Hi
i have had 117 participants in a study and one of my variables has been their age. I used mean as the cut off score which resulted in 32 participants below the mean (mean=17) and the rest above it.
Most of my participants aged around the mean and I decided to put them into only two groups only because they were not widely distributed in terms of their age and their division into finer groups (like three or four groups rather than two) did not result any different results when analyzed by ANOVA, my data analysis approach.
Is my interpretation on using mean as cut off score plausible?
What could be a good set off score in cases when the participants are not widely distributed?
Does any one have a source to learn about this?
I appreciate your help.
Dear Elahe,
IMHO I don't understand why you should set such a cut-off, and above all why it should be set as the mean age... if a certain cut-off has no rational meaning, it is not only useless but dangerous because it may result in a misclassification bias.
This is my research question.
Which factor (age, gender, or interlocutor) has the greatest influence on EFL students’ WTC orally?
I used Repeated measure anova to answer it.
What kind of variables are they? Continuous, categorical? You should be more explicit. How many repeated measures of what?
Dear Elahe,
IMHO I don't understand why you should set such a cut-off, and above all why it should be set as the mean age... if a certain cut-off has no rational meaning, it is not only useless but dangerous because it may result in a misclassification bias.
Agree with Mehmet and Gnesi in regarding the need to clarifying your exact research question and the data type of each variable.
From the information so far it seems you are dichotomizing the age variable in predicting the outcome (WTC). By doing do you are actually losing some statistical power because you somewhat arbitrarily separate the age into different groups/categories, therefore not taking advantage of all the variance in age in predicting the outcome.
I second the other responses above. Dichotomization is generally not recommended. Just keep Age as a continuous variable in your model and avoid this whole question.
I agree with the above, plus I'm not sure why you are using repeated measures ANOVA (you don't mention repeated assessments).
From your brief description, it seems to me you could instead be conducting a regression analysis with age (continuous), gender, and interlocutor (presumably categorical, dummy coded). You could then report the relative contributoin of each of your potential predictors of WTC score.
I agree with Michael. why not regression to determine the strength of the relationship between WTC and a series of other changing variables (age and gender).
Use mean+-1/2sd formula recode mean+1/2sd equal to 1(high), mean-1/2sd equal to 3(low) and between these two as 2(medium) and then use ANOVA test
- Badges
- Science topic