My study has two independent variables which are continuous.
SEI scores
SHA scores
All participants will complete each scale and receive two separate scores (SEI score and SHA score). Demographic information including age (three levels young age, middle age and older adults) and gender (two levels; female and male) were also collected for each participant.
Is this a 2x3 within-subjects design?
My study wants to test...
1- test for a difference between the mean SEI and SHA scores.
2- test whether there is a difference between gender in the SHA scores.
3- test whether there is a difference between age in SHA scores.
4-test whether there is an interaction between age and gender in SHA scores.
To analyse the above questions would I use...
1- a paired-samples t-test test for a difference between the mean SEI and SHA scores.
2-2x3 unrelated ANOVA to test for the difference of SHA scores between age and gender and the interaction between age and gender groups?
I have no idea if this makes sense as I am new to statistics and research. Would there be any suggestions?
Hi Abigal
I think that:
1- test for a difference between the mean SEI and SHA scores.use t-test for corrlated sample.
2- test whether there is a difference between gender in the SHA scores.use t-test for independent samples.
3- test whether there is a difference between age in SHA scores.use ANOVA one way.
4-test whether there is an interaction between age and gender in SHA scores.use ANOVA two-way with interaction.
Hi Abigail
I suggest following approach. If you know a priori that SEI and SHA are statistically independent, please skip 1 and 2.
1. Check the correlation between SEI and SHA. If the correlation is between 0.4 -- 0.7, perform a multivariate analysis by setting up a single model including both the SEI and SHA as outcome measures. This may be possible in SPSS with a 2-way (AGE by Gender) MANOVA: (https://statistics.laerd.com/spss-tutorials/two-way-manova-using-spss-statistics.php). In R, you may want to consider setting up a linear mixed model: (https://www.r-bloggers.com/multivariate-linear-mixed-models-livin’-la-vida-loca/).
2. If the correlation between SEI and SHA is high > 0.7, you may want to double check if it is worth measuring two variables which may be recording the same underlying construct in different ways.
3. If the correlation is small, setting up 2 separate 2-way ANOVAs is the way to go and I agree with your analysis plan. Some researchers would still argue for a single multivariate model which estimates separate error variance for the two response variables. This can be done in R using "nlme" package.
Hi Abigal
I think that:
1- test for a difference between the mean SEI and SHA scores.use t-test for corrlated sample.
2- test whether there is a difference between gender in the SHA scores.use t-test for independent samples.
3- test whether there is a difference between age in SHA scores.use ANOVA one way.
4-test whether there is an interaction between age and gender in SHA scores.use ANOVA two-way with interaction.
Dear Abigail,
It shouldn't be too difficult. Apparently, I think your SEI and SHA scores are two different domains that are within the same scale.
Since both are continuous, when you find associations between two continuous scores, use correlation analysis. When you find associations between continuous and categorical variables with two levels, use t-test. When you find associations between continuous and categorical variables with three or more levels use ANOVA with a post-hoc tests to observe where the difference is. If your design is before and after, use paired t-tests with the variables you wish to associate them.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA