Advertisement
If you have a new account but are having problems posting or verifying your account, please email us on hello@boards.ie for help. Thanks :)
Hello all! Please ensure that you are posting a new thread or question in the appropriate forum. The Feedback forum is overwhelmed with questions that are having to be moved elsewhere. If you need help to verify your account contact hello@boards.ie

Analyses of Means (Total Beginner to Stats here!)

Options
  • 02-11-2011 1:40pm
    #1
    yellow hen
    Registered Users Posts: 4,030 ✭✭✭


    hi,

    I am a total beginner to stats but need to do with a bit of analyis on two datasets over the course of 4 years. I am using software to do the maths part as I havent a breeze :( For 3 of the 4 years, the hyothesis that one group perform better than the other is supported. However for the fourth year there the hypothesis is not supported. I need to look at the data now and see why this wasnt the case for that year. Is there something I should be looking at?

    Sorry, that i cant show the data :(


Comments

  • Options
    #2
    mawk
    Registered Users Posts: 2,738 ✭✭✭mawk


    So the means of the two data sets were significantly different, but consistent to themselves for three years but on the 4th year they become homogenous?

    Then you have to look at eternal factors and quite possibly the age of the participants


  • Options
    #3
    LeixlipRed
    Closed Accounts Posts: 6,081 ✭✭✭LeixlipRed


    Need to give us more information really. Could be a variety of things. What variables are recorded?


  • Options
    #4
    yellow hen
    Registered Users Posts: 4,030 ✭✭✭yellow hen


    LeixlipRed wrote: »
    Need to give us more information really. Could be a variety of things. What variables are recorded?

    thank you both.

    LeixlipRed, the variables are diploma GPA, diploma subject choices and grades and then i have their Masters subject choices and grades. I want to see if there is a correlation between subject choices at diploma level and masters level. Eg. If a student took French Level1 at Dip level, I want show that this would directly impact on their performance at French Level2 during their masters. Analayses of means upheld this hypothesis for every year with the exception of one. I'm stumped now!


  • Options
    #5
    LeixlipRed
    Closed Accounts Posts: 6,081 ✭✭✭LeixlipRed


    Do you have the data for the other variables? Run some t-tests there as well and see what comes up. If I had to guess perhaps the GPA of the first 3 groups is better than the 4th group to a significant level? Could be a million things though. Probably open ended, multiple answers. If you could post the data we could have a look.


  • Options
    #6
    Ostrom
    Registered Users Posts: 3,483 ✭✭✭Ostrom


    yellow hen wrote: »
    thank you both.

    LeixlipRed, the variables are diploma GPA, diploma subject choices and grades and then i have their Masters subject choices and grades. I want to see if there is a correlation between subject choices at diploma level and masters level. Eg. If a student took French Level1 at Dip level, I want show that this would directly impact on their performance at French Level2 during their masters. Analayses of means upheld this hypothesis for every year with the exception of one. I'm stumped now!

    If you're trying to explain why the test was non-significant, it might be worth displaying descriptive stats, or some visual aid for all years. The test alone is pretty useless at this (all it really tells you is the probability of difference due to sample error), whereas a boxplot with mean and variance displayed will show your reader the size of the differences across years. Smaller differences in mean, reduced variance, or a reduction in sample size could all account for a non-significant fourth year result.

    You can also calculate effect sizes for comparison of means, which might be as useful here as the t-test, since you want to know how 'big' the difference is (the t-test wont accurately tell you this).

    Apologies if this is completely patronising (only because I made a similar mistake myself at the start - and you state you are a beginner) but make sure you're choosing the correct test (independent / dependent observations).

    Why is it problematic if the hypothesis is not supported? Non-significant results are important too!


  • Advertisement
Advertisement