Hello all, I am researching the effect of the strength of ties and their type (boundary-spanning or not) on creativity. My strength of ties variable consists of 3 items (for now: closeness, frequency and longevity). Sadly my sample is really small (n=20) and the first question in the survey was to fill in up to 10 names with which the respondents socialise. Only 5 were mandatory so the majority filled only 5 and only one filled all 10 spaces. Now the items are Closeness1 to Closeness10 (for example) and when I want to run a reliability test in SPSS it doesn't allow it saying that there are "too few cases (n=1) for the analysis". I have coded missing data as 999 and tried the MISSING=INCLUDE command yet nothing worked. My question thus is am I overseeing something and if so what and how could I solve this?
Or is this maybe not the way to test for reliability in social networks?
Hope this makes sense.
Thank you for your time.
I am not 100% sure but reliability (internal consistency) analysis is used to test to what extent the separate items measure the same concept. I was using reliability on SPSS for example for Depression subscale of DASS scale. Even 3 items for 20 participants with half of the data missing was enough for me to calculate the reliability on SPSS ( I was just simulating your conditions, number of cases and substantial data missing). Maybe apart from the number of cases it is the problem with the fact that your measurement is nominal rather than ordinal, interval, or ratio? But even then you should have some bizarre, e.g. below zero reliability score...
I am not certain I am answering your question, but following some of Burt's work, running a SNA based on 5 alters is adequate. See e.g. http://www.sciencedirect.com/science/article/pii/0378873384900078. Hence, you could just run your analysis based on the 5 names, and exlude the extra 5 that one respondent gave.
Hi, to run reliability test, there should be no missing (replacing missing with 999) does not mean you handled the missing. and have small sample will not give you trusted reliability.
you may need to increase the sample and to replace missing if there is any.
- Badges
- Science topic
- Similar topics
- Mathematical Sciences
- Statistics
- Statistical Software
- SPSS