Researching the Social World Part 1 Research Methods

1
Researching the Social World(Part 1 Research
Methods)

• An Introduction plus some
• revision of basic terms
• and concepts

2
Course Content

• Quantitative and Qualitative methods
• ANOVA and Multiple Regression
• Observations and Interviews
• Ethics
• Writing Up a Report
• Designing and carrying out your own research

3
Introduction to Research Methods and Data
Analysis in Psychology
Darren Langdridge
Lecturer Laurence Hopkins Course Researching
the Social World

• Benefits of purchasing this book
• Accessible and jargon free
• Covers both qualitative and quantitative methods
  and data analysis - this book will see you right
  through your course!
• Includes activities and study exercises to cement
  knowledge and understanding

0130978329
Available from Blackwells, Liverpool University
4
But dont forget or sell

• Dancey, C.P. Reidy, J. (2002) Statistics
  Without Maths for Psychology 2nd Edition.
  Harlow. Pearson Education Limited
• You will be using this book in many of your
  seminars
• You might also find the following books useful if
  you want a more detailed explanation.
• Field, A. (2000), Discovering Statistics
  using SPSS for Windows. London. Sage
  Publications.
• Smith , J. (2003) Qualitative Psychology A
  Practical Guide to Research Methods. London. Sage
  Publications.

5
AssessmentOverall you have to get 40

• Class Test
• 2 hour test 4 questions
• Essay on quant/qual debate
• Definitions of terms
• Interpreting SPSS
• Reliability/validity question

• Coursework
• Choice of topic and method is yours (options in
  handbook)
• Carried out as a group
• Write up individually
• Hand in Thursday
• Dec 1st

6
Heavy Reminder!

• Plagiarism will result in your work receiving a
  mark of zero. If the plagiarism is major then you
  will not be allowed to resubmit in August. Minor
  cases will resubmit in August but can only get a
  grade E overall.

• This year any work handed in late without an
  extension will receive a mark of zero. There is
  no 10 deduction in the first week of lateness

7
Last year (2005/6) 29 failed

• 18 As, 19, Bs
• 30 got B or above in class test
• 56 got a grade C or above in class test

8
Attendance Participation

• Non- attendance will be followed up i.e. letters,
  phone calls, interviews.
• See me if you need to change seminar group
• Act as if you are interested!!
• Participate in workshops it helps to pass the
  time
• Ask questions they are a really good way of
  getting answers
• Do not be afraid to look foolish!

9
The Websitehttp//hopelive.hope.ac.uk/psychology/
LevelI/rsw/index.htm

• Please get in habit of checking every week
• New useful stuff going on all the time
• Often you will need to print off material for the
  next seminar
• There are SPSS data files there that you can use
  to practice with.

10
Idiots guide to Psychological Research

• Measure them e.g correlational research
• Watch them e.g. observational research
• Ask them e.g. interviews or questionnaires
• Mess about with them and see what happens e.g.
  experimental research

11
The Scientific Method
12
Quantitative/qualitative debateClaimed Features

• Quantitative
• Hard, fixed, objective, value-free, hypothesis
  testing,
• numbers
• Controlled environments
• Reductionist
• Positivist
• Outcomes
• Concerned with causal explanations
• Replicable

• Qualitative
• Soft, flexible, subjective, political,
  speculative,
• symbols (words)
• Focus on natural settings
• Holistic
• Hermeneutic (concerned with meaning
• Process
• Concerned with induction and grounded theory
• Rich detailed data

13
Revision

• 1 2 tailed hypotheses
• Independent dependent variables
• Between within subject/participant designs
• Types of data (continuous or categorical
  nominal, ordinal interval)
• Mean and Standard Deviation
• Significance levels
• Parametric non-parametric tests
• Correlations
• t-tests

14
Hypotheses

• People who smoke cannabis will giggle (for no
  apparent reason) more than a control group.
• Men earn more money than women.
• The more you attend PRI seminars the higher your
  mark will be in the exam.

15
ANALYSIS OF VARIANCE

• ANOVA

16
Back to the t-test

• The t-test is a parametric test (what!) which
  compares TWO conditions. It can be between or
  within participants (i.e. independent or repeated
  measures)
• The t-test is not that different from ANOVA

17
Parametric tests (e.g. t-tests, pearsons
correlation, ANOVA)

• Make estimates about the general population from
  sample statistics.
• Data should be interval, normally distributed and
  there should be homogeneity (similarity) of
  variance between conditions.
• More powerful than non-parametric tests
  (Wilcoxon, Mann-Whitney etc)

18
Comparison of 2 Athletic Clubs at the 100 metres
19
Comparison of the 2 clubs on a memory recall test
20
(No Transcript)
21
(No Transcript)
22
The variance caused by the independent variable
is much larger than the variance occurring due to
random unsystematic fluctuations.
23
Here the variance caused by the independent
variable (i.e. variance between conditions) is
small compared to the variance that is occurring
by random unsystematic fluctuations.
24

• So Analysis of Variance compares the variance
  caused by independent variables (and their
  interactions) with random unsystematic error. If
  this ratio is large then the test will be
  significant.
• Generally, in error bar charts the degree of
  overlap of the confidence limits will give some
  idea as to whether the results are significant or
  not.

25
STATISTICAL SIGNIFICANCE MEMORY TESTS (20
WORDS) Imagine we keep testing 2 groups taken
randomly from the whole population and we keep
repeating this. The null hypothesis (i.e. no
difference between the groups) is true. USUALLY A
14 B 14 (DIFFERENCE 0) A 14 B 13
(DIFFERENCE 1) A 12 B 14 (DIFFERENCE
-2) A 15 B 13 (DIFFERENCE 2) A 13 B 13
(DIFFERENCE 0) AND SO ON I.E. USUALLY THERE
WONT BE MUCH DIFFERENCE HOWEVER OCCASIONALLY
JUST BY CHANCE (REMEMBER THE NULL HYPOTHESIS IS
TRUE) A 17 B 11 (DIFFERENCE 6) A 10 B 17
(DIFFERENCE -7) THE CHANCES(PROBABILITY) OF
GETTING THESE RESULTS IF NULL HYPOTHESIS IS TRUE
IS VERY SMALL BUT IT DOES SOMETIMES HAPPEN. NOW
IMAGINE WE INVESTIGATE AGEING AND MEMORY AND NOW
GROUP X YOUNG AND GROUP Y OLD. WE HPYOTHESISE
THAT THE YOUNG WILL DO MUCH BETTER THAN THE OLD
AND WE GET THESE RESULTS X 16 Y 9
(DIFFERENCE 7) OUR STATISTICAL TEST WILL TELL
US THE PROBABILITY OF GETTING THESE RESULTS IF
THE NULL HYPOTHESIS IS TRUE AND IT WILL SAY THAT
p IS VERY SMALL (MAYBE lt 0.05 OR 5) THEREFORE WE
WOULD REJECT THE NULL HYPOTHESIS SINCE IT IS
UNLIKELY TO BE TRUE. IT IS MORE LIKELY THAT OUR
EXPERIMENTAL HYPOTHESIS IS TRUE I.E. THAT THERE
IS A DIFFERENCE BETWEEN OLD AND YOUNG MEMORY.
26
ANOVA

• The t-test is used when there are just 2
  conditions and when there is only 1 independent
  variable.
• Anova is used when there are 3 or more conditions
  or when there is more than 1 independent
  variable.
• One Way Anova 1 IV Two Way Anova 2 IV
• Can have more than 2 IVs (but we wont be doing
  these so dont worry about it for now!)

27
One Way ANOVA

• One Independent variable
• 3 or more conditions (levels)
• Can be independent (unrelated or between
  participants) or repeated measures (unrelated or
  within participants)

28
Viagra Study

• Field (2000) uses some fictitious data regarding
  a between participant study with 3 conditions or
  levels where the independent variable dose of
  Viagra.
• Placebo
• Low Dose Viagra
• High Dose Viagra
• Dependent Variable ??

29
(No Transcript)
30
Descriptive Statistics
31
F Ratio

• In basic terms the F ratio is the variance caused
  by the treatment (independent variable or effect)
  divided by the residual, unsystematic, sampling
  error.
• The variance is worked out by summing the squared
  deviation of scores from the mean (Sum of
  Squares)
• As a general rule for samples above 10 an F ratio
  of above 5 will probably be significant but SPSS
  will tell you!

32
Anova Summary Table(In this case an unrelated
ANOVA)
33
Levenes Test(for homogeneity of variance)

• This is a test that you dont want to be
  significant i.e. you want the sig (or p) to be
  more than 0.05
• If it is significant then the validity of the
  results would have to be questioned there are
  ways of correcting for it but beyond scope of
  this course

34
Post-hoc tests (e.g. Tukeys Honestly Significant
Difference or HSD) (Only relevant if you have 3
or more levels of the IV.)(SPSS cannot do post
hoc tests on within-participant data)
35
Family Wise Error Rate

• Whenever we do a statistical test using the
  (plt0.05) level of significance there is a 5
  chance of making a Type 1 Error or a 95 chance
  that we wont make a Type 1 Error.
• It can be shown that as we perform more and more
  tests on the same set of data then the chance of
  making a Type 1 error goes up from 5 according
  to how many tests are performed.
• The likelihood of making a Type 1 Error is known
  as the Family Wise Error Rate and can be
  calculated to be equal to 1 (0.95)n where n
  is the number of tests.
• Post hoc tests use more conservative (strict)
  calculations to make sure that there is no more
  probability of making a Type 1 error than in a
  single test.

36
Planned (a priori) Comparisons

• If you decide before the data is analysed that
  you want to compare certain conditions then it is
  better than using post hoc tests since there will
  be less tests performed and if a difference
  exists you are more likely to find it this way.

37
Repeated MeasuresOne Way ANOVA

• Because the participants in each condition are
  the same people this reduces some of the
  unsystematic variation that exists in between
  participants designs.
• This makes these tests more sensitive and
  powerful.
• However as well as the homogeneity of variance
  assumption we have the criteria of SPHERICITY
  that has to be met.

38
Sphericity (only relevant if you have 3 or more
levels of the within-participant independent
variable)

• The variance of the difference between pairs of
  scores are equal for all groups (i.e. A-B A-C
  B-C) Read Field p324 for more details.
• It is tested for by Mauchlys test which (like
  Levenes) you dont want to be significant.
• However if it is significant then there is a
  correction called the Greenhouse-Geisser and it
  is this row in the output that you should use.

39
Within-participants ANOVA
40
Output for Within-Participants ANOVA
41
Two Way Anovas

• In real life there is more than one variable
  affecting our dependent variables
• Anova can handle experimental designs where there
  is more than one IV (multifactorial designs)
• These Anovas can be independent, repeated
  measures (rare) or a mixture of the 2.
• ANOVA can tell us whether the IVs are producing
  a main effect but also whether or not the 2 IVs
  are interacting with each other.

42
Two Way Anova

• Imagine that we are investigating the effect of 2
  IVs on 1 DV
• e.g. the effect of alcohol and also eating a
  kebab on the DV on vomiting at end of night
• Or the effect of gender and also the amount of
  violence in a film on the DV of how much the film
  was liked

43
An interaction between kebabs and alcohol
44
Anova Summary Table
45
An interaction between gender and violent content
of films
46
Anova summary table
47
The effect of the media on body image for males
and females

• Females and males were shown either photos of
  attractive same sex models or photos of
  landscapes.
• They were then asked to rate how they felt about
  their own body.

48
(No Transcript)
49
No interaction

• Here we can see that the males have scored higher
  than the females.
• The landscape condition has generally scored
  higher than the model condition
• With the lines being so parallel it is doubtful
  that there is an interaction,
• However to see if any of the main effects or the
  interaction are significant we have to check with
  the Anova table in SPSS.

50
(No Transcript)
51
Mixed Anova

• In this type of Anova one of the IVs is between
  participants and the other is a
  within-participant variable.
• For example you could investigate the effects of
  alcohol on driving but want to know if gender was
  also an important factor.
• In this case gender would be a between-participant
  IV (Obviously!) but alcohol could be a
  within-participant variable (in this case with 3
  levels sober, a bit drunk and completely psed)
• This would be a 2 x 3 mixed design.

52
(No Transcript)
53
(No Transcript)
54
(No Transcript)
55
(No Transcript)