 Hi, welcome to Sorry I Don't Work at Your Library, the effects of revealing institutional affiliation in a consortium chat service. This is a lightning talk to accompany the poster session at the 2020 Library Assessment Conference. My name is Sabina Pagato and I'm from Scholars Portal at the Ontario Council of University Libraries. And I'm Catherine Barrett and I'm from the University of Toronto Scarborough Library. The Ontario Council of University Libraries, or OKOL, is a library consortium in the province of Ontario, Canada, and Scholars Portal is the digital services arm of OKOL. One of the services that we run is Ask a Librarian, a consortium chat reference service with 15 participating universities across the province. Years ago we began a large research project joint between Scholars Portal and the University of Toronto, which is our largest member to investigate factors relating to satisfaction and dissatisfaction among chat users. One of the findings that we had is that Ask users were just as satisfied with service from other libraries as service from their own library. The first note here is whether users of the collaborative chat service were dissatisfied if they learned they were being served by a chat operator from another institution, and also whether that relationship is mediated by other circumstances or factors that arose within the chat. As part of this project, we reviewed six months of chat transcripts. We only included chats in the study sample if the user and the chat operator were affiliated with different OKOL libraries. We reviewed the chat transcripts for instances where the operator revealed to the user that they were not from their home institution. If they did, we hand coded that a mismatch reveal took place. We also reviewed the transcripts and hand coded the user's question type according to a preexisting coding key from Scholars Portal. We also coded instances of specific operator behaviors like transferring the user to another operator, making a referral, admitting they lacked expertise with the user's question, and saying they could not do something for the user. Finally, we pulled in information about the user type, such as undergraduate student or faculty member from the chat session metadata. Finally, we performed chi-square tests of independence to find out if there were statistically significant associations between institutional mismatch reveals and the other study variables. And there are more details available about the methods and the results of the statistical tests on our poster. So what did we find? Our first and biggest finding is that our users were more likely to be dissatisfied if they learned that they were being served by somebody from another library. So it's important to remember it's not whether there was a mismatch, but whether this mismatch was revealed that was statistically significant in being correlated with user dissatisfaction. We also learned that this response is stronger in graduate students and in response to research-based questions. Since graduate students tend to have complex and specific research needs, this finding may indicate that users expect to receive specialized expertise from their home library and are dissatisfied when operators are not able to immediately meet their needs. Similarly, research questions require the most in-depth knowledge. Users may be dissatisfied when operators are not familiar enough with their library's resources or services to answer their questions. Mismatch reveals may cause them to question the purpose of a chat service that can only answer general questions. Taken together, these results indicate that users will question whether they can get the personalized and specialized help they need through a staffing model that does not pair them up with local librarians. Some operator behaviors also mediate this relationship. Users are more likely to be dissatisfied by institutional mismatch reveals when a transfer does not occur during the chat or when the operator does not reveal a lack of expertise. These are instances where revealing a mismatch may have been done preemptively rather than because the circumstances demanded it. This suggests that context is important when revealing a mismatch. If a mismatch is revealed to provide information about why the user is being pointed to another chat operator with more local knowledge or because the operator does not know the answer to their question, then it is understandable to the user. However, if the reveal seems irrelevant, the user may question the service model or why they are being helped by this particular individual. While this variable did not meet the threshold for statistical significance, not making a referral was also mediating factor with a p-value of .07, which also plays into the trend we've just discussed. It may also indicate a specifically local question was not sufficiently answered but for which no alternatives were provided. So our major takeaway from all this is to pay attention to context. When chatting with a user from another institution, be thoughtful about whether you reveal a mismatch and how you do so if you do choose to do so. At the SQL librarian level, we have begun telling our operators not to proactively reveal if they are chatting with someone from another institution, but rather to wait until there is a circumstance within the chat that makes it necessary or relevant information for the user side. So that's all from our research. You can see more on our poster on the library assessment website. We also have a previous article as discussed earlier that talked a little bit about the mismatch without the mismatch reveal, and then you can contact us with any questions. Thank you.