Methodology · Anxiety Test (LBL-ANX) · v1.0

The research behind the score.

A full accounting of the GAD-7 instrument, scoring algorithm, validation evidence, normative data, archetype derivation, and documented author choices. Standalone — readable without ever opening the tool.

Developed by Abiot Y. Derbie, PhD — cognitive neuroscientist & founder. Reviewed by Eskezeia Y. Dessie, PhD — clinical reviewer. Browser-local. Nothing transmitted. No accounts. Source-cited methodology.

1. What this methodology covers

This document is the full methodology for the LifeByLogic Anxiety Test (LBL-ANX). It is written to be standalone — you should be able to read this document without ever opening the tool itself and understand exactly what the tool measures, how the score is computed, what the bands mean, where the population norms come from, how the archetypes are derived, and where the limits of the instrument are.

The tool implements the GAD-7 (Generalized Anxiety Disorder 7-item scale) developed by Spitzer, Kroenke, Williams, and Löwe in 2006. The 7 items, the 0–3 frequency scoring, and the severity bands (0–4 / 5–9 / 10–14 / 15–21) are all standard, validated, and unmodified. What LifeByLogic adds — and what this methodology must justify — is the interpretive layer on top: a sub-dimension symptom profile (cognitive worry / somatic activation / behavioral interference), a 5-archetype framework matching pattern to evidence-based intervention class, a population-norm comparison, a diagnostic-probability context, and a transdiagnostic flag at score ≥10.

Three components are documented author choices rather than direct derivations from the published literature. They are named explicitly throughout this document and summarized in §12: the sub-dimension item assignments (3/3/1 split), the behavioral×3 multiplier in the archetype-matching logic, and the approximate confirmed-GAD rates per band derived from Spitzer 2006 ROC data.

2. Anxiety framework derivation

Generalized anxiety disorder is, per DSM-5, characterized by persistent excessive worry that is difficult to control and accompanied by somatic symptoms such as restlessness, fatigue, difficulty concentrating, irritability, muscle tension, or sleep disturbance. The condition must be present for at least six months and cause clinically significant distress or functional impairment.

The construct is conceptually distinct from situational anxiety (a normal adaptive response to an actual threat), from panic disorder (recurrent unexpected panic attacks), from social anxiety disorder (fear of social evaluation), and from PTSD (anxiety following traumatic experience). However, these conditions share substantial symptom overlap, and the cognitive core of GAD — uncontrollable worry — is found across psychological disorders and is therefore considered transdiagnostic per Harvey et al. 2004 in the cognitive-behavioral processes literature.

Because the GAD-7 captures the cognitive core of worry along with the somatic and behavioral components most commonly co-occurring with it, the instrument has been shown by Kroenke et al. 2007 to function reasonably well as a screen for panic disorder (sensitivity 74%), social anxiety disorder (sensitivity 72%), and PTSD (sensitivity 66%) at the standard cutoff of ≥10. The tool surfaces this transdiagnostic consideration when the user's score reaches that threshold, with copy that emphasizes that a high score deserves multi-domain follow-up rather than a single label.

Lifetime prevalence of GAD in the US population is approximately 5.7% per Kessler et al. 2005 in the National Comorbidity Survey Replication, with 12-month prevalence around 3.1%. Prevalence in primary care settings is higher, ranging from 7% to 10% depending on the sample. This base-rate variation is important for interpreting the diagnostic probability context — see §10.

3. Why the GAD-7 specifically

The GAD-7 was selected over alternative anxiety instruments for five reasons:

  1. Brevity. Seven items can be completed in 2 minutes. The Beck Anxiety Inventory has 21 items, the Penn State Worry Questionnaire has 16, the State-Trait Anxiety Inventory has 40. For a primary-care or self-screen context, brevity matters more than incremental measurement nuance.
  2. Validation. Cronbach's α = 0.92 in the original validation (Spitzer 2006, n=2,740 primary care). Subsequent validation in general population (Löwe 2008, n=5,030; Hinz 2017, n=5,036), psychiatric samples (Beard & Björgvinsson 2014; Rutter & Brown 2017), and across dozens of cultural and linguistic translations.
  3. Standard cutpoints. The 0–4 / 5–9 / 10–14 / 15–21 severity bands are used by NHS IAPT, VA/DoD clinical practice guidelines, APA Anxiety Disorder Treatment Guidelines, and the World Health Organization. A user's clinician is most likely to use the same cutpoints, making the tool's output directly translatable to clinical conversation.
  4. Permissive licensing. The GAD-7 is published with explicit permission from Pfizer to reproduce, translate, display, or distribute without charge. This matters for a free public tool. Other instruments (e.g., MBI, BDI-II) have restrictive licensing fees.
  5. Transdiagnostic utility. Although designed for GAD, the instrument also screens panic, social anxiety, and PTSD reasonably well per Kroenke 2007. A single 2-minute screen with broad transdiagnostic sensitivity is more useful in self-screen contexts than a narrow GAD-only instrument.

The trade-off: a 7-item instrument cannot distinguish anxiety subtypes with the precision of a 30-item battery. The tool addresses this by surfacing the sub-dimension profile (§7) and by flagging transdiagnostic considerations at high scores rather than over-claiming GAD specifically.

4. The 7 GAD-7 items, justified

The 7 items reproduce the published GAD-7 (Spitzer 2006). The cluster annotation is a LifeByLogic interpretive aid documented in §7. The frequency response scale (0–3) is unchanged.

#ItemClusterScore range
1Feeling nervous, anxious, or on edgeSomatic activation0–3
2Not being able to stop or control worryingCognitive worry0–3
3Worrying too much about different thingsCognitive worry0–3
4Trouble relaxingSomatic activation0–3
5Being so restless that it is hard to sit stillSomatic activation0–3
6Becoming easily annoyed or irritableBehavioral interference0–3
7Feeling afraid as if something awful might happenCognitive worry0–3

Response options: 0 = Not at all, 1 = Several days, 2 = More than half the days, 3 = Nearly every day. The reference period is the past 2 weeks. Total score range: 0–21.

Item 7 ("Feeling afraid as if something awful might happen") is given special handling in the results panel: when the user reports near-daily fear (response = 3), the tool surfaces a contextual note acknowledging that persistent existential fear is exhausting to live with and a known feature of anxiety disorders that responds well to evidence-based treatment. This is not a hard escalation modal — that pattern is reserved for instruments with explicit suicidality items (e.g., PHQ-9 item 9 in the Depression Test). It is a gentle contextual surfacing.

5. Scoring algorithm — pseudocode

The scoring runs entirely in the browser. No server calls, no data persistence, no transmission. Pseudocode:

function scoreAnxietyTest(responses):
    # responses = {1: 0..3, 2: 0..3, ..., 7: 0..3}

    total = sum(responses.values())                    # 0..21
    cognitive = responses[2] + responses[3] + responses[7]   # 0..9
    somatic = responses[1] + responses[4] + responses[5]     # 0..9
    behavioral = responses[6]                                # 0..3

    band = scoreToBand(total)
    # 0..4   -> minimal
    # 5..9   -> mild
    # 10..14 -> moderate
    # 15..21 -> severe

    archetype = matchArchetype(total, cognitive, somatic, behavioral)
    # See §8 for matching logic. First match wins.
    # Behavioral is scaled ×3 to bring it into proportion with cognitive (max 9) and somatic (max 9).

    normGen = (total - 3.0) / 3.5     # z-score vs Hinz 2017 general pop
    normPC = (total - 6.1) / 4.7      # z-score vs Spitzer 2006 primary care
    pctGen = normalCDF(normGen) * 100  # percentile in general pop
    pctPC = normalCDF(normPC) * 100   # percentile in primary care

    dxProb = diagnosticProbabilityByBand(band)
    # band -> approximate confirmed-GAD rate per 100, derived from Spitzer 2006 ROC

    showCareAware = (total >= 10)
    showSeverePrompt = (total >= 15)
    showTransdxNote = (total >= 10)
    showItem7Note = (responses[7] == 3)

    return {
        total, cognitive, somatic, behavioral,
        band, archetype,
        normGen: pctGen, normPC: pctPC,
        dxProb,
        flags: {showCareAware, showSeverePrompt, showTransdxNote, showItem7Note}
    }

The normalCDF function uses the Abramowitz & Stegun approximation, accurate to 7.5×10⁻⁸. Percentile interpretation assumes approximately normal distribution of GAD-7 scores in the general population, which is an approximation — Hinz et al. 2017 documents that the actual distribution is right-skewed (most people score low). The percentile is therefore most accurate for moderate-to-high scores; at low scores it slightly overstates how typical the score is.

6. Severity bands & the cutoff debate

The four severity bands come directly from Spitzer et al. 2006 and are used unchanged:

The cutoff of ≥10 is not the only proposed threshold. Plummer et al. 2016 conducted a systematic review and diagnostic meta-analysis suggesting that a cutoff of ≥8 may optimize sensitivity-specificity balance in some primary-care contexts. Beard & Björgvinsson 2014 found cutoff ≥10 had only 79.5% sensitivity in a heterogeneous psychiatric outpatient sample, with high false-positive rates for anxiety disorders other than GAD specifically.

Despite these debates, ≥10 remains the most-cited cutpoint in clinical practice. The tool uses ≥10 because:

  1. It matches what the user's clinician is most likely to use, making the result directly translatable.
  2. The shift from ≥10 to ≥8 changes which side of the moderate band a score of 8–9 falls on; both are still above the minimal band, so the practical recommendation (consider follow-up) is similar.
  3. NHS IAPT, VA/DoD, and APA all use ≥10. Adopting a non-standard cutpoint would create inconsistency between the tool's output and the user's clinical environment.

7. Sub-dimension symptom profile — author choice

This is a documented author choice. The original GAD-7 is unidimensional in factor analysis (Spitzer et al. 2006 confirmed single-factor structure across primary-care samples; Hinz et al. 2017 replicated in general population). The 3-dimension profile is presented as an interpretive aid for matching pattern to evidence-based interventions, not as a validated subscale.

The 7 items split into three interpretive sub-dimensions based on item content:

The 3/3/1 imbalance reflects the GAD-7's design: Spitzer's team selected items emphasizing the cognitive and somatic core of GAD (worry plus arousal), with behavioral interference under-represented. We do not modify the instrument; we surface the imbalance and use a behavioral×3 multiplier in the archetype-matching logic (§8) to bring the dimension into proportion for pattern detection.

Why surface this profile at all if it is not validated? Because clinical anxiety treatment differentiates strongly between intervention classes:

A single severity score does not surface which intervention class is most appropriate. The sub-dimension profile does. This is the tool's primary differentiator from a generic GAD-7 implementation that delivers only a 0–21 score and a band label.

8. Archetype framework derivation

Five archetypes match symptom profile to evidence-based intervention class. The framework is LifeByLogic-derived, not a published clinical typology, and is explicitly named as an interpretive author choice on both the tool page and this methodology page.

Matching logic, in order (first match wins):

if total < 5:
    return THE_SETTLED

# Behavioral scaled ×3 to compare proportionally
b_scaled = behavioral * 3

if cognitive >= somatic AND cognitive >= b_scaled:
    return THE_WORRIER

if somatic > cognitive AND somatic >= b_scaled:
    return THE_RESTLESS

if b_scaled >= cognitive AND b_scaled >= somatic:
    return THE_REACTIVE

return THE_MULTIDIMENSIONAL

The five archetypes and the evidence underlying their pathway recommendations:

The archetypes are interpretive, not diagnostic. They surface action-relevant patterns; they do not classify users into clinical categories.

9. Normative data sources

Two normative samples anchor the population-norm comparison:

PopulationSampleMeanSDSource
General population (Germany)n = 5,0363.03.5Hinz et al. 2017
Female (Germany)n = 2,6153.23.6Hinz et al. 2017
Male (Germany)n = 2,4212.73.4Hinz et al. 2017
Primary care (US)n = 2,7406.14.7Spitzer et al. 2006
Psychiatric outpatientn = 50211.65.4Beard & Björgvinsson 2014

The general-population norm (Hinz 2017) anchors the user's score relative to people not seeking help. The primary-care norm (Spitzer 2006) anchors the user's score relative to people who consulted a primary-care clinician about symptoms. The two together provide a useful range: a score that is above-average in the general population may still be below the typical primary-care patient.

The psychiatric outpatient norm (Beard 2014) is included for completeness but is not displayed in the tool because it represents a heavily selected population (people already in psychiatric treatment) and most users will not be in that context.

An optional sex-stratified comparison is offered (with explicit opt-out). When sex is provided, the female or male Hinz 2017 norm replaces the general population norm. This is a small but meaningful improvement: anxiety prevalence is higher in women than men across most studies, and using the general-population norm without sex stratification slightly understates how typical a high score is for women and slightly overstates it for men.

10. Diagnostic probability context

Rather than reporting positive predictive value (PPV) or negative predictive value (NPV) — which are population-dependent and difficult to interpret — the tool surfaces an approximate "X of 100" framing derived from Spitzer 2006 ROC data:

BandApproximate confirmed-GAD rate
0–4 Minimal≈ 1 in 100
5–9 Mild≈ 12 in 100
10–14 Moderate≈ 55 in 100
15–21 Severe≈ 85 in 100
These values are documented author choices. They are derived from Spitzer 2006 ROC data (sensitivity 89%, specificity 82% at ≥10 in n=965 patients with structured diagnostic interviews) by reading off score-stratified positive predictive values, then rounded to easy-to-interpret numbers. The exact rate depends on the base rate of GAD in the user's population. Spitzer's primary-care sample had ~7-10% GAD prevalence; general population is ~3-5%; clinical settings can be much higher. The tool flags these values as "approximate, derived from Spitzer 2006."

The framing is preferred over PPV because it is more intuitive ("of 100 people who scored like me, how many had GAD?") and avoids implying that the score itself is diagnostic. The tool consistently positions these rates as illustrative — they are not predictions about the individual user. A user scoring 12 has not been told they have a 55% probability of GAD; they have been told that 55 of every 100 people scoring in their band were confirmed with GAD upon clinical interview in Spitzer's primary-care sample.

11. Reliability & validity evidence

The GAD-7 has the deepest psychometric validation evidence of any brief anxiety screener. Highlights:

Limitations of validation: validation studies use the standard cutoff of ≥10 in primary care; performance in specific subpopulations (e.g., college students, perinatal women, migraine patients, post-stroke patients) varies somewhat. Specificity is lower in heterogeneous psychiatric samples (Beard 2014) because the GAD-7 detects anxiety symptoms broadly, including those associated with panic, social anxiety, and PTSD — which is a feature rather than a bug for a transdiagnostic screen but a limitation for GAD-specific case-finding.

12. Documented author choices

Three components of this tool are not directly derivable from the published literature. They are explicit author choices, made for the reasons documented here. Naming them explicitly is an editorial commitment to transparency.

Two additional editorial choices that shape the user experience but are not numerical:

13. Limitations

  1. Snapshot, not trajectory. GAD-7 measures last-2-weeks symptoms. Anxiety naturally fluctuates. A high score during a stressful week may not indicate GAD; a low score during a calm week may not indicate stable absence of an anxiety problem.
  2. Self-report dependent. Honest self-report is the foundation. Self-criticism, alexithymia, denial, recall bias, or aspirational responding can all distort scores in either direction.
  3. Not diagnostic. GAD diagnosis requires DSM-5 or ICD-11 clinical interview. The screen is sensitive (89% per Spitzer 2006) but not diagnostic.
  4. Cultural variation. GAD-7 was developed in US/European primary care. Validation in non-Western populations exists but cultural expression of anxiety varies, and somatic-vs-cognitive emphasis differs across cultures.
  5. Comorbidity ignored at the score level. A high GAD-7 score may reflect panic disorder, social anxiety, PTSD, or depression rather than GAD specifically. The transdiagnostic flag at ≥10 surfaces this consideration but does not differentiate. Users with specific anxiety patterns (panic, social, trauma-related) should consult a clinician for focused screening.
  6. Sub-dimension scoring is interpretive, not validated. See §7 and §12.
  7. Archetypes are interpretive frameworks. See §8 and §12.
  8. Diagnostic probability values are approximate. See §10 and §12.
  9. Population norms are German general population and US primary care. Users from other populations should interpret the norm comparison with cultural awareness.
  10. The 0–21 scoring is unidimensional by design. A score of 14 with cognitive=9, somatic=4, behavioral=1 is the same total as a score of 14 with cognitive=4, somatic=8, behavioral=2. The sub-dimension profile surfaces this difference but the total score does not.

14. Independent review

This methodology was reviewed by Eskezeia Y. Dessie, PhD for clinical accuracy and care-aware appropriateness. The review covered: instrument selection rationale, severity-band justification, the documented author choices in §12, the care-aware infrastructure (top crisis block at ≥10, item-7 contextual note, severe-band clinician prompt at ≥15), and the limitations section. Any factual or clinical errors are the responsibility of the author (Abiot Y. Derbie, PhD); the review does not constitute clinical endorsement of the tool for diagnostic use.

External clinical advisors are invited to submit corrections via the corrections form. Substantive changes will be logged in the version history (§15) and the methodology page will be updated with the next-review date moved up.

15. Version log

VersionDateChanges
v1.02026-05-06Inaugural public release. Full feature set: GAD-7 instrument, 4 severity bands (Spitzer 2006), 3-dimension symptom profile (author choice), 5-archetype framework (author choice), population-norm comparison vs Hinz 2017 + Spitzer 2006, approximate diagnostic-probability context, transdiagnostic flag at ≥10, care-aware top + bottom blocks, item-7 contextual note, severe-band clinician prompt, 13 peer-reviewed citations.

16. Methodology FAQ

Why use the GAD-7 instead of a longer instrument?

The GAD-7 was selected because it is the most widely-used and -validated brief anxiety screener in the world, with over 40,000 citations. Longer instruments like the Beck Anxiety Inventory (21 items), Penn State Worry Questionnaire (16 items), or State-Trait Anxiety Inventory (40 items) carry more nuance but are too long for a 2-minute self-screen. Spitzer's team specifically designed the GAD-7 for primary-care contexts where brevity matters; their 2006 paper documents how they reduced an initial 13-item pool to 7 by selecting items with the highest correlation to total score, balancing brevity against psychometric performance. The GAD-7 also has CC-BY-style permissive licensing — Pfizer published the instrument with explicit permission to reproduce, translate, display, or distribute without charge. This matters for a free public tool.

Why is the cutoff 10 and not 8?

The standard cutoff of ≥10 comes from Spitzer et al. 2006 and remains the most-cited threshold across the clinical literature. It is used in NHS IAPT (Improving Access to Psychological Therapies), the VA/DoD Clinical Practice Guidelines for Major Depressive Disorder and Generalized Anxiety Disorder, and the APA Anxiety Disorder Treatment Guidelines. A 2016 meta-analysis by Plummer et al. suggested cutoff ≥8 may optimize sensitivity-specificity in some contexts, and Beard & Björgvinsson 2014 found cutoff ≥10 had only 79.5% sensitivity in a heterogeneous psychiatric outpatient sample. The tool uses ≥10 because it is the most clinically-established cutpoint and is consistent with clinical practice the user's clinician is most likely to use. The Limitations section flags this debate explicitly.

Why split the GAD-7 into three sub-dimensions when it is unidimensional?

This is a documented author choice. Spitzer et al. 2006 confirmed single-factor structure for the GAD-7 in their original validation — meaning all seven items load on one underlying anxiety construct. We do not dispute that. The 3-dimension profile (cognitive worry, somatic activation, behavioral interference) is presented as an interpretive aid for matching pattern to evidence-based interventions, NOT as a validated subscale. The clinical literature on anxiety treatment differentiates strongly between worry-focused interventions (CBT, metacognitive therapy), arousal-focused interventions (exercise, breath work, PMR), and irritability-focused interventions (sleep, interpersonal repair). Surfacing where the user's symptoms cluster makes the pathway recommendations more actionable than a single severity score alone. The 3/3/1 item assignment reflects the GAD-7's clinical focus on cognitive and somatic symptoms; behavioral interference is under-represented in the original instrument.

How were the archetypes derived?

The 5 archetypes (Settled, Worrier, Restless, Reactive, Multidimensional) are LifeByLogic-derived interpretive frameworks. They are not a published clinical typology and should not be reported as diagnostic categories. The framework was constructed by mapping the 3 sub-dimension profile patterns to the most evidence-supported intervention classes for those patterns: cognitive-dominant maps to worry-control therapies (Hofmann 2012 meta-analysis), somatic-dominant to body-based interventions (Stonerock 2015 exercise meta-analysis), behavioral-dominant to sleep and interpersonal interventions (Kahn-Greene 2007 sleep-irritability research), no-dominant-pattern to multi-modal protocols (Barlow 2017 Unified Protocol). The 'first match wins' ordering means more specific archetypes are tested before broader ones, so a user with cognitive=somatic but behavioral×3 lower is matched to The Worrier rather than The Multidimensional.

What is the behavioral×3 multiplier?

Another documented author choice. The GAD-7 has 3 cognitive-worry items (max score 9), 3 somatic-activation items (max score 9), but only 1 behavioral-interference item (max score 3). Without rescaling, behavioral interference would never trigger archetype assignment because its raw score can never exceed 3 while cognitive and somatic can reach 9. The behavioral×3 multiplier scales the behavioral dimension to bring it into proportion with the other two, allowing The Reactive archetype to match when behavioral irritability genuinely dominates the user's profile. This is a numerical reconciliation choice, not a claim that behavioral interference is three times as important as the other dimensions.

Why are the 'approximately X of 100' diagnostic probability values approximate?

These values are derived from Spitzer et al. 2006 ROC curve data, which reported sensitivity 89% and specificity 82% at the ≥10 cutoff for n=965 patients with confirmed structured diagnostic interviews. We translated this into score-stratified diagnostic confirmation rates by reading off the score-by-score positive predictive value. The exact PPV depends on the base rate of GAD in the user's population — primary care has roughly 7-10% GAD prevalence (Kroenke 2007), general population is ~3-5% lifetime prevalence (Kessler 2005), and clinical settings can be much higher. We use Spitzer's primary-care sample base rate and round to easy-to-interpret numbers (1, 12, 55, 85 in 100). These values are illustrative and explicitly named as 'approximate, derived from Spitzer 2006' on the tool page.

What if my score is high but I do not feel anxious?

There are several possibilities. First, GAD-7 measures last-2-weeks symptom frequency; if the past two weeks were unusually stressful, your score reflects that snapshot, not a stable trait. Second, the GAD-7 captures somatic symptoms (restlessness, trouble relaxing) that some people attribute to other causes (stimulant intake, sleep deprivation, medical conditions). Third, the instrument is also sensitive to depression-anxiety mixed presentations and to other anxiety disorders (panic, social anxiety, PTSD per Kroenke 2007). A high score reflects 'symptoms consistent with anxiety' rather than 'you have GAD.' If the score does not match your subjective experience, that mismatch is itself useful information to discuss with a clinician — both directions of mismatch (high score / low felt anxiety, or low score / high felt anxiety) carry diagnostic signal.

How does this tool differ from the standard clinical GAD-7?

The 7 items and 0-3 frequency scoring are identical to the standard clinical GAD-7. The validated severity bands (0-4, 5-9, 10-14, 15-21) are identical. The differences are interpretive layers added on top: a sub-dimension symptom profile (cognitive/somatic/behavioral, our author choice), a 5-archetype framework matching pattern to evidence-based intervention class (our author choice), a population-norm comparison vs Hinz 2017 general population and Spitzer 2006 primary care, an approximate diagnostic-probability context derived from Spitzer's ROC data, and a transdiagnostic flag at ≥10 surfacing that GAD-7 also screens panic/social anxiety/PTSD per Kroenke 2007. None of these layers alter the core GAD-7 score. A clinician using this tool sees the same standard GAD-7 result they would see using any GAD-7 implementation.

Is the test valid for cultures outside the US/Europe?

The GAD-7 has been translated and validated in over 30 languages including Spanish, Mandarin, Korean, Arabic, Portuguese, German, French, Japanese, Hindi, and Swahili. Validation studies in non-Western populations generally support the instrument's reliability (Cronbach's α typically >0.85) and validity (factor structure replicates), though specific cutoff values may differ by population. Cultural expression of anxiety also varies — somatic vs cognitive emphasis differs across cultures, and the linguistic framing of 'worry' does not translate identically across languages. The tool uses the original English wording with the standard Spitzer cutoffs. Users from non-Western cultures should interpret the screen with cultural awareness and consult clinicians familiar with their population for diagnostic confirmation.

How was the version 1.0 release determined?

Version 1.0 represents the initial public release with the full feature set: GAD-7 instrument, severity bands, sub-dimension symptom profile, 5-archetype framework, population-norm comparison, diagnostic-probability context, transdiagnostic flag, care-aware crisis resources, item-7 contextual note, severe-band clinician prompt, and 13 peer-reviewed citations. Future versions will track changes to scoring logic, archetype thresholds, normative data sources, or pathway recommendations. Cosmetic changes to copy or styling do not bump the version. The version log section below documents change history.

17. How to cite this methodology

If you reference this methodology in academic, clinical, or professional work, use one of the standard citation formats below.

LifeByLogic. (2026). Anxiety Test methodology: GAD-7 instrument, scoring, archetypes, and limitations (Version 1.0). https://lifebylogic.com/behavior-lab/anxiety-test/methodology/

18. References

  1. Spitzer, R. L., Kroenke, K., Williams, J. B. W., & Löwe, B. (2006). A brief measure for assessing generalized anxiety disorder: the GAD-7. Archives of Internal Medicine, 166(10), 1092–1097. doi.org/10.1001/archinte.166.10.1092
  2. Löwe, B., Decker, O., Müller, S., Brähler, E., Schellberg, D., Herzog, W., & Herzberg, P. Y. (2008). Validation and standardization of the Generalized Anxiety Disorder Screener (GAD-7) in the general population. Medical Care, 46(3), 266–274. doi.org/10.1097/MLR.0b013e318160d093
  3. Hinz, A., Klein, A. M., Brähler, E., Glaesmer, H., Luck, T., Riedel-Heller, S. G., et al. (2017). Psychometric evaluation of the Generalized Anxiety Disorder screener GAD-7, based on a large German general population sample. Journal of Affective Disorders, 210, 338–344. doi.org/10.1016/j.jad.2016.12.012
  4. Plummer, F., Manea, L., Trepel, D., & McMillan, D. (2016). Screening for anxiety disorders with the GAD-7 and GAD-2: a systematic review and diagnostic metaanalysis. General Hospital Psychiatry, 39, 24–31. doi.org/10.1016/j.genhosppsych.2015.11.005
  5. Kroenke, K., Spitzer, R. L., Williams, J. B. W., Monahan, P. O., & Löwe, B. (2007). Anxiety disorders in primary care: prevalence, impairment, comorbidity, and detection. Annals of Internal Medicine, 146(5), 317–325. doi.org/10.7326/0003-4819-146-5-200703060-00004
  6. Beard, C., & Björgvinsson, T. (2014). Beyond generalized anxiety disorder: psychometric properties of the GAD-7 in a heterogeneous psychiatric sample. Journal of Anxiety Disorders, 28(6), 547–552. doi.org/10.1016/j.janxdis.2014.06.002
  7. Hofmann, S. G., Asnaani, A., Vonk, I. J. J., Sawyer, A. T., & Fang, A. (2012). The efficacy of cognitive behavioral therapy: a review of meta-analyses. Cognitive Therapy and Research, 36(5), 427–440. doi.org/10.1007/s10608-012-9476-1
  8. Goyal, M., Singh, S., Sibinga, E. M. S., Gould, N. F., Rowland-Seymour, A., Sharma, R., et al. (2014). Meditation programs for psychological stress and well-being: a systematic review and meta-analysis. JAMA Internal Medicine, 174(3), 357–368. doi.org/10.1001/jamainternmed.2013.13018
  9. Stonerock, G. L., Hoffman, B. M., Smith, P. J., & Blumenthal, J. A. (2015). Exercise as treatment for anxiety: systematic review and analysis. Annals of Behavioral Medicine, 49(4), 542–556. doi.org/10.1007/s12160-014-9685-9
  10. Andrews, G., Basu, A., Cuijpers, P., Craske, M. G., McEvoy, P., English, C. L., & Newby, J. M. (2018). Computer therapy for the anxiety and depression disorders is effective, acceptable and practical health care: an updated meta-analysis. Journal of Anxiety Disorders, 55, 70–78. doi.org/10.1016/j.janxdis.2018.01.001
  11. Kessler, R. C., Berglund, P., Demler, O., Jin, R., Merikangas, K. R., & Walters, E. E. (2005). Lifetime prevalence and age-of-onset distributions of DSM-IV disorders in the National Comorbidity Survey Replication. Archives of General Psychiatry, 62(6), 593–602. doi.org/10.1001/archpsyc.62.6.593
  12. Barlow, D. H., Farchione, T. J., Sauer-Zavala, S., Latin, H. M., Ellard, K. K., Bullis, J. R., et al. (2017). Unified Protocol for Transdiagnostic Treatment of Emotional Disorders: Therapist Guide (2nd ed.). Oxford University Press.
  13. American Psychiatric Association. (2013). Diagnostic and Statistical Manual of Mental Disorders (5th ed.). DSM-5. APA Publishing.