Figure 8.6  Graph used by Hidi et al. (2002)  Figure 6.3

Handling Data

281

A graph very similar to a histogram is the bar chart. Bar charts are often used for 

qualitative or categorical data, although they can be used quite effectively with quantitative 

data if the number of unique scores in the data set is not large. A bar chart plots the 

number of times a particular value or category occurs in a data set, with the height of the 

bar representing the number of observations with that score or in that category. The Y-axis 

could represent any measurement unit: relative frequency, raw count, per cent, or whatever 

else is appropriate for the situation. For example, the bar chart in Figure 8.8 plots the 

number of positive and negative feedback statements before and after the reported 

intervention.

Figure 6.4

 Graph used by Hidi et al. (2002)

Figure 6.5 

Ivens’s box plot

 

Figure 8.7  Ivens’s (2007) box plot 

SCHOOL-BASED RESEARCH

132

holistic scores

1

2

3

4

pre-Intervention

post-Intervention

girl IP

girl IP

+M

boy IP

boy IP

+M

P (1,58) 

= 8.5

P 

< 0.01

Figure 8.6  Graph used by Hidi et al. (2002)

130

120

120

100

90

80

70

60

50

40

30

N

=

77

Study1

771

Study 2

41

Study 3

Study

SWB

Figure 8.7  Ivens’s box plot

09-Wilson-Ch-08.indd   132

8/31/2012   5:41:10 PM

Handling Data

282

Pie charts

A pie chart is a way of summarizing a set of categorical data. It is a circle which is divided 

into segments. Each segment represents a particular category. The area of each segment 

is proportional to the number of cases in that category. For example, O’Brien (2007) 

summarizes the percentages of types of names evaluated by students as more or less 

severe (Figure 6.7).

Scatter plots

A scatter plot is a useful summary of a set of two variables, usually drawn before working 

out a linear correlation coefficient or fitting a regression line (see Chapter 11). Scatter 

plots provide a good visual picture of the relationship between the two variables, and aids 

the interpretation of the correlation coefficient or regression model.

Each unit contributes one point to the scatter plot, on which points are plotted but not 

joined. The resulting pattern indicates the type and strength of the relationship between 

the two variables. Hayes et al. (2007) used a scatter plot to show the number of positive 

and negative comments plotted against on task behaviour (Figure 6.8).

Handling data

133

category. The Y-axis could represent any measurement unit: relative frequency, raw 

count, per cent, or whatever else is appropriate for the situation. For example, the bar 

chart in Figure 8.8 plots the number of positive and negative feedback statements 

before and after the reported intervention.

Pie charts

A pie chart is a way of summarizing a set of categorical data. It is a circle which is divided 

into segments. Each segment represents a particular category. The area of each segment 

is proportional to the number of cases in that category. For example, O’Brien (2007) 

summarizes the percentages of types of names evaluated by students as more or less 

severe (Figure 8.9).

Scatter plots

A scatter plot is a useful summary of a set of two variables, usually drawn before working 

out a linear correlation coefficient or fitting a regression line (see Chapter 11). Scatter 

plots provide a good visual picture of the relationship between the two variables, and 

aids the interpretation of the correlation coefficient or regression model.

Each unit contributes one point to the scatter plot, on which points are 

plotted but not joined. The resulting pattern indicates the type and strength of the 

Negative

Positive

Rates of feedback per min

statements per min

Feedback type

Focus method

2004

Focus method

2005

Satruaon and

Hantop 2000

Data set

1.20

1.00

0.00

0.60

0.40

0.20

0.00

Figure 8.8  Hayes et al. (2007) bar graph

09-Wilson-Ch-08.indd   133

8/31/2012   5:41:10 PM

Figure 6.6 

Hayes et al. (2007) bar graph

Handling Data

283

Figure 6.7 

Pie chart in O’Brien (2007)

Figure 6.8 

Hayes et al. (2007) scatter plot

SCHOOL-BASED RESEARCH

134

Sexual

42%

Family

5%

Family

59%

Family

1%

Individual

24%

Racial

29%

Racial

4%

Sexual

36%

789 EXTREMELY BAD (N = 96)

(green names)

858 NOT AS BAD (N = 96)

(yellow names)

Figure 8. 9  Pie chart in O’Brien (2007)

10

20

30

40

50

60

70

80

90

100

0

−40

−30

−20

−10

0

10

20

30

40

More Postitive

Statements

than Negative.

More Negative

Statements

than Postitive.

Feedback (Negative–Positive) and % on Task

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

* *

*

*

Figure 8.10  Hayes et al. (2007) scatter plot

relationship between the two variables. Hayes et al. (2007) used a scatter plot to 

show the number of positive and negative comments plotted against on task behav-

iour (Figure 8.10).

09-Wilson-Ch-08.indd   134

8/31/2012   5:41:11 PM

SCHOOL-BASED RESEARCH

134

Sexual

42%

Family

5%

Family

59%

Family

1%

Individual

24%

Racial

29%

Racial

4%

Sexual

36%

789 EXTREMELY BAD (N = 96)

(green names)

858 NOT AS BAD (N = 96)

(yellow names)

Figure 8. 9  Pie chart in O’Brien (2007)

10

20

30

40

50

60

70

80

90

100

0

−40

−30

−20

−10

0

10

20

30

40

More Postitive

Statements

than Negative.

More Negative

Statements

than Postitive.

Feedback (Negative–Positive) and % on Task

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

*

* *

*

*

Figure 8.10  Hayes et al. (2007) scatter plot

relationship between the two variables. Hayes et al. (2007) used a scatter plot to 

show the number of positive and negative comments plotted against on task behav-

iour (Figure 8.10).

09-Wilson-Ch-08.indd   134

8/31/2012   5:41:11 PM

Handling Data

284

Working mainly with words

If your data is mainly in the form of words, then you should read about how to analyse 

qualitative data before you start. The next section will help with ideas to present qualitative 

data.

The best way of moving from raw qualitative data, such as interview transcripts or journal 

entries, to meaningful understanding is through becoming immersed in the data. In other 

words, you need to try to look for themes that run through the data and then interpret 

the implications of these themes for your research project. These themes can be either 

discovered or uncovered. Use the constant comparative methods to analyse your data, 

Then read Box 6.1 to find out how Nardi and Steward (2003) analysed their interview 

transcripts and developed headings and codes, which they used in their second reading of 

the text (Figure 6.11).

Activity 6.2 

Constant comparative method: step by step guide

1.  Read  your  notes,  diaries,  interview  transcripts  (recordings),  notes  from 

observations etc. and highlight parts that you think are important ideas. Use 

different colours for different kinds of ‘important’ ideas. So, you interpret the 

text to identify those patterns or themes which underpin what people are 

saying. This is called coding.

2.  These are called temporary constructs. Make a list of them.

3.  Now read through your data again, comparing the data against your list of 

temporary constructs (this is the constant comparative bit).

4.  Now make a grid with the temporary constructs in a column on the left and 

on the right side note the page numbers where the temporary construct is 

mentioned in your data. You can make notes and observations on the grid 

as you do this.

5.  Delete any temporary constructs that are not ‘earning their keep’.

6.  After your second reading make a list of second order constructs that seem 

to explain your data. These ideas should help you to summarize important 

themes in your data.

7 (adapted from Thomas, (2010)

Handling Data

285

Box 6.1 

Inductive Analysis

Data analysis of the interview data 

Immediately after each interview, an Interview Protocol, a condensed account 

of the interview where the interviewees’ statements are reproduced from 

the audio-recording, not with verbatim accuracy but as faithfully and concisely 

as possible, was produced. The interviews were also fully transcribed and the 

contents of the audiotapes were digitized and copied on compact disks. Within 

a spirit of seeking data-grounded theory, as proposed in Glaser and Strauss 

(1967),  but  with  due  attention  to  foundational  theoretical  perspectives,  as 

indicated, for example, by Hammersley (1990), a first level of coding followed 

according to seven wide categories, as set out in Table I. A second-level coding 

of the now Annotated Interview Protocols led to the production of a Code 

System (a gradually enriched, eventually ‘saturated’ version of the preliminary 

one consisting of 36 T, 29 P, 40 C, 30 M, 14 S, 5 Sc, 2 METH categories, giving 

a total of 156). Numerous examples of these categories can be seen in the 

subsequent section of this article. Occurrences of each category in the now 

27 Coded Interview Protocols were recorded in a massive interviewee-by-

category spreadsheet. The frequency of each category is available in the last 

row of the spreadsheet. By examining the spreadsheet horizontally, we could 

identify the codes in which each interviewee scored higher and thus form an 

impression of his/her focal points. By examining the spreadsheet vertically, we 

could identify the codes that featured higher frequencies across the total body 

of interviewees. Subsequently, each category was assigned an ordered pair (x, 

y) as follows: x corresponds to the number of times the category has been 

identified in the Coded Interview Protocols and y to the number of students 

who have referred to the category. Further scrutiny based on validation of 

significance relating to frequency, researcher emphasis, and external theory led 

to the selection of the Pivotal Categories around which we clustered all the 

categories, across the Code System, that were tangentially relevant (covered 

part of the same ground, highlighted a different angle of the same issue, etc.). 

Out of this clustering, five major characteristics of quiet disaffection emerged. In 

the following, we introduce each of the characteristics using the corresponding 

cluster of categories and substantiate using extracts from the interviews. The 

evidence is supported further with references to the classroom observations, 

the student profiles and the relevant literature.

Nardi and Steward (2003: 348–9)

Handling Data

286

Figure 6.9 

Nardi and Steward’s (2003) coding schedule

Maloney and Plaut had a predetermined theory, and their approach was to deductively 

uncover data to support this theory. For example, Maloney (2007) analysed video 

recordings of students’ roles in lessons, using a series of codes derived from previous work 

carried out with teams (Figure 6.10).

Plaut (2006) constructed a conceptual model of confusion (Figure 6.11), which she then 

used to analyse the stimulated recall interviews with teachers

Representing Interactions

If you have used focus groups, you will need to decide whether to transcribe the complete 

group discussions or whether to use abridged transcripts in your analysis. Transcripts 

are useful in that they can provide more than a record of the discussion, and they also 

allow for a more intimate understanding of the content of the talk, the flow of discussion 

and the group dynamics. You could also analyse the linguistic elements such as gestures, 

laughter, sounds of disbelief, gaze, and so on. O’Brien (2007) represents the dynamic 

nature of the dialogue which took place during the focus group interviews in Figure 6.12.

Handling data

137

Name of category

Abbreviation

Content: Interviewees’ statements on:

Conceptual difficulty

C

Difficulties in various mathematical topics and ways of 

coping

Mathematics

M

Nature and significance of mathematics

Performance

P

Own and others’ ability and performance in 

mathematics

Teaching

T

Mathematics teaching including the role of activities, 

teaching styles and teacher personality

Social

S

The role of peers, parents and others in mathematical 

learning

School

Sc

Schooling in general

Methodology

METH

The impact of the researcher’s presence in the 

classroom

Figure 8.11  Nardi and Steward’s (2003) coding schedule

Maloney and Plaut had a predetermined theory, and their approach was to deductively 

uncover data to support this theory. For example, Maloney (2007) analysed video 

recordings of students’ roles in lessons, using a series of codes derived from previous 

work carried out with teams (Figure 8.12).

Plaut (2006) constructed a conceptual model of confusion (Figure 6.13), which she 

then used to analyse the stimulated recall interviews with teachers.

Representing interactions 

If you have used focus groups, you will need to decide whether to transcribe the com-

plete group discussions or whether to use abridged transcripts in your analysis. 

Transcripts are useful in that they can provide more than a record of the discussion, and 

they also allow for a more intimate understanding of the content of the talk, the flow of 

discussion and the group dynamics. You could also analyse the linguistic elements such 

as gestures, laughter, sounds of disbelief, gaze, and so on. O’Brien (2007) represents the 

dynamic nature of the dialogue which took place during the focus group interviews in 

Figure 8.14.

09-Wilson-Ch-08.indd   137

8/31/2012   5:41:12 PM

Handling Data

287

SCHOOL-BASED RESEARCH

138

Table 5  Characteristics of the roles

Role

Code

Features

Positive Roles Chair

Ch

Asks questions and asks others for contributions

Suggests what the group can do

Discussion 

Manager

DM

Starts and/or ends discussions

Makes final decision with or without consultation

Directs the groups; suggests what action to take 

Information 

Manager

IM

Checks on the tasks to be done or validity of evidence

Refers back to the E1 evidence 

Summarises evidence 

Promoter of 

Ideas

PI

Suggests idea–may or may not be acceptable to others

Impatient when discussing ideas other than their own

Wants to get the decision made

Influential 

Contributor

IC

Makes claims with reference to data

Responds to others by posing questions or challenging ideas

Suggests a possible decision

Negative 

Roles

Non-influential 

Contributor

NIC

Responds to others’ comments with agreement or confirming 

points made

Makes suggestions that are ignored by the others

Agrees with the decision that someone else makes

Non-

responsive 

Contributor

NRC

Has own ideas but puts them forward only when asked

May make a different decision to the others

Does not attempt to persuade others to change their minds

Reticent 

Participant

RP

Makes little contribution

May read out E1 evidence but not make any comments

Makes few claims

Distracter

Di

Talks about issues not related to the task

Tells long stories that are marginally related to the discussion

Displays silly behavior

Figure 8.12  Maloney’s (2007) codes for analysing video recordings of lessons

09-Wilson-Ch-08.indd   138

8/31/2012   5:41:12 PM

Figure 6.10 

Maloney’s (2007) codes for analysing video recordings of lessons

Handling Data

288

ANALYSING IMAGES

Photographs can provide a speedy (when using a digital camera) and clear point of 

reference for discussion. They can serve as an illustration to accompany transcripts. You 

can also print out multiple copies for learners to comment on. However, remember that 

viewers don’t always interpret as much or indeed the same things from photographs as 

they would from written excerpts of conversation.

Issues of Representation

Photographs can be interpreted in two ways: firstly, you can focus on the content of any 

visual representation – for example, who is the person in the photograph? Secondly, you 

may want to look at who produced the image, and for whom. Why was this photograph 

taken of this particular person, and then kept by that particular person? 

Handling data

139

Causes of confusion

(side face)

Responses to confusion

(side face)

Name of confusion

(bottom face)

I JUST DON’T GET IT

Types of confusion

(front face)

Figure 8.13  Plaut’s (2006) conceptual model of confusion

Analysing images

Photographs can provide a speedy (when using a digital camera) and clear point of ref-

erence for discussion. They can serve as an illustration to accompany transcripts. You 

can also print out multiple copies for learners to comment on. However, remember that 

viewers don’t always interpret as much or indeed the same things from photographs as 

they would from written excerpts of conversation.

issues of Representation

Photographs can be interpreted in two ways: firstly, you can focus on the content of any 

visual representation – for example, who is the person in the photograph? Secondly, you 

may want to look at who produced the image, and for whom. Why was this photograph 

taken of this particular person, and then kept by that particular person?

09-Wilson-Ch-08.indd   139

8/31/2012   5:41:12 PM

Figure 6.11 

Plaut’s (2006) conceptual model of confusion

Handling Data

289

Concept and network maps

You can carry out a descriptive, first-level analysis of concept maps. This could be followed

by an analysis of nodes which might be grouped according to their proximity to

other items, such as types of roles.

You can also extract information about the links made between words or themes. The 

map, and its first analysis, is a really useful basis on which to come back to people and talk 

about your perceptions of networks, or concept maps, the value and strength of links and 

how you are using them.

Figure 6.12 

OBrien’s (2007) diagram interview interactions

SCHOOL-BASED RESEARCH

140

GROUP-BASED BULLYING

IS

WORSE

INDIVIDUAL-BASED

BULLYING IS WORSE

iii

ii

i

1

2

3

4

i

ii

iii

Orientation i

Orientation ii

Orientation iii

Commensurable evaluations: 13 dimensions

Incommensurable evaluations: 7 dimensions

Indistinguishable evaluations: 10 dimensions

INDIVIDUAL-BASED

BULLYING IS NOT AS BAD

GROUP-BASED BULLYING

IS NOT AS BAD

Figure 8.14  OBrien’s (2007) diagram interview interactions

Concept and network maps

You can carry out a descriptive, first-level analysis of concept maps. This could be fol-

lowed by an analysis of nodes which might be grouped according to their proximity to 

other items, such as types of roles.

09-Wilson-Ch-08.indd   140

8/31/2012   5:41:13 PM

Handling Data

290

жүктеу/скачать 32,4 Mb.

Достарыңызбен бөлісу: