Swedish teenagers’ difficulties and abilities to determine digital news credibility

Reported and measured construct	Question
Background variables	Gender (girl/boy/other identity) Upper-secondary school programme (year & programme) Language at home (foreign language/Swedish) Political party sympathies (1 or more parties)4
Self-reported categories	Items in self-report
Preferred news	Preferred news format(s) (e.g. paper news, radio etc.) Preferred news source (s) (e.g. local newspaper, evening paper etc.)
Self-rated abilities	Source criticism on the Internet (fact-checking ability) Finding information on the Internet (search ability)
Information credibility	Importance of news credibility (credibility importance) Reliability of information on the internet (info reliability)
Sourcing in education	Source critical evaluation in education (sourcing in education)
Categories of civic online reasoning	Test-items
Detecting sponsored material (sourcing)	Screenshot from evening paper 1 (Aftonbladet) Screenshot from evening paper 2 (Expressen) Screenshot from IT journal (Techworld)
Comparing articles (corroboration)	Two articles about weight loss (weight loss) Two articles about the government’s policy on racism (racism)
Scrutinizing comments and images (evidence)	Manipulated photograph of a smoking girl (smoking) Article about the government’s energy goals (energy goals) False information about climate change (climate change) Reader’s comment on incomes (income) Manipulated photograph on daisies in Fukushima (Fukushima)

Comment: The measured constructs refer to abilities that have been identified as important for the detection of false news online. The questions are referred to in the text with the names in parentheses.

4 Since participants were allowed to choose more than one party, we calculated a normalized mean score that spans from 0–1, where 0 represents the most right-winged party (SD) and 1 represents the most left-winged party (V). All parties in between those are given a score based on their position (0–7) on the spectrum (SD-V) divided by 7.

Predicted probabilities of a correct answer on the comment on incomes as a function of internet info reliability and credibility importance
*Comment:* The predicted probabilities of a correct answer on the comment on incomes as a function of internet info reliability (x-axis) and credibility importance (legend). The higher the ratings on credibility importance, the higher the probability of a correct answer, whereas a higher rating on internet info reliability was associated with a lower probability of a correct answer.

Table A1

Estimates of an additive poisson regression model for number of correct answers on expressen, with coefficients denoting the expected log count for a unit increase of the ordinal variables and the expected log count for each category compared with the baseline category for the categorical variables. Search ability was omitted due to perfect correlation with search ability

*Parameter*		*Estimate (expected log count)*	*SE (standard error)*	*Z-ratio*
Intercept		2.221	0.509	4.363^***
Form	Secondary 1^st	Baseline
	Secondary 2^nd	−0.622	0.404	−1.537
	Secondary 3^rd	−0.543	0.358	−1.514
Program	SS	Baseline
	Other	−0.293	0.336	−0.872
	EC	−0.503	0.400	−1.256
	AS	−0.111	0.163	−0.677
	NA	−0.513	0.394	−1.300
Sourcing in school		0.0186	0.0242	0.770
Credibility importance		−0.00198	0.0469	−0.042
Internet info reliability		−0.0448	0.0590	−0.759
Search ability		−0.0682	0.0612	−1.116
Language home	Yes	Baseline
	No	0.223	0.0937	2.378 ^*
Party score		0.00372	0.164	0.023

[i] SS= social sciences, Other= all other programs, EC= economic, AS= aesthetic, NA=natural sciences.

[ii] Model fit: Residual variance: 77.568 on 111 degrees of freedom. AIC: 516.01.

[iii] Significance codes: (^*) .1, ^* <.05, ^**< .01, ^*** <.001

Table A2

Estimates of an additive poisson regression model for incidences of public service radio, with coefficients denoting the expected log count for a unit increase of the ordinal variables and the expected log count for each category compared with the baseline category for the categorical variables. Search ability was omitted due to perfect correlation with fact-checking ability

Parameter		Estimate (expected log count)	SE (standard error)	Z-ratio
Intercept		−2.162	1.780	−1.215
Form	Secondary 1^st	Baseline
	Secondary 2^nd	−0.927	0.794	−1.169
	Secondary 3^rd	−0.267	0.485	−0.552
Sourcing in school		0.163	0.137	1.193
Credibility importance		0.452	0.274	1.652 (^*)
Internet info reliability		−0.0670	0.283	−0.225
Fact-checking ability		0.565	0.283	−1.993^*
Language home	Yes	Baseline
	No	0.554	0.529	1.049
Party score		0.133	0.892	0.149
Gender	Other identity	Baseline
	Girl	−0.287	0.678	−0.423
	Boy	−0.899	0.772	−1.164

[i] Model fit: Residual variance: 66.952 on 113 degrees of freedom. AIC: 138.95.

[ii] Significance codes: (^*) .0986, ^* <.05, ^** <.01, ^*** <.001

Table A3

Estimates of an additive poisson regression model for incidences of public service tv, with coefficients denoting the expected log count for a unit increase of the ordinal variables and the expected log count for each category compared with the baseline category for the categorical variables. Search ability was omitted due to perfect correlation with fact-checking ability

Parameter		Estimate (expected log count)	SE (standard error)	Z-ratio
Intercept		−3.238	1.268	−2.553 ^*
Form	Secondary 1st	Baseline
	Secondary 2nd	−0.243	0.445	−0.584
	Secondary 3rd	−0.106	0.324	−0.744
Sourcing in school		−0.0615	0.0841	−0.731
Credibility importance		0.505	0.0841	2.595 ^**
Internet info reliability		−0.281	0.204	−1.435
Fact-checking ability		0.198	0.221	0.897
Language home	Yes	Baseline
	No	0.728	0.374	1.949 (^*)
Party score		0.268	0.506	0.531

[i] Model fit: Residual variance: 74.078 on 115 degrees of freedom. AIC: 194.08.

[ii] Significance codes: (^*) .0513, ^* <.05, ^** <.01, ^*** <.001

Table A4

Estimates of an additive poisson regression model for incidences of international news, with coefficients denoting the expected log count for a unit increase of the ordinal variables and the expected log count for each category compared with the baseline category for the categorical variables. Search ability was omitted due to perfect correlation with fact-checking ability

Parameter		Estimate (expected log count)	SE (standard error)	Z-ratio
Intercept		−3.488	1.914	−1.822
Form	Secondary 1st	Baseline
	Secondary 2nd	−0.110	0.611	−0.181
	Secondary 3rd	0.488	0.540	0.904
	Primary 9th	−1.138	1.0761	−1.058
Sourcing in school		−0.135	0.115	−1.172
Credibility importance		0.406	0.266	1.528
Internet info reliability		0.472	0.301	1.567
Fact-checking ability		0.108	0.294	0.367
Language home	Yes	Baseline
	No	−0.936	0.443	−2.114 ^*
Party score		0.296	0.801	0.369
Gender	Other identity	Baseline
	Girl	−0.334	0.834	−0.401
	Boy	−0.500	0.848	−0.590

[i] Model fit: Residual variance: 71.741 on 113 degrees of freedom. AIC: 133.74.

[ii] Significance codes: (^*) .0603, ^* <.05, ^** <.01, ^*** <.001

Table A5

Estimates of an additive poisson regression model for incidence of radio as preferred media format, with coefficients denoting the expected log count for a unit increase of the ordinal variables and the expected log count for each category compared with the baseline category for the categorical variables. Search ability was omitted due to perfect correlation with fact-checking ability

Parameter		Estimate (expected log count)	SE (standard error)	Z-ratio
Intercept		−0.868	1.401	−0.620
Form	Secondary 1st	Baseline
	Secondary 2nd	−0.588	0.577	−1.019
	Secondary 3rd	0.172	0.404	0.426
Sourcing in school		0.0990	0.101	0.984
Credibility importance		0.543	0.229	2.368 ^*
Internet info reliability		−0.0687	0.248	−0.276
Fact-checking ability		−0.687	0.224	−3.060 ^**
Language home	Yes	Baseline
	No	−0.0453	0.376	−0.121
Party score		−0.278	0.690	−0.402
Gender	Other identity	Baseline
	Girl	−0.157	0.585	−0.268
	Boy	−0.608	0.647	−0.940

[i] Model fit: Residual variance: 72.644 on 113 degrees of freedom. AIC: 170.64.

[ii] Significance codes: (^*) .100, ^* <.05, ^** <.01, ^*** <.001

Table A6

Estimates of an additive logistic regression model for correct/incorrect answer on racism, with coefficients denoting the log odds of answering correct for a unit increase of the ordinal variables and the log odds of answering correct for each category compared with the baseline category for the categorical variables. Search ability was omitted due to perfect correlation with fact-checking ability

Parameter		Estimate (expected log count)	SE (standard error)	Z-ratio
Intercept		−2.005	3.679	−0.545
Form	Secondary 1st	Baseline
	Secondary 2nd	3.178	2.365	1.344
	Secondary 3rd	0.119	2.0185	−0.059
Program	SS	Baseline
	Other	0.858	1.737	0.446
	EC	1.0355	2.321	−1.256
	AS	2.882	1.280	2.251^*
	NA	1.995	2.340	0.853
Sourcing in school		0.263	0.152	1.729 (^*)
Credibility importance		−0.295	0.356	−0.827
Internet info reliability		−0.598	0.431	−1.388
Fact-checking ability		−0.0434	0.380	−0.114
Language home	Yes	Baseline
	No	0.299	0.649	0.460
Party score		−0.658	1.225	−0.538
Gender	Other identity	Baseline
	Girl	2.415	1.218	1.983^*
	Boy	1.652	1.295	1.259

[i] Model fit: Residual variance: 87.534 on 70 degrees of freedom. AIC: 117.53.

[ii] Significance codes: (^*) .0837, ^* <.05, ^** <.01, ^*** <.001

Table A7

Estimates of an additive logistic regression model for correct/incorrect answer on income, with coefficients denoting the log odds of answering correct for a unit increase of the ordinal variables and the log odds of answering correct for each category compared with the baseline category for the categorical variables. Search ability was omitted due to perfect correlation with fact-checking ability

Parameter		Estimate (expected log	SE (standard error)	Z-ratio
Intercept		2.106	2.609	0.807
Sourcing in school		−0.0521	0.120	−0.261
Credibility importance		1.100	0.421	2.391^*
Internet info reliability		−1.556	0.485	−2.382^*
Fact-checking ability		0.0412	0.465	0.009
Language home	Yes	Baseline
	No	−0.204	0.755	−0.271
Party score		−0.650	1.291	−0.504

[i] Model fit: Residual variance: 60.133 on 78 degrees of freedom. AIC: 74.133.

[ii] Significance codes: (^*) .100, ^* <.05, ^** <.01, ^***<.001

Table A8

Estimates of an additive logistic regression model for correct/incorrect answer on smoking, with coefficients denoting the log odds of answering correct for a unit increase of the ordinal variables and the log odds of answering correct for each category compared with the baseline category for the categorical variables. Search ability was omitted due to perfect correlation with fact-checking ability

Parameter		Estimate (expected log	SE (standard error)	Z-ratio
Intercept		2.579	2.380	1.084
Sourcing in school		0.109	0.146	0.748
Credibility importance		0.570	0.344	1.655(^*)
Internet info reliability		0.228	0.372	0.612
Fact-checking ability		−1.280	0.490	−2.613^**
Language home	Yes	Baseline
	No	0.345	0.604	0.571
Party score		−0.210	0.977	−0.215

[i] Model fit: Residual variance: 81.230 on 78 degrees of freedom. AIC: 95.23.

[ii] Significance codes: (^*) .098, ^* <.05, ^** <.01, ^***<.001

Table A9

Estimates of an additive logistic regression model for correct/incorrect answer on fukushima, with coefficients denoting the log odds of answering correct for a unit increase of the ordinal variables and the log odds of answering correct for each category compared with the baseline category for the categorical variables. Search ability was omitted due to perfect correlation with fact-checking ability

Parameter		Estimate (expected log count)	SE (standard error)	Z-ratio
Intercept		−2.0857	4.0026	−0.521
Form	Secondary 1st	Baseline
	Secondary 2nd	2.213	2.131	1.038
	Secondary 3rd	−0.337	1.614	−0.209
Program	SS	Baseline
	Other	0.642	1.582	0.406
	EC	1.246	1.982	0.628
	AS	3.010	1.190	2.530^*
	NA	3.789	2.240	1.691(^*)
Sourcing in school		0.0208	0.202	0.103
Credibility importance		0.468	0.414	1.131
Internet info reliability		−0.131	0.568	−0.230
Fact-checking ability		0.410	0.439	0.934
Language home	Yes	Baseline
	No	−1.551	0.993	−1.562

[i] Model fit: Residual variance: 61.629 on 98 degrees of freedom. AIC: 85.629.

[ii] Significance codes: (^*) .098, ^* <.05, ^** <.01, ^***<.001

Table B1

Media format and media sources that were measured in the survey. The youths indicated what format and sources they used when consuming news; they could indicate multiple formats and sources.

Media format	Media source
Paper format	Local news
Radio	National news
TV	Evening paper
Mobile phone	Public service radio
Computer	Public service TV
Tablet	International news
	Social media

References

Authors

Metrics

Articles in this issue

DOI: https://doi.org/10.2478/nor-2019-0002 | Journal eISSN: 2001-5119 | Journal ISSN: 1403-1108

Journal RSS Feed

Language: English

Page range: 23 - 42

Published on: Feb 19, 2019

Published by: University of Gothenburg Nordicom

In partnership with: Paradigm Publishing Services

Keywords:

media information literacy

Related subjects:

Social sciences,

Communication science,

Mass communication,

Public and political communication

© 2019 Thomas Nygren, Mona Guath, published by University of Gothenburg Nordicom
This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License.

Volume 40 (2019): Issue 1 (February 2019)

Swedish teenagers’ difficulties and abilities to determine digital news credibility

Figures & Tables

Table 1

Figure 1