Concept

Availability Heuristic

Intro

Our minds judge how common something is by how easily we can think of examples. That shortcut is called the availability heuristic, and it gets us into trouble a lot.

Ask people: do more English words start with K, or have K as the third letter? Most say "start with K." Wrong. There are about three times more K-third-letter words. But our brains store words by first letter, so K-first words come faster, and we treat that speed as evidence of frequency. Daniel Kahneman and Amos Tversky named this in 1973, and Kahneman won a Nobel Prize for the broader work.

This bias drives a lot of bad arguments about religion. "Look at the Crusades, the Inquisition, 9/11; religion causes violence." Those examples are vivid and recent in coverage, so they come to mind first. Less vivid: the 100-million-plus deaths under twentieth-century atheist regimes (Stalin, Mao, Pol Pot, Kim). When you ask for the actual base rate of religious vs secular violence per capita per year, the picture flips. Steven Pinker, a non-Christian, lays this out in The Better Angels of Our Nature.

The same trick works the other way too. Christians cite Mother Teresa, Wilberforce, and Bonhoeffer as proof Christianity produces saints. That is just as availability-biased unless we engage the base-rate question honestly. The cure for the bias is the same in both directions: slow down, ask "what is the comparison group?" and demand the cases that did not make the headlines.

Quick reply: "What does the comparative base rate look like? Until we have it, we have a feeling, not an argument."

In full

The systematic cognitive error of judging the frequency, probability, or importance of an event-class by how easily examples of that class come to mind, rather than by reference to actual base rates. Coined by Amos Tversky and Daniel Kahneman in their landmark paper "Availability: A Heuristic for Judging Frequency and Probability" (Cognitive Psychology 5:207-232, 1973), the availability heuristic is one of the three foundational heuristics of behavioral economics (alongside representativeness and anchoring), and the third and final cognitive-bias entry in the Fallacies folder, sister to Confirmation Bias and Survivorship Bias. Distinct from confirmation bias (which is about how the perceiver weights evidence at hand) and from survivorship bias (which is about which evidence ever reaches the perceiver), the availability heuristic operates on the retrievability of evidence already in memory: events that are recent, vivid, emotionally charged, personally experienced, or heavily media-covered are more easily recalled, and that ease-of-recall is misread as evidence of frequency or importance.

The canonical demonstration is the Tversky-Kahneman 1973 K-letter experiment: subjects were asked whether more English words begin with the letter K or have K as the third letter. Most answered "begin with K." The actual answer is that K-third-position words are approximately three times more common than K-initial words, but K-initial words are easier to retrieve (we mentally index by first letter), so they feel more frequent. The availability of retrieval-cued examples replaced the actual frequency in subjects' judgments. The 1973 paper documented the same pattern across multiple domains: causes of death (subjects overestimated deaths from dramatic causes like tornadoes and underestimated deaths from undramatic causes like asthma); celebrity name recognition (more easily recalled names judged as belonging to more prominent people); and a wide range of probability judgments under uncertainty.

Availability heuristic operates in apologetic discourse along five distinguishable manifestations that recur:

Recency effect, events from the recent past are overweighted relative to historical base rates ("the Crusades happened a thousand years ago, but the 9/11 attack felt like yesterday, therefore Islam is more violent than Christianity historically").
Vividness effect, emotionally striking, personally consequential, or visually memorable events overweighted relative to mundane ones (a single Catholic priest abuse case can dominate moral judgment more than statistical evidence about base rates of abuse across professions).
Media-coverage effect, heavily-covered events feel more frequent ("religious violence is covered constantly, so religion must cause most violence", though the actual quantitative base rate, per Pinker The Better Angels of Our Nature 2011, runs heavily the other way: 20th-century atheist regimes killed an order-of-magnitude more than all religious wars in history combined).
Personal-experience effect, cases known directly weighted higher than statistically representative cases (a Christian who has experienced a clear answered prayer overweights prayer-effectiveness; a Christian who has experienced devastating unanswered prayer overweights prayer-failure; both are availability-biased relative to base rates).
Iconic-event-availability, particular cases become the canonical-cited example for an entire category (Galileo for science-vs-religion conflict; Salem for Christian violence; the Crusades for medieval Christian aggression; ISIS for Islamic violence; Stalin for atheist regime mortality), occluding the comparative base rate of the broader category.

The bias is major-severity because (a) it operates pre-attentively on memory-retrieval, before conscious deliberation can correct it, (b) the iconic-event-availability subform is heavily exploited in popular discourse and political-cultural framing (single dramatic cases stand in for entire categories), (c) media incentives systematically amplify availability bias (vivid stories generate engagement; statistical base-rate stories do not), and (d) it operates symmetrically, both atheist polemic and Christian apologetic deployments are vulnerable to the diagnosis.

Canonical structure

The availability heuristic is not a fallacy in the strict syllogistic sense, like Confirmation Bias and Survivorship Bias, it is a cognitive shortcut for probability judgment under uncertainty rather than an inferential-form error. Its general structure (Tversky-Kahneman 1973 + Kahneman Thinking, Fast and Slow 2011 ch. 12-13):

Stage 1: Observer faces a question requiring estimate of frequency, probability, or importance of event-class E
Stage 2: Observer's System 1 (fast, automatic cognition) substitutes the easier question "how easily can I recall instances of E?" for the harder question "what is the base rate of E?"
Stage 3: Retrieval-ease is read as a proxy for actual frequency
Stage 4: Inference systematically over-represents E if E-instances are recent, vivid, emotionally charged, personally experienced, or media-covered, and systematically under-represents E if E-instances are old, mundane, statistical, second-hand, or media-uncovered

The error: at no stage is a formal logical fallacy committed. The substitution of "what comes easily to mind" for "what is the base rate" is a System-1 cognitive shortcut that the observer typically does not consciously notice. Detection requires slowing down to ask "is the available sample of cases I'm drawing on representative of the actual population?", Kahneman's Thinking, Fast and Slow names this the engagement of System 2 (slow, deliberate, statistical reasoning) to override System 1's availability-driven intuition.

This makes the availability heuristic structurally invisible to the unaided observer (the available examples feel like the universal sample) AND systematically reinforced by media + political-cultural incentives (which reward vivid case-studies and penalize statistical base-rate framings). Correction requires deliberate base-rate-seeking + comparator-population analysis, neither of which is intuitive under conversational pressure.

Diagnostic markers

The availability heuristic is operating when:

Single iconic case stands in for category. Argument cites Galileo as proof of science-religion conflict; cites the Crusades as proof of Christian violence; cites Stalin as proof of atheist-regime tendencies; cites 9/11 as proof of Islamic violence. Detection test: ask "what does the comparative base rate look like?", inability to answer or unfamiliarity with the comparative case set indicates availability bias is operating.
Recency over historical depth. Argument weighted toward post-2000 examples or post-1945 examples without engaging the longer historical record. Detection test: where are the medieval, early-modern, and 19th-century cases that would balance the recency framing? (E.g., the medical-evidence base for Lourdes runs from 1858 onward, not just from 2000 onward; the cumulative-historical Christian-civilization case runs from c. AD 30 onward, not just from the post-2000 American culture-war framing.)
Vividness over statistical evidence. Argument privileges single dramatic case over aggregate statistical data. Detection test: does the argument cite peer-reviewed statistical evidence, or only emotionally-resonant individual cases? (E.g., a single dramatic case of a Christian moral failure does not falsify Christian moral teaching; a single dramatic case of a miraculous healing does not establish miracle-effectiveness; both require the comparative base rate.)
Media-coverage as proxy for prevalence. Argument assumes media-coverage frequency tracks event-frequency. Detection test: ask "what is the relationship between media coverage of this event-class and its actual base rate?", the gap between coverage and rate is precisely where availability bias operates. (E.g., Catholic priest abuse received heavy media coverage post-2002; teacher-of-children abuse rates per the U.S. Department of Education's Hofer 2004 report run several times higher per-capita but receive vastly less coverage; the availability-driven popular sense that priests are uniquely abusive is empirically baseless.)
Personal anecdote over external data. Argument leans heavily on the speaker's personal experience or the experience of a small known circle. Detection test: how representative is the speaker's experience relative to the broader population? (E.g., the speaker's personal experience of an answered prayer is one data point among the millions of prayer-events; the speaker's personal experience of a friend who deconverted from Christianity is one data point among the population of deconversions and reconversions.)

Atheist deployment against Christian apologetics

The availability heuristic operates in atheist apologetic discourse in multiple deployment patterns:

"Religion causes violence, look at the Crusades, the Inquisition, Salem witch trials, ISIS, the Northern Ireland Troubles, the Sri Lankan civil war." The general claim: Christianity (and religion generally) causes more violence than secular alternatives. The availability deployment: the listed cases are the iconic-vivid-recent examples of religious violence; the cited speaker's audience can name them but typically cannot name the comparator base of secular-and-atheist violence (the 100M+ deaths under Stalin / Mao / Pol Pot / Kim Il-sung; the French Revolution Reign of Terror; the Albanian hyperatheist regime; the Khmer Rouge's anti-religious purge). The "religion causes violence" intuition is generated by availability of religious-violence examples, not by base-rate comparison. Anchor refutation in the codex: Religion Causes Violence Objection + Religion Causes Violence Objection; Pinker The Better Angels of Our Nature (2011); Karen Armstrong Fields of Blood (2014); William Cavanaugh The Myth of Religious Violence (Oxford 2009).
"Catholic priests are pedophiles." The general claim: the Catholic priest-abuse scandal demonstrates a systemic Catholic moral failure unique to Catholic clergy. The availability deployment: the John Jay Report (2004) on Catholic abuse received heavy media coverage post-2002 (Boston Globe Spotlight investigation; subsequent national reporting). Less covered: the Hofer 2004 U.S. Department of Education report on teacher-of-children abuse rates (per-capita rates several times higher than the Catholic priest rate during the same comparison window); the comparable base rates in other religious-clerical professions (Protestant pastors; Jewish rabbis; Hindu and Buddhist clerics, the pattern appears across professions involving access to children and trusted-adult positioning, not uniquely in Catholic priests). The Catholic-priest-abuse intuition is availability-biased; the underlying problem (access + trust + power asymmetry + institutional cover-up dynamics) is not Catholic-specific. The Catholic Church's actual reform measures (the 2002 Dallas Charter; 2010 reforms; ~95% reduction in new-allegations rate post-reform per CARA studies) are typically not part of the available case-set.
"Christianity is anti-science, look at Galileo and the Scopes trial." The general claim: Christianity systematically opposes scientific progress. The availability deployment: Galileo (the iconic 17th-century case) and the Scopes trial (the iconic 1925 American case) stand in for the entire history of Christianity-and-science. Less available: the historical record showing that the institutional foundations of modern Western science are largely Christian (Lindberg + Numbers God and Nature 1986; Hannam God's Philosophers 2009 / The Genesis of Science 2011); that the dominant figures of the Scientific Revolution (Copernicus + Kepler + Galileo + Newton + Boyle + Pascal + Faraday + Maxwell + Mendel + Lemaître) were practicing Christians whose religious commitments often motivated their scientific work; that the warfare-thesis (John William Draper 1874; Andrew Dickson White 1896) has been comprehensively dismantled by modern history of science (Lindberg, Numbers, Russell, Brooke, Harrison). The "Christianity vs science" intuition is generated by availability of two iconic cases, not by base-rate engagement with Christian-scientific history. Anchor refutation in the codex: Faith and Reason + Religion Causes Violence Objection + Lindberg + Hannam + Numbers historical material.
"Why don't more miraculous healings happen if God exists?" The general claim: if God answered prayers, we should see more documented healings. The availability deployment: televangelist healing-claim spectacle (Benny Hinn-style) makes Christian healing-claims appear to be the dominant pattern, generating "if those are obvious frauds, the whole category is fraud" availability inference. Less available: the rigorous Lourdes Bureau Médical 70-confirmed-cases out of ~7000+ claims rejection-rate as structural denominator-control (see Survivorship Bias for the reverse availability problem); the Habermas-Licona resurrection minimal-facts case; the Vatican Consulta Medica process; the peer-reviewed Brown et al. STEPP study (2010) and Byrd/Harris cardiac studies. The "religion makes wild healing claims that can't be substantiated" intuition is availability-biased toward the spectacle-end of the spectrum, not the rigorous-investigation end. The corpus's Miracles hub is structured precisely to engage this asymmetry, selecting for the rigorously-investigated end of the spectrum.
"All Muslims are violent extremists, look at 9/11 and ISIS." The general claim (when made by atheist-rhetorical or Islamophobic deployment): Islamic religion uniquely produces violent extremism. The availability deployment: 9/11 and ISIS are recent + vivid + heavily-covered + personally consequential for Western audiences; less available: the ~1.8 billion Muslims worldwide who have not engaged in violent extremism; the Sufi tradition; the moderate Sunni majority; the historical Muslim-Christian-Jewish coexistence in Andalusian Spain, Ottoman Constantinople, Mughal India; the comparable base rates of violent extremism in other religious + secular ideological movements. The Christian apologist should NOT deploy the availability bias against Muslims as if Muslim majority is implicated by ISIS, see the parallel in Christianity-and-violence: the apologist insists on base-rate engagement for Christianity and must extend the same epistemic charity to Islam. The substantive case for Christianity vs Islam is not made by availability-bias deployment but by the case in Religious Pluralism Objection Defeater + Liar Lunatic or Lord + Argument from the Resurrection. Polemical on the position; tender on the 1.8 billion persons.

Christian deployment against atheist arguments

The charge runs symmetrically, availability bias also operates in Christian apologetic discourse, and Christians should be vigilant against it in their own deployments:

"Look at all these conversion testimonies, Christianity is uniquely transformative." The Christian deployment: cite Lee Strobel + Antony Flew + C.S. Lewis + Augustine + Pascal as iconic conversion-testimonies. The availability problem: the available conversion-testimonies are biased toward published authors with platforms; the broader population includes deconversions (also heavily testimonied within the deconversion subculture, see Bart Ehrman, Dan Barker, Sarah Vowell). The Christian apologetic case for transformative power should engage the comparative base rate of life-trajectory change across worldview commitments, not rely on availability of cited celebrity-conversions. (The substantive case can survive base-rate engagement; see Resurrection-Centric Growth Argument on the explosive Jewish-monotheistic-context growth as the historical base-rate case. The availability-of-conversions case is the rhetorically weak version.)
"Look at Mother Teresa, John Paul II, Wilberforce, Bonhoeffer, Christianity produces saints." The Christian deployment: cite iconic Christian moral exemplars. The availability problem: the iconic-saint examples are availability-biased toward published-celebrated cases; the broader population of professed Christians includes the moral failures the atheist polemicist availability-biases toward. The substantive case is the comparative-base-rate case (Tom Holland Dominion 2019 makes this empirically, the modern Western moral framework that condemns Christian-historical moral failures is itself a Christian inheritance), not the iconic-cases-stand-in-for-category move.
"Christianity built the great universities + hospitals + civil-rights movements." The Christian deployment: cite iconic Christian institutional legacies (Oxford, Cambridge, Harvard's Christian foundation; the Hospitaller orders; Wilberforce + abolition; King + civil rights). The availability problem: the iconic institutional cases are availability-biased toward Western-Christian celebrated cases; the broader case requires engaging the comparative-cultural base rate (which institutional structures other religious + secular frameworks built). The Holland / Mangalwadi / Sanneh substantive case (in the religious-pluralism cluster) is the base-rate version; the iconic-institutions move is the availability-biased version.
"Look at the spectacular miracles, Lourdes Bureau approved cases, Eucharistic miracles, Padre Pio." The Christian deployment: cite the most-spectacular miracles in the corpus. The availability problem: the spectacular cases are availability-biased toward dramatic-case-set; the Christian apologetic case for miracles should engage the systematic-vetting structure (Bureau Médical 99%+ rejection rate; Vatican Consulta Medica calibration; Habermas-Licona minimal-facts methodology) rather than relying on availability of spectacular cases. The corpus's Miracles hub-and-spoke is explicitly designed to engage this asymmetry, selecting for evidentially-rigorous cases at three calibrated tiers, not for availability of dramatic claims.
"Look at Stalin + Mao + Pol Pot + Kim, atheism kills." The Christian deployment (against Religion Causes Violence Objection): cite the 100M+ deaths under 20th-c. atheist regimes. The availability problem: while the substantive case is correct (the 20th-c. atheist-regime mortality empirically dwarfs historical religious-war mortality, per Atheist Regime Body Count in the corpus), deploying it via four iconic atheist-regime exemplars without engaging the broader comparative case (other atheist regimes that did NOT produce mass killing, secular-democratic atheist demographics; the Northern European secular-Christian heritage countries) is itself availability-biased. The substantive case requires statistical engagement (mortality-per-regime-year normalized; ideology-specific mortality), not iconic cases.

How to rebut the availability bias when used against you

Three load-bearing rebuttal patterns:

Force base-rate engagement. The standard rebuttal pattern for any availability-driven argument is to ask "what is the base rate?" Force the opponent to specify the population from which their cited cases are drawn, and to compare frequency of cited cases vs frequency of the contrast set. Example: "You cited the Crusades as an example of religious violence, what is the per-capita-per-year mortality of the Crusades vs the per-capita-per-year mortality of (a) 20th-c. secular ideological wars + revolutions, (b) pre-Christian classical-era warfare, (c) post-Christianized Western civilization peace-period? Until we have the comparative base rate, we have no inference about religion-and-violence."
Demand the missing cases. Where the opponent cites N iconic vivid cases of category C, ask "what about cases where the predicted outcome did NOT occur?" Example: "You cite Galileo and Scopes as religion-vs-science cases. What about Copernicus + Kepler + Newton + Boyle + Pascal + Faraday + Maxwell + Mendel + Lemaître, Christian scientists whose religious commitments motivated their scientific work? The omission of these cases is not random; it's availability-biased toward the conflict narrative."
Diagnose the cognitive structure. Once base-rate engagement is forced, name the cognitive structure to make the diagnosis explicit: "The availability heuristic is operating here. The cases you cite are recent / vivid / media-covered / iconic. The cases you don't cite are equally informative for the question we're trying to answer. Tversky and Kahneman 1973 documented this pattern empirically. Once we de-bias by engaging the full case set, the inferential pattern reverses." This shifts the conversation from rhetorical-point-scoring to substantive epistemic engagement, which is where the Christian apologetic case is strongest.

False-fallacy examples (when the availability charge is misapplied)

The availability charge is misapplied in the following pattern types:

Charge: "You're committing the availability heuristic by citing peer-reviewed studies on prayer-effects (Brown et al. STEPP 2010; Byrd 1988; Harris 1999)." Why the charge fails: peer-reviewed statistical studies are precisely the correction to availability bias, not an instance of it. Citing controlled studies engages base rates and methodologically-vetted populations; the charge confuses the diagnosis (anecdote-driven inference) with the corrective (statistical engagement). When an apologist cites a peer-reviewed study, the substantive question is whether the study is methodologically sound, not whether citing it is availability-biased.
Charge: "You're committing the availability heuristic by citing the resurrection of Christ, that's just one case." Why the charge fails: the resurrection's evidential structure is not "one easily-recalled case among many that I'm overweighting"; it's "the unique falsifiable historical claim Christianity stakes itself on (1 Cor 15:14), supported by multiple independent attestation lines (Gospels + pre-Pauline creeds + extra-biblical sources + transformation of skeptics + explosive Jewish-monotheist-context growth)." The case does not stand on availability of similar cases; it stands on its own evidential structure. The criterion of inclusion in the resurrection-evidential case is methodological (multiple attestation, criterion of embarrassment, criterion of dissimilarity, etc.), not availability. See Argument from the Resurrection.
Charge: "You're committing the availability heuristic by citing the Lourdes Bureau cases, those are exactly the cherry-picked-availability cases." Why the charge fails: the Lourdes Bureau methodology is anti-availability by design. The Bureau requires (a) a specified population (every claim received by the Bureau Médical), (b) a documented denominator (~7000+ cases reviewed since 1883), (c) explicit five-criteria vetting (instantaneous + complete + medically inexplicable + persistent + physician-documented), and (d) ~99% rejection rate. The Bureau publishes its rejection criteria and rejected-case patterns as well as approved cases, the methodology is the corrective to availability bias, not an instance of it. When a critic dismisses Lourdes-Bureau-approved cases as availability-biased, they have failed to engage the methodological structure that distinguishes the Bureau-approved cases from the rejected claims.
Charge: "You're committing the availability heuristic by deploying multiple converted-skeptics testimonies (Lewis, Strobel, Flew, Wallace, McGrath), those are the only ones you remember." Why the charge fails: the deployment of converted-skeptics testimonies is structurally anti-availability when used responsibly: the deployment is "credentialed-skeptic-investigators who initially had every motive to debunk the case engaged the evidence rigorously and concluded otherwise." This is the criterion-of-embarrassment-equivalent for conversion testimonies, the conversion happened against the witness's prior availability bias. The cases are cited not because they are easy to recall but because they exemplify the methodological structure of skeptic-engaging-evidence-and-converting. When the charge is misapplied, it confuses (a) "I cite these because they came easily to mind" (availability-biased) with (b) "I cite these because their methodological structure makes them evidentially weighty" (methodologically-grounded).

When the availability charge IS legitimate (against Christian apologetic deployments)

The availability charge does stick when the Christian deployment matches the diagnostic markers:

A Christian apologist cites Lee Strobel's conversion-testimony as if it counts as evidence beyond its own narrative weight, when in fact the case is anecdotal and is one data point among many possible life-trajectory cases (this is availability-biased). The substantive case for Christianity does NOT rest on Strobel's conversion; it rests on the historical-evidential case for the resurrection.
A Christian apologist cites a small number of dramatic miracle-cases (Padre Pio + Lourdes Bureau approvals) without engaging the systematic-vetting structure or comparing to the rejected-claims rate. This is availability-biased, the corrective is to cite the methodological structure of vetting, not just the spectacular outcomes.
A Christian apologist cites Mother Teresa as proof of Christianity's moral fruit without acknowledging that some Christians have made grave moral failures and that the comparative-base-rate question (does Christian commitment systematically produce more moral fruit than secular commitment?) requires statistical engagement. Holland Dominion makes this engagement; the iconic-saint deployment alone does not.
A Christian apologist deploys "look at Stalin + Mao + Pol Pot + Kim, atheism kills" without engaging the broader comparative-base-rate question (atheist-regime mortality vs religious-war mortality vs secular-democratic mortality vs non-ideological warfare mortality). The substantive case in Atheist Regime Body Count handles the base-rate question; deploying iconic-cases alone falls into the availability-biased version of the argument.

The principle: when the Christian deployment is methodologically-grounded (engages base rates + comparators + statistical evidence + methodological vetting structures), the availability charge cannot stick. When the Christian deployment relies on iconic-cases-stand-in-for-category, the charge sticks. Avoid the latter; deploy the former.

Christian scholarly resources

Daniel Kahneman, Thinking, Fast and Slow (Farrar, Straus and Giroux, 2011), the canonical popular treatment of System 1 / System 2 cognition; chapters 12-13 on the availability heuristic. Kahneman won the 2002 Nobel Prize in Economic Sciences for this work (with Tversky deceased and not eligible). Essential reading for engagement with cognitive-bias-driven apologetic challenges.
Amos Tversky & Daniel Kahneman, "Availability: A Heuristic for Judging Frequency and Probability" in Cognitive Psychology 5:207-232 (1973). The original 1973 paper. Republished in Judgment Under Uncertainty: Heuristics and Biases (Kahneman, Slovic, Tversky eds., Cambridge University Press, 1982).
Steven Pinker, The Better Angels of Our Nature: Why Violence Has Declined (Viking, 2011), large-scale statistical engagement with the violence-and-history question; dismantles the availability-driven narrative that secular modernity is more violent than religious past. Christianity-and-violence section engages comparative base rates.
Karen Armstrong, Fields of Blood: Religion and the History of Violence (Knopf, 2014), religion-historian's engagement with the religion-causes-violence claim; argues for the dominant role of state and political violence relative to religious violence in human history.
William Cavanaugh, The Myth of Religious Violence (Oxford University Press, 2009), definitive academic critique of the "religion is uniquely violent" thesis; documents the modernist-state-formation interest in classifying violence as religious.
Tom Holland, Dominion: How the Christian Revolution Remade the World (Basic Books, 2019), historical-survey of Christianity's foundational role in modern Western moral and institutional frameworks; engages availability-biased "Christianity is regressive" narrative with comparative-historical engagement.
David Bentley Hart, Atheist Delusions: The Christian Revolution and Its Fashionable Enemies (Yale University Press, 2009), engages New Atheist availability-biased deployments against Christianity with sustained historical-comparative case.
Rodney Stark, For the Glory of God: How Monotheism Led to Reformations, Science, Witch-Hunts, and the End of Slavery (Princeton University Press, 2003) + Stark's broader sociology-of-religion work, engages Christianity-and-witch-hunts + Christianity-and-science availability narratives with sociological-historical data.
Thomas Gilovich, How We Know What Isn't So: The Fallibility of Human Reason in Everyday Life (Free Press, 1991), accessible cognitive-psychology treatment with substantial chapters on availability + confirmation + survivorship biases.
Edward Feser, The Last Superstition: A Refutation of the New Atheism (St. Augustine's Press, 2008), engages New Atheist availability-biased iconic-cases (Galileo, Crusades, Inquisition) with comparative philosophical-historical case.
John Lennox, Gunning for God: Why the New Atheists Are Missing the Target (Lion, 2011), popular-academic engagement with New Atheist informal-logic moves including availability-biased deployments.