The syntax for this command is:. The CCHS dataset does not contain any string variable. We will then convert that variable back to a numerical format. Now, when we destring, we are replacing the string variable by its numerical counterpart. How you choose to do this in your own dataset depends on how you plan to use the variables.
Will you still have any use for the string variable? If so generate a new one when you destring. Do you just want that variable to not be in string format?
Then replace it with the new one. Outliers deserve their own section because there is often confusion as to what exactly constitutes an outlier. An outlier is NOT an observation with an unusual but possible value for a variable [12] ; rare events do occur. The outliers you should be concerned about are the ones that come from coding error.
How do you tell which is which? Common sense goes a long way here. First, look at your data using the data editor browse. Outliers tend to jump at you. If you have a small dataset, you can also tabulate each of your variables:. Tabulating a variable will give you a list of all the possible values that variable takes in the dataset.
Outliers will be the extreme values. Look at the order of magnitude. Are these values believable? If the dataset is very big, however, it may not be practical to stare at all the values a variable can take. In fact, Stata will not tabulate if there are too many different values. In the CCHS dataset, caseid is the individual id, while hwtghtm is the height in meters.
The graph tells us there are no outliers in this dataset:. Another way to look for outliers is to summarize the observations for a variable, using the detailed option:.
Clearly, there are no outliers. Is it plausible that there really was a 5. Look at the order of magnitude by which this observation would differ from the second largest.
What should you do with such an observation? There are a number of solutions but none is perfect:. Suppose we wish to analyze 2 subsets of the hs1 data separately, males and females. We subset the dataset twice, reloading between, and save the datasets for use later. Suppose that our data file had many, many variables, say variables, but we only care about just a handful of them, id , female , read and write.
We can subset our dataset to keep just those variables as shown below. In survey data, missing values may mean that the surveyor did not ask the question, that the respondent did not answer the question, or that the data are truly missing.
Some datasets have these three cases coded differently; others lump them together. For numeric data, keep in mind that missing data are not the same as a value of zero. This may seem obvious, but I have had many students nonchalantly say "oh, so we can just replace those with zeros I assume you mean the seven omitted observations have no missing values for any of the variables that are used in the model, and perhaps you confirmed that with Code:.
Comment Post Cancel. Hi William, below is my output! Thank you for the output presented with CODE delimiters - it is very readable and reveals a lot about your problem, First, we need to clarify terminology.
To Stata, what Excel would call a "row" in your data is what Stata calls an "observation". So you have far more than or observations in your data. Your dataset appears to consist of observations from authorities.
Each authority provides up to 5 years of quarterly data, so there are up to 20 observations from each authority. Your tab output shows that Stata omitted observations from 60 different authorities from your regression.
0コメント