Which statistical test? Cyberbullying research

Ask here about academic and research issues, like designing studies, recruiting participants, choosing statistics, submitting for publication, etc.
Post Reply
emma1993
Posts: 5
Joined: Thu Apr 02, 2015 11:52 pm

Which statistical test? Cyberbullying research

Post by emma1993 » Sat Jul 04, 2015 3:23 am

Hi all, I am new to this forum and have been participating in people's dissertation surveys posted, but have a small question to ask about some of my own research!

I am putting together a proposal for when I start my Forensic Psychology Masters this September, and have decided I want to focus on the dark triad of personality traits and cyberbullying. I would like to first ask participants if they have ever been a victim/perpetrator of cyberbullying/both or neither - therefore placing participants into four distinct groups. I would also like to look at gender differences between participant's responses to the dark triad personality test - I have looked at the 'dirty dozen' concise measure of the dark triad and would like to use this to test for the three traits and compare results between groups.

I am struggling with which statistical tests I would need to do in order to test for differences between the four independent groups and gender differences in their dark triad scores - any help would be greatly appreciated! My study is very rough at the moment as I'm in the stage of getting together a decent enough study to propose when I start.

Many thanks! :)

User avatar
miriam
Site Admin
Posts: 7834
Joined: Sat Mar 24, 2007 11:20 pm
Location: Bucks
Contact:

Re: Which statistical test? Cyberbullying research

Post by miriam » Wed Jul 08, 2015 12:41 am

I think to do an MSc you will need a certain level of autonomy in your ability to find information and a certain level of knowledge with research design and statistics, so I'd suggest you go look at the flow charts in any basic stats for psychology text book, and/or seek out support at your university.

But basically you are asking whether a variable differs across groups, and that is an ANOVA if the data is parametric and probably a Mann-Whitney U if it is non-parametric. You can use post-hoc analysis to see where the between groups differences lie if you have more than two groups, though I'd suggest comparing bully to not bully, and victim to not victim and victim to bully in three separate tests with compensation for multiple testing in your power calculations, as between groups comparisons can be done with a t-test and be pretty robust.
Miriam

See my blog at http://clinpsyeye.wordpress.com

User avatar
sarahg
Posts: 614
Joined: Sat Dec 10, 2011 1:46 am

Re: Which statistical test? Cyberbullying research

Post by sarahg » Wed Jul 08, 2015 9:35 am

Miriam is right that for an MSc level you need a certain level of autonomy. Additionally, your research project needs to go beyond the level of undergraduate in order to obtain higher marks.

What is your reasoning for wanting to define CB behaviours into those four categories, and then explore gender differences in relation to 'the dark triad'. Quick search suggests that CB is not related to 'the dark triad'. But also CB is extremely hard to define, and putting people into one of those four categories can be very difficult, what about those involved in CB incidences (i.e. sharing a photo, or posting a nasty comment) who may not realise they are perpetrators? These are all things to think about before jumping to the end game. However, a pet hate of mine is people who do not think how they will analyse the data when designing their study, so credit to you for thinking about that before proceeding :)

If you really want to up your game, I'd suggest using some sort of latent class analysis, or at least use a paper which has conducted a LCA to show support for the existing CB groups.

emma1993
Posts: 5
Joined: Thu Apr 02, 2015 11:52 pm

Re: Which statistical test? Cyberbullying research

Post by emma1993 » Wed Jul 08, 2015 11:48 am

Thanks for your replies.

I read textbooks and looked at flow charts etc. before writing this post, but just wanted to seek some clarity on here from people with more research experience than myself. Stats aren't my strong suit, but I didn't jump on here for help without first trying to find out the answer myself. If I was like that I wouldn't feel comfortable applying for an MSc.

ANOVA seems to be the way to go with post hoc testing to see what the differences are, but I may review my topic and do a bit more reading before writing the proposal.

emma1993
Posts: 5
Joined: Thu Apr 02, 2015 11:52 pm

Re: Which statistical test? Cyberbullying research

Post by emma1993 » Wed Jul 08, 2015 6:27 pm

Just to add on from my previous post, my reasoning for this particular research is that I volunteered as a cyberbullying mentor for nearly three years and therefore have experience helping those affected (victims and occasionally perpetrators) and felt that it may be interesting to look at personality types. There currently isn't a lot of research surrounding this topic at all (trolling and the dark triad has been done, but not cyberbullying specifically).

As I say, still very very early days but the stats side was a concern of mine as it may become quite complex to analyse.

User avatar
sarahg
Posts: 614
Joined: Sat Dec 10, 2011 1:46 am

Re: Which statistical test? Cyberbullying research

Post by sarahg » Thu Jul 09, 2015 8:59 am

I don't know much about the dark triad, so apologise...but if you are classifying people into personality groups then you will have purely categorical data and thus an ANOVA would not be suitable. If people will be given a score which suggests they lie somewhere along a personality continuum, then an ANOVA is possible.

Also, a recent study examining the dark triad and cyber-bullying...may be worth a look for you
http://www.sciencedirect.com/science/ar ... 3215001739

emma1993
Posts: 5
Joined: Thu Apr 02, 2015 11:52 pm

Re: Which statistical test? Cyberbullying research

Post by emma1993 » Thu Jul 09, 2015 5:51 pm

Thank you very much for that, I hadn't seen this one :)

Post Reply

Who is online

Users browsing this forum: No registered users and 1 guest