Tell me more ×
Cross Validated is a question and answer site for statisticians, data analysts, data miners and data visualization experts. It's 100% free, no registration required.
up vote 3 down vote favorite

I want to do a meta-analysis with count data for treatement/control cases (abundance of sea polichaetos). My problem is that I need to use zero values (an informative zero value) for the mean and standard deviation for one of the treatement, but R has a problem: "Studies with zero values for sd.e or sd.c get no weight in meta-analysis". I can agroup the case by Family (byvar=Family).

There is an option in metabin, where you could add either a numerical value to each cell frequency for studies with a zero cell count or the character string "TA" which stands for treatment arm continuity correction, but I want to use metacont which don´t have this option.

Can you help me? Thanks!

> datos<-read.table(file="datos_totales.csv",head=TRUE,sep=",",dec=",") 
> datos[10:20,]
Study ControlN ControlMean ControlSD TreatedN TreatedMean TreatedSD       Family
10 Ali.Gmb.S1.Inv       3    893.3333 180.36999        3  1213.33333 410.52812 Capitellidae
11 Ali.Gmb.S2.Inv       3    640.0000 348.71192        3   666.66667 220.30282 Capitellidae
12 Ali.Gmb.S3.Inv       3    426.6667 184.75209        3   920.00000 628.64935 Capitellidae
13 Ali.Cul.S1.Ver       3    213.3333 115.47005        3     0.00000   0.00000 Cirratulidae
14 Ali.Cul.S2.Ver       3    160.0000 144.22205        3    40.00000  40.00000 Cirratulidae
15 Ali.Cul.S3.Ver       3    293.3333 234.37861        3    26.66667  46.18802 Cirratulidae
16 Ali.Cul.S1.Inv       3    653.3333 532.66625        3    13.33333  23.09401 Cirratulidae
17 Ali.Cul.S2.Inv       3    706.6667 335.45988        3    26.66667  23.09401 Cirratulidae
18 Ali.Cul.S3.In        3    666.6667 674.48746        3    80.00000  80.00000 Cirratulidae
19 Ali.Gmb.S1.Vr        3    280.0000  69.28203        3     0.00000   0.00000 Cirratulidae
20 Ali.Gmb.S2.Ver       3    226.6667 220.30282        3     0.00000   0.00000 Cirratulidae

>  library(meta)

> metaanalisis2<- metacont(TreatedN, TreatedMean, TreatedSD, ControlN, ControlMean, 
  ControlSD, data=datos, byvar=Family, label.c="Control", print.byvar=FALSE, 
  label.e="Treatment", studlab=Study)
**Mensajes de aviso perdidos
In metacont(TreatedN, TreatedMean, TreatedSD, ControlN, ControlMean,  :
  Studies with zero values for sd.e or sd.c get no weight in meta-analysis**
share|improve this question
3  
Uwe Ligges on r-help has already given the 'answer': You cannot have standard deviations of 0. Are these missing values? – Bernd Weiss Jul 9 '11 at 18:20
@Bernd nice observation! – suncoolsu Jul 9 '11 at 18:24
1  
@Rosa If you could provide more information about the methods that you (or another scientist) used to calculate SD, it would be easier to help you troubleshoot your problem. It seems that the problem is not in the meta-analysis but in the calculation of summary statistics. – David Jul 10 '11 at 17:49
Hi, thanks for your interest. – Rosa Jul 15 '11 at 13:09
The data are abundance of sea polichaeta with (treatement) and without (control) fishing. Therefore, zero cells values means the impact of fishing on this population. The idea of adding the small number to one of the zero cells in each group of three zero cells is that it doesn't do too much damage to the location of the mean, and it allows me to do the calculation without an undefined result. Mean change from 0 to 0.333. ¿What do you think? Look "Higgins JPT, Green S. 2011. www.cochrane-handbook.org" – Rosa Jul 15 '11 at 13:22

Know someone who can answer? Share a link to this question via email, Google+, Twitter, or Facebook.

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Browse other questions tagged or ask your own question.