Absolutely would have avoided this question in the exam.
This was a tough one, had to google a number of things to figure it out, but ultimately managed it. This link really helped.
Also hadn't used the nested test tool before, so this was good practice.
Solution attached. Just thinking this through, as the chi-square result is statistically significant at all thresholds, we are saying the models are significantly similar and the removal of f38 had no statistically significant effect. This is based on the understanding that the null hypothesis for the chi-square test is that there is no association between the results of the two models and the alternative hypothesis (supported by the results) is that there is an association between the two models.