如何在更大的数据集中为子组的子组生成汇总统计数据（类似于 psych::describeBy）？

Question

New to R (for biostats) here! I have a huge dataset, and am using describe() and describeBy() from the psych package. But I'm also trying to find a way to do basic stats for subgroups within subgroups.

For example, my dataset is about membership within a club, which has Chinese and Indian members. Other variables include gender, age, height, weight, BMI, etcetera.

I have figured out psych::describeBy to look at means and standard deviation for subgroups defined by one variable, eg ethnicity, but I can't figure out how to narrow this down further so that I generate a summary only for Chinese male members.

I tried redefining using the subset() function, and then running describeBy again, eg

chinese <- subset(maindata, chinese=1)
describeBy(chinese, male=1)

But this didn't work, and the results were the same as describeBy(maindata,chinese=1) , rather than the Chinese male subset.

I hope that makes sense.

The only other solution I can think of is to breakdown my main dataset into smaller ones in MS Excel and re-uploading each separately (eg Chinese.xls, Indian.xls), or to create a new variable with defined by a combination of ethnicity-gender, eg Chinesemale=1, Chinesefemale=2, Indianmale=3, Indianfemale=4.

I more or less will need to analyse by these subgroups of subgroups for t-tests and Fisher's exact, so any good package recommendations that would help address these would be appreciated!

Thanks in advance!!

Answer 1

Sample Data

df1 <- data.frame(subject = c(1, 2, 3, 4, 5),
                  chinese = c(1, 1, 1, 0, 0),
                  male = c(1, 0, 1, 0, 1),
                  value = c(45, 23, 84, 11, 12))

Two changes in syntax from your code:

• double equal sign in subset() . You want to keep rows where chinese is equal to 1. You would use a single equal sign if you were assigning a value of 1 to a parameter called chinese .
• In describeBy() , the group parameter gives you different summary statistics for each category in that column (as shown below). You can't use it to subset for male=1.

chinese <- subset(df1, chinese == 1)

describeBy(chinese, group = "male")

 Descriptive statistics by group 
group: 0
        vars n mean sd median trimmed mad min max range skew kurtosis se
subject    1 1    2 NA      2       2   0   2   2     0   NA       NA NA
chinese    2 1    1 NA      1       1   0   1   1     0   NA       NA NA
male       3 1    0 NA      0       0   0   0   0     0   NA       NA NA
value      4 1   23 NA     23      23   0  23  23     0   NA       NA NA
------------------------------------------------------------------------------------------------------------------------------------- 
group: 1
        vars n mean    sd median trimmed   mad min max range skew kurtosis   se
subject    1 2  2.0  1.41    2.0     2.0  1.48   1   3     2    0    -2.75  1.0
chinese    2 2  1.0  0.00    1.0     1.0  0.00   1   1     0  NaN      NaN  0.0
male       3 2  1.0  0.00    1.0     1.0  0.00   1   1     0  NaN      NaN  0.0
value      4 2 64.5 27.58   64.5    64.5 28.91  45  84    39    0    -2.75 19.5

If you only want to see the summary stats for males in the sample, you could add & male == 1 to the subset() :

chinese <- subset(df1, chinese == 1 & male == 1)

describeBy(chinese)

        vars n mean    sd median trimmed   mad min max range skew kurtosis   se
subject    1 2  2.0  1.41    2.0     2.0  1.48   1   3     2    0    -2.75  1.0
chinese    2 2  1.0  0.00    1.0     1.0  0.00   1   1     0  NaN      NaN  0.0
male       3 2  1.0  0.00    1.0     1.0  0.00   1   1     0  NaN      NaN  0.0
value      4 2 64.5 27.58   64.5    64.5 28.91  45  84    39    0    -2.75 19.5