1 Introduction
-
the lack of meeting structure narrows down the discussion;
-
a few vocal participants dominate the discussion, while others only listen even though they have profound views on things;
-
participants use the meeting primarily to complain rather than to improve.
-
a proposal of a generic model of running a retrospective game, which is based on the group creativity model (Nijstad and Paulus 2003) as well as the state-of-the-art in group idea generation;
-
a proposal of a systematic methodological approach to choose and adopt retrospective games that suit the Scrum team the best and have a chance to improve their retrospective practice;
-
an in-depth discussion of the promises and realities of game-based retrospectives, including the advantages and disadvantages of six implemented games as well as eight lessons learned that were identified during our Action Research project;
-
a refinement of the Action Research methodology, which consists of a revised validity system for Action Research studies in software engineering, and a proposal of a new research design borrowed from Case Study.
2 Theoretical and Practical Background
2.1 Agile retrospective
-
What went well, that if we don’t discuss, we might forget?
-
What did not work and how can we improve it?
-
What will we commit to improve in the next iteration?
2.2 Team creativity
-
social loafing (free-riding) – the tendency of individuals to rely on the efforts of others to accomplish the task (Karau and Williams 1993);
-
downward matching – the tendency of individuals to match their performance to the least productive group member (Paulus and Dzindolet 1993);
-
evaluation apprehension – people often refrain from expressing ideas that they think are too controversial, weird, or unrealistic, because they fear they will receive negative evaluations from others (Diehl and Stroebe 1987);
-
production blocking – only one person can speak at a time, while the other group members must sit passively waiting for their turn; in the meantime, they may forget ideas or lose their motivation to share (Nijstad et al. 2003).
2.3 Serious games
-
competitive games, where individuals form strategies that directly oppose to others;
-
cooperative games, where players have interests that are neither completely opposed nor completely coincident; hence, opportunities exist for cooperation but not necessarily for equal reward;
-
collaborative games, where all players work together as a team, sharing a common goal, outcomes, and penalties.
2.4 Retrospective games
Game | Game board | Categories |
---|---|---|
Sailboat | an image of a sailboat, rocks, wind, an island | Island – the team’s goal; Anchor – everything that slows the team down; Rocks – the risks the team might encounter; Wind – everything that helps the team reach their goal |
Mad/Sad/Glad | three columns | Mad – frustrations, issues that have annoyed the team and/or have wasted a lot of time; Sad – disappointments, issues that have not worked out as well as was hoped; Glad – pleasures, issues that have made the team happy |
Mood++ | five columns | * – all categories from Mad/Sad/Glad; Flowers – appreciation to teammates who have done something magnificent for the team or a particular team member; Ideas – suggestions how to improve the teamwork or the process |
Starfish | an image of a starfish with five arms | Stop Doing – activities or practices that do not add value, or even worse, are hindrances to progress; Less Of – activities or practices that have been done and have added value but have required more effort than really needed; Keep Doing – activities or practices that the team is doing well and wants to keep; More Of – activities or practices that are useful but not fully taken advantage of; the team believes that they will bring more value if they are done even more; Start Doing – activities or practices that the team believes will bring value and will improve current processes |
5Ls | five columns | Liked – what did the team really appreciate about the Sprint? Learned – what new things has the team learned during the Sprint? Lacked – what things could the team have done better in the Sprint? Longed For – what things did the team wish for but were not present during the Sprint? Loathed – what things did the team dislike in the Sprint? |
-
each team member writes down on a sheet of paper what he/she appreciates;
-
team members form a circle;
-
one participant sits in the center of the “circle of recognition”;
-
everyone in the circle reads the appreciation feedback to the team member in the center (complete the 360 degrees);
-
team members change in the center until everyone has received feedback.
3 Related work
4 Study design
4.1 Research method
-
Diagnosing. The current organizational situation is diagnosed in order to identify the primary problems that are the underlying reasons for change (Baskerville 1999).
-
Action planning. The researchers together with practitioners plan actions to address the identified problems.
-
Action taking. The planned intervention is implemented.
-
Evaluating. The actors determine whether the expected effects of the intervention were achieved, and whether these effects remedied the problems (Baskerville and Wood-Harper 1998).
-
Specifying learning. The actors reflect on the intervention and its effects and document findings that will add to the body of knowledge of the discipline. They also reach a decision whether to finish the project or to proceed through an additional process cycle (Davison et al. 2004).
The Principle of the Researcher–Client Agreement (RCA) | |
It provides a guiding foundation for an AR project. By specifying the scope and objectives of the project as well as the co-operation between the researcher and client and involving key stakeholders in its creation, the agreement constitutes the research environment and promotes a spirit of shared inquiry. | |
The Principle of the Cyclical Process Model (CPM) | |
It requires the project to be conducted in a precise and sequential manner according to the five phases model by Susman and Evered (1978). | |
The Principle of Theory | |
It emphasizes the importance of applying one or more theories to guide the project activities and relate the findings to existing theories (Davison et al. 2004). | |
The Principle of Change through Action | |
It reflects the essence of Action Research, which is to implement intervention in order to change the current situation and its unsatisfactory conditions. | |
The Principle of Learning through Reflection | |
It guarantees that both researchers and practitioners draw insights from what they have learned and identify implications for other situations and contexts. |
OKE Poland | Dynatrace | SentiOne | |
---|---|---|---|
top-level management | CEO | CTO, COO | |
middle-level management | development lead | ||
low-level management | team leader | team leader | team leader |
4.2 Research questions
4.3 Research context
OKE_A | 1x Lead Programmer & SM | OKE_B | 1x Lead Programmer & SM & Proxy PO |
2x Junior Programmers | 1x Senior Programmer | ||
1x Tester & PO | 3x Programmers | ||
1x Tester | |||
Dyna_A | 1x Team leader & PO & SM | Dyna_B | 1x Lead Developer & SM & PO |
3x Senior Developers | 1x Senior Developer | ||
4x Developers | 3x Developers | ||
2x Junior Developers | |||
Senti_A | 1x PO | Senti_B | 1x PO |
1x Quality Manager | 1x Project Coordinator | ||
1x Product Designer | 4x Senior Research Engineers | ||
2x Senior Developers | 1x Junior Research Engineer | ||
3x Developers | 1x Linguist | ||
1x Tester |
4.3.1 OKE Poland (https://oke.pl)
4.3.2 Dynatrace (https://jobs.dynatrace.pl)
4.3.3 SentiOne (https://sentione.com/)
4.4 Data collection and analysis
AR phase | Data collection technique | Data source | Objective |
---|---|---|---|
Diagnosing | unstructured interview | team leaders and/or scrum masters | To find the adoption of Scrum practices in the participating teams. |
semi-structured interview | team members | To identify retrospective problems in the participating teams. | |
Action taking | participant observation | researchers | To take a close look at how the game-based retrospectives proceed. |
Evaluating | questionnaire | team members | To assess the adopted games. |
focus group | team members | To discuss the results of the intervention and collect “lessons learned” recommendations. |
5 Action research cycle
5.1 Diagnosing
Team; # team members | Scrum compliance | Sprint length; development / maintenance |
---|---|---|
OKE_A; 4 | No Daily Scrum due to the following reasons: • team members usually have only a few hours of overlapping time during the workday, • team members are distributed among different rooms in the office space, • the tester and the team leader are also engaged in other projects. Sprint Planning is held at the beginning of each sprint. Sprint Review and Sprint Retrospective are merged into one meeting, which usually takes place on a regular basis at the end of each sprint. | 1 week; 40% / 60% |
OKE_B; 6 | Daily Scrum is held every day. Sprint Planning is done by the Product Owner and the team leader, while the results are communicated to the developers. Sprint Review meeting is replaced by a Sprint Review report in which every developer describes the functionality that he/she has implemented in the sprint. The report is sent to the Product Owner. Sprint Retrospective is usually skipped, because the team constantly works under time pressure to deliver the increment. | flexible, 1 week during the research; 50% / 50% |
Dyna_A; 10 | Strictly adhere to Scrum, but they are forced to do it. Sometimes a simplified version of Mad/Sad/Glad or Starfish has been used during the Sprint Retrospective, but only a few team members consider this approach useful. We traced the underlying cause of the problem to the replacement of sticky notes for writing directly on the whiteboard, which leads to the situation where only part of the team is involved in writing. Besides, almost half of the team members would not attend retrospectives if their attendance was not obligatory. | 2 weeks; 40% / 60% |
Dyna_B; 5 | Daily Scrum, Sprint Planning and Sprint Review are conducted in accordance with the Scrum Guide. Sprint Retrospective is done at times and there is no specific time for it. The meetings are considered rather boring, because they do not go beyond the typical three retrospective questions. Nevertheless, only one team member would skip the meeting if it was not obligatory. | 2 weeks; 65% / 35% |
Senti_A; 9 | Daily Scrum, Sprint Planning and Sprint Retrospective are conducted in accordance with the Scrum Guide. Moreover, all team members willingly participate in retrospectives. Sprint Review is replaced by a Demo meeting, which is organized every quarter by the Product Team to update the stakeholders about recent key changes and to plan for the next months. In addition, Backlog Refinement is organized every sprint to estimate the cards in the Product Backlog. | 2 weeks; 70% / 30% |
Senti_B; 8 | Daily Scrum, Sprint Planning and Sprint Review are conducted in accordance with the Scrum Guide. Sprint Retrospective is held very rarely, because the team works constantly under the pressure of grant deadlines. In addition, Backlog Refinement is organized every sprint to estimate the cards in the Product Backlog. All meetings are added to the calendar, however they are rescheduled very often. | 2 weeks; 90% / 10% |
-
Do your retrospectives bring added value or are they a waste of time?
-
Are your retrospectives too repetitive?
-
Are your retrospectives thoroughly prepared and well structured?
-
Do all participants contribute to the discussion or it is dominated by a few vocal people?
-
Is there a tendency to view your retrospectives as a chance to complain instead of a chance to improve?
-
Would you attend retrospectives if your attendance was not obligatory?
-
the same ideas come up every time;
-
there is not enough to talk about based on a single Sprint;
-
there is no good discussion on how to improve the teamwork;
-
wasting time on long, ineffectual debates on one issue while neglecting other issues due to a lack of time;
-
difficulties in writing down ideas expressed in long unstructured oral statements;
-
retrospectives are too short to come up with solutions to complex problems, so additional meetings should be planned;
-
it is hard to prioritize issues and focus on the most important ones, because team members want to discuss and explore potential solutions for all of the problems in one meeting.
Retrospective problem | OKE_A | OKE_B | Dyna_A | Dyna_B | Senti_A | Senti_B |
---|---|---|---|---|---|---|
little added value | ++ | + | +++ | ++ | + | n/a |
too repetitive | +++ | ++ | + | +++ | +++ | n/a |
lack of structure / preparation | + | + | + | +++ | n/a | |
unequal participation | ++ | +++ | ++ | + | n/a | |
too many complaints | + | ++ | + | ++ | n/a |
5.2 Action planning
5.3 Action taking
5.4 Evaluating
5.4.1 Questionnaires
5.4.2 Focus groups
-
team codes: all teams, OKE_A, OKE_B, Dyna_A, Dyna_B, Senti_A, Senti_B
-
object codes: game-based retrospective, questionnaire results, all games, Sailboat, Mad/Sad/Glad, Mood++, Starfish, 5Ls, 360 Degrees Appreciation
-
description codes: advantage, disadvantage, recommendation, explanation, like, dislike, improvement, declination, surprising result, unsurprising result, discovery, preference.
unit of meaning | team code | object code | description code |
---|---|---|---|
The approach “different game at each meeting” is not good, because I do not like if something changes and there is a need to learn from scratch. {Another person agreed on this opinion.} | Senti_B | game-based retrospective | dislike, preference |
It is worth changing the game after a longer time of usage. {A few people agreed on this opinion.} | Senti_B | game-based retrospective | recommendation |
One game was unusual. It was about praising each other. {The facilitator recalled that it was 360 Degrees Appreciation.} This game is cool, because it gives a different view, and it is easy to understand. | Senti_B | 360 Degrees Appreciation | like, explanation |
However, if we ran this game all the time, that wouldn’t be good. This game is worth conducting once in a while. | Senti_B | 360 Degrees Appreciation | recommendation |
The variety of games baffled us. | Senti_B | game-based retrospective | dislike |
It’s okay to change games from time to time, because different teammates have different preferences. If we stick to one game, some teammates may feel excluded. | Senti_B | game-based retrospective | recommendation |
It’s worth conducting the same game for a few sprints to fully understand it and to memorize its rules and then switching to another game. {All teammates agreed on this opinion.} | Senti_B | game-based retrospective | recommendation |
Game | Advantages | Disadvantages |
---|---|---|
Sailboat | - clear, distinctive categories - uses metaphors to stimulate creativity - helps team members to align goals | - may become boring if used too often, because the vision and risks rarely change through the project |
Mad/Sad/Glad | - lets bad emotions and toxic feelings out - simple rules - not time-consuming | - Mad category is too similar to Sad category - restricts the discussion |
Mood++ | - makes team members kinder toward each other - allows the team discuss any ideas | - time-consuming |
Starfish | - names of the categories explicitly call to actions | - sometimes it is not obvious to which category an idea should belong - time-consuming |
5Ls | - names of the categories are confusing for non-English natives - fuzzy boundaries between the categories - time-consuming | |
360 Degrees Appreciation | - strengthens team relationships and trust - fosters self-esteem - alleviates conflicts | - a team building activity rather than a Sprint retrospective - precludes constructive criticism |
5.5 Specifying learning
5.5.1 Lesson 1: Retrospective games provide meeting diversity, which encourages new perspectives and helps teams bring up new ideas
5.5.2 Lesson 2: Retrospective games drive participants into a collaborative mindset, but do not lead to breakthrough findings
5.5.3 Lesson 3: Retrospective games encourage equal participation
5.5.4 Lesson 4: Retrospective games organize both the meeting as well as the discussion
5.5.5 Lesson 5: Not all agile team members have an agile mindset
5.5.6 Lesson 6: If participants are not afraid to voice their opinions, data obtained through a focus group is more reliable than through a questionnaire
5.5.7 Lesson 7: Scrum masters should have a toolbox of possible retrospective games and help their teams empirically determine which games are effective for them
5.5.8 Lesson 8: Organizational culture and managers’ mindsets are still significant barriers to the successful adoption of agile practices
6 Threats to validity
-
construct validity reflects to what extent the operational measures truly represent the underlying realities of the intervention;
-
internal validity describes the extent to which the observed effect (intervention outputs) is caused only by the intervention;
-
external validity concerns to what extent the results of the study can be applied to other settings, and to what extent the results are of interest to other people outside of the participating organizations;
-
reliability considers the extent to which the data and the analysis are dependent on the specific researchers;
-
statistical conclusion validity refers to the appropriate use of statistics to infer about the correlation between treatment and outcome.