Group paper topic: Fake News
Word count (excluding references): 1484
, Results
The outcome variable (dependent variable) of our research, belief in fake news, is continuous
and measured on a 7-point Likert scale ranging from ‘Strongly disagree’ (1) to ‘Strongly agree’
(7). As stated before, to assess the effectiveness of our intervention
𝐻 0: 𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑛𝑜 𝑠𝑖𝑔𝑛𝑖𝑓𝑖𝑐𝑎𝑛𝑡
𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑖𝑛 𝑏𝑒𝑙𝑖𝑒𝑓 𝑖𝑛 𝑓𝑎𝑘𝑒 𝑛𝑒𝑤𝑠 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑡ℎ𝑒 𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 𝑎𝑛𝑑 𝑐𝑜𝑛𝑡𝑟𝑜𝑙 𝑔𝑟𝑜𝑢𝑝 needs to be
rejected, so that 𝐻 1: 𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑎 𝑠𝑖𝑔𝑛𝑖𝑓𝑖𝑐𝑎𝑛𝑡
𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑖𝑛 𝑏𝑒𝑙𝑖𝑒𝑓 𝑖𝑛 𝑓𝑎𝑘𝑒 𝑛𝑒𝑤𝑠 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑡ℎ𝑒 𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 𝑎𝑛𝑑 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑔𝑟𝑜𝑢𝑝 is the only other
likely option.
One should explore different significance testing methods and their assumptions before
using a certain test. Firstly, using a Saphiro-Wilk test, it was concluded that the datasets from
questions 11c, 11e, 18c and 18e1 from the survey for both the treatment as well as the control
group were not normally distributed according to the test, in which an α of 0.05 was selected
and the p-values for all relevant datasets were below α, indicating the difference between the
datasets and the general normal distribution is big enough to be statistically significant.
Because of the continuous dependent variable and the non-normality of the data, neither a
2
T-test or a χ -test is appropriate. Instead, a Mann-Whitney U test should be used (Van Steen
2020; Griffin 2013). The chosen significance level is α = 0.05 for all tests. The results are
presented in Table 1.
Question N, x̄ and SD in U P-value Significant Effect
determining belief control and difference in size r
in fake news2 experimental medians
group between the
control and
experimental
group?
11c: "Ik vind dit N (control) = 50 1173.5 0.19706 P-value > α, so -0.01
een betrouwbaar x̄ (control) = 2.56 no significant
nieuwsbericht." SD (control) = 1.41 difference.
N (exp.) = 55
x̄ (exp.) = 2.18
SD (exp.) = 1.28
1
These were the questions used to measure the belief in fake news.
2
In Dutch, as the survey was also in Dutch, to prevent translation and validity issues the original
questions are used.
1
Word count (excluding references): 1484
, Results
The outcome variable (dependent variable) of our research, belief in fake news, is continuous
and measured on a 7-point Likert scale ranging from ‘Strongly disagree’ (1) to ‘Strongly agree’
(7). As stated before, to assess the effectiveness of our intervention
𝐻 0: 𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑛𝑜 𝑠𝑖𝑔𝑛𝑖𝑓𝑖𝑐𝑎𝑛𝑡
𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑖𝑛 𝑏𝑒𝑙𝑖𝑒𝑓 𝑖𝑛 𝑓𝑎𝑘𝑒 𝑛𝑒𝑤𝑠 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑡ℎ𝑒 𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 𝑎𝑛𝑑 𝑐𝑜𝑛𝑡𝑟𝑜𝑙 𝑔𝑟𝑜𝑢𝑝 needs to be
rejected, so that 𝐻 1: 𝑇ℎ𝑒𝑟𝑒 𝑖𝑠 𝑎 𝑠𝑖𝑔𝑛𝑖𝑓𝑖𝑐𝑎𝑛𝑡
𝑑𝑖𝑓𝑓𝑒𝑟𝑒𝑛𝑐𝑒 𝑖𝑛 𝑏𝑒𝑙𝑖𝑒𝑓 𝑖𝑛 𝑓𝑎𝑘𝑒 𝑛𝑒𝑤𝑠 𝑏𝑒𝑡𝑤𝑒𝑒𝑛 𝑡ℎ𝑒 𝑡𝑟𝑒𝑎𝑡𝑚𝑒𝑛𝑡 𝑎𝑛𝑑 𝑐𝑜𝑛𝑡𝑟𝑜𝑙𝑔𝑟𝑜𝑢𝑝 is the only other
likely option.
One should explore different significance testing methods and their assumptions before
using a certain test. Firstly, using a Saphiro-Wilk test, it was concluded that the datasets from
questions 11c, 11e, 18c and 18e1 from the survey for both the treatment as well as the control
group were not normally distributed according to the test, in which an α of 0.05 was selected
and the p-values for all relevant datasets were below α, indicating the difference between the
datasets and the general normal distribution is big enough to be statistically significant.
Because of the continuous dependent variable and the non-normality of the data, neither a
2
T-test or a χ -test is appropriate. Instead, a Mann-Whitney U test should be used (Van Steen
2020; Griffin 2013). The chosen significance level is α = 0.05 for all tests. The results are
presented in Table 1.
Question N, x̄ and SD in U P-value Significant Effect
determining belief control and difference in size r
in fake news2 experimental medians
group between the
control and
experimental
group?
11c: "Ik vind dit N (control) = 50 1173.5 0.19706 P-value > α, so -0.01
een betrouwbaar x̄ (control) = 2.56 no significant
nieuwsbericht." SD (control) = 1.41 difference.
N (exp.) = 55
x̄ (exp.) = 2.18
SD (exp.) = 1.28
1
These were the questions used to measure the belief in fake news.
2
In Dutch, as the survey was also in Dutch, to prevent translation and validity issues the original
questions are used.
1
