6.8.1. Data Preparation
UNISTAT does not use a special data format requiring a major re-arrangement of your data. All you need to do is to enter your data in a standard spreadsheet format where each column contains a different statistic, i.e. odds ratios in one column, probabilities in another column, etc. In most cases, UNISTAT will not require you to make any prior calculations before entering your data.
For instance, if the analysis is based on a 2 x 2 table, you do not need to calculate each cell frequency to perform a meta analysis. UNISTAT will accept data in terms of sums and ratios of frequencies reported by individual studies.
|
Study Name |
a |
b |
c |
d |
a+b |
c+d |
a/[a+b] |
c/[c+d] |
|
Study 1 |
19 |
18 |
164 |
157 |
||||
|
Study 2 |
11 |
13 |
73 |
41 |
||||
|
Study 3 |
17 |
24 |
82 |
57 |
||||
|
Study 4 |
53 |
54 |
0.4313 |
0.2517 |
For an unpaired two-sample analysis with continuous data there are many more types of data entry possibilities.
|
Study Name |
No A |
No B |
Mean A |
Mean B |
Mean Diff |
Std Dev A |
Std Dev B |
Std Err Pooled |
|
S 1 |
43 |
42 |
43.2 |
47.6 |
|
2.43 |
7.89 |
|
|
S 2 |
25 |
42 |
73.7 |
64.2 |
|
|
||
|
S 3 |
37 |
52 |
23.4 |
25.5 |
|
|
||
|
S 4 |
53 |
32 |
5.31 |
6.37 |
|
|
||
|
S 5 |
41 |
32 |
5.43 |
|
||||
|
S 6 |
74 |
96 |
6.67 |
7.81 |
|
|
||
|
S 7 |
87 |
76 |
-4.56 |
|
||||
|
S 8 |
23 |
21 |
7.65 |
|
||||
|
S 9 |
32 |
34 |
2.34 |
|
||||
|
S 10 |
13 |
16 |
23.6 |
4.57 |
(table continued)
|
Var A |
Var B |
t-value |
p-value |
Tails |
Lower Bound |
Upper Bound |
Conf Level |
|
|
S 1 |
|
|||||||
|
S 2 |
|
|||||||
|
S 3 |
12.6 |
32.1 |
|
|||||
|
S 4 |
3.456 |
|
||||||
|
S 5 |
4.567 |
|
||||||
|
S 6 |
0.0756 |
1 |
|
|||||
|
S 7 |
0.1034 |
2 |
|
|||||
|
S 8 |
9.02 |
0.95 |
||||||
|
S 9 |
1.23 |
0.90 |
||||||
|
S 10 |
|
The above tables demonstrate only some of the data types supported by UNISTAT. For a full list of data options see the next section.
You can also mix different types of data in an analysis. For instance, in below hypothetical example, the data from unpaired two-sample analyses are entered in the first five columns and data from 2 x 2 tables entered in the last five columns. UNISTAT’s variable selection interface allows you to define the first nine studies as unpaired samples and studies from S 10 to S 13 as 2 x 2 tables and run a meta analysis on all data.
|
Study Name |
No A |
No B |
Mean Diff |
Std Err Pooled |
Study Name |
A |
B |
C |
D |
|
S 1 |
43 |
42 |
0.02 |
7.68 |
S 10 |
14 |
65 |
41 |
89 |
|
S 2 |
25 |
42 |
-2.32 |
6.98 |
S 11 |
23 |
79 |
37 |
67 |
|
S 3 |
37 |
52 |
-0.68 |
3.40 |
S 12 |
37 |
87 |
17 |
93 |
|
S 4 |
53 |
32 |
0.73 |
5.33 |
S 13 |
41 |
73 |
24 |
77 |
|
S 5 |
41 |
32 |
5.43 |
6.98 |
|
|
|
|
|
|
S 6 |
74 |
96 |
2.66 |
3.45 |
|
|
|
|
|
|
S 7 |
87 |
76 |
-4.56 |
3.21 |
|
|
|
|
|
|
S 8 |
23 |
21 |
7.65 |
5.15 |
|
|
|
|
|
|
S 9 |
32 |
34 |
2.34 |
5.87 |
|
|
|
|
|