SHOW:
|
|
- or go back to the newest paste.
1 | # Date file was retrieved on March 14, 2014, from | |
2 | # http://hkupop.hku.hk/chinese/popexpress/ce2012/dataset/tp1403013.sav | |
3 | ||
4 | ############################ | |
5 | # To load in the data file # | |
6 | ############################ | |
7 | ||
8 | # Install the "foreign" package, to read in SPSS files | |
9 | install.packages("foreign") | |
10 | ||
11 | # Load in the "foreign" package | |
12 | library("foreign") | |
13 | ||
14 | # Read in the data file | |
15 | cy201403 <- read.spss("tp1403013.sav", to.data.frame=T) | |
16 | ||
17 | # Check if file has been read in correctly | |
18 | head(cy201403) | |
19 | ||
20 | ||
21 | ###################### | |
22 | # Basic descriptions # | |
23 | ###################### | |
24 | ||
25 | # Number of subjects = 1017 | |
26 | dim(cy201403)[1] | |
27 | ||
28 | # Number of raters = 998 | |
29 | sum(table(cy201403$CE_rating)) | |
30 | ||
31 | ||
32 | ||
33 | ################## | |
34 | # On raw ratings # | |
35 | ################## | |
36 | ||
37 | # A quick scan of the distribution of the raw ratings | |
38 | table(cy201403$CE_rating) | |
39 | ||
40 | # Mean of raw ratings = 47.4008 | |
41 | mean(cy201403$CE_rating, na.rm=T) | |
42 | ||
43 | # Median of raw ratings = 50 | |
44 | median(cy201403$CE_rating, na.rm=T) | |
45 | ||
46 | # Histogram of the raw ratings | |
47 | hist(cy201403$CE_rating, main="CY Leung's rating (raw), \nMarch 2013", | |
48 | xlab="Raw Scores", ylab="Frequency", right=F, axes=F, | |
49 | breaks=seq(0, 100, 10), col="lightgray") | |
50 | axis(1, at=seq(0, 100, 10)) | |
51 | axis(2, at=seq(0, 300, 50)) | |
52 | ||
53 | # Indicate mean on plot | |
54 | abline(v=mean(cy201403$CE_rating, na.rm=T), col="red", lwd=4) | |
55 | ||
56 | # Indicate median on plot | |
57 | abline(v=median(cy201403$CE_rating, na.rm=T), col="blue", lwd=4) | |
58 | ||
59 | # Add labels | |
60 | text(mean(cy201403$CE_rating, na.rm=T), 200, "Mean", col="red", lwd=4, pos=2) | |
61 | text(median(cy201403$CE_rating, na.rm=T), 200, "Median", col="blue", lwd=4, pos=4) | |
62 | ||
63 | ||
64 | ###################### | |
65 | # On weighed ratings # | |
66 | ###################### | |
67 | ||
68 | # Calculate weighed ratings | |
69 | cy201403$weighedRating <- cy201403$CE_rating * cy201403$weight | |
70 | ||
71 | # Mean of raw ratings = 47.48411 | |
72 | mean(cy201403$weighedRating, na.rm=T) | |
73 | ||
74 | # Median of raw ratings = 41.9557 | |
75 | median(cy201403$weighedRating, na.rm=T) | |
76 | ||
77 | # Histrogram of the weighed ratings | |
78 | hist(cy201403$weighedRating, main="CY Leung's rating (weighed), \nMarch 2013", | |
79 | xlab="Weighed Scores", ylab="Frequency", right=F, axes=F, | |
80 | breaks=seq(0, 500, 10), col="lightgray") | |
81 | axis(1, at=seq(0, 500, 50)) | |
82 | axis(2, at=seq(0, 140, 20)) | |
83 | ||
84 | # Indicate mean on plot | |
85 | abline(v=mean(cy201403$weighedRating, na.rm=T), col="red", lwd=4) | |
86 | ||
87 | # Indicate median on plot | |
88 | abline(v=median(cy201403$weighedRating, na.rm=T), col="blue", lwd=4) | |
89 | ||
90 | # Add labels | |
91 | - | text(mean(cy201403$weighedRating, na.rm=T), 200, "Mean", col="red", lwd=4, pos=2) |
91 | + | text(mean(cy201403$weighedRating, na.rm=T), 100, "Mean", col="red", lwd=4, pos=4) |
92 | - | text(median(cy201403$weighedRating, na.rm=T), 200, "Median", col="blue", lwd=4, pos=4) |
92 | + | text(median(cy201403$weighedRating, na.rm=T), 100, "Median", col="blue", lwd=4, pos=2) |