2

I have a data set for two groups i.e. control group and treatment group. They are two different group from the same socio-economic background. But the treatment group has participated in co-management program with the local forest department. Both of these groups have networked with different local organizations, those who are working for forest management and livelihood improvement. I have around 15 independent variable and 4 dependent variables. They are measured on different types of scale (Likert, rating, binary). But the dependent variable (capacity development) is measured with the 4-point rating scale and we have the total score.

Now, we would like to find out the change in the capacity level of both groups and also what are the important factors that actually contribute to this capacity development process. My first idea was to conduct ANCOVA or MANCOVA to find out the significant mean differences between group, and then run one of the suitable regression models to find out the factors.

Does it sound good? Can anyone help me to find out the best one?

KH. Kabir
  • 21
  • 2
  • Welcome to Cross Validated. You mention 4 dependent variables, but from the rest of your question it looks like you are only interested in one of them (capacity development). Is this correct? In order to increase the chances of getting an answer, you need to show us your data. Start from including a subsample, describing what the independent and dependent variables are and making some plots/summaries. For example, since it looks like all your variables are categorical, heat maps, mosaic plots, contingency tables can be appropriate. continued... – DeltaIV Mar 05 '17 at 11:57
  • ...continued Regarding methods, I don't see how ANOVA could be applied if the dependent variable is an ordinal one. This could be due to my ignorance, of course, but I think that nonparametric tests such as Mann-Whitney or Kruskal-Wallis, and regression methods suited for ordinal variables, such as for example ordered logistic regression, would be more appropriate to your problem. – DeltaIV Mar 05 '17 at 12:10
  • There is four types of capacity i.e. capacity to navigate complexity, the capacity to learn and reflect, capacity for collaboration and capacity to engage in the political process. First of all, I would like to find out the change in this four level between this two groups. Data was collected with a 4-point rating scale. Every capacity aspects had several questions and thus we get the total score for every capacity. The other variables are some demographic variables and variables related to social capital and social networking. – KH. Kabir Mar 05 '17 at 12:11
  • If I understand correctly, each capacity is measured by asking some questions (how many?): each question has 4 possible answers (something like *Strongly disagree*, *Disagree*, *Agree*, *Strongly Agree*). You compute the total score for a capacity by summing the points for each question. Right? Now I think I understand how you could use ANOVA. But I still believe that you should include a sample of your data in your question, if you want to get an answer. – DeltaIV Mar 05 '17 at 12:19
  • Yes, you get it correctly and the scale was Never, sometimes, most of the time and always. I have total 18 questions for all types of capacity. For the first type of capacity, I have 4 questions, four 2nd-5, for 3rd-5 and for the final one I have 4. I have done the reliability test for all questions and get a value of .75. I do not know how to add data or send you the some sample of data set.!! – KH. Kabir Mar 05 '17 at 12:25
  • You cannot send me data and you don't need to. You could include a table, as done in [this question](http://stats.stackexchange.com/questions/265543/survival-analysis-age-censoring-bias-in-independent-categorical-variables). If you use R, you could include the results of commands such as `str(mydata)` and `head(mydata)` in your question, with `mydata` being the data frame holding your data. You could include[mosaic plots](http://stats.stackexchange.com/questions/153446/r-how-to-interpret-mosaic-and-association-plots) showing the relationships among your variables. If you don't know how to[1/2] – DeltaIV Mar 05 '17 at 13:44
  • [2/2] use the site features in order to include images, code, etc. in your question, see [here](http://stats.stackexchange.com/help/formatting) and in general study the Help center. – DeltaIV Mar 05 '17 at 13:46
  • Sorry I am not using R. I am using spss. Then how can I share all the things with you? – KH. Kabir Mar 06 '17 at 23:37

0 Answers0