As the second part of the lecture on qualitative data analysis we discussed the need to cross-validate the collected insights. In this presentation I show what are the different approaches to data triangulation and how I applied them in my research work.
2. Any bias inherent in particular data sources,
investigator and method would be neutralised when
used in conjunction with other data sources,
investigators and methods.
(Creswell, 1994: 174)
3. Any bias inherent in particular data sources,
investigator, and method would be neutralised
when used in conjunction with other data sources,
investigators and methods.
(Creswell, 1994: 174)
It assumes that data from different methods will
corroborate one another, where the choice of
methods is intended to investigate a single social
phenomenon from different vantage points.
(Denzin,1970; Brannen, 2005)
4. Data collected from different methods
cannot be simply added together
to produce a unitary or rounded reality.
(Brannen 2005: 176)
5. - corroboration: the same results are derived from both
qualitative and quantitative methods
- elaboration: qualitative data analysis explains how the
quantitative results can be applied
- complementarity: qualitative and quantitative results
differ but when put together they generate coherent
insights
- contradiction: qualitative data and quantitative
findings conflict
7. supporting
communication
at work
Szostek, Agnieszka Matysiak, et al. "Understanding the implications of social
translucence for systems supporting communication at work." Proceedings
of the 2008 ACM conference on Computer supported cooperative work.
ACM, 2008.
8. context
social ways to initiate
communication in
face-to-face settings
technical ways to initiate
communication in
mediated settings
9. study objective
- what is a successful way to achieve
visibility of one’s communicative
state?
- what else is required to make a system
become socially translucent?
21. participants equally often indicated status
using 4,3,2 or 1 slider
availability slider alone used only 1% of times
all sliders used equally often
availability messages: to indicate availability
contextualized availability messages: to indicate
unavailability
contextual messages: to remain ambiguous
22. co-discovery interviews with RGT
- manual setting of availability as a way to control
‘professional image
- AvBOX well depicting unavailability sufficiently
ambiguous
- StatusME uninformative or privacy threatening
- need for awareness to know by whom and how
often their status was checked
- need for accountability to notify that
communication was poorly timed
23. 731 logged interactions
485 with AvBox
246 with StatusME
StatusME (Mean)
AvBox (Mean)
Week 1 3.76 2.26
Week 2 2.75 1.48
Week 3 5.58** 2.81**
Overall 4.03* 2.18*
* significant at p < .01 ** significant at p < .005
29. - Theoretical Triangulation: looking at the research situation from
different theoretical perspectives
- Methods Triangulation:
- one researcher using two or more research techniques
(within and between quantitative-qualitative techniques);
- two or more researchers using the same research
technique;
- two or more researchers using two or more research
techniques.
30. - Theoretical Triangulation: looking at the research situation from
different theoretical perspectives
- Methods Triangulation:
- one researcher using two or more research techniques
(within and between quantitative-qualitative techniques);
- two or more researchers using the same research
technique;
- two or more researchers using two or more research
techniques.
- Data Triangulation: combining qualitative and quantitative data
within the same method
32. - sequential implementation: the researcher
collects both quantitative and qualitative data
in phases
- concurrent implementation: the researcher
collects both quantitative and qualitative data
at the same time
33. - equal priority: the same weight is given to
quantitative and qualitative data
- dominant priority: priority is give to either
quantitative and qualitative data
34. - integration of quantitative and qualitative data occurs
at different stages of the research process:
- during data collection
- during data analysis
- during interpretation
- or in combination of places
35. - sequential explanatory strategy: using
qualitative results to explain and interpret
the findings of a primarily quantitative study
- sequential exploratory strategy: using
quantitative data to support qualitative
findings
- concurrent triangulation strategy: running
both abovementioned strategies in parallel
to cross-validate or corroborate the
obtained results
36. - within qualitative methods triangulation:
combining different qualitative methods, eg.
observations, interviews and creative workshops to
validate the results
- quantification of qualitative data: running
quantitative analysis of the qualitative data
37. dealing with
email overload
Szóstek, Agnieszka Matysiak. "‘Dealing with My Emails’: Latent user needs in
email management." Computers in Human Behavior 27.2 (2011): 723-729.
42. - too many emails in the inbox
- too many folders
- too many emails that do not require response
- using email as task manager
- checking email at different times of the day
49. - evaluate which quality differentiates two chosen inbox
concepts from the third one, e.g.: ‘Managing this (traditional)
inbox is effortless as it doesn’t allow but also doesn’t require
any action from me.”
- after determining a particular quality define its other polar, e.g.:
‘Managing inboxes allowing for restructuring emails like the
two I designed might require quite some effort, so I can end up
spending more time arranging my emails rather than
answering them.’
- finally assess which of these qualities is a positive quality in the
context of inbox design, e.g.: positive - effortless and negative –
effortful
51. qualitative analysis
- qualitative content analysis of the narratives
- all paired comparisons open-coded while preserving their
positive or negative affiliation by two independent coders
- formulation of two main categories defining two distinct
phases in email management: email organization and email
retrieval
- identifying three types of user needs for each category
- mapping each statement to a relevant category from the
classification scheme
- creation of two mappings for statements pointing at a causal
dependency between two needs
52. quantitative analysis
- choosing two indices to compare the relative salience of the
identified needs: importance and dominance
- importance measured by the order in which one need was
mentioned in relation to all other needs
- dominance computed based on a normalized index ranging
from 0 to 1, where a value of 0 identified a need reported first
and averaged it for all references related to the same group of
needs
55. reliable inbox structure
‘It has a structure that remains consistent over time, so I
don’t need to learn it over and over again.’
‘The structure should not be complicated and have too
many rules, because if I forget them I can have
difficulties finding an email that I am looking for.’
56. no obligation to classify
‘It doesn’t force me to annotate my messages right
away. I wouldn’t know how to classify many emails
right after their arrival.’
‘It forces me to classify messages. It might be very
difficult to categorize many emails right away as it is
difficult to imagine what an email may imply in the
future.’
57. contextual email information
‘These inboxes allow to visualize thematic priority rather
than priority of each individual email, give context to a
project or an activity and help me to see if I follow
everything per theme.’
‘There is no relationship with my activities visualized. It is
difficult to see what was the last information in a
discussion or where this information is located. It loses
the continuity between emails, so it requires extra effort
to find the email I am looking for.’
58. sorting flexibility
‘Sorting only according to the arrival date lacks the
overview regarding the problems and cases; it takes
into account only one attribute of emails (time).’
‘Sorting according to tasks allows to quickly get an
overview of different cases, shows more than one
view on a specific case; uses two or more attributes of
emails at the same time - like subject and time.’
59. possibility to annotate information
‘An automatic structure of the inbox implies no effort to
organize my emails.’
‘Emails can be arranged, I can change their order and
have them grouped in a customized rather than
predefined way. The structure is more flexible, I can
change it if I want to but I don’t have to do so if I don’t
want to.’
60. efficient search engine
‘It is easy to remember when an email arrived, which
gives a good starting point for email search.’
‘Annotating emails results in higher awareness of their
content and therefore it gives better means to
memorize and to search old emails.’