Beta coefficient for one of the detection covariates is not showing up in model summary

[Basavaraj Mulage]

Hi all,

I am having trouble with one of the detection covariates when fitting single season occupancy models using "Unmarked" in R. Have tried fitting models accounting for false positives 

Background : I have collected species (Indian grey wolf) presence absence data via interview based surveys in an occupancy framework, where I had laid out 25 sq km grid across my study area (67000 sq. km)  and within each grid I had 5 spatial replicates ( 5 respondents belonging to different age, sex, education and professional classes and were 1 km apart), in this manner I have collected data for 2027 grids out of 3064 grids altogether. 

I prepared detection matrix with each row representing a site and each column as replicates ( 2027*5). I am using proportion of each LULC class in a grid  as site covariates and  age, gender, education and profession as detection/observation covariates.I am trying to fit single season occupancy models, and beta coefficients for "Farmer" isn't showing up in model summary , only "Shepherd" and 'Others" are showing up I did not have this problem earlier when I had fit models for a smaller study area approx for 300 grids earlier. Now I have the full data set and trying to fit the model and it's throwing this error. any help is highly appreciated

detection covariates 

Age - has sub classes like 0-15, 16-19 etc  (6 levels)

Gender - Male and Female (2 levels)

Education - illiterate, primary etc  (6 levels)

Profession -  Farmer, shepherd, others (3 levels)

Call:
occuFP(detformula = ~profession, FPformula = ~1, stateformula = ~1,
    data = occframeFP_wolf)

Occupancy:
 Estimate     SE     z P(>|z|)
    -0.06 0.0514 -1.17   0.243

Detection:
                   Estimate     SE     z  P(>|z|)
(Intercept)          -0.424 0.0438 -9.70 3.04e-22
professionOthers      0.180 0.0784  2.30 2.16e-02
professionShepherd    0.303 0.1101  2.75 5.88e-03

false positive:
 Estimate   SE     z P(>|z|)
    -11.4 9.16 -1.25   0.212

Pcertain:
 Estimate     SE    z   P(>|z|)
     2.33 0.0836 27.9 3.19e-171

AIC: 9005.724 

[Josh Twining]

Hi Basavaraj,

This isn't an error, this is just how unmarked (and most linear modelling functions) handle factors. 

The first level (farmers in this instance) is made a reference level and contained within the intercept. The coefficient estimates of other levels in the factor (others and shepard) are relative to the reference level. So for example, the results you shared are suggesting that Shepards have slighter higher probability of detecting wolves conditional on z=1 than farmers.

If you are interested in exploring the effect of profession on detection probability in an easier to interpret way, you could make marginal effects plots to visualize how p varies as a function of profession, i.e.,  you can make predictions for each of these levels using the predict function, and then use your top model to predict detection probability for the three levels, and plot these predictions

Josh

--
*** Three hierarchical modeling email lists ***
(1) unmarked (this list): for questions specific to the R package unmarked
(2) SCR: for design and Bayesian or non-bayesian analysis of spatial capture-recapture
(3) HMecology: for everything else, especially material covered in the books by Royle & Dorazio (2008), Kéry & Schaub (2012), Kéry & Royle (2016, 2021) and Schaub & Kéry (2022)
---
You received this message because you are subscribed to the Google Groups "unmarked" group.
To unsubscribe from this group and stop receiving emails from it, send an email to unmarked+u...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/unmarked/895c42e1-40b3-41da-8de8-0a1f1f2ece4dn%40googlegroups.com.

--

Joshua P. Twining, PhD,

Assistant Professor,

126 Nash Hall,

Oregon State University,

Oregon, 

97331

Email: joshuap...@gmail.com / joshua....@oregonstate.edu 

Mob: (607) 220 7547

[Jim Baldwin]

Just to echo what Josh wrote, if you want explicit coefficients for all three levels, use "~ 0 + profession".

Jim

To view this discussion visit https://groups.google.com/d/msgid/unmarked/CA%2Bg%2BShmRDyq9MCVmfSq57xRmgBZGR9ey8KjaHXazSywvBu%2Bgkg%40mail.gmail.com.

[Marc Kery]

Or: "~ profession - 1" 🙂

---

From: unma...@googlegroups.com <unma...@googlegroups.com> on behalf of Jim Baldwin <jbald...@gmail.com>
Sent: Tuesday, November 4, 2025 09:33
To: unma...@googlegroups.com <unma...@googlegroups.com>
Subject: Re: [unmarked] Beta coefficient for one of the detection covariates is not showing up in model summary

 

To view this discussion visit https://groups.google.com/d/msgid/unmarked/CAJCKiwwmUynr_6rp2WFt3Dw-ah3yhKvCjpew0P3ikHyFHwYiXA%40mail.gmail.com.

[Basavaraj Mulage]

thank you Joshua, Marc and Jim for clarification