CLINICAL PERSPECTIVES ON

ANXIETY, MOOD AND STRESS

GENS 0005
LECTURE TOPIC ONE (C): WHAT IS
SCIENCE?

DR. JOSHUA M. BRODERICK

What are the main types of experiments (procedures)?

True experiment
Correlational study
Quasi-experimental study
Case study

What is a correlational study?

Theory: Being inactive causes depression

Hypothesis: The fewer hours your spend per week being
active the greater your depression will be
Dependent variable: We only measure variables and see if
they are correlated
We will measure how many hours per day people are
not active (behaviour)
Operationalise = Number of hours spent active
We will measure how depressed people are (selfreport questionnaire
Operationalise = Score (0-100) on self-report
questionnaire

What is a correlational study?

Depression and Activity Negative Correlation

Depressio
n
(0-100)

Note: This is a graphical

representation of a negative
correlation!!!

Activity (hours per day)

As the amount of depression decreases the amount of activity increases;
OR as the amount of activity increases, depression decreases (negative
correlation)

What is a correlational study?

Depressio
n
(0-100)

Note: This is a graphical

representation of a positive
correlation!!!
This is a positive
correlation
Activity (hours per day)

As the amount of depression increases the amount of activity increases (a

positive correlation) this finding would does not support our hypothesis we
may want to reformulate our theory!

What is a correlational study?

Note: The fewer participants we have (in any type of study) the less
valid our study findings become. The less likely it is we will observe an
effect, the less likely it is our findings are representative of what might
be observed in the real world

What does a correlational study NOT tell me?

Correlation does not equal causation (very important)
Perhaps a third variable influences the correlation
Variable A causes variable B
Perhaps variable C causes both A and B

B
A

What does a correlational study NOT tell me?

Correlation does not equal causation (very important)
Perhaps A causes B or B causes A (how would we know?)
Require an experiment!

B
A

What does a correlational study NOT tell me?

Correlation does not equal causation (very important)

Positive correlation with reading ability (A) and amount of books
in the home (B)
B causes A: Having lots of books causes children to be good
readers!
C causes A and B (Intelligent parents cause children to
be good readers via genetics AND the parents buy many
books)
A causes B (Children who are good readers have more
books because they ask for them because they are
already good readers)

http://io9.com/5724738/if-you-want-to-live-longer-then-walk-faster

What is a quasi-experimental study?

nghin cu bn thc nghim

Sometimes we are unable to do experiments for ethical or practical

reasons
In this case we must do another kind of research
Research where subjects are selected based on a pre-existing
iu kin sn c
condition phin mun
lo lng
E.g., Depression, anxiety
It would be unethical to give people these conditions
phn b ngu nhin vi iu kin thc nghim cng nhiu cng tt
Quasi-experiments still involve random allocation to experimental
conditions as much as possible for confounding variables but not
cho bin s gy nhiu - l bin s cung cp mt gii thch khc ca
random selection
mi lin h gia bin s c lp v bin s ph thuc
Quasi-experimental studies are like true experiments without random
selection (restricted inferences to wider population)
What do you need to do to control for the loss of random selection?

Measure potential confounding variables (e.g., age, sex, intelligence,

socio-economic status, health)
There are a potential infinite number of confounds, making this a
difficult process

Independent variables-bin s c lp/khng ph thuc - Cc bin s khng phi l o lng kt cc ca nghin cu, m nh nghin cu mong mun cc
bin s ny c th gii thch s c/khng xut hin ca bin s ph thuc
Dependent variable: Bin s l mi quan tm ch yu trong nghin cu - v.d.bnh/khng bnh trong nghin cu bnh chng, nghin cu on h,..khi/khng
khi trong th nghim lm sng - m nh(...)

What is a quasi-experimental study?

Identical
twin studies allow for a quasi-experimental investigation of the
bn cht, tnh cht c tnh
tm l
role of nature and nurture on various psychological processes, abnormal
behaviours, mental illnesses

These studies are quasi-experimental because the twins are selected (not
at random)

What is a single case study?

Investigate one individual case and write a report on it

Advantages:

You get a lot of information

Information is often both qualitative and quantitative
Quantitative (observable and measurable)
Qualitative (not easily quantified)
E.g., tell me about your experience as a refugee

Disadvantages:

Is it representative or just a one off finding?

Considered less scientific (not relying upon sound statistical principles)

What factors result in poor science being conducted?

Errors occurring when conducting/designing the experiment
This issue speaks to validity
Validity is the extent to which a concept, conclusion or measurement is
well-founded and corresponds accurately to the real world
Basically, is what we are concluding from our experiment valid based on
what we have done (i.e. how we have conducted and designed the
experiment)

What factors result in poor science being conducted?

Not enough constants in the experiment
Remember a constant is a factor that is the same between groups

Research Question: Does studying impact on exam performance?

Dependent variable: Score on university examination (score: 0-100)
Independent variable: Amount of time spent studying (8 vs. 1 hour p. week)
Imagine two scenarios:
A) In addition to allocating people to two conditions, researchers allow
one group to sleep more, socialise less (drink less), and to cheat on
the examinations
B) Researchers ensure that both groups sleep the same amount,
socialise the same amount, and do not cheat

Which of the two scenarios represents poorer science?

Why can we not conclude anything about amount of time studying from
Scenario A?
The more constants you take into account the better your science is;
conversely the less constants you take into account the poorer your
science is.

What factors result in poor science being conducted?

Inappropriate control group
Without a control group you are not truly manipulating anything
Without a decent control you cannot answer your research question
Research Question: Does alcohol impair driving ability?
Dependent variable: Score on driving test (score: 1-5)
Independent variable: Amount of alcohol in system (see scenarios)
Imagine two scenarios:
A) Researchers design a study such that there are two groups of
people who are allowed to drink one is allowed to drink a lot and the
other a little
B) Researchers design a study such that there are two groups one
group is allowed to drink and the other is not allowed to drink at all
Which of the two scenarios represents poorer science?
Why is Scenario B better suited to answer our research question
Good science should have appropriate control groups to address a
research question

What factors result in poor science being conducted?

Bias
Bias is a subjective (i.e. not objective) opinion about anything (person,
group, idea, thing) about whether it is good or bad, which importantly
influences how you interact, respond and deal with it
Bias is like prejudice nh kin, kin ko da theo kinh nghim hoc l l
Bias is always one sided (science is open minded)
Different types of bias include:
Personal bias
Measurement bias
Sampling bias

The more bias in a piece of scientific research the poor it is

Bias makes data untrustworthy
Bias makes data inaccurate/non-representative of the real world
Bias represents a huge problem for science as consumers of science you
should be on the look out for it!
Bias occurs often in news, magazines, talk shows. People present one
set of data in a particular way regardless of its validity in order to
support their view

What factors result in poor science being conducted?

Errors occurring when conducting/designing/reporting the experiment
Personal bias
A scientific finding that occurs in the context of an individual or group of
researchers attempting to manipulate or report (or not report) the outcome
in a way which suits their bias

What factors result in poor science being conducted?

Errors occurring when conducting/designing/reporting the experiment
Sampling bias

Take home message

Keep an open mind

Follow the process
Be creative
Have fun
Work hard