living, thinking and writing
|
|
| PRACTICE living, thinking and writing |
|
|
Sunday, January 12, 2003 Backfire of a poor survey design. I've been sitting two days with my SPSS window open on my laptop, checking the survey data I collected and Chen Hao entered last year. It is 433 copies of questionnaire, each with 404 variables. I must salute Chen Hao for his unbelievable patience and apologize for my unthoughtful design. That's really too much, and the worse thing is we did not agree on how to code and imput data at the beginning, which cost me unnecessary work (and enormously) now to correct it. Can't believe my poor correlation result (.60 between knowledge scores and centrality values, but only significant for countries on center layer of world system), I decide to take at a look at the raw data--which is in dusty, heavy and big box that I haven't touched for one year. The research project is a long ordeal, I started a proposal on the spring of 2001 and collected data on the fall of the same year. On the second year, God sent me an angel, Chenhao, to help me with the data entry. We got some rough results and put together an abstract, which is later accepted by IAMCR of 2002 in Barcelona, but unfortunately I could not finish the full paper since I was too much engaged in teaching and some other unpleasant stuff, partly also because I was not very happy with the result. I decided to finish it during this break when I was informed that the department set a higher standard on preliminary papers. I chose this unfinished paper instead of another four or five conference papers (and they are finished!), because I believe this is a good topic and the research approach is integreted, also because I wanted to break a Chinese professors' spell that Chinese students can only study China issues. Not even half of the data being corrected, I've already raised my correlation rate by 10%. It is a pleasant output, but the cost is too high. A bad researhc design (I should have the respondents imputing their answers on computer), bad-wording questions (which results on overlapping answers) and lack of consistency on data entr not only greatly distorted the results at first place, and also takes us too much extra time to fix it. I'm not sure if I want to do any survey again, but this experience gave me an unforgettable lesson, and I will be very careful next time when I start. Well, have to go back working on the other 200 or so copies of questionnaires. posted by lmeimei @2:56 PM| permanent link| | |
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||
