Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Page MenuHomePhabricator

Analysis Request for Editor Growth due to Campaign Activity in Sub-Saharan Africa
Closed, ResolvedPublic

Description

Name for main point of contact and contact preference
@ifried

What teams or departments is this for?
Campaigns

What are the details of your request? Include relevant timelines or deadlines

We have seen notable growth in the number of editors in Sub-Saharan Africa between 2019 and 2020. We want to know to what extent this growth can be attributed to campaign activity (e.g., #1lib1ref, Wiki Loves Africa, Gender Gap, etc). In other words, what percentage of editors in SSA Africa participated in campaign activities in the 2019-2020 period?

Are you asking about a specific metric? Is there a date after which the analysis will no longer be useful? Please provide any timeline/relevant deadlines, requested formats, examples, links to documentation, or other relevant information that would help us understand your request.

Yes, we want to see the breakdown of new editors in Sub-Saharan Africa based on:

  • Newcomers -- percentage of editors that participated in a campaign within one month of registering accounts (broken down by month, between Sept 2019 & Oct 2020)
  • All participants -- percentage of all editors in SSA that participated in a campaign (broken down by month, between Sept 2019 & Oct 2020)

How will you use this data or analysis?

We will use this data to better understand the current potential rate of growth among newcomers activated via campaigns in Sub-Saharan African. This will help us establish a baseline understanding, so we can track trends or changes over time. We can also have a better understanding of our potential impact in editor onboarding and retention as we develop more tools to support campaign organizers and participants.

Is this request urgent or time sensitive?

Medium-level priority. It would be good to get some insights as soon as we can, but it's not extremely urgent.

Data to be presented as two numbers (what percentage of editors in SSA Africa participated in campaign activities in the 2019-2020 period? what percent were newcomers?) and 1-2 related campaign SSA editor to SSA editor charts.

Event Timeline

ldelench_wmf moved this task from Triage to Current Quarter on the Product-Analytics board.

Question: what percentage of editors in SSA Africa participated in campaign activities in the 2019-2020 period?

Steps

    • Clean data
  • get clean csv formatting: json to csv, align columns etc
  • data quality: names
    • Query
  • Query Geoeditor data
    • Calculate

Prep Question: How many 2019-2020 campaign participants edited in September 2021?

  • 5%

Prep Question: How many 2019-2020 campaign participants edited in September (09/01-09/30) 2021 from SSA?
*.4% (243)

Question: what percentage of editors in SSA Africa participated in campaign activities in the 2019-2020 period?

  • .044%

Question: % of 2019/2020 campaign participants that edited in Sept 2021 from SSA who registered their account in 2019-2020?

  • 61%

charts: interesting data as far as edit_counts, number of campaigns editors participated in, number of wikis

Notes on data shared above by @Iflorez:

  • Clarification -- "How many 2019-2020 campaign participants edited in September 2021?": This refers to a group of about 59,000 unique editors who participated in various campaign events (checklist of campaigns found above) between January 1, 2019 and December 31, 2020. The participants were found by looking at data found in the Programs & Events Dashboard, Wiki Loves data, grants data, and other recommendations from the campaigns team. The end result was a finding that between Sept 1-30, 5% of these participants edited in September 2021.
  • Observation -- "% of 2019/2020 campaign participants that edited in Sept 2021 from SSA who registered their account in 2019-2020?": The 61% number indicates a potentially that campaigns potentially provide a strong pipeline for new account creation in SSA. A potential next step could be explore the level of general editing and campaign activity associated with these new accounts along with long-term retention rate. Maybe this could be a new ticket.

SSA = registered users that participated in campaigns in 2019-2020 AND were active as editors in 09/2021 AND edited from an SSA (Sub-Saharan Africa) country
All = registered users that participated in campaigns in 2019-2020 AND were active as editors in 09/2021


% of "all" participants contributing to more than one campaign: 47%
% of "SSA" participants contributing to more than one campaign: 70%

Top countries represented in the "SSA" group:

CountryEditor count, binned
Nigeria100-150
Ghana20-100
South Africa10-20
Tanzania10-20
Benin10-20
Kenyaunder 10

How many participants edited in more than one wiki?
61% - "all"
54% - "SSA"

Wiki participation:
Majority of SSA participants participate in 5 wikis or less; Few editors participate on 5+ wikis
Majority of all participants participate in 5 wikis or less; Though, there are many editors that participate on 5+ wikis

Top wikis represented -"all" group, desc

wikidatawiki
enwiki
commonswiki
metawiki
frwiki
eswiki

Top wikis represented -"SSA" group, desc

enwiki
metawiki
commonswiki
wikidatawiki
frwiki
swwiki

Edit Groups seen in the "SSA" group, desc

Edit countNumber of editors with this edit count, binned
0-5050
100-30020-50
300-60020-50
1k-2k20-50
50-10020-50
300-60010-20
600-1k10-20
2k-3k10-20

Questions:

  • How representative are the two groups that we are looking at when compared to editors overall or to SSA editors overall or to active editors overall or to active editors in SSA overall?
  • How specific to September are these results? Would we see different data and trends if we carried this out at another time period?
  • September was a key month for Wiki Loves Monuments. How should we consider campaign calendars when we query given that certain large campaigns have peak activity on particular months?
  • How does this data differ from data related to participants from different campaigns?
  • How different are the results if we query for data immediately after a campaign ends?

@ldelench_wmf I think it would be a good idea to organize a meeting to share the findings that Irene has collected so far on campaign activity in SSA.

Tagging @fnartey & @Astinson: The early findings show the vast majority of SSA campaign activity from this small sample size (which is just for Sept 2021) occurring in Ghana and Kenya, with no edits that we know of coming from the 4 countries we have selected for research purposes. This doesn't necessarily mean that we shouldn't conduct research in those countries, or that the small sample size represents all trends, but we should be aware of this dynamic and discuss pros/cons. It would be good to share in our next design research check-in. Thanks!

@ifried am not worried about the edits not coming from those countries: we know that the organizers are there, and they are running events -- especially in DRC, Uganda and Rwanda -- the challenge is often seeing their contributions -- of organized activity in general -- and then also we know that the organizers have some serious efficacy gaps in editing events in SSA (its hard to identify productive & persistent audiences, retaining those audiences is challenging, and leveling them up to the right skills is also challenging). We are also missing some of the most important campaigns for the region so far, WikiLoves Africa, Wiki Loves Women and WikiGap (at least according to the checklist).

@Astinson & @ifried

I've updated the checklist of data included in this investigation. WikiLoves Africa, Wiki Loves Women and WikiGap data are included in the data collected.

A note from Jaime on the GDI team:
In the SSA group, as far as edit counts, the largest group is editors that have made 0-50 edits. We may be looking at micro-contribution editors in that group...we may also be looking at different bandwidth use or other impacts of internet access/plans and editors that save less often to save on data use.

If this aspect is picked up for further investigation in the future, it may be useful to better understand the edits and see whether these edits were single image uploads, micro-contributions, or comprised of single edits with many components.