I am using IBM SPSS version 21 as a statistical analysis software. My research is about comparing 2 diferent populations. Let's say it's group A and group B.
Each group has variables changing between 2 different timelines : T0 and T1, and these variables are qualitative. Let's say one of these variables is called X.
X is coded either 0 for no, or 1 for yes.
As a qualitative variable, the frequency of X is calculated in percentage by SPSS.
The difference between the frequencies in T0 and T1 is calculated manually by this formula : ( Freq of X(T0) in group A - Freq of X (T1) in group A )/ Freq of X(T0) in group A * 100
So we obtain the variation between these 2 timelines in percentage.
My question is : how do i compare the variation of group A versus the variation of group B between these two timelines (T0 and T1) using SPSS ?
Once you have the variation percentages for both groups, you can compare them using various statistical methods: Independent Samples T-test or Mann-Whitney U Test (If your data violates the assumptions of normality and/or homogeneity of variances).
Howell, D. C. (2013). Statistical Methods for Psychology (8th ed.). Cengage Learning
Field, A. (2013). Discovering Statistics Using IBM SPSS Statistics (4th ed.). Sage Publications.
Thank you so much for your answer.
How do I calculate the variation using SPSS? I can do it manually but I don't know if SPSS can do it.
Hello Mariam Mabrouk. If you had only one group, would you use McNemar's Chi-square? Are you essentially asking how to compare the McNemar Chi-square results for two independent groups? Thank you for clarifying.
Bruce Weaver, McNemar's test is typically used for paired data, where you have two measurements on the same subjects. However, if you have two independent groups and you want to compare them, you might consider using other statistical tests like the chi-square test of independence or Fisher's exact test if your data is sparse.
Jos Feys, The Cochran-Mantel-Haenszel (CMH) test is a statistical test used to analyze the association between two categorical variables while controlling for the effect of a third categorical variable, often called a stratifying variable. It's particularly useful when dealing with matched or paired data across different levels of a third variable.
Agresti, A. (2018). Categorical Data Analysis. John Wiley & Sons.
Dr Amit Kumar, I fear you may not have read my previous reply as carefully as you might have done. I asked IF there was only one group, would Mariam use McNemar's test? And if so, is she interested in comparing the McNemar tests for her two independent groups. I did not suggest that one can or should use McNemar's test when comparing two independent groups. I hope this clarifies things. ;-)
PS- If Mariam's response is yes to both of my questions, using GENLIN in SPSS with generalized estimating equations (GEE) is one way to include both group (between-Ss) and time (repeated measures) in one model. That would require a long file format with two rows per person, one for each time point. It would also require a unique ID variable with the same value on each row for a given subject.
Dr Amit Kumar Jos Feys Bruce Weaver
Thank you all for your answers.
To answer you, I do know that McNemar's test compares dichotomic variables between 2 paired groups, however it is not the test that best describes what I want to achieve. The Cochan test compares categorial variables in independant groups based on a third variable, but in my database, I have 2 groups: group A and group B, along with 2 variables : Variable X0 in timeline T0 (T0 for example is year 2022) and variable X1 in timeline T1 (T1 being the year 2023). And the following document clarifies exactly what I want to do.
I want to compare the last line of the table: the percentage of variation between the two groups to see if in the group A for example, the variation (increase or decrease) was significantly higher than the one in group B.
My research is mainly about comparing the variation of the frequency of hypoglycemia between patients and controls between the years 2022 and 2023. Did Hypoglycemia decrease in frequency? How much percent? and Is this decrease significantly different than the decrease in controls ? Those are the questions I want to answer, but I do not know the test to use because it is a complicated matter and I do not know if it is even usable in SPSS or not.
Mariam Mabrouk , this sentence in your most recent post caught my eye:
My research is mainly about comparing the variation of the frequency of hypoglycemia between patients and controls between the years 2022 and 2023.
Do you have new groups of patients and controls in 2023, or are they the same patients & controls you had in 2022?
Bruce Weaver
They are the same. The prupose of this comparison is to eliminate other factors that could influence hypoglycemia, like in the patients' groups, it can be due to the different circumstances of the follow-up between the years 2022 and 2023. What if patients had less hypoglycemia because their disease simply got less severe for example? That's the purpose of the presence of controls in my study.
Thank you for clarifying, Mariam Mabrouk. In that case, I see that you have two possible ways to account for the correlated nature of the data:
1) Using GENLIN with generalized estimating equations (GEE); or
2) Using GENLINMIXED with occasions clustered within patients.
In both cases, I would likely choose a logit model.
For option 1, there are some relevant examples here:
https://stats.oarc.ucla.edu/spss/library/spss-librarypanel-data-analysis-using-gee/
For option 2, this book has some relevant info:
https://books.google.ca/books/about/Multilevel_Modeling_of_Categorical_Outco.html?id=i9D2Hu2OFW0C&redir_esc=y
Cheers,
Bruce
- Badges
- Science topic
- Similar topics
- Papers