Questions tagged [overdispersion]
Overdispersion is when there is greater variability than there 'ought' to be in the data. Eg, the variance of counts is often greater than the mean, whereas the variance of a Poisson should equal the mean.
269 questions
0 votes
0 answers
37 views
Residual issues in binomial GLMM when including random effect
I'm working on a dataset of ~2900 fish, where the visually estimated sex was compared to the true sex. In about 10% of the cases (≈260 fish), the estimation was wrong (deviation = TRUE). I'd like to ...
- 165
2 votes
0 answers
100 views
Preventing Shrinking towards the mean in Bayesian Modelling
I am working on a model where I am aiming to predict influenza in geographic regions. In my case I aiming to predict influenza cases in high resolution regions using low resolution data. I am using a ...
- 151
3 votes
0 answers
86 views
No indication data is overdispersed but model fit improves when accounting for it
In glmmTMB I've noticed on several occasions that I can have a standard Poisson glmer with no indication of overdispersion using the check_overdispersion, ...
16 votes
2 answers
868 views
How can variance and mean be equal if they have different unit of measurement?
Suppose the number of road accidents follows a Poisson distribution. Suppose there is a mean of 4 accidents in a given city in a given time period. The variance would be measured in $\text{accidents}^...
- 3,403
1 vote
0 answers
82 views
Different ratios (3755 vs 1.04) for overdispersion ratio using different methods - check_overdispersion vs manually - why?
I have a model and when testing for overdispersion I used manual method and also check_overdispersion from the Performance package, and got massively different ratios (although both were overdispersed....
- 23
5 votes
1 answer
225 views
How to model overdispersed data (days to germination) in a small experimental dataset?
I'm working with a small dataset (n=36) from a greenhouse experiment and want to test the effect of soil inoculum on days to germination. I'm having trouble finding a model that fits my data. The data ...
- 83
2 votes
2 answers
243 views
How to decide between a hurdle model or negative binomial distribution for count data using glmmTMB?
I am running some GLMMs in the glmmTMB package. My responses are based on seedling data where in each sub-plot I have (1) presence/ absence of seedlings (0 or 1) and (2) a abundance count. I am trying ...
- 171
1 vote
0 answers
56 views
How to model Endogeneity in a Zero-Inflated process
I have a multivariate dataset consisting of household level variables such as education, location of the household, occupation, income, consumption etc. The regressand variable is number of social ...
- 11
0 votes
2 answers
188 views
How does glm.nb() in R parametrize the Negative Binomial distribution with $\mu$ and $\theta$?
I'm working with the glm.nb() function in R, which fits a Negative Binomial distribution using the parameters $\mu$ (mean) and $\theta$ (overdispersion parameter). The variance is given by: $$ \text{...
0 votes
0 answers
39 views
Statistical Inquiry. Overdispersion issue or abundance of 0s
Trying to go over some data and its half presence absence while the other half is count data. The model I've been working on lately is a mix of lmer and glmer.nb as that seems to be the approach for ...
- 1
0 votes
0 answers
36 views
How to model frequency data with mixed effects?
I have some frequency data that I am not sure how to model I have two independent experiments (replicate: A, B), where I treat neurons (10 per treatment) with the same drug but for different time ...
4 votes
2 answers
220 views
Why do I get different standard errors when i group the data before fitting a Quasi-Poisson GLM for counts with an offset = log(population)?
Is it theoretically correct to group the data before fitting a quasi-Poisson or should I leave it ungrouped? Commentary as well as an example in R code is posted below. I noticed I get different ...
- 99
1 vote
0 answers
54 views
modelling overdispersed binomial data using generalised linear models
I've run a series of tests where each test involves answering 10 questions which are scored correct/wrong, and there are roughly 36 participants per test divided more or less evenly into 3 treatment ...
4 votes
1 answer
232 views
Validity of AIC When Comparing Models with Varying Dispersion Parameters
I'm currently making a binomial model with a logit link, which is parameterised as a quasibinomial since I'm allowing it to calculate the dispersion parameter. I was wondering, since changes to the ...
- 95
7 votes
2 answers
626 views
How much dispersion is too much for quasipoisson regression?
Quasipoisson regression goes beyond standard poisson regression in taking into account overdispersion (whereby the dependent variable's variance is much greater than its mean). This is explained at ...
- 13.1k