Nested design help
Locked
 
|
Dear list: I have a student conducting a study for a thesis where he is looking at firemen/women’s’ intentions to use safety devices. He has a pre measure and a post measure (intervention between pre and post)-a repeated measure (within group design). However, the firemen/women are nested within battalions. Each battalion is made up of 4-6 station houses. The firemen/women work out of station houses- so it seems that should be the level of analysis. I haven’t used a nested design in many years and would appreciate any direction in regard to texts or articles that would help me figure out the design so that we can then analyze the data within spss. Advice appreciated. mfs Martin F. Sherman, Ph.D. Professor of Psychology Director of Masters Education in Psychology: Thesis Track Loyola University Maryland Department of Psychology 222 B Beatty Hall 4501 North Charles Street Baltimore, MD 21210 |
|
|
Martin, I assume the DV is at the person level, the fire-person, rather than the station house. True? Suppose so, the persons are nested with in firehouses within battalions. Was the intervention at the person or station-house level? Does battalions matter? How many battalions were there? 1, 5, 20? Gene Maguin From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Martin Sherman Dear list: I have a student conducting a study for a thesis where he is looking at firemen/women’s’ intentions to use safety devices. He has a pre measure and a post measure (intervention between pre and post)-a repeated measure (within group design). However, the firemen/women are nested within battalions. Each battalion is made up of 4-6 station houses. The firemen/women work out of station houses- so it seems that should be the level of analysis. I haven’t used a nested design in many years and would appreciate any direction in regard to texts or articles that would help me figure out the design so that we can then analyze the data within spss. Advice appreciated. mfs Martin F. Sherman, Ph.D. Professor of Psychology Director of Masters Education in Psychology: Thesis Track Loyola University Maryland Department of Psychology 222 B Beatty Hall 4501 North Charles Street Baltimore, MD 21210 |
|
|
Gene: He is only looking at 4-6 battalions and each battalion has about 4-6 stations. Total number of stations will be about 16-20 and the intervention (designed to increase safety use during overhaul-masks) is at the station level. So the level of analysis will be ?? mfs From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Maguin, Eugene Martin, I assume the DV is at the person level, the fire-person, rather than the station house. True? Suppose so, the persons are nested with in firehouses within battalions. Was the intervention at the person or station-house level? Does battalions matter? How many battalions were there? 1, 5, 20? Gene Maguin From: SPSSX(r) Discussion [hidden email] On Behalf Of Martin Sherman Dear list: I have a student conducting a study for a thesis where he is looking at firemen/women’s’ intentions to use safety devices. He has a pre measure and a post measure (intervention between pre and post)-a repeated measure (within group design). However, the firemen/women are nested within battalions. Each battalion is made up of 4-6 station houses. The firemen/women work out of station houses- so it seems that should be the level of analysis. I haven’t used a nested design in many years and would appreciate any direction in regard to texts or articles that would help me figure out the design so that we can then analyze the data within spss. Advice appreciated. mfs Martin F. Sherman, Ph.D. Professor of Psychology Director of Masters Education in Psychology: Thesis Track Loyola University Maryland Department of Psychology 222 B Beatty Hall 4501 North Charles Street Baltimore, MD 21210 |
|
Martin Long time since I did one of these (British Crime Survey 1982, Undergraduate Income and Expenditure, 1983). Click on Help > Command Syntax Reference . . .and then have a look at pp723 ff under FILE TYPE. I think your student needs NESTED, but there’s always a workaround. John F Hall (Mr) Email: [hidden email] Website: www.surveyresearch.weebly.com Skype: surveyresearcher1 Phone: (+33) (0) 2.33.45.91.47 From: SPSSX(r) Discussion [mailto:[hidden email]] On Behalf Of Martin Sherman Gene: He is only looking at 4-6 battalions and each battalion has about 4-6 stations. Total number of stations will be about 16-20 and the intervention (designed to increase safety use during overhaul-masks) is at the station level. So the level of analysis will be ?? mfs From: SPSSX(r) Discussion [[hidden email]] On Behalf Of Maguin, Eugene Martin, I assume the DV is at the person level, the fire-person, rather than the station house. True? Suppose so, the persons are nested with in firehouses within battalions. Was the intervention at the person or station-house level? Does battalions matter? How many battalions were there? 1, 5, 20? Gene Maguin From: SPSSX(r) Discussion [hidden email] On Behalf Of Martin Sherman Dear list: I have a student conducting a study for a thesis where he is looking at firemen/women’s’ intentions to use safety devices. He has a pre measure and a post measure (intervention between pre and post)-a repeated measure (within group design). However, the firemen/women are nested within battalions. Each battalion is made up of 4-6 station houses. The firemen/women work out of station houses- so it seems that should be the level of analysis. I haven’t used a nested design in many years and would appreciate any direction in regard to texts or articles that would help me figure out the design so that we can then analyze the data within spss. Advice appreciated. mfs Martin F. Sherman, Ph.D. Professor of Psychology Director of Masters Education in Psychology: Thesis Track Loyola University Maryland Department of Psychology 222 B Beatty Hall 4501 North Charles Street Baltimore, MD 21210 |
|
|
In reply to this post by msherman
Martin, I think how the data are analyzed depends on what your student wants to know. I'll assume the question is just whether intentionality increased. Persons are within stationhouses and stationhouses are within battalions. It seems to me that battalions are irrelevant and could be ignored. Next, is your student interested in just these stationhouses or in this sample of stationhouses from the population of Chicago stationhouses? If he says, 'Just these', then I think the analysis is just a between (stationhouses)-within (time) anova. He'd hope for a significant time effect and nothing else. But, if he says, 'This sample', then I think stationhouse becomes a random factor. So now, it would be good to compute the ICC at both time points, which you can do with the mixed command by fitting an intercept only model to the timepoint data. Mixed intentionspre/random=intercept | subject(stationhouse) covtype(id). Suppose the ICCs are large enough (large undefined). Then you have several choices. Choice one, I think, is an ordinary between-within model with the between factors (stationhouses) as random. I have never used this specific model structure and I don't know how it will work. Glm intentionspre intentionspost by stationhouse/random=stationhouse/wsfactor time(2). Choice two is a two point growth curve model (Singer and Willett is a good reference). If you assume the regression of intentions on time is constant across groups, you have: Mixed intentions with time/fixed=time/random=intercept | subject(stationhouse) covtype(id). However, I think you are really interested in that regression coefficient. You want it positive, large, and significant. So use this: Mixed intentions with time/fixed=time/random=intercept time | subject(stationhouse) covtype(un). I think a repeated measures model with a random intercept is also possible but I am unsure how to set it up. Others on the list are more knowledgeable. I hope that they will reply because I'd like to see how to set such a model up. Gene Maguin From: Martin Sherman [hidden email] Gene: He is only looking at 4-6 battalions and each battalion has about 4-6 stations. Total number of stations will be about 16-20 and the intervention (designed to increase safety use during overhaul-masks) is at the station level. So the level of analysis will be ?? mfs From: SPSSX(r) Discussion [hidden email] On Behalf Of Maguin, Eugene Martin, I assume the DV is at the person level, the fire-person, rather than the station house. True? Suppose so, the persons are nested with in firehouses within battalions. Was the intervention at the person or station-house level? Does battalions matter? How many battalions were there? 1, 5, 20? Gene Maguin From: SPSSX(r) Discussion [hidden email] On Behalf Of Martin Sherman Dear list: I have a student conducting a study for a thesis where he is looking at firemen/women’s’ intentions to use safety devices. He has a pre measure and a post measure (intervention between pre and post)-a repeated measure (within group design). However, the firemen/women are nested within battalions. Each battalion is made up of 4-6 station houses. The firemen/women work out of station houses- so it seems that should be the level of analysis. I haven’t used a nested design in many years and would appreciate any direction in regard to texts or articles that would help me figure out the design so that we can then analyze the data within spss. Advice appreciated. mfs Martin F. Sherman, Ph.D. Professor of Psychology Director of Masters Education in Psychology: Thesis Track Loyola University Maryland Department of Psychology 222 B Beatty Hall 4501 North Charles Street Baltimore, MD 21210 |
| Free forum by Nabble | Edit this page |