Where the online purchasing differences appear
Brazil's 2024 TIC Domicílios estimates show a larger urban–rural percentage-point difference in participation in online shopping than in purchases of books, magazines or newspapers among people who already shop online. The first difference is 18.5415 percentage points; the second is 5.1329 points. These are descriptive comparisons between two indicators with different denominators. They do not establish that one obstacle matters more than another, or that removing an obstacle would produce a particular increase in reading purchases. The value of the comparison is that it separates participation in a sales channel from the composition of purchases made through it.
For publishing, those distinctions change the question. A person who does not buy anything online and a person who buys online but reports no books, magazines or newspapers both fall outside the category's online purchasers. Their observed purchasing situations differ. The survey supports describing those situations at group level, but it does not identify the reasons behind them. This study compares the official estimates for urban and rural areas, household-income bands and broad regions, retains the source's separate sampling margins, and avoids converting the differences into a claim about individual reading interest. The measures concern purchases during a twelve-month reference period, not lifetime participation or a purchasing journey followed over time.
The population begins with Internet users aged ten and older
The survey's target population is people aged ten or older living in permanent private households in Brazil. Its coverage includes urban and rural areas. That definition excludes an interpretation in terms of adults alone, all residents regardless of housing circumstances, or households as purchasing units. The individual is the unit represented by these indicators. Household characteristics such as income and location classify that individual; they do not turn the published percentage into the fraction of households buying a product. This matters particularly when several residents share an income or make purchases for other household members.
Internet users are people who used the network at least once in the three months before interview. H2 therefore does not use everyone aged ten or older as its denominator. People outside the recent-user definition are outside both indicators examined here. The twelve-month purchase reference period and three-month Internet-use filter perform different functions and should not be collapsed into a single time window. Fieldwork took place from March through August 2024, so the purchase questions refer to the preceding twelve months for each respondent, rather than a common January-to-December transaction ledger. The observations describe the population represented by the survey in that edition, not every online sale occurring within the calendar year.
Two indicators, two nested denominators
H2 asks whether an Internet user bought or ordered products or services online in the previous twelve months, even if payment was not made online. Its affirmative share includes online orders paid through another channel. The denominator is all Internet users within the relevant subgroup, not only users who answered yes or no after discarding other response categories. Nationally, the published affirmative estimate is 46.1460%, with a 95% sampling margin of 2.4170 percentage points. This is participation in online purchasing broadly defined, not participation specifically in book retail or digital publishing.
Published indicator H3 starts with the users who reported an online purchase or order. It asks which product categories they bought, and the selected category combines books, magazines and newspapers. Nationally, that estimate is 21.1243%, with a margin of 3.9378 points. It would be incorrect to report that figure as the share of all Brazilian Internet users buying reading materials online. There is also a numbering trap in the documentation: the published H3 category comes from questionnaire H5, item I. Questionnaire H3 itself concerns reasons for not purchasing. Checking the wording and routing, rather than matching question numbers mechanically, establishes that the category question is asked only of H2 purchasers. Multiple product-category answers are possible, so category shares need not sum to one hundred.
The urban–rural contrast is larger for shopping participation
Among urban Internet users, online purchasing participation is estimated at 48.1888%, with a margin of 2.7592 percentage points. The corresponding rural estimate is 29.6473%, with a margin of 5.5525 points. Subtracting rural from urban gives the descriptive H2 difference of 18.5415 points. These figures retain the same recent-Internet-user definition within each location group. They do not compare all urban residents with all rural residents, and they do not adjust for the groups' different distributions of age, household income, education or other characteristics. Those compositional differences are part of the observed comparison.
Among online purchasers, the urban books/magazines/newspapers estimate is 21.4876%, with a margin of 4.0967 points; the rural estimate is 16.3547%, with a margin of 7.1709 points. The resulting H3 difference is 5.1329 points. Its smaller numerical size is the central descriptive finding, but the two differences are not estimates of successive losses from a common fixed population. The rural purchasers are a selected subset of rural Internet users, just as urban purchasers are a subset of urban users. In addition, the rural category estimate has a wide margin. Reporting that uncertainty alongside the point estimate is essential. No test here establishes whether the difference between the two urban–rural gaps is statistically distinguishable from zero.
Household income shows a substantial participation gradient
For Internet users in households with monthly income up to one minimum wage, H2 participation is 23.0957%, with a margin of 3.2029 points. It is 40.5338% ± 4.6044 points for more than one to two minimum wages, 55.7145% ± 5.9047 for more than two to three, and 58.7409% ± 6.8339 for more than three to five. The two higher bands have estimates of 71.1905% ± 8.5896 for more than five to ten minimum wages and 80.8063% ± 12.0574 for more than ten. The point estimates increase across these six reported positive-income bands, although the margins become wider at the upper end.
The contrast between the highest band and up to one minimum wage is 57.7106 percentage points. Choosing the more-than-five-to-ten band instead gives 48.0948 points. Both comparisons describe estimated online-shopping participation within recent Internet users; neither estimates the causal effect of additional household income. The bands use total monthly income from household residents, not personal earnings, disposable income or income adjusted for household size. People in households of different sizes can therefore share the same income classification while having different resources per person. No-income, unknown-income and nonresponse groups are retained separately in the full table. They are not folded into the lowest positive-income band or imputed as zero earnings.
Reading-category purchases do not increase steadily across every income band
Conditional on online purchasing, the books/magazines/newspapers estimate is 9.4046% ± 3.1843 points in the up-to-one-minimum-wage group. It is 15.7935% ± 6.2847 for more than one to two minimum wages, 20.6989% ± 9.8090 for more than two to three, and 30.5067% ± 10.7519 for more than three to five. The estimates in the two higher bands are 29.5509% ± 11.3651 and 29.1111% ± 16.4446. Unlike the H2 point estimates, this sequence does not increase at every step. Its upper-band differences are small relative to their separate published margins, but that observation is not a substitute for a formal comparison.
Highest minus lowest gives a descriptive conditional-category difference of 19.7065 percentage points. Comparing more than five to ten minimum wages with the lowest band gives 20.1463 points. Both are numerically smaller than their corresponding H2 contrasts. This does not establish a universal relationship between income and literary purchasing, or show that the richest group has lower demand than the adjacent groups. The highest-income H3 margin alone is 16.4446 points, and the category combines several different products. Presenting a precise league table would exaggerate what these estimates resolve. A responsible summary emphasizes the broad contrast between lower-income purchasers and the upper bands while retaining uncertainty and avoiding a claim that every successive increase in income changes purchasing probability.
Regional estimates add context without becoming a ranking
The broad-region H2 estimates are 53.8735% ± 5.7078 points in the Southeast, 34.3257% ± 3.1615 in the Northeast, 55.4991% ± 5.1765 in the South, 35.5374% ± 4.0568 in the North and 36.1552% ± 4.5584 in the Center-West. These estimates add geographical detail to the national and urban–rural results. A region contains both urban and rural residents, as well as different income and demographic compositions. The regional breakdown is therefore not an independent explanation for the area or income differences. The tables do not provide a joint adjustment that would disentangle those overlapping classifications.
For books, magazines or newspapers among online purchasers, the corresponding regional estimates are 22.2645% ± 7.1411 points in the Southeast, 19.9132% ± 4.5550 in the Northeast, 21.8440% ± 5.8863 in the South, 14.5987% ± 5.4505 in the North and 20.2591% ± 6.6908 in the Center-West. Their point estimates occupy a narrower range than the H2 regional estimates, but this article does not infer an ordering of underlying population rates. A publishing strategy based on a ranking of these conditional estimates would need more evidence. The survey offers broad descriptive context, not estimates for particular cities, delivery routes, retailers or publishers. Keeping all five regions visible is more informative than selecting only the most contrasting pair.
Purchasing reading-related products is not the same as reading
The H3 category is explicitly books, magazines or newspapers. No split among those products is available in the selected indicator. The results cannot identify whether the income or location pattern arises mainly from book buying, periodical buying, or their combination. They also do not distinguish print and digital formats, fiction and nonfiction, novels and other books, or any particular genre. Calling the measure ebook adoption or romance demand would replace the observed outcome with an unmeasured one. The category's exact wording belongs in tables, charts and short summaries, not only in a methodological note at the end.
Purchase participation also differs from expenditure, quantities and reading activity. Someone purchasing one eligible item and someone purchasing many can both contribute one affirmative response to this category. A purchase could be intended for another person; the indicator does not show that the buyer read the material. Reading borrowed books, using library collections, reading existing possessions or accessing freely available material can occur without an eligible purchase. Conversely, a person reporting no reading-category online purchase may buy the same products offline. For these reasons the study does not classify nonpurchasers as nonreaders or as uninterested in literature. Its contribution is a channel-specific description of purchasing participation, complementary to but distinct from studies of reading frequency or the concentration of household expenditure.
Why 2024 remains the matched reference year
A newer survey edition does not automatically provide a newer observation for every indicator. The 2025 TIC Domicílios index includes H2 online-purchasing participation, but not the H3 product-category table used here. More importantly, the official 2025 collection report explains that module rotation produced a reduced electronic-commerce module containing only the question about purchasing products or services online. The 2025 questionnaire confirms that design: its commerce module contains H2 without the product-category question. The vintage decision therefore rests on affirmative documentation about collection, not simply on failing to locate a webpage.
For the paired question, 2024 is the latest verified edition with both measures. Combining 2025 H2 with 2024 H3 would join participation in one period to category composition in another, without establishing that the latter remained unchanged. That would not answer a same-edition comparison and could obscure the module's rotation. This study consequently holds both indicators at 2024 and makes no year-to-year change claim. The source files were checked on October 30, 2026, which is the retrieval date rather than the period of respondents' purchases. Retaining that distinction allows a reader to understand why a recent analysis uses an earlier observation year and what evidence would be required for an eventual matched update.
Sampling margins come from the survey design
TIC Domicílios uses a stratified, three-stage cluster probability sample: selection of primary sampling units constructed from census sectors, selection of households and selection of residents. The methodological report describes 88 strata. Weights reflect selection probabilities, nonresponse adjustments and calibration to population estimates from PNAD Contínua. The individual calibration may consider sex, age, area, survey strata, activity status and education. These features mean that the published estimates cannot be treated as simple unweighted percentages from an independent random sample of equal-probability respondents. The weighting and clustering are part of the evidence, not optional refinements added by this analysis.
For individual estimates, the producer uses replication with 200 replicate weights respecting the sample design. Published sampling margins correspond to a 95% confidence level and are calculated as 1.96 times the estimated standard error. This article retains those supplied margins, in percentage points, for each displayed percentage. The collection report records interviews in 23,856 households across 634 municipalities, including 21,170 households with TIC Domicílios individual interviews and the remainder with TIC Kids Online Brasil interviews. None of those totals is the unweighted base for a particular H2 or H3 subgroup. Using the overall interview count to recreate a rural or high-income category margin would ignore both routing and survey design. The shared collection does not change the age-ten-and-older universe of the indicators examined here.
What the calculations deliberately leave unresolved
The arithmetic contrasts subtract one subgroup's published percentage from another within an indicator. The analysis retains the source values at four decimal places for verification, while a shorter narrative can display one decimal. Separate margins are attached to the original estimates, not to a subtraction, ratio or difference between gaps. Estimating uncertainty for those derived quantities requires the relevant joint sampling information, including covariance. The tables alone do not supply that information. No significance test is therefore claimed for the urban–rural or income contrasts, and comparing whether separate intervals overlap is not used as a replacement test.
No joint purchase share is created by multiplying H2 and H3. Their routing establishes nested conceptual populations, but a publishable derived estimate with uncertainty would still require careful verification of compatible weighting, numerators and joint design information. That extra quantity is unnecessary to the descriptive question answered here. Nor are the two indicators interpreted as a causal decomposition into access barriers and preferences. Observed purchaser selection can change the composition of the conditional population. Delivery conditions, payment options, prices and tastes may be relevant subjects for other studies, but these two tables do not identify their separate effects. The narrower conclusion is useful precisely because it remains tied to measured group differences rather than claiming to explain how those differences arose.
A reusable record of estimates, definitions and limits
The accompanying aggregate table contains all 34 selected estimates: national total, urban and rural areas, five regions and nine income-response categories for each of the two indicators. Every estimate retains its source sampling margin and workbook location. The descriptive-contrasts file records the urban–rural comparison and the two specified higher-income-versus-lowest-band comparisons without adding inferential statistics. The reproduction script rebuilds those calculations from the published aggregate extract and checks the frozen source archive's hash when original-source extraction is requested. Original workbook packages remain available from the producer; they are not republished as this study's downloads. No respondent-level records are included.
The methodological report explains that a dash represents item nonresponse and that a rounded zero must not automatically be read as no responses. The selected cells are numeric, but the absence of an explicit small-cell suppression rule in the reviewed material is not evidence that every subgroup estimate is precise. Unknown-income and nonresponse groups have particularly wide margins and should remain visible rather than be silently discarded. Dataset-specific metadata license the table package under CC BY 4.0, distinct from the website footer's CC BY-SA declaration. Reuse should credit NIC.br, link the applicable license and identify the selection and calculations as modifications. The resulting record supports a limited, auditable conclusion: the observed group contrasts in online-shopping participation are numerically larger than the corresponding conditional reading-category contrasts, without establishing causation or statistical significance for that comparison.
Online purchasing in urban and rural Brazil
TIC Domicílios 2024 · People aged 10+ in permanent private households. Estimate and source 95% sampling margin.
Full chart and source
Scroll horizontally to inspect the chart. ↔
Chart and values
Scroll horizontally to see all columns.
| Indicator | Area | Estimate (%) | 95% margin (pp) | Denominator |
|---|---|---|---|---|
| H2 | Urban | 48.1888 | 2.7592 | Base: recent Internet users |
| H2 | Rural | 29.6473 | 5.5525 | Base: recent Internet users |
| H3 | Urban | 21.4876 | 4.0967 | Base: recent Internet users who purchased/ordered online |
| H3 | Rural | 16.3547 | 7.1709 | Base: recent Internet users who purchased/ordered online |
Purchase recall: previous 12 months. Internet use: previous 3 months. Different denominators. Margins do not test differences between groups or panels. 2024 is the latest verified matched pair; the 2025 module contains H2 only. Source: NIC.br / Cetic.br, TIC Domicílios 2024, tables H2 and H3 (v1.1). Alisson Bento.
Data & downloads
Scroll horizontally to see all columns.
| Group | Estimate (%) | 95% margin (pp) | Proportion source cell | Margin source cell |
|---|---|---|---|---|
| Total | 46.1460 | 2.4170 | H2!C5 | H2!C5 |
| Urban | 48.1888 | 2.7592 | H2!C6 | H2!C6 |
| Rural | 29.6473 | 5.5525 | H2!C7 | H2!C7 |
| Southeast | 53.8735 | 5.7078 | H2!C8 | H2!C8 |
| Northeast | 34.3257 | 3.1615 | H2!C9 | H2!C9 |
| South | 55.4991 | 5.1765 | H2!C10 | H2!C10 |
| North | 35.5374 | 4.0568 | H2!C11 | H2!C11 |
| Center-West | 36.1552 | 4.5584 | H2!C12 | H2!C12 |
| Up to 1 minimum wage | 23.0957 | 3.2029 | H2!C31 | H2!C31 |
| More than 1 to 2 minimum wages | 40.5338 | 4.6044 | H2!C32 | H2!C32 |
| More than 2 to 3 minimum wages | 55.7145 | 5.9047 | H2!C33 | H2!C33 |
| More than 3 to 5 minimum wages | 58.7409 | 6.8339 | H2!C34 | H2!C34 |
| More than 5 to 10 minimum wages | 71.1905 | 8.5896 | H2!C35 | H2!C35 |
| More than 10 minimum wages | 80.8063 | 12.0574 | H2!C36 | H2!C36 |
| No income | 27.3527 | 27.9793 | H2!C37 | H2!C37 |
| Does not know income | 39.8944 | 10.4564 | H2!C38 | H2!C38 |
| No income response | 62.8992 | 20.5397 | H2!C39 | H2!C39 |
Denominator: recent Internet users aged 10+ in permanent private households. Source: NIC.br, TIC Domicílios 2024, version 1.1. Margins are separate 95% sampling margins, not errors for contrasts. Income uses total monthly household income.
Scroll horizontally to see all columns.
| Group | Estimate (%) | 95% margin (pp) | Proportion source cell | Margin source cell |
|---|---|---|---|---|
| Total | 21.1243 | 3.9378 | H3!I5 | H3!I5 |
| Urban | 21.4876 | 4.0967 | H3!I6 | H3!I6 |
| Rural | 16.3547 | 7.1709 | H3!I7 | H3!I7 |
| Southeast | 22.2645 | 7.1411 | H3!I8 | H3!I8 |
| Northeast | 19.9132 | 4.5550 | H3!I9 | H3!I9 |
| South | 21.8440 | 5.8863 | H3!I10 | H3!I10 |
| North | 14.5987 | 5.4505 | H3!I11 | H3!I11 |
| Center-West | 20.2591 | 6.6908 | H3!I12 | H3!I12 |
| Up to 1 minimum wage | 9.4046 | 3.1843 | H3!I31 | H3!I31 |
| More than 1 to 2 minimum wages | 15.7935 | 6.2847 | H3!I32 | H3!I32 |
| More than 2 to 3 minimum wages | 20.6989 | 9.8090 | H3!I33 | H3!I33 |
| More than 3 to 5 minimum wages | 30.5067 | 10.7519 | H3!I34 | H3!I34 |
| More than 5 to 10 minimum wages | 29.5509 | 11.3651 | H3!I35 | H3!I35 |
| More than 10 minimum wages | 29.1111 | 16.4446 | H3!I36 | H3!I36 |
| No income | 5.1986 | 19.2496 | H3!I37 | H3!I37 |
| Does not know income | 22.9832 | 15.1899 | H3!I38 | H3!I38 |
| No income response | 14.7046 | 19.2604 | H3!I39 | H3!I39 |
Different denominator: recent Internet users aged 10+ who bought or ordered online in the previous twelve months. Source: NIC.br, TIC Domicílios 2024, version 1.1. Margins are separate 95% sampling margins, not errors for contrasts. Income uses total monthly household income.
Scroll horizontally to see all columns.
| Indicator | Area | Estimate (%) | 95% margin (pp) | Denominator |
|---|---|---|---|---|
| H2 | Urban | 48.1888 | 2.7592 | Base: recent Internet users |
| H2 | Rural | 29.6473 | 5.5525 | Base: recent Internet users |
| H3 | Urban | 21.4876 | 4.0967 | Base: recent Internet users who purchased/ordered online |
| H3 | Rural | 16.3547 | 7.1709 | Base: recent Internet users who purchased/ordered online |
Purchase recall: previous 12 months. Internet use: previous 3 months. Different denominators. Margins do not test differences between groups or panels. 2024 is the latest verified matched pair; the 2025 module contains H2 only. Source: NIC.br / Cetic.br, TIC Domicílios 2024, tables H2 and H3 (v1.1). Alisson Bento.
Evidence & method
How we calculated it
- Secondary descriptive analysis of official NIC.br version 1.1 H2/H3 proportions and sampling margins. Age 10+ permanent-private-household population; recent Internet use in preceding three months.
- Published H3 maps to questionnaire H5 item I, restricted to H2 purchasers. Twelve-month purchasing includes online ordering even when paid offline.
- All 34 selected aggregate proportions and separate 95% margins retained; differences use unrounded source table values. No covariance-free inference or multiplication of nested percentages.
- 2024 is the latest verified matched pair: 2025 rotating commerce module contains participation only.
- Stratified three-stage probability survey with calibrated weights and 200-replicate individual uncertainty; margins equal 1.96 times design-based standard errors.
What the data cannot tell us
- Different nested denominators; conditional purchasers differ compositionally from all Internet users. No causal or significance claim for the comparison of gaps.
- Category combines books, magazines and newspapers; no format, genre, expenditure, reading-frequency or offline-purchase inference.
- Household income bands are not personal or equivalized income; subgroup composition and uncertainty vary.
- Highest-income and residual-income categories have wide margins; sample subgroup counts are not available in the selected tables. No ranking or invented suppression threshold.
- Respondent-relative twelve-month purchase periods during March–August 2024 fieldwork; not a calendar-year sales census.
Original sources
- TIC Domicílios 2024 individual tables and dataset license
TIC Domicílios 2024 individual tables and dataset license: Version 1.1 proportions, totals and margins. Dataset explicitly licensed CC BY 4.0.
Accessed October 30, 2026
- H2: online purchasing among Internet users
H2: online purchasing among Internet users: Recent Internet users aged 10+; purchases or orders in previous twelve months.
Accessed October 30, 2026
- H3: product categories among online purchasers
H3: product categories among online purchasers: Books, magazines OR newspapers; denominator restricted to online purchasers.
Accessed October 30, 2026
- 2024 methodological report
2024 methodological report: Private-household population, three-stage sample, calibration, 200 replicates and 95% margins; printed pages 31–41.
Accessed October 30, 2026
- 2024 questionnaire
2024 questionnaire: H2 purchase routing; published H3 corresponds to questionnaire H5 item I, not questionnaire H3.
Accessed October 30, 2026
- 2024 collection report
2024 collection report: March–August fieldwork and survey collection counts, not subgroup bases.
Accessed October 30, 2026
- 2025 collection report
2025 collection report: Reduced commerce module contains only the online-purchase question; matched H3 absent.
Accessed October 30, 2026
- 2025 questionnaire
2025 questionnaire: Commerce module has H2 only; confirms no matched category update.
Accessed October 30, 2026
- Creative Commons Attribution 4.0
Creative Commons Attribution 4.0: Applicable license linked by dataset metadata; attribution, license link and modifications notice required.
Accessed October 30, 2026
Cite & reuse this study
Alisson Bento. Online shopping and reading-category purchases in Brazil: two different income and rural gaps. Original secondary analysis of NIC.br/Cetic.br TIC Domicílios 2024. Sources retrieved October 30, 2026. https://alissonbento.com/research/brazil-online-shopping-reading-purchases/
Source: Núcleo de Informação e Coordenação do Ponto BR, TIC Domicílios 2024 individual tables, version 1.1, CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Original selection, calculations and explanation: Alisson Bento. Modifications: selected indicators and groups, retained source margins, calculated descriptive differences. No endorsement implied. Dataset-specific CC BY differs from website-footer CC BY-SA; see rights download.
Corrections or editorial questions