Hey, I'm attempting to discuss my GLM results in an APA format, however, I am struggling with the language used. This is because from my understanding the size of the intercept number does not increase or decrease the size of the effect? Therefore, it is hard to discuss.
Here is my code.
glm(formula = counter_prop ~ appear_selected * label_selected,
family = quasibinomial(link = "logit"), data = subset(aggDataOut,
identityResponse == 1))
Deviance Residuals:
Min 1Q Median 3Q Max
-0.85593 -0.13662 -0.01543 0.09408 1.00935
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -0.87439 0.03792 -23.060 < 2e-16 ***
presume_selectedvegan -0.79385 0.06049 -13.123 < 2e-16 ***
label_selectedNL 0.25689 0.05201 4.939 1.03e-06 ***
appear_selectedvegan:label_selectedNL 0.09171 0.08161 1.124 0.262
---
Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1
(Dispersion parameter for quasibinomial family taken to be 0.04270381)
Null deviance: 42.151 on 582 degrees of freedom
Residual deviance: 25.051 on 579 degrees of freedom
AIC: NA
Number of Fisher Scoring iterations: 4
Here how I have tried to interpret the results ;
This is a general linear model investigates the impact on food choice from two characterstics of the food, whether the food is presumed to be vegan or not and If the food has a vegan label or not, but only for the identity response “1 – not vegan".
This model finds a significant negative effect for food choice that is presumed to be vegan (est = -0.79385, Std. Error = 0.06049, P < 0.05). There is also a positive significant effect for food without a vegan label (est = 0.25689, Std.Error = 0.05201, P < 0.05 ), however, this effect is smaller? (not sure how to word this).
Also, because I am conducting 6 GLM for this analysis, one for each identity response, does this mean that I have to multiply the confidence value (normally 0.05) by 6? If so how do I do this in R? 