Belal M. Hijji, RN. PhD,

1
An Examination of the Routine Blood Transfusion
Knowledge Questionnaire Through Item Analysis
Belal M. Hijji, RN. PhD, Arwa Owies, RN. PhD
Faculty of Nursing, Philadelphia University,
Jordan
Presented in
The second International Conference for Allied
Health Care Professionals, Physicians,
Physiotherapists, and Technicians, April 7-8
2010, Alexandria, Egypt
2
Contents

• Background
• Aim and objectives
• Methods
• Results
• Summary of results and discussion
• Conclusion
• Acknowledgment
• References

3
Background
4
The Routine Blood Transfusion Knowledge
Questionnaire (RBTKQ) was developed (Hijji, 2007)
as a measure to investigate nurses level of
knowledge and practice of blood transfusion
The RBTKQ includes has true-false items (three),
multiple-choice (MC) (20), and multiple-response
(MR) (10)
For the purpose of this analysis, five sections
of the RBTKQ that are relevant for inclusion (A)
is about issues related to patient preparation
before blood bag collection, (B) is concerned
with blood bag collection section, (C) is related
to pre-transfusion initiation nursing activities,
(D) addresses post-transfusion initiation
activities, and (E) is related to complications
of blood transfusion
5
Since its inception, the RBTKQ has not undergone
any form of item analysis to evaluate its
psychometric properties
An examination of empirical item response data
maybe appropriate in detecting item flaws
(Crocker and Algina, 1986), and provides useful
information about test-item quality (Oermann
Gaberson, 1998).
6
AIM
7
Principal Aim
The main aim of this study was to determine the
overall quality of the items and the test
Objectives A. Identify the items difficulty p
and discrimination D levels B. Investigating
the performance of distractors C. Find out
whether the items discriminate between nurses
whose scores place them in the top 30 of all
nurses and those whose scores place them in the
lower 30 of all nurses. This figure provides a
stable index of differences between high and low
ability groups (Crocker Algina, 1986)
8
Methods
9
Data Collection
Responses to the item were taken from the RBTKQ
that was filled out by a random sample of 305
nurses (95.3). Analysis was performed for p, D,
and for the distractors. Because the p value is
unavailable with MR items (http//www.cat.ilstu.ed
u/services/opscan/opFAQStatQ.php), each one of
these was broken down to a number of sub-items
equal to the number of its options. Based on a
nurses selection of an option, a label of yes
or no, or true or false was assigned. The
outcome of this procedure culminated in a total
of 89 items for the whole questionnaire.
10
Data Analysis
scoring
For most items, one point was awarded for the
correct answer and 0 for the incorrect one. The
maximum possible score was 57 (100) points
Scores were arranged in descending order from the
highest to the lowest
The number of nurses subpopulations was 91 for
each group. The upper groups scores ranged from
32 (56) to 40 (70) lower group from 8 (14) to
27 (48). The mean knowledge score for the sample
is 29.6 (SD 4.14). A passing score of 34 (60)
(one standard deviation above the mean) was set
up.
11

• For each item, data analysis details
• the p value (whole sample). The p value is set at
  0.5 to 0.85 (Lin et al., 1999).
• the p value of each option selected by both
  groups
• the difference in proportions D between nurses
  subgroups selecting each option. A D value of
  0.20 or above is appropriate (Crocker and Algina,
  1986 Brown, 1983) for the correct option.
• the number of omits

12
Results
13

• Due to time restrictions, there will be partial
  presentation of the results

14
Table 1. Computations of item analysis indices
for sections A and C
Lower 30 Upper 30 Percentages
Section A
Options
p Omit F T Key Item
.47 1-1 55-28 35-62 F 1
.60-.31 .40-.69
.29 -.29 D (h l)
.30 3-2 35-24 53-65 F 2
.38-.26 .61-.71
.12 -.13 d
.33 1-5 41-22 49-64 F 3
.45-.24 .55-.76
.21 -.21 d
Section C
.89 0-0 4-19 87-72 T 1
.04-.21 .96-.79
-.17 .17 d
.81 0-0 78-66 13-25 F 2
.86-.73 .14-.27
.13 -.13 d

1 Difficulty index for the top and lower 30 of
nurses 2 This is the difference between the
proportions of nurses subpopulation. The h
stands for high scorers l for low scorers
15
Table 2. Computations of item analysis indices
for section D
D
Lower 30 Upper 30 Percentages
Options
p Omit F T Key Item
.33 0-1 52-71 39-19 T 1
.57-.79 .43-.21
-.22 .22 d
.83 0-1 13-15 78-75 T 2
.14-.18 .86-.82
-0.04 .04 d
.68 0-1 68-52 23-38 F 3
.75-.57 .25-.43
.18 -.18 d
.47 0-1 50-36 41-54 F 4
.55-40 .45-.60
.15 - .15 d
.71 0-1 14-44 77-46 T 5
.15-.49 .85-.51
-.34 .34 d
16
Table 4. Computations of item analysis for
sections A, B, and C
Lower 30 Upper 30 Percentages
Section A
Options
p Omit E D C B A Key Item
.12 0-1 -- 22-4 7-13 5-7 57-66 D 4
.24-.05 .08-.14 .05-.08 .63-.73
.19 -.06 -.03 -.10 d
Section B
.51 1-5 - 2-3 31-47 1-10 56-26 A 1
.03-.09 .34-.52 .01-.11 .62-.29
-.06 -.18 -.10 .33 d
.46 0-1 11-16 29-25 2-8 1-6 48-35 A 2
.12-.18 .32-27 .02-.09 .01-.08 .53-.38
-.06 .05 -.07 -.07 .15 d
.09 2-1 6-1 10-37 11-7 6-6 36-40 C 3
.07-.01 .11-.41 .12-.08 .09-.08 .40-.44
.06 -.30 .04 0.01 -.04 d

1 Omits were considered incorrect responses
(Oermann Gaberson, 1998). Here omits were
applied to option D.
17
Table 4. (Continued.)
Section C
Options
p Omit E D C B A Key Item
.30 0-0 5-6 4-3 39-15 8-16 35-51 C 1
.05-.07 .04-.03 .43-.16 .09-.18 .38-.56
-.02 .01 .27 -.09 -.18 d
.38 0-1 0-3 0-9 21-32 19-25 51-22 A 2
0-.03 0-.11 .23-.35 .21-.27 .56-.24
-.03 -.11 -.12 -.06 .32 d
.08 1-1 0-0 13-7 2-8 19-30 56-45 D 3
.14-.08 .03-.10 .21-.33 .62-.49
.06 -.07 -.12 .13 d
.90 1-0 88-70 0-3 1-5 0-7 1-6 E 4
.97-.77 0-.03 .01-.05 0-.08 .01-.07
.20 -.03 -.04 -.07 -.06 d
.27 13-10 - 7-13 25-14 16-22 30-33 C 5
.22-.25 .27-.15 .18-.24 .33-.36
-.03 .12 -.06 -.03 d
.70 1-0 - 0-4 81-50 3-15 6-22 C 6
0-.04 .89-.55 .03-.16 .07-.24
-.04 .34 -.13 -.17 d
18
Table 5. Computations of item analysis indices
for sections D
Lower 30 Upper 30 Percentages
Options
p Omit E D C B A Key Item
.32 3-1 - 4-12 8-15 31-21 37-37 B 1
.08-.14 .09-.16 .34-.23 .41-.41
-.06 -.07 .11 0 d
.05 1-5 9-2 1-3 2-5 4-12 74-64 E 2
.10-.02 .02-.09 .02-.05 .04-.13 .81-.70
.08 -.07 -.03 -.09 .11 d
.14 1-2 - 4-6 8-10 20-17 48-42 C 3
.05-.09 .10-.13 .25-.22 .59-.55
-.04 -.03 .03 .04 d
.45 1-2 - 8-8 61-30 15-17 6-34 C 4
.10-.11 .67-.33 .16-.19 .07-.37
-0.01 .34 -.03 -.30 d
.45 0-4 - 13-12 28-23 41-37 9-15 B 5
.14-.13 .31-.25 .45-.41 .10-.21
.01 .06 .04 -.11 d

1 Item not applicable to 22 nurses 2 Item not
applicable to 76 nurses
19
Table 5. Computations of item analysis indices
for sections E
Lower 30 Upper 30 Percentages Item
Options
p Omit E D C B A Key
.50 0-0 - 11-15 59-34 5-29 16-13 C 1
.12-.16 .65-.37 .05-.32 .18-.14
-.04 .28 -.27 .04 d
.45 0-2 14-17 7-21 51-31 11-15 8-5 C 2
.15-.19 .08-.23 .56-.34 .12-.16 .09-.08
-.04 -.15 .22 -.04 .01 d
.09 0-3 0-1 4-6 9-5 3-13 75-63 C 3
0-.01 .04-.10 .10-.05 .03-.14 .82-.69
-.01 -.06 .05 -.11 .13 d
.29 3-2 - 44-15 27-41 15-21 2-12 D 4
.48-.16 .30-.45 .16-.23 .05-.15
.32 -.15 -.07 -.10 d
.65 0-3 - 1-6 6-31 79-39 5-12 B 5
.01-.10 .07-.34 .87-.43 .05-.13
-.09 -.27 .44 -.08 d

20
Summary of Results and Discussion

• Item difficulty
• Eight items (27) had p value of .50 to .85
  (neither too difficult nor too easy)
• Two items (6) had p value of gt.85 (easy)
• 20 items (67) had p value of lt .50 (difficult).
  This could be, most probably, because nurses did
  not master the content
• All distractors were selected by nurses and only
  a few of them, sometimes, omitted responding to
  some items.
• Sixteen items (53) had a D value of .20. This
  indicates that the items were especially easier
  for the top group and more difficult to the lower
  group.
• The last two results are good indicators of the
  quality of item (Osterland, 1998, Oermann
  Gaberson, 1998). Functional distractors decrease
  the chances that low ability nurses select the
  correct response by guessing which would
  introduce measurement error (Crocker Algina,
  1986)

21

• Fourteen items (47) had a D value of lt .20.
• The test was not designed to maximise variance
  among groups. In essence, D relies on a
  fundamental assumption, which is that examinees
  who exhibit mastery on the subject matter are
  presumed to be more likely to respond correctly
  to an item than low ability examinees (Osterland,
  1998). In this study in fact, none of the nurses
  per se mastered the content of interest.
• The differences between both groups for the
  distractor were, mostly, negative values. This
  indicates that fewer upper group nurses selected
  the distractor than did the lower group nurses.
• This is another positive feature of items
  (Osterland, 1998)
• However, the differences for some distractors in
  eight items were scant ( -.04). For two items,
  the differences were positive indicating that
  higher ability nurses selected distractors more
  often than low ability nurses.
• In this case, Osterland (1998) recommends
  reviewing the options for possible improvement.

22

• For a number of items (n 13, 43), nurses
  responses were widely scattered among the
  responses alternatives, and/ or the correct
  responses attracted less attention than the
  distractor(s).
• This result may indicate that nurses could not
  understand the items or that the items covered
  unfamiliar content. Therefore, these items may
  need to be reviewed (Crocker Algina, 1986).

23
Conclusion

• This study has highlighted the need for
• Jordanian nurses to master the content related
  knowledge and practice of blood transfusion.
• Examining the questionnaire items for possible
  improvement
• However, no educator should retain, revise, or
  eliminate an item on the basis on item statistics
  alone (Oermann Gaberson, 1998).

24
Acknowledgements
25
We are immensely grateful to the Ministry of
Health, all nursing staff, and hospitals
management. We express our gratitude to
Philadelphia University Deanship of Scientific
Research and Postgraduate Studies for funding the
study.
26
References
27

• Brown, F. (1983). Principles of educational and
  psychological testing (3rd ed.). New York Holt,
  Rinehart and Winston.
• Crocker, L. and Algina, J. (1986). Introduction
  to classical modern test theory. Belmont CA
  Wadsworth.
• Hijji, B. (2007). Knowledge and practice of blood
  transfusion A survey of nurses in Abu Dhabi,
  United Arab Emirates. Unpublished PhD thesis, the
  University of Ulster.
• Lin, L. Tseng, H. and Wu, S. (1999). Item
  analysis of the registered nurse licensure exam
  taken by nurse candidates from vocational nursing
  high schools in Taiwan. Proc. Natl. Sci. counc.
  ROC (D). 9(1) 24-31.
• Oermann, M. and Gaberson, K. (1998). Evaluation
  and testing in nursing education. New York
  Springer.
• Osterland, S. (1998). Constructing Test Items
  Multiple-Choice, Constructed-Response,
  Perfromance, and Other Formats. 2nd edition.
  London Kluwer.

28
Thank You
Principal Investigator bhijji_at_philadelphia.edu.jo
bhijji1_at_gmail.com