Help on deciding the test I need to use

[k-nabhani]

Hi,

I am just starting off with Spss and statistics for my dissertation project. I would appreciate it if anybody can explain what type of test should I use for the following scenario:

My study aims to see if anxiety has an effect on the spread of misinformation online.

I am going to measure the anxiety levels of participants with the Dass 21 questionnaire which I think is an ordinal scale as it has a four point selection from 'doesn't apply to me' to 'strongly apply to me'. however this questionnaire also measures for stress and depression, would it effect my results if I ignore these scores and just employ the anxiety scores?

After the test the participants will be split into two group which are normal level anxiety and high level anxiety. They will then be presented with fake and real headlines and asked which are they most likely to share online.

DV: anxiety

IV: sharing activity

what design is this?

what's the best test to use here?

an explanation would be highly appreciated.

Thanks

 

[Karabiner]

Anxiety is not your DV but the predictor (IV). You did not explain how "sharing activity" is
elicted and measured (e.g. do participants read 100 fake and 100 real hedalines and in
each case indicate whether they would share them yes/no). And the most important
information (sample size) is missing.

With kind regards

Karabiner

 

[k-nabhani]

Hi Karabiner,

Thank you for your reply really appreciate it.

Each participant would be presented with 3 fake and 3 real news headlines and would have to select either yes I would share this online’ or ‘no I would not share this online ’.

sample size 93

 

[Karabiner]

I would not be too happy with a categorization into allegedly "normal" and "high"
anxiety, since participants slighltly above vs. slightlly below your cutoff would be
placed into different categories, while at the same time someone with a value just
above cutpoint ("slightly elevated anxiety") would be placed into the same category
as someone with extreme anxiety. That seems nonsensical. Also, categorizing destroys
information.
https://discourse.datamethods.org/t/categorizing-continuous-variables/3402

If you planned to categorize by using the sample median, and not by using normative
values from the general population, then you should skip that categorizing thing anyway,
because category labels would be misleading, and the results cannot be well
generalized.

You did not explicitly state your research question(s), so I have to guess. Maybe you could
use a repeated-measures analysis of variance, with the repeated-measures factor
"type of stimulus" (fake versus real) and the uncategorized anxiety score as interval
scaled predictor. The interaction effect between "type of stimulus" and "anxiety" tells you
whether the difference between fake and real news with regard to sharing is different
for different levels of anxiety. If there is no interaction effect, then the main effect of anxiety
tells you whether more anxious participants generally tend to share more [or less, respectively]
then less anxious participants, and the main effect of "type of stimulus" tells you
whether there is a difference between fake and real headlines with regard to sharing.

If your research question is different from what I supposed, then please clarify.

With kind regards

Karabiner

 

[k-nabhani]

Hi Karabiner,

Thanks again. My research question is:

Are anxious people more likely to share fake news online?