Rosenberg Self-Esteem Scale (RSES)

The RSES was designed to capture “a positive or negative attitude toward the self as a totality” operationalising the theoretical construct of global self-acceptance (Rosenberg, 1965). The RSES was designed to capture “a positive or negative attitude toward the self as a totality” operationalising the theoretical construct of global self-acceptance (Rosenberg, 1965). The RSES is grounded in Rosenberg’s conceptualisation of self-esteem as a fundamental dimension of self-evaluation that encompasses an individual’s overall sense of worth and personal value.

The RSES comprises two subscales distinguishing between instrumental self-evaluation and intrinsic self-worth:

• Self-Competence (SC): assesses confidence in personal capabilities and sense of efficacy, reflecting the instrumental dimension of self-evaluation. This subscale captures beliefs about one’s ability to achieve goals and handle challenges effectively, with items focusing on feeling capable, having good qualities, and being able to do things as well as others.
• Self-Liking (SL): measures intrinsic self-worth and personal acceptance, representing the affective dimension of self-evaluation. This subscale reflects fundamental feelings about oneself as a person, independent of specific achievements, with items addressing overall satisfaction with oneself, feelings of worth, and positive self-regard.

The scale is suitable for individuals aged 12 years and above, spanning adolescence through late adulthood across diverse populations, from community samples to clinical groups presenting with various mental health conditions (Sinclair et al., 2010; Orth & Robins, 2014). Its brevity and accessibility make it particularly suitable for routine clinical assessment and repeated measurement contexts where participant burden must be minimised.

Clinical Utility of the RSES

Extensive research supports the clinical relevance of RSES scores, with low self-esteem consistently linked to depression, anxiety, and psychological distress (Gray-Little et al., 1997). Furthermore, longitudinal evidence indicates that low self-esteem prospectively predicts later depressive symptoms, independent of prior mood states (Orth et al., 2008). The scale also shows meaningful relationships with life satisfaction, interpersonal functioning, and treatment outcomes across various psychological interventions. These associations highlight the role of self-esteem as a transdiagnostic factor that may contribute to both the development and maintenance of psychological difficulties. 

Clinicians can utilise the RSES for multiple purposes within therapeutic contexts. As a screening tool, it helps identify low levels of self-esteem which may maintain psychological difficulties, particularly in cognitive-behavioural case formulations for depression and anxiety. For treatment planning, it can indicate the need for specific self-esteem interventions, such as self-compassion modules or cognitive restructuring targeting negative self-beliefs. The scale proves particularly valuable for tracking therapeutic progress, as it can detect meaningful change over treatment periods (Sinclair et al., 2010).

RSES Total Score and Subscale Scores

The scale yields a single total score ranging from 0 to 30, with higher scores indicating more positive self-esteem. Subscale scores range from 0-15 and are listed below:

Self-Competence (5 items: 3, 4, 5, 7, 9): assesses confidence in personal capabilities and sense of efficacy, reflecting beliefs about one’s ability to achieve goals and handle challenges effectively.

Self-Liking (5 items: 1, 2, 6, 8, 9, 10): measures intrinsic self-worth and personal acceptance, representing fundamental feelings about oneself as a person independent of specific achievements.

RSES Interpretive Guidelines

Score interpretation follows empirically derived percentile guidelines based on the total score of the community sample reported in Sinclair et al. (2010):

• Very low scores fall at and below the 14th percentile and indicate significant self-esteem difficulties associated with elevated psychological distress and increased vulnerability to mood disorders.
  
• Low scores range from the 15th to 29th percentiles, representing below-average self-esteem that may contribute to psychological difficulties and warrant therapeutic attention.
  
• Average scores fall between the 30th and 70th percentiles, representing typical self-esteem levels consistent with community norms and psychological well-being.
  
• High scores fall between the 71st and 85th percentile, indicating notably positive self-esteem that surpasses approximately two-thirds of community adults, though very elevated scores may warrant additional assessment to distinguish healthy self-regard from potentially inflated presentations.
  
• Very high equates to at and above the 86th percentile, which, while potentially reflecting genuine self-confidence, may indicate potentially defensive or unrealistic self-appraisal.

RSES Age-Based Normative Reference Groups and Clinically Meaningful Change

These guidelines are used in combination with two age-based normative comparison groups, meaning that depending on the age of the client, whether they are in the 25U or 26O group, the percentile distributions adjust accordingly. For tracking progress, changes of 3 points are considered clinically meaningful, representing approximately .5 standard deviations (Sinclair et al., 2010).

RSES Plots

On first administration, a horizontal comparison bar graph is shown for the total score. This comparison graph illustrates the client’s score relative to community and treatment seeking reference groups. 

A bar chart displays the subscale raw scores for visual comparison. 

When the assessment is administered multiple times, two longitudinal line graphs are generated to track changes in the total score and subscales over time.

RSES Content Validity

The RSES demonstrates robust construct validity through extensive factor analytic and item response theory investigations. Convergent validity is well-established through strong correlations with theoretically related constructs, including life satisfaction, extraversion, and positive affectivity (Schmitt & Allik, 2005). Discriminant validity is evidenced by appropriate correlations with distinct constructs; the scale correlates negatively with neuroticism whilst showing minimal associations with cognitive abilities or social desirability measures. Criterion validity is supported by the scale’s ability to differentiate between clinical and community samples and its predictive utility for therapeutic outcomes.

Internal Consistency of the RSES

Internal consistency reliability consistently meets acceptable standards across diverse populations. Meta-analytic evidence indicates Cronbach’s α ≈ .80 (Gray-Little et al., 1997), with more recent studies reporting values ranging from α = .85 in U.S. community samples (Sinclair et al., 2010) to α = .88 in international samples (Wongpakaran & Wongpakaran, 2012). The scale demonstrates excellent internal consistency across demographic subgroups, with coefficient alpha values typically ranging from .84 to .95, indicating strong item coherence regardless of age, gender, or cultural background. Test-retest reliability shows strong temporal stability, with correlations of r = .85-.88 across 1-2 week intervals (APA, 2023) and r = .63 across seven months (Rosenberg, 1965), suggesting that the scale captures relatively stable individual differences in self-esteem whilst remaining sensitive to meaningful change.

RSES Dimensionality

Factor analytic studies show evidence for the RSES as a unidimensional measure of global self-esteem. Although some investigations suggest a two-factor structure, these factors correspond to positively and negatively worded items rather than substantively distinct constructs. Early interpretations assigned various labels to these factors (e.g., “positive self-regard” versus “self-deprecation”), but subsequent analyses have demonstrated that this apparent dimensionality is the result of method effects associated with item wording direction rather than meaningful psychological distinctions (Huang & Dong, 2012; Marsh, Scalas, & Nagengast, 2010). When method effects are appropriately modelled through bifactor analysis, the scale demonstrates a unidimensional structure with a single general self-esteem factor accounting for the majority of item variance (Hyland et al., 2014). Principal components analysis typically reveals that the first component explains approximately 55% of item-level variance, with item loadings consistently exceeding .60 across diverse samples (Sinclair et al., 2010). This unidimensional structure supports the use of a single total score for clinical interpretation and research applications.

RSES Total Score Normative Data

Normative data for the total score are available from Sinclair et al. (2010), who examined 503 U.S. adults recruited to approximate the characteristics of the general U.S. population, with a mean (SD) age of 44.7 (16.3), ranging from 18-87 and comprising 51.9% females. Differences between male and female scores were minimal, with a mean(SD) of 22.43(6.21) for males and 22.79(5.41) for females. Clinical samples typically demonstrate substantially lower means, with depressed populations often scoring in the 12-15 range, highlighting the influence of psychological distress on scale scores. Age effects are pronounced and indicate gradual increases in self-esteem from adolescence through older adulthood. As a result, NovoPsych has adjusted the normative reference groups by age, described in more detail below.

Based on Sinclair et al. (2010), NovoPsych has established interpretive guidelines using the mean and standard deviation of the total score (M = 22.62(5.80)). The guidelines below provide clinicians with empirically grounded benchmarks for score interpretation. The Very Low range (1st-14th percentiles) represents significant self-esteem difficulties observed in fewer than 14% of community adults. The Low range encompasses the 15th-29th percentiles, indicating below-average self-esteem. Average self-esteem is represented by percentiles between the 30th and 70th, reflecting typical community functioning. High self-esteem corresponds to the 71st to the 85th percentiles, indicating notably positive self-regard that exceeds approximately seven-tenths of the normative sample. Very High self-esteem which, while potentially reflecting genuine self-confidence, may indicate potentially defensive or unrealistic self-appraisal, is represented by scores on the 86th percentile and above.

                                                                                                        Very Low     1st-14th
                                                                                                         Low                 15th-29th
                                                                                                         Average        30th-70th
                                                                                                         High                71st-85th
                                                                                                         Very High     86th+

To account for age differences in average self-esteem across the lifespan, NovoPsych applied these interpretive guidelines to two age-based normative reference groups. These two groups are 25 years old and younger (25U), and those older than 25 years (26O). The total score means and standard deviations for these groups are: 25U = 19.76(5.57), 26O = 23.30(5.46).

Clinical norms for comparison to the Sinclair et al. (2010) community sample are reported here from a sample of treatment-seeking adults (n = 12,106) by NovoPsych (Bartholomew et al., 2025), with a mean(SD) total score of 15.09(6.99). This treatment seeking sample had a mean age(SD) of 35.68(13.14) years, comprising 55.47% females, 35.89% males, with 8.64% not reporting. The lower mean score compared to community norms demonstrates the expected pattern of reduced self-esteem among individuals seeking psychological treatment, with the treatment seeking sample mean falling approximately 1.3 standard deviations below community norms.

RSES Subscale Norms, and Item Attribution Clarification

The literature contains inconsistencies in how RSES subscale items have been attributed to the Self-Competence and Self-Liking dimensions. The original and correct item assignments derive from Tafarodi and Milne (2002), who conducted the foundational factor analytic work establishing these subscales. According to their research, self-competence comprises items 3, 4, 5, 7, and 9, and self-liking comprises items 1, 2, 6, 8, and 10.

However, subsequent studies, including the widely-cited work by Sinclair et al. (2010), do not follow Tafarodi and Milne’s subscale attribution methodology but instead, inadvertently used different item groupings (items 1-5 for Self-Competence and items 6-10 for Self-Liking). This discrepancy means that the subscale-level normative data reported in Sinclair et al. (2010) were calculated using incorrect item assignments that do not align with the subscale structure established by Tafarodi and Milne (2002).

Assess your patients with the Rosenberg Self-Esteem Scale using a NovoPsych account

• For Psychologists & Mental Health Clinicians
• Send Assessments to Patient’s Phone
• Receive Comprehensive Reports
• Access Over 100 Validated Assessments​
• Instant Psychometric Scoring
• Track Symptoms​
• Inform Treatment​