Handling missing values

[Vera Schmidt]

Hi guys,
I have a big problem with missing values in my data set.

Briefly: cross-sectional data, reason for the missing values is especially the irregulary interview turnus.

Missing	Total
436	4357
436	4,357
197	4,357
197	4,357
2,327	4,357
0	4,357
58	4,357
441	4,357
0	4,357
585	4,357
2,1	4,357
0	4,357
10	4,357
189	4,357
198	4,357
228	4,357
226	4,357
225	4,357
229	4,357
2	4,357
62	4,357
31	4,357
0	4,357
0	4,357
192	4,357
669	4,357
0	4,357
1,536	4,357
488	4,357
1,465	4,357
488	4,357
1,465	4,357
488	4,357
1,465	4,357
488	4,357
1,465	4,357
488	4,357
1,465	4,357
488	4,357
1,465	4,357
516	4,357
1,483	4,357
488	4,357
1,465	4,357
32	4,357
462	4,357
459	4,357
456	4,357
793	4,357
1,482	4,357
798	4,357
1,492	4,357
461	4,357
460	4,357
406	4,357
785	4,357
952	4,357
463	4,357
462	4,357
12	4,357
1,191	4,357
1,191	4,357
1,191	4,357
1,191	4,357
1,191	4,357
522	4,357
522	4,357
522	4,357
522	4,357
522	4,357
2,805	4,357
2,805	4,357
2,838	4,357
2,807	4,357
2,958	4,357
2,959	4,357
2,961	4,357
2,96	4,357
2,007	4,357
2,006	4,357
2,009	4,357
2,038	4,357
2,035	4,357
848	4,357
1,498	4,357
1,968	4,357
1,968	4,357
1,968	4,357
1,974	4,357
2,54	4,357
2,171	4,357
1,121	4,357
1,106	4,357
1,122	4,357
2,301	4,357
2,298	4,357
2,298	4,357
2,298	4,357
2,301	4,357
2,304	4,357
2,301	4,357
1,899	4,357
2,692	4,357
3,759	4,357
3,762	4,357
3,766	4,357
3,758	4,357
3,758	4,357
3,758	4,357
3,76	4,357
3,76	4,357
3,76	4,357
3,76	4,357
1,956	4,357
2,548	4,357
1,973	4,357
1,986	4,357
3,572	4,357
2,919	4,357
2,355	4,357
2,355	4,357
2,382	4,357
2,911	4,357
1,899	4,357
29	4,357
29	4,357
29	4,357
29	4,357
3,759	4,357
3,759	4,357
809	4,357
0	4,357
0	4,357
0	4,357
0	4,357

I decided to drop first all missing values with more than 50 percent missing. After that I dropped all missing values. Finally I got 120 observations in total.

What do you think: Can I do so? I could not find any rule of thumb in dropping missing values with more than 50 percent missing.

You can see that I have a lot of variables. For this reason I going to do a factor analysis in the next step. But in my opinion I need to carry missing values first.

Looking forward to hear from you! Many thanks in advance.

Vera

[Leonardo Guizzetti]

You may not like the answer but your question is too vague to give specific advice. I suggest that you spend some time learning about missing data and imputation (-help mi intro-) as as a starting point and then refer to a good textbook on the topic.

[George Ford]

why are they missing?