Week 4

Descriptive Research
Questions

PSYC4045A
29 March 2021 https://ulwazi.wits.ac.za/courses/14308
Objectives 1. Explore the characteristics of a
descriptive research question
2. Apply the logic of inquiry to a
descriptive research question
and be guided by an example
3. Conduct & interpret
descriptive statistical analyses
using SPSS
4. Critically evaluate descriptive
research using a research
article
Prescribed Reading:
• Laher, S. & Botha, A. (2012). Methods of sampling. In C. Wagner,
B.Kawulich & M. Garner. Doing social research: A global context
(Pp.86-100). London: McGraw-Hill
• Field, A. (2013). Discovering statistics using IBM SPSS statistics. Sage.
Chapters 1 (18-26), 3, 5 (131-148)
• Laher, S. (2016). Ostinato Rigore: Establishing methodological rigour in
quantitative studies. South African Journal of Psychology, 46, 316-327
• Laher, S., Subjee, S., & Bemath, N. (2018). An Exploration of Muslim
females’ perceptions of Major Depressive Disorder. Mental Health,
Religion & Culture, 21, 625-642 – compulsory for Wednesday lecture
 THE QUESTION
WHAT ARE ATTITUDES TOWARDS MENTAL ILLNESS
AMONGST SOUTH AFRICAN COMMUNITY
MEMBERS?
The logic of inquiry
 Approaching research questions
•What is the nature of the question?
•Interpretive
•Descriptive
•Relational, group differences, causal, discursive, etc
•What is the paradigm that informs the question?
•Positivist
•Interpretive
•Critical Social Science
•What are the best ways to obtain information about the social world (methodology)?
Paradigms
Paradigm Ontology Epistemology Methodology Associated approaches

Positivism The world is material, Possible to obtain objective Quantification of Quantification

observable measurable. conclusive knowledge phenomena for Experimental methods
Numerical and linguistic about reality that is comparison, prediction and
abstractions of phenomena generalisable. control.
reflect real phenomena

Interpretivism A definitive universal Proxy or perceived Systematic re-organisation Phenomenology

reality does exist but it will knowledge can be obtained and interpretation of data Ethnography
be interpreted but only through (frequently texts) for Grounded theory
idiosyncratically. Language interpretation/vicarious understanding specific Narrative analysis
has a proximal relationship engagement. meanings.
to reality but tells us more
about the interpreter than
it does about reality itself.

Constructionism The human world is Knowledge is inextricably Attempt to understand the Critical discourse analysis
constituted and presented bound to language and organisation of language as (CDA)
through language and signs. To understand a means to deciphering the Discursive psychology
signs. Humans use human action, we need to work that it is doing. Conversation analysis
language to do social understand their uses of Genealogy
action. language. Narrative analysis
TYPES OF QUANTITATIVE RESEARCH DESIGNS

Quantitative NON - EXPERIMENTAL

RESEARCH DESIGNS

research EXPLORATORY OR PRE-

EXPERIMENTAL RESEARCH
designs may be DESIGNS

classified into QUASI - EXPERIMENTAL

RESEARCH DESIGNS
four broad
categories: TRUE - EXPERIMENTAL
RESEARCH DESIGNS
 Classifying research designs
This is done using the following 3 criteria:

Comparison of the outcomes of the group receiving the treatment (experimental

group) with the outcomes in a control group of subjects

Change or manipulation of the IV: a deliberate change in the level of the IV by the
researcher

Random assignment: split the sample so that each subject has an equal, non-zero
probability of being in either the control group or the experimental group i.e. assigning,
in some random fashion, subjects in the sample to either the control or the
experimental group
 NON PRE QUASI TRUE
IV Manipulated No Yes Yes Yes

Control No No Yes Yes

Group

Random assignment No No No Yes

Cross-sectional Single Group Posttest- Posttest only Posttest only

Research design only design Nonequivalent control control group design
group design
Correlational Longitudinal Research Single Group Pretest- Pretest-posttest contro
research design posttest design Pretest-posttest non- group design
equivalent control
Ex-post facto research group design Solomon 4-group
design design
 Nonexperimental Designs
Advantages Limitations
• easy to implement • no control group (can only establish
• cost and time efficient association between variables)
• useful for descriptive purposes • no manipulation of IV (no/weak
directionality established)
• no random assignment (non-
• Repeated measures and spuriousness cannot be demonstrated)
replication may allow for trends to
be determined. • many possible threats to internal
validity - therefore does not allow for
causal inference
Cross-sectional Designs
• Involves the collection of information from any given sample of
population elements only once
• In single cross-sectional designs, information is obtained from one
sample of respondents only once
• In multiple cross-sectional designs, information is obtained from 2 or
more samples of respondents but the information is obtained from each
sample only once.
 The logic of inquiry
✔ ✔

✔
 Our question for this lecture:

Unit of analysis WHAT ARE ATTITUDES TOWARDS

MENTAL ILLNESS AMONGST SOUTH
AFRICAN COMMUNITY MEMBERS?

• What is the level at which we are analyzing our

data?
• Social artifacts
• Individuals?
• Families?
• Communities?
• Groups?
• Countries?
Sampling

(www.managementparadise.com)
Sampling

✔ ✔ ✔ ✔

✔
SAMPLIN Our question for this lecture:
WHAT ARE ATTITUDES TOWARDS MENTAL ILLNESS

G AMONGST SOUTH AFRICAN COMMUNITY MEMBERS?

• Units of analysis • Size of Sample

• Individuals but who? • Items x 3
• Professionals, patients, • Sample size calculators
community members • Bigger is better
• Type of sampling: • Access & Gatekeeping
• Nonprobability • Ethics
• Convenience • Vulnerability
• Snowball • Consent
• Volunteer
• Confidentiality
• Homogenous?
• Anonymity
• Mixed purposive
• Feedback
• Debriefing
 The logic of inquiry
✔ ✔
✔

✔
Candidate data collection techniques
• How can we optimally ‘extract’, source and/or produce information
that reflects the units of analysis?

Our question for this lecture:

WHAT ARE ATTITUDES TOWARDS MENTAL ILLNESS
AMONGST SOUTH AFRICAN COMMUNITY
MEMBERS?
QUESTIONNAIRES
• Develop your own, adapt or use • Items must be clear
as is • Not double barrelled or
• Questions or statements ambiguous
• Open or closed ended • Have short items
• Scales or vignettes • Avoid negative items
• How many sections? • Avoid biased items
• Demographics • Formatting of q.nairre
• Clear instructions • Response formats
• Order of items / scales • Translation
• Response rates
 QUESTIONNAIRES

Advantages: Limitations:
• can be administered to large numbers of • application from one population /
people demographic area to another
• time and cost efficient • illiteracy
• provide a type of 'anonymity' • low rate of questionnaire – return
• yield valuable descriptive information • ambiguous replies
about broad trends • depends on the truthfulness of the
subject
Task 1 – Questionnaire design & Ethics
• Complete the attitudes towards MI questionnaire yourself
• Reflect on your experience of completing the questionnaire.
• Consider the participant information sheets (PIS) on the
questionnaire provided. Are they adequate?
• Evaluate the questionnaire provided in terms of the criteria
discussed for questionnaire construction and presentation

Our question for this lecture:

WHAT ARE ATTITUDES TOWARDS MENTAL ILLNESS AMONGST
SOUTH AFRICAN COMMUNITY MEMBERS?
 The logic of inquiry
✔ ✔ ✔
✔

✔
Data entry
• Recommended that the data be initially entered into Excel
• Each respondent /questionnaire should be given an identity
number
• Rows contain the information for the respondents
• Columns contain the questions
• Thus initial data file should be a digital representation of the
questionnaires
• Do not do any calculations or reversals manually
Coding your data
gender male 1
female 2

home language Afrikaans 1

English 2
IsiNdbele 3
IsiXhosa 4
isizulu 5
Sepedi 6
Sesotho 7
Setswana 8
Siswati 9
Tshivenda 10
Xitsonga 11
Other 12
 Cleaning data
• Find and eliminate data entry and other errors
• Examine missing data and perhaps account for it in some way
(statistically)
• Double check entries
• Check for anomalous entries – using SPSS functions
Entries out of the range of the item’s scale
Entries that do not make sense
Usually use frequencies for this
Variables and scale items
• In general, a number of items are used to measure a variable
• Some of the items may have meanings that are in the opposite
direction of the others.
Need to:
• Reverse the relevant items so that the scores mean the same
thing
• See orientation session video on ‘What is reverse scoring’
Importing data into SPSS
• Open SPPS Version 27
• Cancel “Open File” screen that appears
• Go to “File” menu and select “Open,” then “Data”
• Find your file and double click to open.
• You might have to change the file format from SPSS to Excel or
whatever your file is saved in
• Click on the appropriate worksheet to import
• Once the dataset is open, go to “Variable View” to adjust the dataset
properties
 The logic of inquiry
✔ ✔ ✔
✔

✔
Data Analysis
• In order to analyse your data you need to be familiar with basic
statistical concepts
• Please see the video and notes on ‘Statistical concepts orientation’ in the
Orientation week session in Modules
• You will need to install SPSS Version 27, licence the program and
become familiar with the functions in SPSS
• Please see the video on ‘Installing SPSS for Windows’ in the Orientation week
session in Modules
• You also need to know the basics of using SPSS
• Please see the video on ‘Conducting analyses on SPSS’ in the Orientation week
session in Modules
 Data Analysis
• What is my research question?

Our question for this lecture:

WHAT ARE ATTITUDES TOWARDS MENTAL ILLNESS AMONGST
SOUTH AFRICAN COMMUNITY MEMBERS?

• What is/are my variables?

• Attitudes towards MI
• What is the scale of measure of my variable/s?
• Measured using a Likert type scale – at least interval
• Always remember: What does a high score mean?
• High score = more negative attitudes towards MI
 Proposed analysis if you use the total attitude score:

Descriptive Statistics
Std.
N Minimum Maximum Mean Deviation Skewness Kurtosis
  Statistic Statistic Statistic Statistic Statistic Statistic Std. Error Statistic Std. Error
AttMI 100 22 71 43.39 10.441 .534 .241 .178 .478
Valid N 100
               

In terms of attitudes towards mental illness, scores ranged between 22 and 71 with a
mean of 43.39 and a standard deviation of 10.44 (see Table ?). The skewness
coefficient is in the appropriate ranges (see Huck, 2009), indicating that attitudes
towards mental illness is normally distributed. The sample does not appear to have
strong positive or negative attitudes towards mental illness.
Histogram for Attitudes towards Mental Illness
Remember: higher scores = more negative views –

What can you conclude about attitudes towards mental illness?

Proposed analysis if you use individual items:
 The logic of inquiry
✔

✔ ✔ ✔
✔

✔
✔

✔
 Task 2 – Engaging with data analysis
• Open ‘MI Tut Data reversed totalled’ in SPSS
• Organise your data in the ‘Variable View’ pane
• Use an appropriate set of descriptive statistic/s to summarise the:
• language groupings
• education groupings
• BMI_Total
• Why did you choose this/these statistics in each case?
• Interpret each of your analyses
• Generate and interpret an appropriate graph for the religious affiliation
variable
 Task 2 continued
• Calculate the appropriate descriptive statistics for the:
• Familiarity with mental illness total
• Attitudes towards mental illness total
• Interpret your results
• Would you be able to conduct parametric statistics using these variables?
Substantiate your answer using a graph as well as appropriate statistics
ASESSING THE QUALITY OF RESEARCH

There are four key issues that typically require consideration when
evaluating a piece of quantitative research:

Internal validity: the degree of certainty about whether the hypothesized

cause is responsible for the hypothesized effect

External validity: the degree to which the research can be applied to

other contexts

Statistical conclusion validity: the degree to which the correct statistical

procedures have been applied and correct interpretations made

Measurement validity: the degree to which the conceptualization and

operationalisation are appropriate
Internal validity
• Internal Validity is the degree of validity of statements made about whether X (the IV
or hypothesized cause) causes Y (the DV or hypothesized effect).
• This is particularly relevant to causal (experimental) designs, where it is very
important to rule out the threat of extraneous variables.
• Internal validity is concerned with ruling out plausible rival hypotheses that
jeopardize statements made about whether X causes Y
• Internal validity threats refer to factors within the context in which the research is
conducted which may affect its validity
External validity
• External validity is the degree of generalisability of findings from a piece
of research to other situations, events and settings i.e. the degree of
generalisability or representativeness
• It is important because external validity determines the degree to which
results can be applied to other contexts and populations.
Task 3 – Article Analysis
1. Read the article on MDD in Muslim females by Laher, et al.
2. Use the slide below to identify the elements discussed in this
lecture
 Variables Scales of
Knowledge of measure
MDD Etiology + Care
Nature of Paradigm Unit of analysis Familiarity MDD
– nominal
the Individuals Others - interval
Attitudes to Analysis
question
Interpretivi MDD
Qual – Thematric
st Etiology MDD Analysis – Braun &
Exploratory
Positivist Care & Clarke
&
Management of Quan –
Descriptive
Sampling MDD
– Mixed Collection Etiology + Care –
perceptions methods Nonprobability, Qual – semi nominal –
& attitudes Qual-Quan convenience structured frequencies +%s
towards approach Qual – 12 interview Others – interval
MDD in Muslim, Indian Constructs/
Muslim Sequential Quan – - Min & max values,
Exploratory females concepts Questionnaire – means, SD’s,
females
Design Quan – 62 Perceptions of Demographics + skewness
Muslim, Indian, MDD adapted PPIQ
female

Possibilities and limits of inferences/conclusions/recommendations

 Task 4 – Further practice with data analysis

1. Describe the sample in terms of age.

2. Was this a balanced sample in terms of gender? Substantiate your answer.
3. Identify the scales of measure for age, religious affiliation, levels of schooling, knowledge
of someone with mental illness, knowledge of mental illness item and attitudes towards
mental illness score.
4. Which were the dominant etiologies identified for mental illness in this sample. Conduct
and interpret an appropriate analysis to answer this question.
5. Would you recommend using a parametric inferential analytic technique on the total
BMI? Substantiate your answer
6. Would there be an association between gender and knowledge of someone who has
depression? Conduct and interpret an appropriate analysis to answer this question.
7. How confident would you be of the result you found in 5?
8. What is the relationship between the familiarity of MI score and attitudes towards
depression score?