Significance testing in publication trend across research areas over time

[Stephan Krayter]

Hello,

i am struggling with a statistical problem. I am working with publication data from the Web of Science. For a specific search strategy I created a variable with the numbers of publications for the TOP 10 strongest research areas and another category for the remaining research areas, so 11 categories in this variable. Then i have a time variable with 6 categories (1960´s, 1970´s up to 2010´s).

In the data I can see, that there is an increasing trend in general publication. For example there are 708 publications across all research areas in the 60s, 1081 in the 70s and so on. Then I have trends for single research areas. For example 38 publications in Psychology in the 60´s, 69 publications in the 70´s and so on. I want to verify if a trend for a single research area is significant (for example some trend is stronger and some trend is weaker then the overall increase).

I know that I should use some kind of significance testing or trend analysis, but I dont know which test to choose nor to calculate them in STATA.

I hope someone can help me out. I would appreciate it very much.

Kind regards,
Stephan

[Bruce Weaver]

Perhaps some of the examples on this UCLA page will be helpful.

[Phil Bromiley]

If you have an area numerical variable and a numerical variable for time, you could do
regress pub i.area##c.time

Then run

regress,coefl

and it will tell you how to refer to the coefficients to do whatever tests you want.

[Stephan Krayter]

I would like to thank you for your help. I think, i could solve the problem with your help.