SPSSX Discussion

Time Series comparison of models

Classic

List

Threaded

2 messages Options

Mike P-5

Time Series comparison of models

Hi All,

Hoping a Time Series Expert can help me out here,

I've collected data over a period of days for a control group and an
exposed group, when plotting via sequence charts there is an upwards
trend in the exposed group. Is there a way I can compare the two
different groups to see if there is a significant difference over time?

I know a t-test would look for the overall means, but I'm more concerned
with the time component making a difference to the two series?

Does this make sense? I've had a look on the net an was wondering if
pooled cross-sectional time series was what I wanted? Can this be done
in SPSS?

Sample data below

control Exposed n_control n_exposed Day_ Date_
.19 .17 120 10
1 1
.15 .25 350 124
2 2
.12 .22 690 482
3 3
.13 .19 349 82
4 4
.14 .30 375 68
5 5
.08 .31 331 180
6 6
.11 .26 407 303
7 7
.15 .22 328 192
8 8
.18 .29 168 83
9 9
.08 .17 198 96
10 10
.17 .13 45 55
11 11
.27 .33 48 58
12 12
.18 .31 42 19
13 13
.04 .13 53 15
14 14
.06 .50 34 47
15 15
.16 .31 47 54
16 16
.15 .23 53 55
17 17
.20 .35 50 38
18 18
.16 .42 50 52
19 19
.16 .22 51 21
20 20
.17 .17 61 21
21 21
.06 .20 29 41
22 22
.08 .34 38 47
23 23
.00 .33 39 41
24 24
.15 .34 47 46
25 25
.05 .33 46 41
26 26
.10 .57 39 11
27 27
.24 .40 49 17
28 28
.08 .25 23 29
29 29

Mike

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email
______________________________________________________________________

Mike P-5

Re: Time Series comparison of models

I'm only asking as I thought 29 points might be two many for a repeated
measures design??

But advice contrary to this is also appreciated

Mike

-----Original Message-----
From: Michael Pearmain
Sent: 20 November 2006 17:16
To: [hidden email]
Subject: Time Series comparison of models

Hi All,

Hoping a Time Series Expert can help me out here,

I've collected data over a period of days for a control group and an
exposed group, when plotting via sequence charts there is an upwards
trend in the exposed group. Is there a way I can compare the two
different groups to see if there is a significant difference over time?

I know a t-test would look for the overall means, but I'm more concerned
with the time component making a difference to the two series?

Does this make sense? I've had a look on the net an was wondering if
pooled cross-sectional time series was what I wanted? Can this be done
in SPSS?

Sample data below

control Exposed n_control n_exposed Day_ Date_
.19 .17 120 10
1 1
.15 .25 350 124
2 2
.12 .22 690 482
3 3
.13 .19 349 82
4 4
.14 .30 375 68
5 5
.08 .31 331 180
6 6
.11 .26 407 303
7 7
.15 .22 328 192
8 8
.18 .29 168 83
9 9
.08 .17 198 96
10 10
.17 .13 45 55
11 11
.27 .33 48 58
12 12
.18 .31 42 19
13 13
.04 .13 53 15
14 14
.06 .50 34 47
15 15
.16 .31 47 54
16 16
.15 .23 53 55
17 17
.20 .35 50 38
18 18
.16 .42 50 52
19 19
.16 .22 51 21
20 20
.17 .17 61 21
21 21
.06 .20 29 41
22 22
.08 .34 38 47
23 23
.00 .33 39 41
24 24
.15 .34 47 46
25 25
.05 .33 46 41
26 26
.10 .57 39 11
27 27
.24 .40 49 17
28 28
.08 .25 23 29
29 29

Mike

______________________________________________________________________
This email has been scanned by the MessageLabs Email Security System.
For more information please visit http://www.messagelabs.com/email
______________________________________________________________________