Positive-Negative Semantic Differential (PN-SMD)

1. Abstract

The Positive-Negative Semantic Differential (PN-SMD) is a psychometric instrument designed to measure relationship satisfaction and subjective relationship quality through a bivariate, two-dimensional evaluative framework. Developed by Richard E. Mattson, Ronald D. Rogge, Matthew D. Johnson, and Frank D. Fincham, the PN-SMD challenges the longstanding psychometric convention of conceptualizing marital and romantic satisfaction as a unidimensional, bipolar continuum ranging from extreme dissatisfaction to extreme satisfaction. Grounded in the Evaluative Space Model (ESM) formulated by John Cacioppo and Gary Berntson, the scale posits that positive and negative relationship evaluations are functionally distinct, neurobiologically dissociable, and capable of co-occurring independently.

The PN-SMD consists of 14 items divided evenly into two orthogonal subscales: the Positive Semantic Dimension (7 items: Interesting, Full, Sturdy, Enjoyable, Good, Friendly, Hopeful) and the Negative Semantic Dimension (7 items: Bad, Lonely, Discouraging, Boring, Empty, Fragile, Miserable). Each item presents an evaluative adjective describing the respondent's romantic relationship, assessed on an 8-point Likert-type intensity scale ranging from 0 ("Not at all") to 7 ("Completely"). By evaluating positive and negative qualities separately, the instrument produces independent indices of relationship positivity and negativity, unlocking nuanced typologies such as ambivalence (high positivity, high negativity) and indifference (low positivity, low negativity) that cannot be detected by conventional unidimensional scales like the Dyadic Adjustment Scale (DAS) or the Couples Satisfaction Index (CSI).

Extensive psychometric investigations have established robust empirical support for the PN-SMD. Confirmatory factor analyses across diverse relationship cohorts consistently favor a two-factor oblique or orthogonal model over a single-factor bipolar structure. The subscales demonstrate high internal consistency, with Cronbach's alpha coefficients routinely exceeding .90 for both positive and negative dimensions across longitudinal samples. The scale shows high convergent validity with established measures of relationship functioning, incremental validity in predicting relationship dissolution and chronic stress, and distinct patterns of neurobiological and affective reactivity.

2. Keywords

Positive-Negative Semantic Differential, PN-SMD, relationship satisfaction, Evaluative Space Model, ambivalence, marital quality, romantic relationships, dyadic adjustment, psychometrics, bivariate evaluation, Richard E. Mattson, Frank D. Fincham

3. Authors

The Positive-Negative Semantic Differential was developed and validated through collaborative programs of research led by:

• Richard E. Mattson, Ph.D. — Professor of Psychology and Director of the Marriage and Family Studies Laboratory, Department of Psychology, Binghamton University (State University of New York), Binghamton, NY, USA.
• Ronald D. Rogge, Ph.D. — Associate Professor of Clinical Psychology, Department of Clinical and Social Sciences in Psychology, University of Rochester, Rochester, NY, USA.
• Matthew D. Johnson, Ph.D. — Professor of Psychology and Director of the Healthy Relationships and Families Lab, Department of Psychology, Binghamton University, Binghamton, NY, USA.
• Frank D. Fincham, Ph.D. — Eminent Scholar and Director of the Family Institute, College of Human Sciences, Florida State University, Tallahassee, FL, USA.
• David A. Paldino, M.S. and Emily K. B. Baker-Davidson, Ph.D. — Contributing research associates across initial psychometric validation studies.

4. Purpose

The primary purpose of the Positive-Negative Semantic Differential (PN-SMD) is to provide an empirically rigorous, theoretically sophisticated measurement of romantic relationship quality that disaggregates positive and negative evaluative processes. For over six decades, relationship science relied almost exclusively on omnibus, unidimensional scales—such as the Marital Adjustment Test (MAT; Locke & Wallace, 1959) and the Dyadic Adjustment Scale (DAS) (Spanier, 1976). These legacy instruments assume a zero-sum, bipolar continuum where the presence of positive feelings necessitates the absence of negative feelings, and vice versa. However, clinical realities and psychological research regularly demonstrate that romantic partners often harbor complex, simultaneous mixtures of warmth, affection, resentment, and distress.

In clinical practice, diagnosing couple distress solely on a unidimensional continuum obscures the specific relational dynamics driving distress. For instance, a couple presenting with moderate scores on a traditional relationship inventory might be categorized as mildly dissatisfied. In contrast, the PN-SMD reveals whether this intermediate score is driven by high ambivalence (abundant passion and affection coupled with high hostility and volatility) or apathy/indifference (the absence of active hostility paired with an emotional vacuum and depleted affection). These two profiles require radically divergent clinical interventions: the ambivalent couple requires emotion regulation, de-escalation of conflict, and cognitive reframing of negative interactions, whereas the indifferent couple requires emotional rekindling, shared novel activities, and behavioral activation.

In basic and applied relationship research, the PN-SMD serves several critical objectives:

• Isolating Trajectories of Decay: Longitudinal investigations reveal that positive and negative relationship qualities change at different rates and respond to distinct stressors. The erosion of relationship positivity often predicts quiet disengagement and emotional drifting, whereas spikes in relationship negativity predict overt hostility, destructive communication patterns, and acute dissolution.
• Capturing Evaluative Ambivalence: By scoring both dimensions independently, researchers can compute explicit ambivalence indices (e.g., Griffin's ambivalence formula or fold-over regression models) to evaluate how relational ambivalence influences psychological well-being, physiological cardiovascular reactivity, immune functioning, and decision-making regarding breakup or divorce.
• Reducing Measurement Redundancy: Many traditional marital scales confound relationship satisfaction (an internal evaluative sentiment) with relationship behaviors, conflict frequency, sexual activity, and shared philosophy. The PN-SMD employs pure semantic differential adjectives, isolating evaluative sentiment from behavioral correlates and eliminating item overlap with external criteria.

5. Psychological Construct

The construct assessed by the PN-SMD is bivariate relationship evaluation, characterized as two distinct neuro-affective and cognitive processing channels: relationship positivity and relationship negativity. Rather than treating subjective satisfaction as a single scalar value, the PN-SMD conceptualizes evaluation as a two-dimensional coordinate system within an evaluative space.

The Positive Semantic Dimension (Positivity)

The positive dimension reflects the degree to which an individual views the romantic relationship as possessing rewarding, stimulating, fortifying, and emotionally enriching qualities. It captures positive reinforcement, intimacy, companionship, shared joy, and optimism. Within the PN-SMD, this construct is operationalized through seven explicit semantic markers:

• Interesting: Captures cognitive and interpersonal engagement, curiosity, novelty, and the antithesis of relational stagnation.
• Full: Reflects emotional completeness, relational richness, and perceived depth of shared experiences.
• Sturdy: Measures relational resilience, dependable security, trust, and structural stability against external adversities.
• Enjoyable: Reflects hedonic pleasure, playfulness, day-to-day positive affect, and shared recreational satisfaction.
• Good: Serves as a global, moral, and experiential affirmation of the relationship's overall benevolence and constructive worth.
• Friendly: Reflects warmth, companionship, alliance, interpersonal gentleness, and authentic fondness between partners.
• Hopeful: Assesses prospective confidence, forward-looking optimism, shared long-term aspirations, and constructive future orientation.

The Negative Semantic Dimension (Negativity)

The negative dimension captures the degree to which an individual evaluates the romantic partnership as draining, aversive, isolating, fragile, and emotionally distressing. Importantly, relationship negativity is not merely the mathematical subtraction of positivity; it represents the activation of threat, alienation, conflict, and behavioral inhibition systems. The PN-SMD operationalizes this dimension via seven distinct negative adjectives:

• Bad: Global experiential appraisal of toxicity, moral or emotional wrongness, and fundamental dissatisfaction.
• Lonely: Captures emotional abandonment, unmet attachment needs, dyadic isolation, and perceived emotional disconnection despite physical partnership.
• Discouraging: Assesses demoralization, chronic exhaustion, perceived futility of improvement efforts, and loss of relational agency.
• Boring: Reflects emotional apathy, monotony, lack of stimulation, and relational atrophy.
• Empty: Indicates the subjective perception of relational hollow core, lack of substance, emotional starvation, and hollow routine.
• Fragile: Assesses structural vulnerability, walking on eggshells, constant threat of disintegration, and lack of foundational safety.
• Miserable: Captures acute suffering, psychic distress, sorrow, and intense unhappiness stemming directly from the partnership.

Evaluative Space Quadrants

Plotting an individual's scores on the two 0–49 coordinate axes produces four distinct relational configurations:

1. Congruent Satisfaction (High Positivity, Low Negativity): Characterized by thriving, flourishing partnerships with high intimacy and minimal aversive transactions.
2. Congruent Distress (Low Positivity, High Negativity): Characterized by severe distress, overt hostility, high conflict, emotional alienation, and heightened risk of separation.
3. Ambivalence (High Positivity, High Negativity): Highly volatile, high-intensity relationships marked by profound passion and deep attachment co-occurring with intense conflict, resentment, and chronic stress.
4. Indifference / Apathy (Low Positivity, Low Negativity): Relationships marked by emotional disengagement, "roommate syndrome," functional coexistence devoid of active warfare but stripped of romance, excitement, and warmth.

6. Theoretical Framework

The theoretical architecture of the PN-SMD is anchored primarily in the Evaluative Space Model (ESM) proposed by John T. Cacioppo and Gary G. Berntson (1994, 1999). Classical psychometric and attitudinal paradigms (e.g., Thurstone, Likert, Osgood's semantic differential) treated attitude as a bipolar continuum ranging from -1 to +1. In contrast, the ESM demonstrates that the underlying affect system is organized into two distinct physical and neural substrates: one dedicated to processing appetitive/reward-related stimuli (positivity) and one dedicated to processing defensive/threat-related stimuli (negativity).

According to the ESM, these two motivational channels possess distinct operational characteristics:

• Positivity Offset: In the absence of explicit stimuli (low activation of both systems), the appetitive system exhibits a slight positive bias, motivating organisms to explore novel environments and initiate social contact.
• Negativity Bias: Negative stimuli produce steeper psychological and physiological activation curves than positive stimuli of equal objective magnitude. Threat, pain, and social rejection demand immediate behavioral responses to prevent harm.
• Modes of Evaluative Activation: The positive and negative evaluative systems can operate in reciprocal activation (one increases while the other decreases), uncoupled activation (one shifts while the other remains baseline), or coactivation (both increase or decrease simultaneously).

When applied to romantic partnerships by Mattson, Rogge, Johnson, and Fincham, the ESM exposes the limitations of unidimensional instruments. In standard bipolar inventories, an individual who rates a relationship as neutral cannot be distinguished from someone experiencing severe coactivation (intense love combined with intense hate) or complete coinhibition (zero affection combined with zero overt conflict). The PN-SMD operationalizes this theoretical insight by severing the semantic poles of Charles Osgood's semantic differential method (Osgood, Suci, & Tannenbaum, 1957). Instead of pairing "Good" versus "Bad" on a single bipolar continuum, "Good" is assessed on an unipolar scale of intensity, and "Bad" is evaluated on an independent unipolar scale of intensity.

The theoretical framework also aligns with attachment theory (Bowlby, 1982; Hazan & Shaver, 1987). Anxious attachment styles frequently yield high coactivation (intense craving for intimacy paired with acute vigilance for betrayal, leading to ambivalence), whereas avoidant attachment styles frequently manifest as coinhibition or elevated negativity coupled with suppressed positivity.

7. Validity

The psychometric validity of the PN-SMD has been rigorously examined across multiple empirical investigations involving dating, engaged, newlywed, and long-term married cohorts.

Construct and Structural Validity

In the seminal validation study by Mattson, Paldino, and Johnson (2007; Psychological Assessment), the construct validity of bivariate assessment was formally compared against traditional unidimensional scoring. Confirmatory factor analysis confirmed that treating positive and negative semantic evaluations as two distinct factors provided a significantly superior fit to the data compared to a unidimensional model. The correlation between the positive and negative subscales ranged from moderate to moderately high (typically r = -.40 to -.65), empirically confirming that while the dimensions share an inverse relationship, they share only 16% to 42% of common variance, demonstrating clear empirical independence.

Convergent and Concurrent Validity

In an extensive follow-up across five independent samples totaling over 2,000 participants (Mattson, Rogge, Johnson, Baker-Davidson, & Fincham, 2013), the PN-SMD exhibited robust convergent validity with gold-standard relationship scales:

• Strong convergent correlations with the Dyadic Adjustment Scale (DAS): Positive dimension correlated positively (r ≈ .68 to .78), while the negative dimension correlated inversely (r ≈ -.65 to -.76).
• Correlations with the Couples Satisfaction Index (CSI): Demonstrating that while the PN-SMD captures core satisfaction variance, its dual scores provide distinct explanatory power.
• Differential associations with observed behavioral interactions: The negative dimension correlated strongly with observed marital hostility, demand-withdraw communication cycles, and criticism during problem-solving discussions (coded using the Rapid Marital Interaction Coding System; RMICS). Conversely, the positive dimension uniquely correlated with observed warmth, validation, humor, and constructive social support.

Incremental and Predictive Validity

The PN-SMD demonstrates marked incremental validity over omnibus unidimensional scales. Mattson et al. (2013) demonstrated that when predicting relationship dissolution over longitudinal follow-ups (up to 2 years), entered together, relationship negativity significantly predicted breakup and divorce even after controlling for baseline global satisfaction. Furthermore, individuals with ambivalent profiles (high positive, high negative) exhibited distinct risk profiles: they did not dissolve relationships at the immediate rate of distressed couples, but they exhibited substantially higher chronic interpersonal stress, depressive symptomology, and sleep disruption.

Subsequent doctoral and master's research at institutions such as Auburn University (German, 2013; Cunningham, 2010) corroborated the scale's predictive utility. German (2013) established that the PN-SMD successfully differentiated between marital disharmony (active behavioral conflict) and marital disaffection (gradual loss of emotional attachment), showing that the negative dimension is heavily tied to overt disharmony, while the positive dimension is the primary indicator of disaffection.

8. Reliability

The Positive-Negative Semantic Differential demonstrates exceptional psychometric reliability across internal consistency, split-half, and test-retest metrics.

Internal Consistency

Internal consistency estimates for both subscales are consistently high across diverse demographic groups, cultural adaptations, and relationship types:

• Positive Semantic Dimension: Cronbach's alpha (α) values routinely range from .91 to .95. McDonald's omega (ω) coefficients similarly range between .92 and .96, indicating that the seven positive adjectives possess excellent homogeneity and item-total correlations exceeding .70.
• Negative Semantic Dimension: Cronbach's alpha (α) values consistently range from .89 to .94. McDonald's omega (ω) coefficients range between .90 and .94, confirming strong internal coherence among the seven negative adjectives.

Test-Retest Reliability and Stability

Longitudinal stability studies documented in Mattson et al. (2013) indicate that the PN-SMD displays expected temporal stability over short-to-medium intervals while maintaining sensitivity to genuine relationship changes:

• Over a 6-to-8-week test-retest interval, intraclass correlation coefficients (ICC) were .84 for the Positive Semantic Dimension and .81 for the Negative Semantic Dimension in stable non-clinical couples.
• Over a 1-year longitudinal interval, stability coefficients were moderately strong (r ≈ .60 to .72), reflecting natural evolutionary shifts in relationship dynamics while retaining significant rank-order consistency.
• Standard error of measurement (SEM) indices remain low (SEM < 2.5 on each 49-point scale), indicating high precision of measurement at both individual and group levels.

9. Factor Analysis

The structural validity of the PN-SMD has been established through both Exploratory Factor Analysis (EFA) and Confirmatory Factor Analysis (CFA), as well as modern Item Response Theory (IRT).

Confirmatory Factor Analysis (CFA) Model Comparison

Mattson et al. (2007, 2013) conducted systematic model comparisons across competing structural configurations:

1. Model 1: One-Factor Bipolar Model. All 14 items load onto a single latent relationship quality factor (with negative items reverse-coded). This model exhibited poor fit: Comparative Fit Index (CFI) < .82, Tucker-Lewis Index (TLI) < .79, Root Mean Square Error of Approximation (RMSEA) > .14, and Standardized Root Mean Square Residual (SRMR) > .10.
2. Model 2: Two-Factor Orthogonal Model. Positive items load on a Positivity factor, negative items load on a Negativity factor, with the factor covariance constrained to zero. Fit improved considerably but remained suboptimal due to the omission of the true natural inverse correlation.
3. Model 3: Two-Factor Oblique Model. Positive items load exclusively on the Positivity factor, negative items load exclusively on the Negativity factor, and the latent factors are allowed to freely correlate. This model demonstrated superior fit across all indices: $\chi^2/df < 2.8$, $\text{CFI} = .97$ to $.99$, $\text{TLI} = .96$ to $.98$, $\text{RMSEA} = .045$ to $.058$ ($90%\text{ CI } [.038, .067]$), and $\text{SRMR} = .034$.

Factor Loadings

Standardized factor loadings ($lambda$) under the two-factor oblique CFA model are robust and statistically significant ($p < .001$) across all items:

• Positive Factor Loadings:
  • Interesting: $lambda = .76 – .82$
  • Full: $lambda = .84 – .89$
  • Sturdy: $lambda = .78 – .84$
  • Enjoyable: $lambda = .85 – .90$
  • Good: $lambda = .88 – .93$
  • Friendly: $lambda = .79 – .85$
  • Hopeful: $lambda = .81 – .87$
• Negative Factor Loadings:
  • Bad: $lambda = .85 – .91$
  • Lonely: $lambda = .74 – .80$
  • Discouraging: $lambda = .82 – .88$
  • Boring: $lambda = .72 – .79$
  • Empty: $lambda = .84 – .89$
  • Fragile: $lambda = .75 – .81$
  • Miserable: $lambda = .86 – .92$

Item Response Theory (IRT) Findings

Samejima's Graded Response Model applied to the PN-SMD demonstrated that the negative items possess exceptionally high discrimination parameters ($a > 2.0$), offering peaked information at moderate to high levels of latent relationship distress. Conversely, positive items provide maximal discrimination at low to moderate levels of relational flourishing. This confirms that measuring both sets of adjectives independently provides a comprehensive band of measurement information across the entire latent spectrum.

10. Instrument / Measurement Tool

• Instrument Name: Positive-Negative Semantic Differential (PN-SMD)
• Authors: Richard E. Mattson, Ronald D. Rogge, Matthew D. Johnson, and Frank D. Fincham
• Test Type: Self-report psychometric questionnaire; semantic differential rating scale
• Target Population: Adults (18+ years) involved in romantic relationships (dating, cohabiting, engaged, or married)
• Administration Format: Paper-and-pencil or computer-administered / web-based survey
• Completion Time: Approximately 2 to 3 minutes
• Number of Items: 14 items (7 positive evaluative adjectives, 7 negative evaluative adjectives)
• Subscales:
  • Positive Semantic Dimension (PSD): Items 1 through 7
  • Negative Semantic Dimension (NSD): Items 8 through 14
• Response Scale: 8-point Likert-type intensity scale:
  • 0 = Not at all
  • 1 = A tiny bit
  • 2 = A little
  • 3 = Somewhat
  • 4 = Mostly
  • 5 = Very
  • 6 = Extremely
  • 7 = Completely
• Scoring Instructions:
  • Positive Semantic Dimension Score: Sum the ratings of all 7 positive items (Range: 0 to 49). Higher scores reflect greater relationship positivity, warmth, and vitality.
  • Negative Semantic Dimension Score: Sum the ratings of all 7 negative items (Range: 0 to 49). Higher scores reflect greater relationship distress, hostility, and alienation.
  • Important Scoring Note: Do NOT reverse-score the negative items to create an omnibus single total score, as doing so destroys the bivariate metric and reintroduces unidimensional bipolarity. The two scores must be retained as independent coordinates $(P, N)$.
  • Ambivalence Index Calculation (Optional): Evaluative ambivalence can be computed using Griffin's formula: $\text{Ambivalence} = \frac{P + N}{2} – |P – N|$, or via polynomial regression with response surface analysis.

11. Permissions & Fee and Test Year

• Year of Initial Publication: 2007 (Initial formulation and validation in Psychological Assessment); expanded and standardized in 2013 (Personal Relationships).
• Copyright & Ownership: The authors (Richard E. Mattson, Ronald D. Rogge, Matthew D. Johnson, and Frank D. Fincham) retain academic intellectual property rights.
• Fee and Licensing: The scale is available free of charge for non-commercial academic, research, and educational purposes. Researchers and clinicians may utilize the instrument without paying royalty fees provided that appropriate bibliographic credit and citation are accorded in all published works.
• Commercial Use: Any commercial deployment, inclusion in proprietary clinical platforms, or use in sponsored profit-making assessments requires prior formal written authorization from the primary authors.
• Official Institutional Repositories: Scale PDF copies and scoring instructions have been made publicly available via university research portals:
  • Binghamton University Marriage and Family Studies Lab: https://www.binghamton.edu/marriage-lab/pdfs/pn-smd.pdf
  • Family Institute Research Repository: http://fincham.info/papers/PNSMD_MS%20updateD.pdf

12. References

• Cacioppo, J. T., & Berntson, G. G. (1994). Relationship between attitudes and evaluative space: A critical review, with emphasis on the separability of positive and negative substrates. Psychological Bulletin, 115(3), 401–423. https://doi.org/10.1037/0033-2909.115.3.401
• Cacioppo, J. T., Gardner, W. L., & Berntson, G. G. (1999). The affect system has parallel and integrative processing components: Form follows function. Journal of Personality and Social Psychology, 76(5), 839–855. https://doi.org/10.1037/0022-3514.76.5.839
• Cunningham, S. K. B. (2010). Reevaluating sexual satisfaction through an expectancy lens: Validation of the Sexual Activity Frequency and Expectation Scale (Master's thesis). Auburn University. Electronic Theses and Dissertations. https://etd.auburn.edu/handle/10415/2229
• German, N. M. (2013). Assessment of disharmony and disaffection (Doctoral dissertation). Auburn University. Electronic Theses and Dissertations. https://etd.auburn.edu/handle/10415/3697
• Mattson, R. E., Paldino, D., & Johnson, M. D. (2007). The increased construct validity and utility of assessing relationship quality using separate positive and negative dimensions. Psychological Assessment, 19(1), 146–151. https://doi.org/10.1037/1040-3590.19.1.146
• Mattson, R. E., Rogge, R. D., Johnson, M. D., Baker-Davidson, E. K. B., & Fincham, F. D. (2013). The positive and negative semantic dimensions of relationship satisfaction. Personal Relationships, 20(2), 328–355. https://doi.org/10.1111/j.1475-6811.2012.01412.x
• Osgood, C. E., Suci, G. J., & Tannenbaum, P. H. (1957). The measurement of meaning. University of Illinois Press.
• Spanier, G. B. (1976). Measuring dyadic adjustment: New scales for assessing the quality of marriage and similar dyads. Journal of Marriage and the Family, 38(1), 15–28. https://doi.org/10.2307/350547

13. Items of the Scale