Web Traffic Analysis

Background

This investigation uses website traffic to proxy the degree to which domestic violence occurs in each state. We use this proxy in order to capture the portion of the population that seeks help outside of law enforcement, as well as that which seeks help within. Many studies fail to capture the former population in studying domestic violence during the pandemic period, as data from other resources such as shelters and hotlines are decentralized and often unavailable.  We can be sure that analyzing website traffic for domestic violence trends may not capture the number of survivors with low access (for geographic reasons or social reasons) to the internet and those who do not know of the existence of these websites. This type of analysis is also limited because these websites are likely accessed not just by survivors, but also by researchers, legal aids, and others that work in DV prevention and programming and therefore inflate the web traffic numbers.

Trends

The trends from this analysis can be broadly defined into 4 categories, offering potential insight into the domestic violence situation happening in each state:

• No change: OH, FL, MD

  • The DVC website traffic in these states showed no significant change after the lockdown compared to prior.

• Major uptick: WI, AL, NC, VA, PA

  • The DVC website traffic in these states showed a major increase after their state’s lockdown. We hypothesize that the constant proximity between survivors and abusers due to the lockdown may be the reason for this phenomenon.

• Down then up: WA, CA, UT, AZ, KS, OK, MO, WV, GA, DC, NY, CT, MA

  • The DVC website traffic in these states showed a downward trend prior to a state’s lockdown and a resurgence after. The resurgence may be explained by close contact during shelter-in-place orders, but it is unclear why the downturn prior occurs.

• Primarily down: ID, NM, TX, IL, IN, SC, NJ

  • The DVC website traffic in these states showed a slump in traffic after a state’s lockdown, with no major signs of resurgence. We hypothesize this is due to the constant close quarters that survivors and abusers share because of the stay-at-home orders.

  • This finding is concerning because the constant slump may indicate that the domestic violence situation is worse in these states with survivors increasingly unable to seek help, despite the increasing rate of DV. Owing to the method of proxy, we were unable to secure complete data for some states, listed below:

• No data: OR

• Lack of meaningful data: MT, NV, CO, MN, LA, MS, MI, TN, KY, DE, RI, NH, VT, ME

  • Some states simply did not issue a stay-at-home order at all, so they fall out of the purview of this analysis:

• No stay at home order issued: WY, ND, SD, NE, IO, AK

Analysis

While these graphs may offer some insight into a state’s domestic violence situation subsequent to a stay-at-home order, they are lacking in some respects. For example, a more meaningful analysis may include a moving average to determine whether an uptick is significant. Other variables, such as average age and internet penetration, should be controlled for to provide a clearer picture. These graphs serve to gain a general sense of the situation, rather than determine a causal effect of the stay-at-home orders on state domestic violence situations.