Skip to content

US written drama and live theatre · 2022 SPPA

Reading Plays and Attending Theatre Are Different Audiences

Play reading among nonmusical theatre attendees is 10.66%; nonmusical attendance among play readers is 20.41%. The numerator is the same, but the audiences differ. These annual 2022 observations do not identify the same play, a sequence or a causal effect.

Published
Share
Share this page
EmailCite & reuse this study
Key findings as text
Reusable imagesReading Plays and Attending Theatre Are Different Audiences — 0.49%: Both activities on the common complete-case base. 39 of 6,372 unweighted respondents; weighted share 0.4940%, with approximate 95% replicate interval 0.2450%–0.7431%. Annual recall before July 2022.Download sharing image (JPG) ↓

Key findings

0.49%

Both activities on the common complete-case base

39 of 6,372 unweighted respondents; weighted share 0.4940%, with approximate 95% replicate interval 0.2450%–0.7431%. Annual recall before July 2022.

10.66% / 20.41%

Play reading among attendees / attendance among play readers

Nonmusical stage attendance. Approximate 95% CIs 5.68%–15.65% and 11.49%–29.34%, respectively; unweighted conditioning bases 336 and 158.

10.61%–11.12%

Missing-counterpart range for reading among all stage-positive respondents

339 stage-positive respondents include 3 missing reading counterparts. Extreme assignment bounds are not sampling confidence intervals.

What the same respondents reported

Reading a play and watching a performance are two ways of encountering dramatic writing, but a survey question about one cannot stand in for a question about the other. In the same 6,372 complete-case respondents to the 2022 U.S. Survey of Public Participation in the Arts, the weighted share reporting both play reading and attendance at a nonmusical stage play was about 0.49%. The share reporting play reading without that attendance was 1.93%, while the share reporting attendance without play reading was 4.14%.

The overlap looks different when the denominator changes. Among respondents reporting nonmusical theatre attendance and answering the play-reading question, 10.66% reported reading a play. Among respondents reporting play reading and answering the attendance question, 20.41% reported attending a nonmusical play. These estimates describe overlapping annual activities, not whether someone read and saw the same work.

This reference makes that distinction usable through a downloadable joint table, explicitly named conditional percentages, replicate-weight uncertainty and missing-response bounds. The observations concern the twelve months before July 2022 interviews. They are neither current 2026 attendance rates nor evidence that theatre displaces reading. A theatre programmer, literary journalist or reading researcher can use the table to choose the appropriate measure before describing an audience.

What the two questions actually ask

The source is the July 2022 Current Population Survey arts supplement, sponsored by the National Endowment for the Arts and administered by the U.S. Census Bureau. Its target population was civilian, noninstitutionalized household adults aged eighteen or older. One eligible person was randomly selected in the participating households, and the supplement assigned respondents one of two question cores and two of four modules.

Both primary questions used here belong to Core 1. The play-reading item asks whether the respondent read any plays during the last twelve months. Its instructions include reading in books, magazines or newspapers, in paper form or online. It therefore does not measure the purchase of a play script, the completion of a whole play, book reading alone, reading for a particular purpose, or the number of different playwrights encountered.

The attendance item asks about going to a live performance of a nonmusical stage play during the last twelve months, excluding elementary or high-school performances. It does not ask which play was performed, whether a script was read before or after attendance, whether a ticket was bought, or whether attendance occurred in the same month as any reading.

These differences are central to interpretation. A person may read dramatic writing without attending a performance. Someone attending a play may encounter that work entirely through actors, staging and sound. A reader may also attend a different play, so even a yes answer to both items does not identify a page-to-stage pathway for one title. The table records co-participation by a person within an annual recall period, rather than an adaptation funnel.

Unlike the separate book-reading item, the play-reading item does not repeat an exclusion for work or school. It must not be labelled an exclusively leisure-reading measure.

The fieldwork took place from 16 to 25 July 2022. The recall period therefore includes approximately July 2021 to July 2022, when pandemic conditions could still affect live events. The analysis does not adjust attendance to a hypothetical unrestricted year. It preserves the observation period instead of treating an unusual historical setting as ordinary current behaviour.

Start with the joint table, rather than separate headlines

The primary complete-case base contains respondents aged eighteen or older with a positive arts-supplement weight and an explicit yes or no response to both questions. There are 6,372 such records. The play-reading answer is yes for 158 of them, the nonmusical-attendance answer is yes for 336, and both answers are yes for 39. These are unweighted sample counts, supplied to make the evidence behind the estimates visible.

The weighted joint table has four mutually exclusive cells. About 93.44% report neither activity, 1.93% report play reading only, 4.14% report nonmusical theatre attendance only, and 0.49% report both. The four estimates sum to 100% on this one common complete-case base. No row is an estimate of the number of performances, scripts, books, tickets or visits.

The two discordant cells together account for about 6.07% of the common base. That is the share reporting exactly one of the two activities. It is not a count of people who rejected the other activity, or proof of an exclusive cultural identity. A no answer applies to the specified twelve-month question; it does not establish a lifetime preference.

Separate marginal percentages cannot supply this joint table. Knowing the overall play-reading rate and the overall attendance rate does not reveal how many people belong to both groups. The overlap has to be observed in linked answers from the same respondents. Multiplying the two marginal rates would impose independence, which is neither assumed nor established here.

The distinction matters even when two published rates look superficially similar. A pair of small percentages might represent almost the same people, largely different groups, or many intermediate arrangements. The data needed to distinguish those possibilities are the joint observations, not just two bars copied from separate charts.

Why 10.66% and 20.41% can both be correct

The numerator for both conditional estimates is the weighted joint group reporting both activities. One denominator contains respondents who attended nonmusical theatre; the other contains respondents who read plays. Because those audiences have different sizes, their conditional percentages differ even though the numerator is unchanged.

Among the 336 unweighted attendance-positive respondents in the common base, the weighted estimate of play reading is 10.66%. Its approximate 95% replicate-weight interval runs from 5.68% to 15.65%. Among the 158 unweighted play-reading respondents in that base, the weighted estimate of nonmusical theatre attendance is 20.41%, with an interval from 11.49% to 29.34%.

Those intervals are deliberately prominent. Only 39 unweighted respondents contribute to the joint yes group. Reporting several decimal places would not make that small observed intersection more precise. Two decimal places help connect the article to its reproducible calculations; the uncertainty, not the displayed digits, sets the appropriate level of confidence in interpretation.

The difference between the inverse conditional estimates is 9.75 percentage points, with an approximate paired replicate-weight interval of 4.53 to 14.98 points. The difference is computed within each replicate, so the covariance between the two estimates is retained. It is not calculated by pretending that the page and stage audiences are independent samples.

Even so, this difference is an asymmetry between two conditional descriptions, not an effect of reading or attendance. It does not say that reading a play makes attendance twice as likely, that theatres recruit a particular proportion of readers, or that one audience is more committed. Statements about causes, conversion or motivation would require different questions and a different research design.

How uncertainty was calculated

The calculation uses the arts-supplement weight and the 160 published Census replicate weights. Records are joined by the documented household questionnaire and person occurrence keys, with a one-to-one merge required. Every retained respondent must have all the replicate weights. The supplied full-sample replicate weight is checked against the main supplement weight before estimation.

The official technical appendix describes balanced repeated replication with a Fay coefficient of 0.5. For each percentage, the same numerator and denominator definition is recalculated under all 160 replicate weights. The variance is four divided by 160 times the sum of squared differences between replicate estimates and the full-sample estimate. The displayed intervals use the estimate plus or minus 1.96 standard errors.

Core 1 requires an expansion factor of two when producing population totals. That common factor cancels from every ratio used here, both in the main estimate and in each replicate. This study consequently reports percentages and unweighted sample counts, not a newly estimated number of U.S. adults in each audience.

These intervals address sampling variability represented by the replicate design. They do not include every source of error. Recall errors, question interpretation, proxy interviewing, nonresponse, the edited nature of the released data and unusual conditions during the recall period remain relevant. A design-based standard error cannot turn two differently worded activities into an identical concept.

The analysis also avoids treating an interval as a forecast range for a particular theatre or reading group. The national household survey is not a sample of that organisation’s customers. A local programme could have a much larger or smaller overlap. Its own audience research would be needed to estimate that relationship directly.

Missing answers are a separate problem

The complete-case table uses explicit yes and no answers. Refusals, unknown answers and other nonresponse codes are not silently converted into no. That rule prevents an apparently simple graph from hiding a decision about people whose second answer is unavailable.

For each conditional percentage, the download package supplies an additional sensitivity calculation. Start with everyone who gives a yes answer to the conditioning question, including those whose answer to the other question is missing. The lower bound assigns all those missing other answers to no; the upper bound assigns them to yes. The result is a range compatible with the observed answers under those two extreme assignments.

A yes answer to either Core 1 item already establishes that the respondent received that question core. The conditional bounds therefore do not require guessing the assignment of respondents whose entire core is unobserved. They condition on an actually recorded yes answer and vary only the missing counterpart. This is a narrower and auditable calculation.

These bounds are not sampling confidence intervals and do not estimate the probability that missing respondents participated. They show how far the displayed conditional result could move if every unresolved counterpart went one way or the other, holding the observed conditioning group and survey weights fixed. They should be read alongside the complete-case intervals rather than combined into a single unexplained error bar.

The primary attendance-positive group contains 339 unweighted respondents, including three with a missing play-reading answer. Assigning those three counterparts to no or yes gives weighted play-reading bounds of 10.61% to 11.12%, compared with the 10.66% complete-case estimate. The play-reading-positive group has no missing attendance counterparts, so its corresponding bound remains 20.41%. That narrow missing-counterpart sensitivity does not remove sampling uncertainty or solve nonresponse elsewhere in the survey.

More elaborate missing-data models might use additional assumptions about respondents and their answers. None is fitted here. Preserving explicit unknowns is more informative than creating a precise-looking answer whose dependence on an imputation model remains invisible.

A secondary check using musical theatre

The declared secondary comparison replaces nonmusical attendance with the survey’s separate live musical-stage-play question. The play-reading definition remains unchanged. The complete-case base is slightly different, so the musical and nonmusical joint tables are displayed separately rather than added together.

For the musical comparison, 6,374 respondents have valid answers to both items. The play-reading-positive group contains 158 unweighted respondents, the attendance-positive group contains 719, and their intersection contains 50. The weighted joint share is about 0.71% of that common base. Musical attendance is reported by about 10.39%, while play reading is reported by about 2.42%.

The conditional estimate of musical attendance among play readers is 29.40%, with an approximate 95% interval of 19.92% to 38.88%. The inverse estimate of play reading among musical attendees is 6.85%, with an interval of 4.30% to 9.40%. These values again use the same joint numerator with different audience denominators.

This secondary result does not identify a better correspondence between a script and a performance. The reading question does not separate musical scripts from nonmusical plays, and the attendance questions do not name titles. A reader could answer yes after reading one kind of play and attending another kind. The secondary table demonstrates the consequence of changing the performance category, not the conversion rate of musical-theatre readers.

Neither table represents all live theatre. The two attendance groups can overlap, and combining their marginal percentages would double-count respondents who attended both kinds of performance. A future analysis of a combined attendance measure would need its own respondent-level union and explicit interpretation. It is not inferred from the sums of the displayed estimates.

What this adds to existing arts research

The association between literary participation and engagement with other arts is established territory. The National Endowment for the Arts has long published reading and attendance statistics, and its 2022 comprehensive report describes the broader survey, participation rates, demographic patterns and historical context. International cultural surveys also publish interrelated participation indicators. This study does not claim to discover that cultural audiences can overlap.

There is also direct precedent for presenting the two directions of a cultural audience comparison. The NEA's 2004 report Reading at Risk, Table 3, relates literary reading to broad performing-arts attendance and reports inverse audience shares. Reversing a conditional denominator is therefore not a new analytical invention here. The present contribution is the narrower, explicitly worded 2022 play-reading and nonmusical-theatre case, with paired replicate uncertainty and missing-counterpart bounds.

Its narrower contribution is a transparent page-and-stage case: a complete four-cell joint table, both directions of the conditional comparison, their paired replicate uncertainty, and a separate treatment of missing counterparts. The distinction between reading plays and attending a performance stays attached to every result. The downloadable intermediate tables make the calculation available to another researcher rather than requiring trust in a headline.

This question also differs from the site’s analysis of fiction among book readers. A play-reading item is not a measure of all books or fiction, while theatre attendance concerns a live performance. It differs from same-day time-use analysis as well: these are two annual recall questions, without a diary sequence, activity duration or evidence of simultaneous participation.

Those boundaries make the reference useful for a specific editorial decision. If the intended claim concerns access to written dramatic work, the reading item is relevant. If it concerns live cultural attendance, the performance item is relevant. If it concerns their overlap, linked answers are necessary. Choosing one of those questions first prevents a familiar artistic relationship from becoming an unsupported statistical substitution.

A practical way to reuse the tables

Begin by retaining the source year and target population. Then state whether the percentage is a joint share of the common complete-case base, reading among attendees, or attendance among readers. Include the unweighted conditioning count and uncertainty when quoting a conditional estimate. A short label such as “audience overlap” is insufficient unless the denominator is supplied next to it.

Do not describe the joint yes group as people who saw an adaptation of a play they had read. The survey identifies no common title, order of activities or motivation. Do not use the figures as a 2026 theatre-market estimate, a sales forecast, a literacy assessment or a causal argument about screens and books. The historical annual questions cannot answer those questions.

The public recipe supplies exact official input fingerprints, selected column definitions, the frozen analytical plan, aggregate outputs and reproduction code. The full person-level source files are obtained from Census through their published public-use routes rather than bundled with the article. No respondent identifiers or individual answer records appear in the public derived tables.

The strongest conclusion is also the most usable one: a page audience and a stage audience are separate measured groups. In this fixed 2022 case, their common numerator supports two substantially different conditional percentages. Keeping the wording, base and uncertainty visible turns that difference into a reliable reference instead of an apparent contradiction.

Play reading and nonmusical attendance overlap in one small group

Three positive joint cells shown on a 0–5% scale; the neither-activity cell (93.44%) is not plotted. Same 6,372 complete cases, not three separate audience samples. Approximate 95% CI. US civilian noninstitutionalized household adults 18+; annual recall before July 2022 interviews. Reading plays includes print/online; stage item excludes elementary/high-school shows. Not necessarily the same work or leisure-only reading.

Full chart and source
Three positive joint cells shown on a 0–5% scale; the neither-activity cell (93.44%) is not plotted. Same 6,372 complete cases, not three separate audience samples. Approximate 95% CI. US civilian noninstitutionalized household adults 18+; annual recall before July 2022 interviews. Reading plays includes print/online; stage item excludes elementary/high-school shows. Not necessarily the same work or leisure-only reading. Plotted weighted estimates with approximate 95% intervals: both activities 0.49% (0.24–0.74); play reading only 1.93% (1.49–2.36); nonmusical attendance only 4.14% (3.45–4.82).

Scroll horizontally to inspect the chart. ↔

Chart and values

Scroll horizontally to see all columns.

Complete declared estimates: primary nonmusical and secondary musical comparisons
Stage comparisonMeasure and denominatorUnweighted denominator nUnweighted numerator nWeighted estimateUnitStandard error (pp)95% CI lower95% CI upper
Nonmusical stage play · primaryNeither activity / complete cases6372591793.440559%0.44809192.56230194.318816
Nonmusical stage play · primaryPlay reading only / complete cases63721191.926126%0.2211611.4926502.359602
Nonmusical stage play · primaryStage attendance only / complete cases63722974.139283%0.3494313.4543984.824168
Nonmusical stage play · primaryBoth activities / complete cases6372390.494032%0.1270630.2449880.743075
Nonmusical stage play · primaryPlay reading / complete cases63721582.420158%0.2629081.9048582.935458
Nonmusical stage play · primaryStage attendance / complete cases63723364.633315%0.3805493.8874385.379192
Nonmusical stage play · primaryPlay reading among attendees3363910.662594%2.5424525.67938915.645800
Nonmusical stage play · primaryAttendance among play readers1583920.413197%4.55373511.48787629.338517
Nonmusical stage play · primaryExactly one activity / complete cases63724166.065410%0.4171405.2478156.883004
Nonmusical stage play · primaryAttendance among readers minus reading among attendees6372—9.750602percentage points2.6658224.52559114.975614
Musical stage play · secondaryNeither activity / complete cases6374554787.905321%0.51842886.88920388.921440
Musical stage play · secondaryPlay reading only / complete cases63741081.707687%0.2174271.2815302.133845
Musical stage play · secondaryStage attendance only / complete cases63746699.675783%0.4551768.78363710.567929
Musical stage play · secondaryBoth activities / complete cases6374500.711208%0.1397540.4372910.985125
Musical stage play · secondaryPlay reading / complete cases63741582.418896%0.2627621.9038832.933909
Musical stage play · secondaryStage attendance / complete cases637471910.386991%0.4691159.46752511.306457
Musical stage play · secondaryPlay reading among attendees719506.847105%1.3018024.2955739.398637
Musical stage play · secondaryAttendance among play readers1585029.402186%4.83709919.92147238.882901
Musical stage play · secondaryExactly one activity / complete cases637477711.383470%0.50363310.39635012.370591
Musical stage play · secondaryAttendance among readers minus reading among attendees6374—22.555081percentage points3.84984915.00937730.100785

All 20 estimates retained. Most values are percentages; the inverse-conditional difference is in percentage points. Unweighted sample counts do not reproduce weighted estimates. Musical/nonmusical groups overlap and must not be summed. For the inverse-conditional difference, n denotes the common complete-case analysis base, not a single ratio denominator; its two component conditioning bases differ.

Same complete-case base: n=6,372. Zoomed horizontal scale: 0–5%; neither activity = 93.44% (95% CI 92.56–94.32), not plotted. Nonmusical theatre. Points show weighted estimates; whiskers are approximate 95% replicate-weight CIs, not missing-response bounds. Source: Census/NEA 2022 SPPA. Original secondary analysis: Alisson Bento.

Download chart (SVG) ↓ · Download chart (PNG) ↓

One joint numerator, two different audience denominators

Same joint numerator with different conditioning denominators. 95% BRR/Fay 0.5 intervals; separate missing-counterpart bounds are not CIs. US civilian noninstitutionalized household adults 18+; annual recall before July 2022 interviews. Reading plays includes print/online; stage item excludes elementary/high-school shows. Not necessarily the same work or leisure-only reading.

Full chart and source
Same joint numerator with different conditioning denominators. 95% BRR/Fay 0.5 intervals; separate missing-counterpart bounds are not CIs. US civilian noninstitutionalized household adults 18+; annual recall before July 2022 interviews. Reading plays includes print/online; stage item excludes elementary/high-school shows. Not necessarily the same work or leisure-only reading. Nonmusical attendance among play readers 20.41% (95% CI 11.49–29.34; n=158); play reading among nonmusical attendees 10.66% (95% CI 5.68–15.65; n=336).

Scroll horizontally to inspect the chart. ↔

Chart and values

Scroll horizontally to see all columns.

Complete declared estimates: primary nonmusical and secondary musical comparisons
Stage comparisonMeasure and denominatorUnweighted denominator nUnweighted numerator nWeighted estimateUnitStandard error (pp)95% CI lower95% CI upper
Nonmusical stage play · primaryNeither activity / complete cases6372591793.440559%0.44809192.56230194.318816
Nonmusical stage play · primaryPlay reading only / complete cases63721191.926126%0.2211611.4926502.359602
Nonmusical stage play · primaryStage attendance only / complete cases63722974.139283%0.3494313.4543984.824168
Nonmusical stage play · primaryBoth activities / complete cases6372390.494032%0.1270630.2449880.743075
Nonmusical stage play · primaryPlay reading / complete cases63721582.420158%0.2629081.9048582.935458
Nonmusical stage play · primaryStage attendance / complete cases63723364.633315%0.3805493.8874385.379192
Nonmusical stage play · primaryPlay reading among attendees3363910.662594%2.5424525.67938915.645800
Nonmusical stage play · primaryAttendance among play readers1583920.413197%4.55373511.48787629.338517
Nonmusical stage play · primaryExactly one activity / complete cases63724166.065410%0.4171405.2478156.883004
Nonmusical stage play · primaryAttendance among readers minus reading among attendees6372—9.750602percentage points2.6658224.52559114.975614
Musical stage play · secondaryNeither activity / complete cases6374554787.905321%0.51842886.88920388.921440
Musical stage play · secondaryPlay reading only / complete cases63741081.707687%0.2174271.2815302.133845
Musical stage play · secondaryStage attendance only / complete cases63746699.675783%0.4551768.78363710.567929
Musical stage play · secondaryBoth activities / complete cases6374500.711208%0.1397540.4372910.985125
Musical stage play · secondaryPlay reading / complete cases63741582.418896%0.2627621.9038832.933909
Musical stage play · secondaryStage attendance / complete cases637471910.386991%0.4691159.46752511.306457
Musical stage play · secondaryPlay reading among attendees719506.847105%1.3018024.2955739.398637
Musical stage play · secondaryAttendance among play readers1585029.402186%4.83709919.92147238.882901
Musical stage play · secondaryExactly one activity / complete cases637477711.383470%0.50363310.39635012.370591
Musical stage play · secondaryAttendance among readers minus reading among attendees6374—22.555081percentage points3.84984915.00937730.100785

All 20 estimates retained. Most values are percentages; the inverse-conditional difference is in percentage points. Unweighted sample counts do not reproduce weighted estimates. Musical/nonmusical groups overlap and must not be summed. For the inverse-conditional difference, n denotes the common complete-case analysis base, not a single ratio denominator; its two component conditioning bases differ.

Nonmusical theatre. Points show weighted estimates; whiskers are approximate 95% replicate-weight CIs, not missing-response bounds. Source: Census/NEA 2022 SPPA. Original secondary analysis: Alisson Bento.

Download chart (SVG) ↓ · Download chart (PNG) ↓

Data & downloads

Scroll horizontally to see all columns.

Complete declared estimates: primary nonmusical and secondary musical comparisons
Stage comparisonMeasure and denominatorUnweighted denominator nUnweighted numerator nWeighted estimateUnitStandard error (pp)95% CI lower95% CI upper
Nonmusical stage play · primaryNeither activity / complete cases6372591793.440559%0.44809192.56230194.318816
Nonmusical stage play · primaryPlay reading only / complete cases63721191.926126%0.2211611.4926502.359602
Nonmusical stage play · primaryStage attendance only / complete cases63722974.139283%0.3494313.4543984.824168
Nonmusical stage play · primaryBoth activities / complete cases6372390.494032%0.1270630.2449880.743075
Nonmusical stage play · primaryPlay reading / complete cases63721582.420158%0.2629081.9048582.935458
Nonmusical stage play · primaryStage attendance / complete cases63723364.633315%0.3805493.8874385.379192
Nonmusical stage play · primaryPlay reading among attendees3363910.662594%2.5424525.67938915.645800
Nonmusical stage play · primaryAttendance among play readers1583920.413197%4.55373511.48787629.338517
Nonmusical stage play · primaryExactly one activity / complete cases63724166.065410%0.4171405.2478156.883004
Nonmusical stage play · primaryAttendance among readers minus reading among attendees6372—9.750602percentage points2.6658224.52559114.975614
Musical stage play · secondaryNeither activity / complete cases6374554787.905321%0.51842886.88920388.921440
Musical stage play · secondaryPlay reading only / complete cases63741081.707687%0.2174271.2815302.133845
Musical stage play · secondaryStage attendance only / complete cases63746699.675783%0.4551768.78363710.567929
Musical stage play · secondaryBoth activities / complete cases6374500.711208%0.1397540.4372910.985125
Musical stage play · secondaryPlay reading / complete cases63741582.418896%0.2627621.9038832.933909
Musical stage play · secondaryStage attendance / complete cases637471910.386991%0.4691159.46752511.306457
Musical stage play · secondaryPlay reading among attendees719506.847105%1.3018024.2955739.398637
Musical stage play · secondaryAttendance among play readers1585029.402186%4.83709919.92147238.882901
Musical stage play · secondaryExactly one activity / complete cases637477711.383470%0.50363310.39635012.370591
Musical stage play · secondaryAttendance among readers minus reading among attendees6374—22.555081percentage points3.84984915.00937730.100785

All 20 estimates retained. Most values are percentages; the inverse-conditional difference is in percentage points. Unweighted sample counts do not reproduce weighted estimates. Musical/nonmusical groups overlap and must not be summed. For the inverse-conditional difference, n denotes the common complete-case analysis base, not a single ratio denominator; its two component conditioning bases differ.

Scroll horizontally to see all columns.

Missing-counterpart extreme-assignment bounds
Stage comparisonObserved positive conditioning itemUnweighted denominator nMissing counterpart nWeighted missing-counterpart share (%)Lower bound (%)Upper bound (%)
Nonmusical stage play · primaryStage-attendance yes33930.51601810.60757311.123592
Nonmusical stage play · primaryPlay-reading yes15800.00000020.41319720.413197
Musical stage play · secondaryStage-attendance yes730111.5083446.7438278.252171
Musical stage play · secondaryPlay-reading yes15800.00000029.40218629.402186

Four ranges hold each observed conditioning-positive group and weights fixed; missing other-item answers are all assigned no versus yes. Not sampling confidence intervals.

Evidence & method

How we calculated it

  1. Official July 2022 CPS arts supplement and 160 Census replicate weights are exact-hash checked and merged 1:1 by documented questionnaire/person keys. The analysis uses positive-weight adults 18+ with explicit yes/no answers to both Core 1 items.
  2. Primary play-reading item PEC1Q15C and nonmusical attendance item PEC1Q6A share the same respondent base. A separate musical-attendance comparison uses PEC1Q5A and its own complete-case base; groups must not be added.
  3. Ratios and the inverse-conditional difference are recomputed under every replicate. Fay coefficient 0.5 yields variance 4/160 times the sum of squared replicate departures. Approximate 95% intervals use ±1.96 SE. Common Core 1 expansion cancels in ratios; no weighted population counts are reported.
  4. Four missing-counterpart ranges assign unresolved other-item responses to no versus yes within observed conditioning-positive groups. These identification ranges are distinct from sampling confidence intervals. No respondent identifiers or individual records are redistributed.

What the data cannot tell us

  • Interviews on 16–25 July 2022 ask about the preceding 12 months, approximately July 2021–July 2022. These are not current 2026 behaviour or an unrestricted post-pandemic counterfactual.
  • Play reading can occur in books, magazines, newspapers, paper or online. It is not book-reading-only or necessarily leisure-only; unlike the separate book question, this item does not repeat a work/school exclusion.
  • Nonmusical live-stage attendance excludes elementary/high-school performances. No common title, sequence, purchase, ticket, completion, motivation or adaptation funnel is identified.
  • Small joint samples and wide replicate intervals remain visible. Sampling intervals do not remove recall, nonresponse, question-interpretation, proxy, edited-data or pandemic-context uncertainties. Musical and nonmusical groups can overlap; no causal displacement or recruitment inference.

Original sources

  1. Census July 2022 CPS arts-supplement dictionary and official public-use data

    Official Census July 2022 arts-supplement public-use microdata. Questions PEC1Q15C/PEC1Q6A are both Core 1; positive-weight adults 18+ and explicit yes/no complete cases. Raw input obtained separately, not bundled.

    Accessed October 31, 2026

  2. Census July 2022 SPPA replicate-weight SAS verification source

    Official Census 160 replicate weights and SAS verification source. Full-sample weight and one-to-one key matching checked; scaling cancels within ratios.

    Accessed October 31, 2026

  3. NEA 2022 SPPA technical appendix

    NEA 2022 SPPA technical appendix: random question-core allocation, BRR Fay 0.5, 160 replicates and source error/context limits.

    Accessed October 31, 2026

  4. NEA 2022 SPPA comprehensive report

    NEA 2022 SPPA comprehensive report supplies source definitions and broader context; the exact new page/nonmusical pair with paired uncertainty and counterpart bounds is a bounded extension, not a first-ever claim.

    Accessed October 31, 2026

  5. NEA Reading at Risk, June 2004, Table 3, printed page 6

    NEA Reading at Risk (2004), Table 3, printed page 6, already demonstrates inverse broad literary/performance audience shares. Prior precedent is credited; no 2002-to-2022 trend inferred.

    Accessed October 31, 2026

Cite & reuse this study

Alisson Bento. Reading Plays and Attending Theatre Are Different Audiences. Original secondary analysis of the 2022 US Survey of Public Participation in the Arts, Census/NEA. https://alissonbento.com/research/us-play-reading-stage-audience-overlap-2022/

Reuse the original aggregate tables and figures with credit to Alisson Bento and Census/NEA. Retain 2022, the two exact activity definitions, denominators, unweighted bases, sampling intervals and separate missing-counterpart bounds. No source report images, logos or individual respondent records are bundled.

Corrections or editorial questions

Pass it on.

Reading Plays and Attending Theatre Are Different Audiences | Alisson Bentoalissonbento.com

WhatsAppFacebookXTelegramRedditThreadsEmailSMS