Questions

If you have any questions or comments, please ask away and I will try my best to get back to you.

16 thoughts on “Questions

  1. Hi Alex.
    I have a quick question.
    Can you tell me how a 2X2 ANOVA can have a significant interaction but no main effect?
    thanks.
    SPSS stuff :s

    1. Hi R,

      Sorry in advance if any of this sounds like stuff you already know.

      Interaction refers to the combined effects of factors (IV’s) on the dependent variable (DV), if an interaction effect is present, the impact of one factor depends on the level of the other factor. To express this using an example, imagine that im running an ANOVA testing the effect of fatigue and gender on attention. Ive found that there is a significant interaction between gender and fatigue on attention, as shown in this diagram;

      Image and video hosting by TinyPic

      This reflects a (Completely made up!) scenario in which, females manage to maintain the attention better as their fatigue increases, however, it quickly drops off and decreases below the level of males attention at around 75% fatigue, however, males attention drops off below female attention at this point. This is a significant interaction because males and females are not reacting in the same way to fatigue in terms of their attention. If this was a non significant interaction, the graph would look like this; (Note that this is not to scale, both genders attention drops by the same amount between fatigue points)

      Image and video hosting by TinyPic

      (Also note, that just because the lines of the interaction graph cross, this doesnt always mean there is an interaction but for this example, lets just assume it is).

      A main effect is the overall effect of one independent variable on the dependent variable. In this case there will be one main effect of gender on attention and another main effect of fatigue on attention. To make this similar to your case, there could be no main effect between gender and attention and no main effect between fatigue and attention. However, there is a significant interaction between the two IV’s, this isn’t necessarily a bad thing, sometimes a significant interaction can tell you more about whats actually happening in the data than main effects can.

      I hope this has answered you question, if your still confused, I found this great resource that might be able to explain this better; http://www4.uwsp.edu/psych/stat/13/anova-2w.htm Possibility five should be able to explain this.

      P.S. Thanks for the first question! : )

  2. Hi Alex!
    Thank you very much for your detailed response. It was very helpful and hope it will aid me in my psychology exam tomorrow!
    A little bit confusing with the graphs going up and then down, but I applied it to my own study in which the IVs were gender and weather (overcast and sunny) on happiness ratings. My graph turned out a simple cross showing an interaction but no main effect! So thank you! You’ve helped me a lot

    Also, I hope you don’t mind answering- I have suggested that likert scales are non parametric due to the data being ordinal. I am aware this is debated about-May I ask what you’re viewpoint is? (A brief simple answer would do :))

    Last question (I promise!), can you briefly explain the Mauchleys test to me, what I’d it used for?
    Thank you ever so much. You have been a wonderful help! (I’ve only just started learning about stats so things get a bit complicated in my head. Haha!).

    Thanks again,
    R.

    1. Mauchleys test for sphericity is testing for homogeneity of variance within a repeated measures design. This test assesses whether the hypothesis that variances of the differences between conditions are equal. If you have a significant value (>.05), there are significant variances of differences. If there are significant variances of difference you need to use The Greenhouse-Geiser correction.

      Hope this helps

  3. Sorry R, Ive just realised that part of my response vanished. Regarding psychometric tests, it all depends on whether your focusing on the single answers to questions or an accumulated score for the test. The accumulated score for a test is more likely to acquire a normal distribution of scores, making it parametric, whereas a single question wont. However, it also depends on what your measuring if your asking psychologists questions about whether they like psychology, the overall score isnt likely to be parametric as you probably wont achieve a normal distribution of scores. Gun to my head though, I would cautiously call them parametric.

    There are tons of literature on this though, well worth a read in my opinion : )

  4. Hi Alex, just stumbled across your blog!

    I really like what you’re doing with it, I want mine to be a similar thing really, something that is personal but that readers would also pick up tips from.

    Can you recommend any other psych blogs?

    1. Hi Hannah, Its great to find someone with similar interests. Off the top of my head Ive found this blog to be quite interesting; http://stevenjoris.wordpress.com

      I also recommend the BPS social psychology website, they have a blog on there which anyone can contribute to (I recently wrote an article for them). If you have twitter, I find there are many more psychologists on there. Im @abbarkerpsych if you are.

  5. Thanks for the suggestions, I’ll take a look now! I love twitter for psychology, my university has an account which a different person curates every week which has been brill so far! @salfordpsych I just followed you, by the way!

  6. Hi alex
    i am also msc student at unıversıty of nottıngham. can you give me some advice for project? i am really confused what to do,and i am interested in clinical and neuropsychology

    1. Hi Seyda,

      Thank you for your comment. Have you recently started your MSc? I always think its a good idea to start thinking about your project as early as possible even if at this stage its just some preliminary ideas. However, the trickier part is thinking these initial ideas. I would recommend reading broadly about a topic that you are interested in and trying to find a “gap” in the research. Once youve found this gap you can then plan your research to investigate this. In this way your research will be novel and further knowledge and understanding.

      Hope this helps
      Alex

  7. hi Alex
    actually i need your help to fined some articles about suicide attempts in adolescents, plus the best conditions for the seclusion rooms in the mental health centers.. could you help me please ?

    Thank you
    Duaa

    1. Hi Duaa,

      I must admit I do not know too much about that area, however, I recommend conducting a literature search to identify relevant studies. To plan your literature search i suggest you map out your research question to identify search terms, I have done something similar in this tutorial https://abbarker.wordpress.com/2013/04/25/meta-synthesis/
      After that I suggest you search for these terms using a boolean search strategy so for example if you were searching for suicide attempts and adolescents, search “suicide attempts” AND (adolescents AND teenagers, etc etc). Its up to you where you search for these but i would recommend EMBASE, MEDLINE, Psychinfo, Web of science and science direct to be thorough.

      Hope that helps
      Alex

    1. Hi Hussein,

      Thanks for the question. A synthesis refers to the bringing together of two topics. A meta-synthesis refers to bringing together two or more “things” from the same topic or area of interest to create a new perspective or viewpoint on the topic or area of interest.
      I hope that helps.
      Alex

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s