What to do if both dependent variable and independent variable are Likert Scale type of variable?

[Du Viet Nguyen]

Hello, my second post here.
I'm working on my thesis. I created questionnaire based on validated scale and received roughly 50 respondents. Both my dependent variable and independent variables are represented by 7 point-Likert-scale type of questions. Independent variables are categorized into two groups of factor (factors). One group consists of 5 questions. the other consists of 4 questions.
What is the best way for me to analyse my data? My Analysis of Quantitative Data class seems to be not enough for me at this point.
Thanks so much!

[Andrew Musau]

The estimation method typically depends on the nature of your outcome variable. Therefore, see either

help ologit

or

help oprobit

For a 7-point scale, linear regression may also be acceptable if one is willing to assume that the categories are spaced equidistant on the latent scale.

help regress

[Richard Williams]

With only 50 cases you can’t be too picky. I would either use use regress or ologit and would probably treat the independent variables as continuous.

it sounds like you have multiple measures of your concepts, so you may be able to create scales out of them.

[Du Viet Nguyen]

With only 50 cases you can’t be too picky. I would either use use regress or ologit and would probably treat the independent variables as continuous.

it sounds like you have multiple measures of your concepts, so you may be able to create scales out of them.

Thank you for your respond. Could you give me more detail on creating scales out of independent variables?

[Richard Williams]

Some basic scaling techniques are described at

https://www3.nd.edu/~rwilliam/stats2/l23.pdf

[Du Viet Nguyen]

Some basic scaling techniques are described at

https://www3.nd.edu/~rwilliam/stats2/l23.pdf

Thank you for your guidance. The document was a joy to read.

[Du Viet Nguyen]

With only 50 cases you can’t be too picky. I would either use use regress or ologit and would probably treat the independent variables as continuous.

it sounds like you have multiple measures of your concepts, so you may be able to create scales out of them.

Dear Prof. Williams,
I'm currently working on ologit, yet i have difficulty in testing for model fit, as guided by this article. i always encounter a "invalid name" error when I execute the ologitgof command with my independent variables. Do you have any ideas what might be wrong? is it because the independent variables i'm testing is likert-scale type of data? I also told to use The Akaike information criterion (AIC) to test for model to refine my independent variable list. Is it the same process?
thank you very much,

[Richard Williams]

Do you have the programs installed? If so, please show your code and output so we can see exactly what problems you are encountering.

[Du Viet Nguyen]

Do you have the programs installed? If so, please show your code and output so we can see exactly what problems you are encountering.

Dear Prof. Williams,
This is what the output look like:

. ologitgof interaction_IN_school interaction_local_gen, tableHL tablePR
_p2.599999904632568 invalid name
r(198);

By the way, I also have another problem with working with Ordered logistic regression model. My framework consist of two elements of study abroad experience: Interaction with local (1) and cross-cultural exposure (2), that could affect the response variable (Intercultural proficiency). Each element consists of 4 variables. I am wondering, whether my OLR model should consist of both elements, or should I run 2 separate models for 2 elements independently. The reason is that, if both elements are included in a single OLR model, variables with have different coefficient values and some are not statistical significant. Meanwhile, if 2 separate models are tested, all covariates have significant p value.
Thank you for your time. I am looking forward to your reply.
Best regards,