CoChat
Sign Up
Cited by 110Literacy and Educational PracticesReading and Literacy DevelopmentSchool Choice and Performance

The link between fiction and teenagers’ reading skills: International evidence from the OECD PISA study

Open access
Jerrim, John, Moss, Gemma·2019·British Educational Research Journal
DOI 10.1002/berj.3498
Open this paper in CoChat — free accountOpen PDF

Abstract

Abstract It is well known that children who read more tend to achieve higher scores in academic reading tests. Much less is known, however, about the link between reading different types of text and young people's reading performance. We investigate this issue using the Programme for International Student Assessment (PISA) 2009 database, exploring the association between the frequency with which teenagers read five different types of text (magazines, non‐fiction, fiction, newspapers and comics) and their PISA reading scores. Analysing data from more than 250,000 teenagers from across 35 industrialised countries, we find evidence of a sizeable ‘fiction effect’; young people who read this type of text frequently have significantly stronger reading skills than their peers who do not. In contrast, the same does not hold true for the four other text types. We therefore conclude that encouraging young people to read fiction may be particularly beneficial for their reading skills. Interventions encouraging fiction reading may be especially important for boys from disadvantaged socio‐economic backgrounds, who are less likely to read this text type.

Integrity signals

Registry

Crossref · 10.1002/berj.3498

Cites

29 of 29 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

4 of 15 cited sources checked · 1 supported · 0 flagged · accuracy not yet measured

  • Premisecitation
    § 1. Introduction

    There has been rather less research on how reading different text types is associated with young people's reading skills.

    “Yet there has been rather less research on how reading different text typesfor instance the frequency with which young people read fiction books compared to nonfiction books -is associated with young people's reading skills (McGeown et al, 2015;Sullivan, 2015).”
    Quote in the paperNot a substantial claim1 of 2 sources checked
    • cites McGeown et al, 2015 source
    • cites Sullivan, 2015 source

      The Introduction says there is extensive evidence linking reading frequency to reading attainment, but notes a shortage of compelling longitudinal evidence. The paper also examines books and newspapers, yet the passages found do not say that relatively little research compares how different text types, such as fiction and nonfiction, relate to young people’s reading skills (searched: text types, fiction, nonfiction, books, newspapers). (LLM, conf=0.72)

      Experimental — accuracy not yet measured
  • Premisecitation
    § 3. Methods

    Young people who are higher achievers tend to read more.

    “The motivation for including this variable is that it is well-known that young people who are higher achievers tend to read more (Sullivan and Brown 2015).”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Sullivan and Brown 2015 source
  • Findingown data
    § 1. Introduction

    For text types other than fiction, null effects are typically found.

    “Moreover, the same is not true for the reading of other text types, where null effects are typically found.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.
  • Findingown data
    § 1. Introduction

    The strong association between reading fiction and reading skills holds true across almost every OECD country.

    “This holds true across almost every OECD country.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.
  • Findingown data
    § 2. Data

    There is a reasonable degree of association between young people who frequently read fiction and non-fiction books (correlation coefficient = 0.52).

    “The one notable exception is that there does seem to be a reasonable degree of association between young people who say they frequently read fiction and non-fiction books (correlation coefficient = 0.52).”
    Quote in the paperSubstantial claim
  • Classified as probably not a substantial claim. 35
  • May combine two claims. 39
  • The quote only partly states this claim. 3

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation
    § 1. Introduction

    Sullivan and Brown (2015) found that reading books and newspapers at age 16 is positively associated with performance in numeracy, spelling and vocabulary tests, while frequently reading comics or magazines typically had either a null or negative effect.

    “They found that reading books and newspapers at age 16 is positively associated with performance in numeracy, spelling and vocabulary tests, while frequently reading comics or magazines typically had either a null or negative effect.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Sullivan and Brown (2015) source

      Split into 2 atoms — 0 confirmed, 1 partially supported, 0 disputed, 0 misattributed, 1 unverifiable. Expand the claim to see per-atom verdicts.

      Experimental — accuracy not yet measured

  • May combine two claims.

    Premisecitation
    § 1. Introduction

    Sullivan and Brown's evidence is limited because it did not separate fiction from non-fiction books, used a sample born in 1970, and was specific to England.

    “However, this evidence is limited by the fact that it did not separate out the effect of reading fiction compared to non-fiction books, was based upon a sample of young people born in 1970 (hence referring to the situation around 50 years ago) and was specific to the English national context.”
    Quote in the paperSubstantial claim1 of 1 sources checked
    • cites Sullivan and Brown (2015) source

      The Data and variables section says the cohort was born in a single week of 1970. The Model 3 section measures book reading without distinguishing fiction from non-fiction. But Data and variables says the cohort included people born in England, Scotland and Wales, so describing the study as specific to England is inaccurate. (LLM, conf=0.90)

      Experimental — accuracy not yet measured

  • May combine two claims.

    Findingown data
    § 2. Data

    Polychoric correlations between the frequency of reading the five text types are generally quite low, suggesting that young people who report reading newspapers and magazines more frequently are not the same young people who read fiction a lot.

    “Interestingly, these correlations are generally quite low, suggesting that young people who report reading newspapers and magazines more frequently are not the same young people who say that they read fiction a lot.”
    Quote in the paperSubstantial claim

  • The quote only partly states this claim.

    Findingown data
    § 1. Introduction

    Frequency of reading fiction is strongly associated with teenagers' reading skills after a wide range of potential confounding characteristics are taken into account.

    “We find strong evidence that reading hold even after a wide range of potential confounding characteristics have been taken into account.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § 2. Data

    It is common for teenagers to read some text types several times a week (e.g. newspapers) but not others (e.g. comics and non-fiction books).

    “This illustrates how it its common for teenagers to read some of these text types several times a week (e.g.Newspapers) but not others (e.g.comics and non-fiction books).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § << Table 4 >>

    In model M3, a one-category increase in frequency of reading fiction is associated with a seven-point (0.07 SD) improvement in PISA reading scores; for non-fiction, the analogous association is four points.

    “In particular, a one-category increase in the frequency with which young people read fiction continues to be associated with a seven-point (0.07 standard deviation) improvement in PISA reading scores.For non-fiction, the analogous association is four points.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • May combine two claims.

    Findingown data
    § 3. Methods

    Even in countries where prior reading achievement and parental reading attitudes and behaviours are available and controlled, similar results are found; the positive fiction effect remains but is slightly reduced (parameter estimate declines by around ten percent).

    “To preview our key findings, even in countries where prior reading achievement and parental reading attitudes and behaviours are available and have been controlled, we continue to find similar results.Most notably, we continue to find a positive fiction effect, although slightly reduced compared to our main analysis (the parameter estimate declines by around ten percent).”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • May combine two claims.

    Findingown data
    § << Table 7 >>

    For fiction, the gap between each parameter estimate is reasonably evenly spread, providing evidence of a dose-response relationship in which the magnitude of the association increases monotonically from the least to the most frequent reading categories.

    “For fiction, the gap between each of the parameter estimates is reasonably evenly spread, and there is thus evidence of a 'dose-response' relationship (i.e. the magnitude of the association increases monotonically as one moves from the least to most frequent categories).”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § << Table 7 >>

    There is essentially no relationship between PISA reading scores and the frequency with which teenagers read the other four types of text, and no evidence of a comparable dose-response relationship.

    “Once more we continue to observe there to be essentially no relationship between PISA reading scores and the frequency with which teenagers read the other four types of text, and no evidence of a comparable dose-response relationship.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § << Table 4 >>

    In model M3, estimated coefficients for variables of interest fall by more than 50 percent between M2 and M3; associations for newspapers, magazines and comics become weakly associated (less than two PISA points per category increase).

    “Specifically, the estimated coefficients for our variables of interest fall by more than 50 percent between M2 and M3, with the association between reading three of the five text types (newspapers, magazines and comics) only weakly associated with young people's performance on the PISA reading test (less than two PISA points per each category increase).”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given
    § 5. Conclusions

    Exploration of the PISA data for lower and middle-income countries suggests that similar findings may not hold in these settings.

    “Our exploration of the PISA data for lower and middle-income countries suggests that similar findings may not hold in these settings.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § << Table 4 >>

    In model M1, links between reading skills and frequency of reading magazines and comics are substantially weaker (less than a three-point increase), with comics not statistically significant at conventional thresholds.

    “The link between reading skills and frequency of reading magazines and comics is substantially weaker (less than a three-point increase), with the latter not reaching statistical significance at conventional thresholds.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • May combine two claims.

    Findingown data
    § << Table 5 >>

    In model M5, a positive and statistically significant coefficient is recorded for fiction in 34 out of 35 OECD countries, generating an average effect of around six PISA points (0.06 SD) for a one-category increase in reading frequency.

    “In particular, note how a positive and statistically significant coefficient is recorded for fiction in 34 out of the 35 OECD countries, generating an average effect of around six PISA points (0.06 standard deviations) for a one-category increase in reading frequency.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • May combine two claims. The quote only partly states this claim.

    Findingown data
    § << Appendix Table A1 >>

    In all seven countries, the fiction parameter estimates remain of moderate size and statistically significant at the five percent level after controlling for prior achievement.

    “Yet in all seven countries the fiction parameter estimates remain of moderate size and statistically significant at the five percent level.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § << Table 3 >>

    Girls report reading fiction books and magazines much more often than boys; for both text types, girls are around half a category further along the five-point frequency scale.

    “Starting with gender, girls report reading fiction books and magazines much more often than boys; for both text types, girls are around half a category further along the five-point frequency scale.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § << Table 3 >>

    Boys are more likely than girls to read comic books (0.57 category difference) and newspapers (0.19 category difference).

    “On the other hand, boys are more likely than girls to read comic books (0.57 of a category difference) and newspapers (0.19 of a category difference).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § << Table 5 >>

    In model M5, the only strong and consistent evidence of a positive effect occurs for reading fiction books and not for other text types.

    “The first key point from Table 5 is that the only strong and consistent evidence of a positive effect occurs for the reading of fiction books and not for the other text types.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • May combine two claims.

    Findingown data
    § << Table 5 >>

    In model M5, coefficients for non-fiction, magazines, newspapers and comics take a range of positive, negative and null values across the 35 countries, with no clear and consistent pattern; OECD average parameter estimates for these text types are all around zero.

    “In contrast, the coefficients for the other text types take a range of positive, negative and null values across the 35 countries, with no clear and consistent pattern emerging.Moreover, the OECD average parameter estimates for non-fiction, magazines, newspapers and comics are all around zero, once again suggesting that frequently reading these text types is not associated with the development of”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • May combine two claims.

    Findingown data
    § << Table 4 >>

    Adding controls in model M2 does not lead to any major change to estimated coefficients or substantive results.

    “The addition of a selection of control variables in model M2, including whether the child attended pre-school, whether they have ever repeated a year at school and family structure, does not lead to any major change to the estimated coefficients or our substantive results.”
    Quote in the paperNot a substantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Additional information

    The educational career questionnaire included a question attempting to capture young people's prior performance in tests of their home language.

    “This included the following question, which attempted to capture young people's prior performance in tests of their home language:”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    Parameter estimates for the other four text types are all smaller in magnitude and inconsistent across countries.

    “In contrast, parameter estimates for the other four text types are all smaller in terms of magnitude and inconsistent across countries.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    Across OECD countries, the association between fiction and PISA reading subdomains is: Access & Retrieve 6.5, Integrate & interpret 7.5, Reflect & evaluate 6.5, Continuous texts 7.6, Non-continuous texts 5.4.

    “OECD average 6.5 1.0 7.5 0.9 6.5 0.9 7.6 0.9 5.4 0.9”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Additional information

    In Italy, 90% of the total sample completed the parent questionnaire.

    “Italy 30,876 27,641 90%”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    In Portugal, each category increase in fiction reading is associated with an increase of 4.9 PISA points without controlling for prior achievement and parental variables, falling to 4.2 when they are included.

    “each category increase in fiction reading is associated with an increase of 4.9 PISA points without controlling for these variables, falling to 4.2 when they are included.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Additional information

    In Germany, 64% of the total sample completed the parent questionnaire.

    “Germany 4,979 3,178 64%”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Additional information

    In South Korea, 99% of the total sample completed the parent questionnaire.

    “South Korea 4,989 4,935 99%”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Additional information

    In Denmark, 60% of the total sample completed the parent questionnaire.

    “Denmark 5,924 3,536 60%”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    Across OECD countries, the average regression coefficient for comic book reading frequency in Model 1 is 1.2 PISA points and is not statistically significant.

    “Model 1 | 17.3* | 1.1 | 9.9* | 1.2 | 5.8* | 1.1 | 2.6* | 1.2 | 1.2 | 1.1”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Additional information

    In Chile, 92% of the total sample completed the parent questionnaire.

    “Chile 5,663 5,183 92%”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    Across OECD countries, the average regression coefficient for newspaper reading frequency in Model 1 is 5.8 PISA points.

    “Model 1 | 17.3* | 1.1 | 9.9* | 1.2 | 5.8* | 1.1”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims. The quote only partly states this claim.

    Findingown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    Across OECD countries, when the five text types are included jointly in a single model, the average regression coefficient for non-fiction is 0.5 PISA points and is not statistically significant.

    “OECD avg. | 6.4* 0.7 | 0.5 | 0.8”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Additional information

    In Hungary, 98% of the total sample completed the parent questionnaire.

    “Hungary 4,605 4,520 98%”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    The average R-squared of the model across countries is 0.775.

    “The average R 2 of the model across countries is 0.775.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Additional information

    In New Zealand, 75% of the total sample completed the parent questionnaire.

    “New Zealand 4,606 3,451 75%”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Additional information

    In Portugal, 77% of the total sample completed the parent questionnaire.

    “Portugal 6,263 4,853 77%”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    After controlling for both prior reading achievement and parental reading attitudes and behaviours, a sizeable, positive and statistically significant fiction effect continues to be found in these countries, though slightly attenuated.

    “Hence, although slightly attenuated, we continue to find evidence of a sizeable, positive and statistically significant fiction effect in these countries even after both prior reading achievement and parental reading attitudes and behaviours have been controlled.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    The analogous figures are 4.1 to 3.7 in Hungary and 6.3 to 5.7 in Italy.

    “Similarly, the analogous figures are 4.1 to 3.7 in Hungary and 6.3 to 5.7 in Italy.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    Across OECD countries, the average regression coefficient for magazine reading frequency in Model 1 is 2.6 PISA points.

    “Model 1 | 17.3* | 1.1 | 9.9* | 1.2 | 5.8* | 1.1 | 2.6* | 1.2”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    Across OECD countries, the fiction effect increases monotonically with reading frequency: compared to never reading fiction, reading a few times a year is associated with 9.0 points, once a month with 13.9 points, several times a month with 20.1 points, and several times a week with 26.2 points.

    “Fiction | 9.0* | 2.5 | 13.9* | 2.9 | 20.1* | 3.0 | 26.2* | 3.4”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingown data
    § Additional information

    A total of 12 OECD countries conducted the educational career questionnaire.

    “A total of 12 OECD countries conducted the educational career questionnaire (Austria, Belgium, Hungary, Ireland, Iceland, Italy, Latvia, Mexico, New Zealand, Poland, Portugal and the Slovak Republic).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 1. Introduction

    The association between frequency of reading and higher attainment has been repeatedly demonstrated in the academic literature.

    “The association between frequency of reading and higher attainment has been repeatedly demonstrated within the academic literature, with many studies exploring the role motivation plays in their interaction (Baker et al, 2000;Krashen, 2004;Topping et al, 2007;Twist et al, 2007;Taboada et al 2009;Schiefele, et al, 2012).”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Baker et al, 2000

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Krashen, 2004

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Topping et al, 2007

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation
    § 1. Introduction

    Differences in the frequency with which young people read may help to explain gender and socio-economic achievement gaps.

    “Research suggests that differences in frequency with which young people read may help to explain gender and socio-economic achievement gaps (Brozo et al, 2007;Cremin et al, 2014;Krashen et al 2012;Sullivan and Brown, 2015).”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites Brozo et al, 2007 source
    • cites Cremin et al, 2014 source
    • cites Krashen et al 2012

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductreasoning
    § 3. Methods

    In model M5, all five text-type variables are included simultaneously and school fixed-effects are included to ensure between-school differences are not confounding results.

    “By including the various text type variables in the model together, we will be able to better establish whether it is fiction or non-fiction reading that is generating the positive effect.We have also chosen to include school fixed-effects within this model to ensure that between-school differences are not confounding our results.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given
    § 3. Methods

    The extent to which the preferred model provides causal estimates depends on the untestable selection on observables assumption.

    “In particular, the extent to which our preferred model provides causal estimates depends upon the (untestable) selection upon observables assumption.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: The preferred model's causal estimates depend on the untestable selection on observables assumption.

  • May combine two claims.

    Study design and conductown data
    § 3. Methods

    The quality of the prior reading achievement variable is likely to vary somewhat across countries, but it seems to act as a reasonable proxy for prior reading achievement in Portugal and Latvia.

    “Although our investigations have suggested that the quality of this variable is likely to vary somewhat across countries, it nevertheless seems to act as a reasonable proxy for prior reading achievement in Portugal and Latvia (see Appendix A for further details).”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Interpretationnone given
    § 5. Conclusions

    The focus of the paper has been 15-year-olds within developed OECD countries only.

    “the focus within this paper has been 15-year-olds within developed (OECD) countries only.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data
    § << Table 4 >>

    Table 4 provides some evidence that there is a 'fiction effect', and that reading fiction may hold particular importance for the development of teenagers' reading skills.

    “Table 4 hence provides some evidence that there is indeed a 'fiction effect', and that reading this type of text may hold particular importance for the development of teenagers' reading skills.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Study design and conductreasoning
    § 5. Conclusions

    The PISA 2009 survey asked adolescents about a selection of their reading habits, with young people likely engaging in a wider range of reading activities, including digital literacy.

    “the PISA 2009 survey asked adolescents about a selection of their reading habits, with young people likely engaging in a wider range of reading activities (including the growing area of digital literacy).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductreasoning
    § 2. Data

    Participation rates across most countries are quite high, typically around 90 percent at school and pupil levels.

    “Participation rates across most countries is therefore quite high, typically around 90 percent at the school 1 and pupil levels.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data
    § 5. Conclusions

    The type of text young people read potentially matters for their academic achievement, with the reading of fiction most likely to be beneficial.

    “Our key conclusion is therefore that the type of text young people read potentially matters for their academic achievement, with the reading of fiction most likely to be beneficial.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Premisecitation
    § 5. Conclusions

    The frequency with which young people read is associated with their reading skills.

    “It is well-known that the frequency with which young people read is associated with their reading skills, with an extensive array of academic research illustrating this relationship (Kirsch et al, 2002;OECD, 2011aOECD, , 2011b;;Baker et al, 2000;Krashen, 2004;Schiefele, et al, 2012;Brozo et al, 2007;Cremin et al, 2014;Sullivan and Brown, 2015).”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites Kirsch et al, 2002

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites OECD, 2011a

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

    • cites Baker et al, 2000

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Study design and conductown data
    § 5. Conclusions

    The paper conducted sensitivity analyses considering how its results would change if other potentially important variables, most notably prior achievement, were included in its models.

    “Although we have conducted sensitivity analyses considering how our results would change if other potentially important variables are included in our models (most notably prior achievement -see Appendix A)”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Study design and conductnone given
    § 5. Conclusions

    The PISA questionnaire relies upon self-reported data from young people about their reading habits.

    “the PISA questionnaire relies upon self-reported data from young people about their reading habits.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Study design and conductreasoning
    § 2. Data

    Final student weights and balanced-repeated-replication (BRR) weights are applied throughout the analysis to account for the complex PISA survey design.

    “To account for the complex PISA survey design, final student weights and balanced-repeated-replication (BRR) weighted are applied throughout our analysis.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing). May combine two claims.

    Study design and conductreasoning
    § 5. Conclusions

    Such self-reports likely contain measurement error, the degree of which is currently unknown.

    “It is likely that such reports contain measurement error -the degree of which is currently unknown.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data
    § 5. Conclusions

    The paper's estimates refer to conditional associations only and do not necessarily reveal cause and effect.

    “our estimates refer to conditional associations only and do not necessarily reveal cause and effect.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductreasoning
    § 3. Methods

    Although the final models control for a wide array of factors and have a high R2 (around 0.8 in most countries), it is impossible to rule out potentially important omitted variables.

    “Although we have been able to control for a wide array of factors, and the R 2 of our final models is high (around 0.8 in most countries), it is impossible to rule out the possibility that there are potentially important omitted variables.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning
    § 5. Conclusions

    Greater attention to volitional reading, with schools setting aside quality time to read and talk informally about self-chosen texts, may be beneficial for the development of young people's reading skills.

    “our results suggest that greater attention to volitional reading, with schools setting aside quality time to read and talk informally about self-chosen texts, may be beneficial for the development of young people's reading skills.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data
    § << Table 5 >>

    The apparent advantages of non-fiction reading in Table 4 were actually driven by overlap between fiction and non-fiction readers; once included in the same model, only the fiction effect remains.

    “Consequently, it seems that the apparent advantages of non-fiction reading previously identified in Table 4 was actually driven by the overlap between fiction and non-fiction readers.Once these variables are included in the same model only the fiction effect remains, suggesting the previous positive non-fiction estimate was due to models M1 to M4 not accounting for the frequency with which young”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • May combine two claims.

    Interpretationown data
    § << Table 6 >>

    The link between fiction and reading performance is not simply due to fiction readers tending to spend more time reading per week; particular types of skill might be developed from frequently reading fiction.

    “This in turn suggests that the link between fiction and reading performance is not simply due to fiction readers tending to spend more time reading per week, but that particular types of skill might be developed from frequently reading this form of text.”
    Quote in the paperSubstantial claim
    • The authors state this as an assumption: There is a linear relationship between reading frequency and young people's PISA reading scores.

  • Classified as probably not a substantial claim. May combine two claims.

    Study design and conductcitation
    § 2. Data

    Five plausible values are used, models are estimated five times and parameter estimates and standard errors are pooled according to Rubin's rules.

    “We follow recommended practise in using these five plausible values throughout this paper, estimating each model five times (once using each plausible value) and then pooling the parameter estimates and standard errors according to 'Rubin's rules' (Rubin 1987).”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites Rubin 1987 source

  • May combine two claims.

    Study design and conductown data
    § 3. Methods

    The main analysis has been unable to control for prior achievement in reading and parental attitudes, engagement and encouragement in reading activities.

    “Specifically, we believe that there are two potentially important factors our main analysis has been unable to control for: (a) prior achievement in reading and (b) parental attitudes, engagement and encouragement in reading activities.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationown data
    § << Appendix Table A2 >> << Appendix Table A3 >>

    Omitting parental variables in the main analysis including all OECD countries is likely to have a small impact, and the substantive finding of a positive and important fiction effect is likely to remain intact.

    “Overall, these results therefore suggest that the impact of omitting these variables in our main analysis including all OECD countries is likely to be small, and our substantive finding of a positive and important fiction effect is likely to remain intact.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Interpretationnone given
    § Additional information

    The correlation of the prior achievement measure with PISA scores is likely to be attenuated to close to zero in situations where it contains significant measurement error.

    “The intuition behind these investigations is that its correlation with PISA scores is likely to be attenuated to close to zero in situations where it contains significant measurement error.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S4 (based).

    Study design and conductnone given
    § Additional information

    The quality of the information provided by the prior achievement question is dependent upon young people being able to recall and accurately report such information.

    “For instance, the quality of the information provided is dependent upon young people being able to recall and accurately report such information.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationown data
    § << Appendix Table A4 >> << Appendix Table A5 >> << Appendix Table A6 >>

    The findings support the conclusion that there is a positive effect of young people frequently reading fiction books, which is not the case for other text types.

    “Again, we interpret these findings as supporting our conclusion that there is a positive effect of young people frequently reading fiction books, which is not the case for other text types.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Study design and conductreasoning
    § Additional information

    Two specific issues stand out as limitations: (a) a lack of prior reading achievement measures and (b) a lack of information with respect to parental reading attitudes, behaviours and support.

    “Two specific issues stand out: (a) a lack of prior reading achievement measures and (b) a lack of information with respect to parental reading attitudes, behaviours and support.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Study design and conductnone given
    § Additional information

    Depending upon the country context, it is possible that young people may be reporting different measures depending upon their school.

    “Moreover, depending upon the country context, it is also possible that young people may be reporting different measures depending upon their school.”
    Quote in the paperNot a substantial claim
68 of 68

  • The authors state this as an assumption.

    There is a linear relationship between reading frequency and young people's PISA reading scores.

    “This imposes an assumption of there being a linear relationship between reading frequency and young people's PISA reading scores.”

  • The authors state this as an assumption.

    The preferred model's causal estimates depend on the untestable selection on observables assumption.

    “the extent to which our preferred model provides causal estimates depends upon the (untestable) selection upon observables assumption.”

  • The paper relies on this without support.

    Different types of text are freely available in the developed world, and the vast majority of 15-year-olds targeted by PISA are still in school there.

    “This is due to our interest in literacy skills within the developed world, where different types of text are freely available, and where the vast majority of 15year-olds targeted by the PISA survey are still in school (which is not the case within some non-OECD countries).”

  • The paper relies on this without support.

    There has been much less large-scale quantitative research into whether what young people read matters for their academic achievement.

    “Yet there has been much less large-scale quantitative research into whether what young people read matters for their academic achievement.”

  • The paper relies on this without support.

    Most existing large-scale databases do not capture this level of detail in variables measuring the frequency with which young people read, often reducing this to a sole indicator such as total reading time.

    “Most existing large-scale databases do not capture this level of detail in variables which measure the frequency with which young people read, often boiling this down to a sole indictor such as total reading time.”

  • The paper relies on this without support.

    Fiction books typically require readers to digest large amounts of long and continuous text.

    “Given that fiction books typically require readers to digest large amounts of long and continuous text”
6 of 6

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Not every coefficient falls by more than half in model M3

    Minor

    The claim applies to all five text types, but magazines fall from 2.0 to 1.5 and comics from 1.9 to 1.1 between M2 and M3. Those declines are 25% and about 42%, respectively.

  • The paper gives conflicting counts of countries with both auxiliary measures

    Minor

    The methods section says Portugal is the only country with both a usable prior-achievement measure and parental-questionnaire information. The appendix instead identifies Italy, Portugal and Hungary as having both.

  • The typical prior-achievement adjustment is overstated

    Minor

    The appendix describes reductions of typically 10-15%, but five of the seven before-and-after estimates in Table A1 decline by only about 4-7%. Only Poland and the Slovak Republic have reductions near the stated range.

3 of 3

Cites

Flagged first, then by citations

Research works printed but not linked: 4

  1. Clark C., 2015, Children's and young people's reading in 2015 report
  2. Kirsch I., 2002, Reading for change: Performance and engagement across countries
  3. Sullivan A., 2015, The life‐long benefits of reading for pleasure, The School Librarian, 63, 5
  4. Twist L., 2007, Readers and reading: The national report for England 2006

Cited by

Sorted by citations

Related

Authors
Jerrim, J., & Moss, G.
Affiliations
University College London (GB)
Corresponding author
Jerrim, John
ORCID
Venue
British Educational Research Journal
Publisher
Wiley
Published
2018-12-26
Volume
45
Issue
1
Pages
181-200
ISSN
  • 0141-1926 (ISSN-L)
  • 1469-3518
DOI
10.1002/berj.3498
OpenAlex
W2906286245
Cited by
110
References
26
Citations by year
2026: 10 · 2025: 14 · 2024: 21 · 2023: 20 · 2022: 16 · 2021: 16 · 2020: 10 · 2019: 3
FWCI
23.48
Type
journal article
Language
en
Open access
Licence
onlinelibrary.wiley.com/termsAndConditions#vor (vor)
Locations
  • British Educational Research Journal (publishedVersion)
  • UCL Discovery (University College London) (submittedVersion)
Funders
Economic and Social Research Council (ES/I028501/1)
Topics
  • Social Sciences › Education › Literacy and Educational Practices
  • Social Sciences › Psychology › Developmental and Educational Psychology › Reading and Literacy Development
  • Social Sciences › Education › School Choice and Performance
Keywords
fiction reading, reading skills, adolescent reading, reading for pleasure, text types, reading achievement, PISA, reading intervention, socioeconomic deprivation, boys, comics, newspapers, magazines, nonfiction reading

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-11.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-11.

Cited sources are checked by gpt-6-sol.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review