SPSS Statistics

 View Only
Expand all | Collapse all

Chi-Square Test of Independence or Goodness of Fit?

  • 1.  Chi-Square Test of Independence or Goodness of Fit?

    Posted Fri February 02, 2024 12:45 PM

    Hello,

    I'd like to ask some advice around the appropriate statistical test to use. I'm currently conducting research around the impact of a restraint minimisation model on the frequency of physical restraint. My hypotheses are: 

    1. The frequency of physical restraint techniques will be significantly lower in the three months following the implemented changes, in comparison to the three months before.
    2. The duration of physical restraint incidents will significantly reduce in the three months following the implemented changes, in comparison to the three months before.
    3. The frequency of injuries will be significantly lower in the three months following the implemented changes, in comparison to the three months before.

    This study has one categorial independent variable with 2 levels (pre-implementation; post-implementation). There are 2 categorical dependent variables, each with 2 levels: the frequency of restraint (restraint was applied; restraint was not applied) and the frequency of injuries (injury was observed; injury was not observed). There is a third continuous dependent variable, which is the duration of restraint in minutes.

    For hypotheses 1 and 3, I thought that the most appropriate test is Chi-Square Test of Independence as I am hypothesising that the variables are dependable on each other. Whilst for hypothesis 2, I reckon that an Independent T-Test would be appropriate to compare duration of restraint.

    Looking forward to any advice!

    Many thanks,

    Susan



    ------------------------------
    Susan Hoang
    ------------------------------


  • 2.  RE: Chi-Square Test of Independence or Goodness of Fit?

    Posted Wed February 14, 2024 02:07 AM

    Hi Susan

    If you aim to test the independence of two attributes, then you can use Chi Square test of independence, provided the data should be categorical or the the frequency (or number) related to the case. However, if you want to test whether there is any significant different between (averages) two continuous variables, then you can apply the independent t test for the case. Also, incase of paired data, for example, if we have the scores obtained before and after a procedure and if we want to check whether the procedure is effective or not, then in that case, we can use the Paired t test.



    ------------------------------
    Bindu Krishnan
    Senior Statistician
    IBM SPSS Statistics
    ------------------------------