It's my first time doing this and I'm very confused about how to enter data into SPSS and which data to select to perform the tests on to meet these assumptions of normal distribution and homogeneity before doing anova ...
My research includes 1 experimental group (divided into 2 subgroups based on number of times they performed an activity -- 1 subgroup is for students who did the activity a total of 10-19 times and 1 subgroup is for students who did the activity a total of 20-29 times)
And there is a control group for those who did not do the activity at all
The aim is to see the effect of the number of times the activity was done (0 or 10-19 or 20-29 times) on oral fluency (measured as words per minute by subtracting the pretest score from the post test score)
For example, if they spoke 80 words in the pretest and 100 words in the post test, 100 minus 80 is 20 which is their improvement score
I cannot figure out how my data should be entered into SPSS and which data to select for analysis
Any help would be so greatly appreciated!
Why have you measured the number of times the activity was performed only to lose almost all of the information by collapsing it into three categories? A person who did it 19 times is in group 1, just like a student who did it 10 times, while a student who did it 20 times is in group 2 along with those who did it 29 times.
I would start with a scatterplot of the number of words post-test versus number of times, with a scatterplot smoother (lowess, say) to look at the relationship. Adjusting for number of pre-test words comes a little later.
Your test is called a paired t-test. Full details on all of this should be at this link:
https://www.google.com/search?rlz=1C1CHBF_enUS874US874&ei=Z-MfXv26CcO1tgWLg5GIBA&q=paired+sample+t-test+assumption+tests+data+entry+SPSS&oq=paired+sample+t-test+assumption+tests+data+entry+SPSS&gs_l=psy-ab.3...26929.33825..34894...0.3..0.254.1468.0j10j1....2..0....1..gws-wiz.......0i71j0i333.9u5cxlBd0Wk&ved=0ahUKEwi9ipOdoYfnAhXDmq0KHYtBBEEQ4dUDCAs&uact=5
if something is missing from this discussion just google it like I did. Best, D. Booth
Just to add to David Eugene Booth good answer:
To check for normal distribution and homogeneity of variance you can use Levene's test. If the test comes out significant (rule of thumb, p
Look at the residual diagnostic plots.
Having word **counts** somewhere in the range of 100, the normal distribution is a reasonable assumption.
Formal tests on assumptions are meaningless at best and harmful at worst.
What strikes me more ist the arbitraryly looking classification of your independent variable (number of repetitions of the activity). You could likely learn more if you had not classified this. I'd talk to a local statistician. It might be fruitful to consider a Poisson- or negative binomial regression model, and you should have a look at possible effects of the baseline value (as I think it is likely that people who spoke already many words in the pretest might not [need to or be able to] learn as many new words .. or it is the other way around).
Amir Nakar I'd do something a little different. It really doesn't matter about the 2 groups by themselves since d=sc1-sc2 is the group of interest. Thus the question is how is d distributed?. We don't that that group size so I would suggest a normal probability plot or Kolmogorov-Smirnov test. Assuming those are ok I would test Ho t=[(d-mean(d))/SE(d)] i.e. t0 with the stat package.all of them should be in my link above. Best, D. Booth
Why have you measured the number of times the activity was performed only to lose almost all of the information by collapsing it into three categories? A person who did it 19 times is in group 1, just like a student who did it 10 times, while a student who did it 20 times is in group 2 along with those who did it 29 times.
I would start with a scatterplot of the number of words post-test versus number of times, with a scatterplot smoother (lowess, say) to look at the relationship. Adjusting for number of pre-test words comes a little later.
- Badges
- Science method
- Similar topics
- Mathematics
- Statistics
- Statistical Modeling
- Analyses of Variance
- ANOVA