When posting images to social media sites, preference should be given to images in which people figure prominently, as these images resonate most with users. When posting images of the Captioned Photo variety, it is essential that the image include some sort of illustration to engage the user’s interest.
Method and Analysis
I collected a random sample of 20 images from the NYPL Instagram using systematic random sampling. Beginning with the second most recent image posted, I took every third image until I had a sample of 20 images. The #LiteraryMarchMadness posts (for an example, see [here]) were not included in the population of images sampled. The sample was collected on 21 March, 2016 in the evening (~1800/6 pm UTC-4). The number of images belonging to each of Hu, Manikonda and Kambhampati’s (2014) categories is tabulated in Appendix 1 and depicted in Figure 1 below.
Five categories (Selfie, Fashion, Activity, Captioned photo, and Friends) accounted for 95% of the sample. The category with the greatest number of posts was Selfie (6 posts), followed by Friends (4 posts). The Selfie and Friends categories accounted for the top 50% of the sample.
The prominence of the Selfie category is not surprising. Given the degree to which “selfies”, or self portraits, have become ingrained in social media culture–in no small part due to mobile technology–it makes sense that this sort of image would resonate with users. It is also unsurprising that the next most represented category is Friends, as the two categories are related in that they both relate to people, who are at the heart of social networking.
It is perhaps unsurprising that the Captioned Photo category is represented in the same degree as the Fashion and Activity categories, as this kind of image has gained popularity on the Internet and social media sites in particular. However, the frequency of Captioned Photo images in this sample is also interesting because the sample is drawn from a library’s Instagram account, and two of the three Captioned Photo images (numbers 8 and 16 respectively) are images of text documents, and only one (number 15) is what many would consider to be a meme or captioned photo. The images of text documents, while interesting to a librarian or research-oriented user of the library, are not visually engaging, and I would expect that this sort of image would not receive as favorable a response as would an image in the Selfie, Fashion, Activity, or Friends categories, or even as compared to a more memetic image such as the third Captioned Photo image from the sample (number 15). This expectation is supported by the response to the three Captioned Photo images, as interpreted by numbers of likes and comments:
|Post||Number of Likes||Number of Comments|
|8 (image of text from book)
|15 (Valentine’s Day meme)||1,092||90|
|16 (image of text from plaque)
There were no images in the categories of Food and Pet in the sample. Due to the small sample size, it is impossible to say whether this is a result of the nature of the Instagram account from which the sample was drawn (i.e. the New York Public Library’s Instagram), the bias of the person running the account, or an actual tendency for these categories to be less popular than the others.
Appendix 1: Counts by category
- Friends (4)
- Food (0)
- Gadget (1)
- Captioned photo (3)
- Pet (0)
- Activity (3)
- Selfie (6)
- Fashion (3)
Appendix 2: Sample