Imputation methods (Imputointimenetelmät), kevät 2017
Imputointimenetelmät/Imputation Methods, spring 2017
Lecturer
Targets and general points on imputation _ Yleistä ja tavoitteet
This course gives a rather deep introduction to imputation methods, that is, when missing or deficient values are replaced with 'as good as possible proxies.' The course material is mainly in English but the Finnish can be used if the attendees wish. The course consists of seven events, three hours each, so that they include both the theory (basics of surveys, imputation methods) and applications with very real data with missing values (I think that it does not matter much if not participating in one event but the first two ones are important to attend). A lot of methods are tried and some are observed more successful than some others. The report of the attempted methods is the basis to get the credits from the course. We use mainly SAS but some methods of SPSS are also tested (if someone is very good in R, it is possible to use but not SPSS tools of course). To understand something about SAS is useful but all support is given during the training, that is, very ready-made SAS codes are given. Naturally, basic knowledge in statistical methods is required but many things such as statistical distribution measures, linear regression and logistic/probit regression are explained and thus learned well enough to be used in imputation. The material will be sent by email for the registrants, but the main training data file will be downloaded here.
The course is going to be held in Period 4 on Mondays and/or Thursdays from 16.15 til 18.35/18.45 depending on the possible break or not. The first time will be Monday 20th March, the second Thursday 23th March, the third Monday 27th March, the fourth Thursday 30th March, and the last three ones on Thursdays (6 April, 20 April and 27 April), all in the SSKH IT class.
I hope that you know basic statistical methodology, including something about statistical models. If you have participated in survey methodology course, that is fine, but it is not necessary. It is however good to introduce yourself to some basics in survey methodology, such as missingness, auxiliary variables and item response (rate)..
The number of the credits is five but if more work is done, 1 or 2 more credits are quite esialy possible, even until eight but it requires much extra work. The introduction material in Finnish with English dictionary in its annex can be found from my free e-book: http://bookboon.com/fi/surveymetodiikka-ebook.
Welcome
Tämä kurssi on imputointimenetelmistä, jossa mennään oikeaan käytäntöön. Kyse on siis puuttuvan tai muuten huonon tiedon paikkaamisesta hyvällä korvikearvolla tai useilla. Sellainen tilanne tulee vastaan melkein kaiken tosielämän datan kanssa, valitettavasti. Joskus kiusa voi olla liiankin suuri eikä aina ole varmaa vastausta siihen mitä olisi hyvä tehdä (vaikka jonkin ratkaisun saa aina aikaan mutta laadusta ei ole takeita). Surveymetodiikan kurssilla emme kovinkaan paljon ehtineet tätä osaa käsitellä vaikka kirjassani (ks. kotisivultani nettiosoite) on melko laaja selostus metodologiasta. Siis kannattaa katsoa ennen ja kurssin aikanakin sitä. Teen varsinaisen kurssimateriaalin englanniksi mutta jotain puhun suomeksi. Kurssi tulee melko pitkälle suoritetuksi tekemällä harjoitukset, alkaen ihan yksinkertaisista ja jatkaen kohti monimutkaisempia jotka saattavat käytännön elämässä olla parempia. Joskus yksinkertainen metodi on riittävä. SAS-koodit siis annetaan mutta SPSS:n melko automaattista imputointia verrataan omiin tuloksiin kurssin loppupuolella. Älä pelkää jollei sinulla ole kokemusta SAS:sta mutta kirjaani kannattaa johdannoksi katsoa.
Kurssi kelpaa varmaan mihin tahansa tilastotieteessä tai muussa metodologiassa ml. jatko-opintoihin.
Tervetuloa.
Material
The lecture material thus will be sent by email to each registrant (before each event) but here you can find something as well.
for those who have no idea about the software
The training data set as the SAS file. Please copy this to your own folder.
Lecture material pdf