Two populations behind one nonreading label
Mexico's 2025 reading survey distinguishes people who read only on social media from people who report neither social-media reading nor the other materials covered by the survey. Among literate adults with known ages, these two groups give different profiles of their main reason for not reading the surveyed materials. The distinction is useful precisely because the broader label “nonreader of MOLEC materials” can otherwise conceal substantial differences in people's circumstances and reading practices.
In the social-media-only group, 38.1% named lack of interest, motivation or enjoyment as their main reason. In the group reporting neither category, 24.3% selected that answer. Health problems showed the opposite descriptive pattern: 8.8% in the social-media-only group and 30.7% in the neither group. These are weighted estimates within each group, not percentages of all adults in Mexico, and they do not demonstrate that social-media reading changes people's interests or health.
Age composition is central to understanding the comparison. Adults aged sixty and over account for 19.7% of the social-media-only group but 56.5% of the neither group. Consequently, a comparison of reasons is also a comparison between populations at different stages of life. This article presents that composition alongside the complete reason distributions rather than trying to attribute the differences to a single behaviour. The analysis describes people and their reported circumstances; it does not rank their reading or diagnose their motivation.
A new national survey, not a continuation of the old headline
The Instituto Nacional de Estadística y Geografía, or INEGI, collected MOLEC 2025 interviews from June 2 to June 27, 2025. The redesigned survey covers people aged twelve and over nationally, including urban and rural settings. Earlier MOLEC headline figures referred to a different coverage framework, and the 2025 instrument added important verification questions and a separate social-media reading screen. A change between headline percentages from the two frameworks would therefore mix changes in measurement and coverage with any change in behaviour.
The population here is deliberately narrower than the published national twelve-plus headline: literate adults aged eighteen or older with a known age, living in the survey's occupied private-dwelling population. The analysis does not include children, residents outside that household coverage, or people who failed the literacy filter. It is not a statement about the share of every Mexican resident who encounters written language. It also does not provide a national literacy-rate analysis.
The public selected-person file contains 4,240 records, of which 4,038 pass the literacy filter. Of those, 3,609 have known ages that place them in the adult population. Five selected-person records have the special unknown-age code, including three literate respondents; they are not silently classified as ninety-eight-year-olds or as adults. A separate code identifying people aged ninety-seven and over is included in the adult population. These coding distinctions matter because an apparently straightforward numerical age filter could otherwise misclassify unknown observations.
How the mutually exclusive groups were constructed
The groups come from individual survey responses, not from subtracting rounded percentages in a press release. For books, magazines, newspapers and comics or manga, the survey first asks a direct reading question. A respondent who answers no receives a verification question that explicitly allows less complete reading or clarifies the material. A yes at either stage counts as reading that material. A blank verification response after a direct yes is a legitimate questionnaire skip, not evidence that the person did not read.
A respondent enters the no-MOLEC-material domain only after negative direct and verification answers for all four types, together with a negative answer about internet pages, forums or blogs excluding social media. The social-media question then divides that domain into two nonoverlapping groups. An affirmative social-media response identifies social-media-only reading; a negative response identifies neither category. The social-media question includes reading for information, communication or entertainment, with examples such as WhatsApp, Facebook and X.
The resulting adult samples contain 533 social-media-only respondents and 317 respondents in the neither group. Their selected-person weights represent approximately 12.60 million and 7.98 million people respectively. These totals remain estimates under the survey's population definitions. They should not be interpreted as platform audience counts, users of a particular application, or people who never read in any context. Someone can encounter other kinds of written material that the instrument does not enumerate, and the survey's category boundaries should remain visible whenever the results are summarized.
One main reason, not a checklist of every barrier
Question P37 asks for the main reason the respondent does not read books, magazines, newspapers or comics. It accepts one response. The questionnaire routes people who read only on social media to this question, as well as people who report no surveyed reading category. The actual file agrees with the reading-screen classification: all 890 eligible respondents across the full age range have a populated reason, and no ineligible respondent has one. Restricting to known-age adults leaves 850 respondents.
Because this is a single-main-reason question, each person's answer represents a prioritization rather than an inventory. Someone who cites school or work might also experience tiredness, limited money or difficulty finding interesting material. Someone who reports a health problem might also lack time. Those additional possibilities are not measured by this answer. A smaller percentage for a reason does not establish that the underlying difficulty is absent from the rest of the population.
The public descriptor includes an explicit “does not know” response even though the displayed paper questionnaire lists the substantive categories. That unknown response remains in each group's denominator and appears separately in the table. It accounts for 3.4% of the social-media-only group and 7.0% of the neither group. No separate refusal code was observed, and structural blanks outside the question's route are not recoded as either a refusal or a substantive reason. Including unknown responses prevents the answer shares from being silently inflated by selective exclusion.
The complete reason profiles
Lack of interest, motivation or enjoyment is the largest estimated category for social-media-only adults, at 38.1%. School or work time pressure follows at 24.5%, and household chores or caring responsibilities account for 13.5%. Preference for other activities is reported by 9.2%, health problems by 8.8%, lack of money by 2.3%, and another reason by 0.4%. The remaining 3.4% is the explicit unknown category. Small rounding differences may prevent the displayed one-decimal figures from adding to exactly one hundred.
Among adults reporting neither surveyed reading category, health problems account for 30.7%, followed by lack of interest, motivation or enjoyment at 24.3%. School or work time pressure represents 17.6%, household chores or caring 9.1%, preference for other activities 6.4%, lack of money 2.7%, and another reason 2.2%. Unknown responses account for 7.0%. The two distributions thus differ in their overall shape, not just in whether time or interest has the larger share.
The least frequent answers require particular restraint. For example, the social-media-only “other reason” estimate is based on three respondents, while lack of money is based on ten. Publishing the complete distribution makes these categories visible, but does not make them precise estimates of narrowly defined experiences. No free-text responses are reproduced, and no individual explanation is inferred from the aggregate. The table supports an overview of the answer profile, not a reliable ranking of every small difference between groups.
A preselected time-pressure contrast
Before comparing the reason estimates across groups, the analysis specified one focal contrast: the weighted share choosing school or work time pressure, social-media-only minus neither. The estimates are 24.4908% and 17.5578%, a difference of 6.9330 percentage points. In ordinary language, school or work was chosen as the main reason by about seven percentage points more of the first group than the second. This is an unadjusted descriptive contrast, not an estimated effect of social-media reading.
School and work time pressure remains distinct from time spent on household chores or caring. Combining those categories after inspecting the results would change the preselected question and could obscure different circumstances. Likewise, the difference is expressed in percentage points rather than as a percentage increase. It compares shares with different group denominators, and neither group's sample is a matched counterpart to the other.
The point contrast was reproduced independently from the validated reason-question route, separately from the original reading-indicator construction. The full set of sixteen reason cells was also checked using that alternate selection. This provides a useful calculation audit, but agreement between calculations does not solve every sampling issue. In particular, the public design variables contain a singleton stratum that requires an explicit variance convention. The main text therefore does not use a significance label or suggest that independent arithmetic verification turns a descriptive difference into a causal or definitive population finding.
Age composition is part of the result
The social-media-only group is distributed across ages eighteen to thirty-nine, forty to fifty-nine, and sixty-plus at 34.9%, 45.3% and 19.7%. The corresponding shares in the neither group are 10.3%, 33.2% and 56.5%. Older adults make up a majority of the second group but about one fifth of the first. These differences are not a minor footnote: they materially change the kinds of circumstances represented in each reason distribution.
The survey does not tell us, from these aggregates alone, what the school/work contrast would be if the groups had the same age structure, health, educational background or labour-market circumstances. Nor does the prominence of the health response establish a particular diagnosis, disability type or accessibility requirement. It is a self-reported main reason, collected within a specific questionnaire route. The analysis does not claim that age explains all differences or that an age adjustment would leave none.
No adjusted model was predeclared for this study, so none is added to manufacture a behaviour-specific interpretation. The estimand is intentionally composition-inclusive: it describes the populations actually represented by the two groups. A future standardized analysis would need a separately justified target population, a clear covariate plan and checks that the data support its comparisons. It would still be an observational analysis, not an experiment establishing whether social-media reading causes more or less engagement with books or other materials.
What the weights and validation establish
The calculations use FAC_ELE, the expansion factor for the selected respondent, rather than a dwelling or household weight. Each reason share is the sum of those weights for the response divided by their sum for the group, with the explicit unknown category retained. Positive weights are present for every selected-person record. Raw sample sizes are reported separately from weighted population estimates so readers can distinguish the number of interviews from the number of people those interviews represent.
As a national validation among literate respondents of all surveyed ages (twelve and over), the same classification produces 79.0858% reading MOLEC materials, 13.0561% reading only on social media and 7.8581% reporting neither. Rounded to one decimal, these reproduce INEGI's published 79.1%, 13.1% and 7.9%. This check is not used to manufacture the adult estimates. It shows that the record-level classification aligns with the official headline before the independently defined adult restriction is applied.
The design variables identify 1,201 primary sampling units across ninety-nine strata. Estimates for adult subpopulations must preserve that full sampled structure, including sampled units whose respondents contribute zero to the target domain. Treating the 850 eligible adults as a simple random sample would ignore clustering, stratification and unequal selection. A weight-only effective sample-size calculation, if inspected as a diagnostic, would not repair that problem or substitute for design-based variance estimation. For that reason, no simple-random-sampling standard errors are presented here.
Secondary appendix: uncertainty under explicit conventions
INEGI's design documentation describes an ultimate-cluster variance calculation. One public design stratum, however, contains only one observed primary sampling unit. Its four respondents are outside the target domain. Their observed contributions to the linearized difference are zero, but that fact does not show that the unit was sampled with certainty or that the population stratum has no uncertainty. The inspected producer documentation does not supply a singleton-handling rule, validated collapsed variance strata or replicate weights that resolve this issue.
Two standard noncertainty conventions were specified before inspecting the existing interval calculations, after point estimates and diagnostics had already been seen; this is an amendment, not an original preregistration. Under grand-mean centering, the singleton contribution is mechanically zero here because the joint ratio-difference influences sum to zero. This “adjust” convention is therefore ineffective as a safeguard for this particular contrast; it must not be described as proof of certainty. An alternative replaces the singleton contribution with the average contribution across the ninety-eight nonsingleton strata. That “average” convention makes an unverified assumption that their variance contributions provide a reasonable guide for the singleton.
For transparency, the normal-approximation ninety-five-percent sensitivity intervals are 0.14 to 13.73 percentage points under grand-mean centering and 0.10 to 13.76 under average-stratum imputation. They are conditional calculations, not unconditional or officially prescribed confidence intervals. The marginally positive lower limits do not justify a significance headline, and no p-value is supplied. Python calculations reproduced the formulas documented for the R survey package; an R survey fit was not executed. The secondary appendix records the effect of explicit conventions without pretending that the unresolved design information has become known.
Reading the evidence without overextending it
The reference periods are not uniform across materials. Books refer to the preceding twelve months, magazines to three months, and newspapers to the preceding week. The screening for comics or manga, internet pages or blogs, and social-media reading does not specify an equivalent common retrospective period. Consequently, “neither” is shorthand for the survey-defined classification, not evidence that someone read no text throughout an entire year. The new verification questions also make it inappropriate to equate the 2025 classifications with older headline categories without a dedicated comparability analysis.
The complete profiles offer a more careful starting point for discussion than a single label for nonreading. Time, interest, health and care responsibilities appear in different proportions across populations whose ages also differ substantially. They suggest questions that an author, educator or reading initiative might investigate, but they do not measure demand for a particular genre, the impact of an intervention, the quality of social-media content or the effectiveness of a platform. No outcome for a proposed reading programme can be calculated from these responses alone.
The latest project information checked on 30 October 2026 still identified the 2025 release, with a next publication scheduled for 18 November 2026. These are therefore observations from June 2025, not a measurement of current 2026 behaviour. INEGI produced the source data; this independent selection, calculation and explanation is by Alisson Bento. The downloadable materials contain aggregate results, methods and code, while individual records remain local to the analysis. Attribution does not imply INEGI endorsement, and neither the calculations nor the written interpretation should be mistaken for an official institutional position.
Main reasons for not reading surveyed materials
Mexico · MOLEC 2025 · Literate adults aged 18+ with known age. Weighted within-group share (%). Social-media-only: n = 533; Neither MOLEC materials nor social media: n = 317 (Unweighted sample).
Full chart and source
Scroll horizontally to inspect the chart. ↔
Chart and values
Scroll horizontally to see all columns.
| Main reason | Social-media-only (%) | Neither (%) | Social-media-only respondents | Neither respondents |
|---|---|---|---|---|
| Lack of interest, motivation or enjoyment | 38.1 | 24.3 | 216 | 90 |
| Preference for other activities | 9.2 | 6.4 | 55 | 25 |
| No time because of school or work | 24.5 | 17.6 | 131 | 48 |
| No time because of household chores or caring | 13.5 | 9.1 | 65 | 26 |
| Lack of money | 2.3 | 2.7 | 10 | 9 |
| Health problems | 8.8 | 30.7 | 42 | 95 |
| Other reason | 0.4 | 2.2 | 3 | 8 |
| Does not know | 3.4 | 7.0 | 11 | 16 |
Literate adults aged 18+ with known age; FAC_ELE weighted percentages. Social-media-only n = 533; neither n = 317. One main answer; unknown retained. Small response cells are imprecise. Neither means neither MOLEC materials nor social media under the survey screens. No unconditional confidence intervals are asserted.
Age composition of the two reading groups
Mexico · MOLEC 2025 · Literate adults aged 18+ with known age. Weighted within-group share (%). Social-media-only: n = 533; Neither MOLEC materials nor social media: n = 317 (Unweighted sample).
Full chart and source
Scroll horizontally to inspect the chart. ↔
Chart and values
Scroll horizontally to see all columns.
| Age | Social-media-only (%) | Neither (%) |
|---|---|---|
| 18–39 | 34.9 | 10.3 |
| 40–59 | 45.3 | 33.2 |
| 60+ | 19.7 | 56.5 |
Weighted within-group percentages among literate adults with known ages; EDAD = 98 excluded, EDAD = 97 included as 97+. Composition is descriptive, not adjustment.
Data & downloads
Scroll horizontally to see all columns.
| Main reason | Social-media-only (%) | Neither (%) | Social-media-only respondents | Neither respondents |
|---|---|---|---|---|
| Lack of interest, motivation or enjoyment | 38.1 | 24.3 | 216 | 90 |
| Preference for other activities | 9.2 | 6.4 | 55 | 25 |
| No time because of school or work | 24.5 | 17.6 | 131 | 48 |
| No time because of household chores or caring | 13.5 | 9.1 | 65 | 26 |
| Lack of money | 2.3 | 2.7 | 10 | 9 |
| Health problems | 8.8 | 30.7 | 42 | 95 |
| Other reason | 0.4 | 2.2 | 3 | 8 |
| Does not know | 3.4 | 7.0 | 11 | 16 |
Literate adults aged 18+ with known age; FAC_ELE weighted percentages. Social-media-only n = 533; neither n = 317. One main answer; unknown retained. Small response cells are imprecise. Neither means neither MOLEC materials nor social media under the survey screens. No unconditional confidence intervals are asserted.
Scroll horizontally to see all columns.
| Age | Social-media-only (%) | Neither (%) |
|---|---|---|
| 18–39 | 34.9 | 10.3 |
| 40–59 | 45.3 | 33.2 |
| 60+ | 19.7 | 56.5 |
Weighted within-group percentages among literate adults with known ages; EDAD = 98 excluded, EDAD = 97 included as 97+. Composition is descriptive, not adjustment.
Evidence & method
How we calculated it
- Public INEGI MOLEC 2025 selected-person microdata; national occupied private-dwelling population, literate known-age adults aged 18+.
- Record-level mutually exclusive classification uses direct and verification reading responses; social-media split P2F. P37 is one main reason, unknown code 9 retained.
- FAC_ELE weighted ratios; focal school/work difference chosen before cross-group estimates. Full profiles and age composition descriptive, no adjustment or causal interpretation.
- All 16 reason cells and weighted denominators independently verified; national 12+ group shares reproduce official rounded 79.1/13.1/7.9.
- Full sampled strata/PSU structure retained for conditional secondary Taylor sensitivity. One unresolved singleton prevents unconditional official-design inference; Python formulas, not an executed R fit.
What the data cannot tell us
- Five selected-person ages are unknown, including three literate respondents; known-age adult estimand excludes them. Literacy and private-dwelling coverage constrain generalization.
- Main reasons are single-response priorities, not all barriers. Small answer cells can be imprecise; unknown is retained.
- Age composition differs strongly; no independent social-media effect, causal claim, genre demand or intervention benefit is estimated.
- Material reference windows differ; redesigned national 2025 categories cannot be spliced into earlier urban adult headline trends.
- Conditional singleton sensitivities rely on explicit conventions; average-stratum exchangeability is unverified, zero influence is not certainty, and barely positive interval endpoints are not a significance headline.
Original sources
- INEGI MOLEC 2025 project
INEGI MOLEC 2025 project: Project, national redesign and distribution discovery.
Accessed October 30, 2026
- INEGI MOLEC 2025 public CSV microdata
INEGI MOLEC 2025 public CSV microdata: TMOLEC local-only analysis; selected-person observations.
Accessed October 30, 2026
- INEGI MOLEC 2025 file descriptor
INEGI MOLEC 2025 file descriptor: Codes, age 98 unknown, age 97 topcode, FAC_ELE/UPM_DIS/EST_DIS.
Accessed October 30, 2026
- INEGI MOLEC 2025 questionnaire
INEGI MOLEC 2025 questionnaire: Printed pages 10, 14, 16: literacy, screening, social-only routing and single main reason.
Accessed October 30, 2026
- INEGI MOLEC 2025 sample design
INEGI MOLEC 2025 sample design: Selected-person weighting and ultimate-cluster variance, equations 17–20. No singleton resolution identified.
Accessed October 30, 2026
- INEGI MOLEC 2025 results
INEGI MOLEC 2025 results: All-age national rounded group shares used only as validation; released 18 November 2025.
Accessed October 30, 2026
- INEGI MOLEC 2025 RNM metadata
INEGI MOLEC 2025 RNM metadata: Public-use access, attribution, design and universe.
Accessed October 30, 2026
- INEGI free-use terms
INEGI free-use terms: Own licence: attribution and nonmisleading reuse, no implied official endorsement.
Accessed October 30, 2026
- Survey package author documentation on singleton PSUs
Survey package author documentation on singleton PSUs: Methodological diagnostic only; ad hoc adjustment is not INEGI approval.
Accessed October 30, 2026
Cite & reuse this study
Alisson Bento. Different reasons behind nonreading: social-media-only and neither groups in Mexico. Independent adult-domain analysis of INEGI, Módulo sobre Lectura (MOLEC) 2025. Sources accessed 30 October 2026. https://alissonbento.com/research/mexico-reading-barriers-social-only-2025/
Source: INEGI. Módulo sobre Lectura (MOLEC) 2025. Bases de datos. Reused under INEGI’s own free-use terms, https://www.inegi.org.mx/inegi/terminos.html. Calculations and explanation: Alisson Bento. Independent analysis; no official endorsement or logos. Only derived aggregates and code are distributed.
Corrections or editorial questions