Looking at survey, I see 12 Likert-scored item, but nothing that was Yes or No.

The simplest approach would be to see whether you can create a scale from your 12 items, and if so then compare the two groups via a t-Test. The first stage in creating a scale would be to calculate coefficient alpha (also known as Cronbach's alpha) to determine how well the 12 items fit together (i.e., their "internal consistency").

What aspects of the two groups do you want to compare? Do you believe that these 11 items were all measuring the same basic thing? Can you say more about the sampling and if the data are independent? For example, were the zoo keepers clustered in different groups (like by zoo, or by section in the zoo), and were some of the other respondents at different zoos, from the same family, etc. Answers to these will help guide people to provide useful answers.

Daniel Wright, I would like to understand if there was a statistical significance in the perceptions of the zoo keepers vs. the visitors, and birds vs. mammals.

Zoo keepers did the surveys separately from the visitors and all data was collected from the same zoo. Zookeepers each took one survey on the 5 species listed at the bottom of the survey. I'm not sure if that is considered a repeat measure though.

The surveys were done on 5 different species (3 mammal and 2 bird species). Unfortunately I didn't collect any data on if the visitors were from the same groups or not. Also due to research limitations, I didn't collect demographic data, it is just a pilot study.

So each human responding about all 5 species, or was each person assigned (randomly?) as single species? Also, I don't understand "I would like to understand if there was a statistical significance ... ." To you mean you want to see if the views of the two groups of humans differ for the 5 types of species? Another question would be if the associations among those items are similar for the 10 groups (i.e., the 2 groups of humans by the 5 groups of non-humans). This looks like an interesting study!

Daniel Wright,

For zookeepers each keeper responded to all 5 species (there were only 12 keepers in the whole zoo). Each day that I surveyed visitors I randomly selected those passing by and viewing the animal on that day.

~ " "Also, I don't understand "I would like to understand if there was a statistical significance ... ." To you mean you want to see if the views of the two groups of humans differ for the 5 types of species? " "

Yes I want to see if the views of the two groups of humans differ from one another, and if the views of the humans differ if it was a bird or a mammal.

It was a fun study to design and implement!

Thanks Suzie Halvorson . That is a more complex design, but it makes sense. I have to go teach (I mean, sit in the same chair since this is all online at the moment), but I will think more later. Dan Oh, one more question. We people generally giving answers at one end of the scales.?

@Suzie I presume that the visitors have taken the survey for any one of the five species, but the 12 keepers have taken the survey for all the five species. I opine that it would be a good idea to analyse the data separately for each species. However, this decreases your sample size (anyway this is a pilot).

Before going to the analysis part, the fundamental concerns with the questionnaire are the possibility for acquiescence bias and the rationale for why two questions were made dichotomous. If your internal consistency reliability is good, you could probably think of deriving a summary score from the questionnaire. If you believe that the questions doesnt measure a single underlying latent construct, it is better to analyse each question separately.

If tou intend to measure summary scores, you can analyse the differences in perceptions of zookeepers between species by a Friedman's test. For comparison between visitors and keepers, Mann Whitney could be employed.

If you intend to analyse the items in the questionnaire separately, you can go with Chi square to compare between visitors and keepers; for inter species perception comparison within keepers, Cochran's Q test can be used.