Information Technology Reference
In-Depth Information
Participants' education had significant influence on
clarity
Q
8
and
insights
Q
9
.
In particular, oce clerks reported to have a clearer goal understanding (o-
clerks=3.98, i-clerks=3.05, p=.031) and get more new insights into process think-
ing (o-clerks=4.05, i-clerks=3.30, p=.022). In all cases, the confidence intervals
left no doubt about the effect.
6
Interpretation of Results
We can identify three types of variables. Those that support their hypothesis,
those that do not support their hypothesis, and those that just missed rigor
scientific standards. We consider the latter ones as conditionally supportive and
argue that a slightly larger sample set would have made the difference.
This claim is based on the t-test in Table 2. It indicates a significant dif-
ference for
talking
V
1
,
fun
Q
3
,and
reviews
V
6
due to method. The confidence
intervals do not allow acceptance with scientific rigor. That means, we can-
not rule out with 95 percent probability that the actual effect size is zero.
For example,
talking
V
1
time is significantly higher (p=.044) in t.BPM sessions
(t.BPM=4.65min, int=3.49min) but the confidence interval includes zero (lb=-
0.19min, ub=2.52min). In this case we miss rigor acceptable levels by twelve
seconds. The rest of the discussion is structured according to the hypothesis
decomposition in Fig. 2.
The engagement variables to measure activation indicate a positive effect
through method. Participants in t.BPM sessions did spend more time talking
(
F
0
.
5
(1
,
15) = 3
.
34,
η
2
=0
.
18) and significantly more time thinking (
F
0
.
5
(1
,
15) =
24
.
93,
η
2
=0
.
62) about their process. They also report more fun (
F
0
.
5
(1
,
16) =
3
.
24,
η
2
=0
.
17) in t.BPM sessions. We reject
H
02
and argue that
H
01
and
H
03
might be rejected with a bigger sample size.
The engagement variables measuring the dimension of identification did not
hold. People did not report significantly more motivation or commitment to their
solution due to the method used. We assume a ceiling effect for
motivation
Q
4
.
Participants got off from class, plus a chocolate bar and a cinema voucher for com-
pensation. On a five point Likert scale we could not find a statistically relevant
difference in
motivation
Q
4
due to the method applied (t.BPM=4.45, int=4.37).
For
commitment
Q
5
we found in Section 5.5 that it significantly raises with rep-
etition (p=.004). Thus, we assume that commitment (as operationalized by us)
indicates self-confidence that raises with due to the learning effect. We do neither
reject
H
04
nor
H
05
.
The variables that operationalize the aspect of validated results show a mixed
picture. We note more reviews (
F
0
.
5
(1
,
15) = 3
.
95,
η
2
=0
.
21) and significantly
more corrections (
F
0
.
5
(1
,
15) = 46
.
3,
η
2
=0
.
76) due to the method. We reject
H
07
and argue that
H
06
might be accepted with a sightly larger sample size. We
conclude that t.BPM provokes more feedback in process elicitation sessions.
The competencies required for result validation rely on perceived measures.
We see that people report significantly more insights into process thinking
(
F
0
.
5
(1
,
15) = 5
.
36,
η
2
=0
.
25) in t.BPM sessions but the goal clarity does not
