowe redefined social desirability as a normal personality process driven by a generalized need for approval and the avoidance of interpersonal disapproval.

In contemporary psychometric theory, socially desirable responding is widely recognized as a multi-component phenomenon, bifurcated by Delroy L. Paulhus into two primary dimensions:

1. Self-Deceptive Enhancement (SDE): An unconscious, honest, but unrealistically positive cognitive self-bias, wherein individuals genuinely believe their inflated, self-aggrandizing evaluations.
2. Impression Management (IM): A deliberate, conscious, and context-dependent manipulation of self-descriptions designed to project a socially sanctioned public persona and elicit approval from evaluators.

The SDRS-5 predominantly assesses the Impression Management (IM) facet, operating through the endorsement of behavioral and emotional exemplars that represent culturally valued virtues that are, in reality, virtually impossible to maintain unconditionally. The scale operates on the diagnostic principle of statistical improbability. The five items present statements about everyday interpersonal behaviors, emotional reactions, and social interactions:

• Virtuous Claims (Over-claiming desirable traits): Endorsing statements such as “I am always courteous even to people who are disagreeable” and “No matter who I'm talking to, I'm always a good listener” requires claiming a superhuman level of patience, self-restraint, and interpersonal warmth. In everyday human behavior, virtually all individuals occasionally experience impatience or conversational disengagement. Consequently, absolute endorsement of these behaviors reflects a pronounced bias toward positive self-presentation.
• Denial of Common Transgressions (Repudiating universal human flaws): Rejecting statements such as “There have been occasions when I took advantage of someone,” “I sometimes try to get even rather than to forgive and forget,” and “I sometimes feel resentful when I don't get my way” requires denying ubiquitous emotional experiences. Virtually every adult has experienced petty resentment, vengeful impulses, or minor opportunistic behaviors. Strongly denying ever experiencing these normative internal states indicates an active effort to conceal personal imperfections.

By contrasting the endorsement of improbable perfection against the denial of common moral failings, the SDRS-5 captures the respondent's baseline threshold for defensive self-presentation across evaluative testing situations.

Theoretical Framework

The theoretical architecture of the SDRS-5 rests upon the intersection of Social Cognitive Survey Methodology and Personality-Based Need for Approval Theory.

The Need for Approval Paradigm

The primary theoretical foundation is rooted in Crowne and Marlowe’s (1960, 1964) conceptual model. Crowne and Marlowe posited that self-report responses do not solely reflect factual recall of one’s history; instead, they function as operant interpersonal behaviors aimed at securing social approval and mitigating vulnerability. In any testing environment, an implicit social evaluative threat is present. Individuals differing in their underlying need for approval respond to this threat differentially: those with elevated approval dependency utilize defensive coping mechanisms, filtering self-relevant information through a lens of normative conformity. The SDRS-5 assumes that this personality trait manifests consistently across varied survey items, creating an identifiable “response set” that distorts true-score variance.

Cognitive Processing Models of Survey Answering

The instrument also integrates the four-stage cognitive model of survey response proposed by Tourangeau, Rips, and Rasinski (2000), which delineates the psychological operations respondents execute when encountering an item:

1. Comprehension: Decoding the linguistic meaning of the statement.
2. Retrieval: Accessing relevant memories, behavioral instances, and autobiographical events from long-term memory.
3. Judgment: Integrating the retrieved information to formulate an internal evaluation.
4. Response Formatting/Editing: Mapping the internal evaluation onto the available response options.

Within this framework, the SDRS-5 operates precisely at the fourth stage: response editing. Even if an individual retrieves memories of feeling resentful or being rude to a disagreeable person, the presence of an elevated social desirability set induces a cognitive editing filter. The respondent suppresses the veridical memory and consciously or semi-consciously selects an option (e.g., “Definitely False”) that aligns with social norms. The five items of the SDRS-5 are engineered with minimal ambiguity, maximizing the likelihood that variation in scores reflects this editing process rather than comprehension or retrieval errors.

Validity

The construct, convergent, discriminant, and predictive validity of the SDRS-5 has been extensively documented in psychometric literature across diverse populations.

Convergent Validity

During its initial validation in the Medical Outcomes Study, Hays et al. (1989) evaluated the performance of the SDRS-5 against the original 33-item Marlowe-Crowne scale across large patient cohorts ($N > 2,000$). The product-moment correlation between the 5-item SDRS-5 and the full 33-item parent instrument was exceptionally high, consistently exceeding $r = .85$ ($p < .001$), with some subsamples reaching $r = .92$. This demonstrates that the extreme reduction in item length (an 85% item reduction) did not compromise the representation of the core Marlowe-Crowne construct. In subsequent research, the SDRS-5 demonstrated robust positive correlations ($r = .50$ to $.65$) with other established impression management measures, such as the Paulhus Deception Scales (PDS) Impression Management subscale.

Discriminant Validity

Discriminant validity was established by examining the relationships between the SDRS-5 and various substantive health and psychological constructs. Hays and colleagues demonstrated that while the SDRS-5 correlated moderately with psychological well-being measures ($r ≈ .20$ to $.35$), its correlations with purely physical health dimensions—such as physical functioning, role limitations due to physical health, and pain severity—were near zero ($r < .10$). This differentiation confirms that the SDRS-5 does not measure general somatic health or functional capacity, but specifically reflects psychological and evaluative bias. Furthermore, its modest correlation with self-deceptive enhancement indicates that the SDRS-5 is distinct from general optimism or positive affectivity.

Construct and Criterion-Related Validity

Empirical studies utilizing experimental demand characteristics have provided strong criterion validity evidence. When participants are administered survey batteries under explicit “fake good” instructions (e.g., simulating a job application or custody evaluation), mean scores on the SDRS-5 demonstrate substantial and statistically significant elevations ($d > 1.00$) compared to baseline or “honest responding” conditions. Conversely, under anonymous testing conditions with guaranteed confidentiality, SDRS-5 scores demonstrate normal distributional properties without floor or ceiling truncation.

Reliability

Evaluating the internal consistency of short scales presents unique psychometric challenges, as Cronbach’s alpha is mathematically constrained by the total number of items ($k$). Despite containing only five items, the SDRS-5 demonstrates satisfactory reliability across diverse empirical investigations:

• Internal Consistency: In the primary validation sample reported by Hays, Hayashi, and Stewart (1989), the Cronbach’s alpha for the SDRS-5 was .68. In clinical subsamples with chronic medical conditions (hypertension, diabetes, coronary heart disease, and depression), alpha coefficients remained remarkably stable, ranging from .66 to .68. Applying the Spearman-Brown prophecy formula demonstrates that a hypothetical 33-item instrument possessing the average inter-item correlation of the SDRS-5 would yield an alpha exceeding .88, confirming the high psychometric efficiency of the selected five items.
• Item-Total Correlations: Corrected item-total correlations for each of the five items consistently fall within acceptable ranges (.32 to .52). Item 4 (“I sometimes feel resentful when I don't get my way”) and Item 1 (“I am always courteous even to people who are disagreeable”) routinely display the strongest discrimination indices.
• Test-Retest Stability: In longitudinal panels surveyed over intervals of one to two years within the Medical Outcomes Study, test-retest reliability coefficients for the SDRS-5 ranged between $r = .65$ and $r = .75$, demonstrating that the scale captures an enduring, trait-like response disposition rather than transient, state-dependent fluctuations.

Factor Analysis

The structural dimensionality of the SDRS-5 has been investigated using both exploratory (EFA) and confirmatory factor analysis (CFA) across numerous empirical studies.

Exploratory Factor Analysis (EFA)

Principal components analysis and exploratory maximum likelihood factor analyses conducted by Hays et al. (1989) revealed a robust single-factor solution. The first unrotated eigenvalue accounts for a substantial majority of the common variance, fulfilling Reckase’s criterion for essential unidimensionality. Factor loadings for the five items onto this general social desirability dimension typically range from .50 to .75, as summarized below:

• Item 1 (“Always courteous…”): Loadings typically range between .52 and .64.
• Item 2 (“Took advantage…”): Loadings typically range between .48 and .59.
• Item 3 (“Try to get even…”): Loadings typically range between .55 and .68.
• Item 4 (“Feel resentful…”): Loadings typically range between .60 and .74.
• Item 5 (“Always a good listener…”): Loadings typically range between .50 and .62.

Confirmatory Factor Analysis (CFA)

Subsequent CFA evaluations have examined single-factor versus bifactor or method-effect models. A pure unidimensional model provides adequate fit, although slight residual covariance is frequently observed between the two positively keyed items (Items 1 and 5) due to common method variance associated with item directionality. When modeling a single substantive trait factor along with an orthogonal method factor for item directionality, model fit indices are exceptional:

• Comparative Fit Index (CFI) > .98
• Tucker-Lewis Index (TLI) > .97
• Root Mean Square Error of Approximation (RMSEA) < .05
• Standardized Root Mean Square Residual (SRMR) < .03

These findings substantiate that the SDRS-5 operates as an essentially unidimensional metric suitable for summing into a single composite score.

Instrument / Measurement Tool

• Instrument Name: Socially Desirable Response Set 5-Item Scale (SDRS-5)
• Alternative Titles: RAND SDRS-5, MOS Social Desirability Scale
• Authors: Ron D. Hays, Toshi Hayashi, and Anita L. Stewart (1989)
• Construct Assessed: Socially desirable response bias (Impression Management / Need for Approval)
• Administration Format: Self-administered pencil-and-paper questionnaire, digital survey (web/mobile), or structured interview
• Administration Time: Under 1 minute (approximately 30 to 60 seconds)
• Target Population: Adolescents and adults (typically aged 14 and older) across general, clinical, and community populations
• Number of Items: 5 items
• Item Keying:
  • Positively Keyed (Virtuous Claims): Item 1, Item 5 (Reverse scored in continuous scoring)
  • Negatively Keyed (Normative Transgressions): Item 2, Item 3, Item 4
• Response Scale: 5-point Likert-type scale:
  • 1 = Definitely True
  • 2 = Mostly True
  • 3 = Don’t Know
  • 4 = Mostly False
  • 5 = Definitely False
• Scoring Protocols: Researchers typically employ one of two scoring methods depending on analytical objectives:
  • Dichotomous (Extreme Response) Scoring: This is the classic scoring algorithm originally formulated by Hays et al. (1989) for identifying extreme social desirability. In this protocol, one point is assigned for each item answered in the socially desirable extreme:
    • For Item 1 and Item 5: Score 1 if “Definitely True” (1); score 0 otherwise (2, 3, 4, or 5).
    • For Item 2, Item 3, and Item 4: Score 1 if “Definitely False” (5); score 0 otherwise (1, 2, 3, or 4).
    • Total dichotomous score ranges from 0 to 5, representing the total count of extreme socially desirable responses.
  • Continuous (Linear) Scoring: For continuous modeling and structural equation modeling, items can be scored on a 1-to-5 continuum:
    • Reverse score Item 1 and Item 5 so that higher values reflect greater social desirability ($1 → 5, 2 → 4, 3 → 3, 4 → 2, 5 → 1$).
    • Keep Item 2, Item 3, and Item 4 unreversed ($1 → 1, 2 → 2, 3 → 3, 4 → 4, 5 → 5$), where 5 corresponds to Definitely False (the desirable response).
    • Sum the five items (range: 5 to 25) or compute the mean across answered items (range: 1.0 to 5.0).

Permissions & Fee and Test Year

The SDRS-5 was published in 1989 through the RAND Corporation's Medical Outcomes Study. As part of RAND's public health research dissemination mandate, the instrument was placed in the public domain. It is accessible free of charge for non-commercial academic research, clinical investigations, and educational applications without requiring explicit written licensing permission from the authors or RAND, provided that proper bibliographic citation is accorded to the original publication (Hays, Hayashi, & Stewart, 1989).

Information and public documentation regarding the instrument are maintained on the official RAND Health Care Survey Tools portal.

References

• Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability independent of psychopathology. Journal of Consulting Psychology, 24(4), 349–354. https://doi.org/10.1037/h0047358
• Crowne, D. P., & Marlowe, D. (1964). The approval motive: Studies in evaluative independence. John Wiley & Sons.
• Hays, R. D., Hayashi, T., & Stewart, A. L. (1989). A five-item measure of socially desirable response set. Educational and Psychological Measurement, 49(3), 629–636. https://doi.org/10.1177/001316448904900315
• Paulhus, D. L. (1984). Two-component models of socially desirable responding. Journal of Personality and Social Psychology, 46(3), 598–609. https://doi.org/10.1037/0022-3514.46.3.598
• Paulhus, D. L. (1991). Measurement and control of response bias. In J. P. Robinson, P. R. Shaver, & L. S. Wrightsman (Eds.), Measures of personality and social psychological attitudes (pp. 17–59). Academic Press. https://doi.org/10.1016/B978-0-12-590241-0.50006-X
• Reynolds, C. R. (1982). Convergent and divergent validity of the Revised Children’s Manifest Anxiety Scale. Educational and Psychological Measurement, 42(4), 1205–1212. https://doi.org/10.1177/001316448204200428
• Tourangeau, R., Rips, L. J., & Rasinski, K. (2000). The psychology of survey response. Cambridge University Press. https://doi.org/10.1017/CBO9780511819322

13. Items of the Scale (Questionnaire)