risky shift: data analysis week 6 practical. week 6 practicalrisky shift week 1 week 2 week 3 week 4...

21
RISKY SHIFT: DATA ANALYSIS Week 6 Practical

Upload: meadow-willcox

Post on 31-Mar-2015

227 views

Category:

Documents


3 download

TRANSCRIPT

Page 1: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RISKY SHIFT:

DATA ANALYSIS

Week 6 Practical

Page 2: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

WEEK 6 PRACTICALRISKY SHIFT

WEEK 1

WEEK 2

WEEK 3

WEEK 4

WEEK 5

WEEK 6

WEEK 7

WEEK 8

WEEK 9

WEEK 10

LECTURE 1 PRACTICAL

NONPARAMETRICS 1 1ST PRACTICAL

NONPARAMETRICS 2 1ST ANALYSIS IN SPSS

SAMPLING DISTRIBUTIONS

1ST ANALYSIS BY HAND

HYPOTHESIS TESTING

2ND PRACTICAL

RELATED T-TEST

2ND ANALYSIS IN SPSS

INDEPENDENT T-TEST

INDEPENDENT ANOVA

DEPENDENT ANOVA

2ND ANALYSIS BY HAND

3RD PRACTICAL

3RD ANALYSIS IN SPSS

NO PRACTICAL

NO LECTURE NO PRACTICAL

Page 3: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

LEARNING OUTCOMES

BY THE END OF THE SESSION, YOU SHOULD BE ABLE TO:

Think about the meaning(s) of your results, how they relate to past research and how they could be flawed.

Make a graph to show the results with Excel 2007.

Use SPSS to test the first, second and third experimental hypotheses of Risky Shift experiment and produce related graphs.

Make a start on writing up your RESULTS and DISCUSSION sections for your lab report.

RISKY SHIFT

Page 4: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTS

The first 2 columns show participant id and group membership.

These columns show mean risk score at PRE, GROUP and POST.

RISKY SHIFT

Page 5: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTS

First, we need to see whether there is any difference at all between the

three groups, hence Friedman’s test.

In SPSS, Friedman’s test is hidden away underneath

nonparametric tests > K related samples.

Q1: Is the average level of risk different across the PRE, GROUP and POST assessments?

RISKY SHIFT

Friedman’s testRATINGS + WITHIN Ss + 3(+) GROUPS

Page 6: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTS

Q1: Is the average level of risk different across the PRE, GROUP and POST assessments?

RISKY SHIFT

Friedman’s testRATINGS + WITHIN Ss + 3(+) GROUPS

SPSS would like to know what the test variables are, and in our case we

have three.

Put premean, groupmean and postmean all into the test variable

box.

Under Test Type, make sure the Friedman test is ticked because there

are other tests you can do.

Page 7: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTS

Q1: Is the average level of risk different across the PRE, GROUP and POST assessments?

RISKY SHIFT

Friedman’s testRATINGS + WITHIN Ss + 3(+) GROUPS

SPSS provides us with the mean ranks for each of the three groups.

The test statistics box confirms our N, the chi-square value, the degrees

of freedom (groups – 1) and the significance level.

Ranks

2.68

1.57

1.75

pre mean1_5

grp mean1_5

post mean1_5

Mean Rank

Test Statisticsa

48

39.767

2

.000

N

Chi-Square

df

Asymp. Sig.

Friedman Testa. X2 (2) = 39.77, p < .001

The significance of this first test gives us the justification for going on to do the subsequent tests.

Page 8: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTS

Q2: Is the average level of risk recorded in the PRE assessment different from the risk recorded in the GROUP assessment?

RISKY SHIFT

Wilcoxon testRATINGS + WITHIN Ss + 2 GROUPS

Second, we need to see whether there is any difference between PRE and GROUP, hence Wilcoxon test.

In SPSS, Wilcoxon test is hidden away underneath nonparametric

tests > 2 related samples.

Page 9: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTS

Q2: Is the average level of risk recorded in the PRE assessment different from the risk recorded in the GROUP assessment?

RISKY SHIFT

Wilcoxon testRATINGS + WITHIN Ss + 2 GROUPS

Again, SPSS would like to know what the test variables are, and in our case

we have a pair of variables.

Put premean and groupmean into the test pair(s) box.

Under Test Type, make sure the Wilcoxon test is ticked because there

are other tests you can do.

Page 10: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTS

Q2: Is the average level of risk recorded in the PRE assessment different from the risk recorded in the GROUP assessment?

RISKY SHIFT

Wilcoxon testRATINGS + WITHIN Ss + 2 GROUPS

Ranks

37a 22.38 828.00

4b 8.25 33.00

7c

48

Negative Ranks

Positive Ranks

Ties

Total

grp mean1_5 -pre mean1_5

N Mean Rank Sum of Ranks

grp mean1_5 < pre mean1_5a.

grp mean1_5 > pre mean1_5b.

grp mean1_5 = pre mean1_5c.

Test Statisticsb

-5.177a

.000

Z

Asymp. Sig. (2-tailed)

grp mean1_5- pre

mean1_5

Based on positive ranks.a.

Wilcoxon Signed Ranks Testb.

SPSS provides us with the positive, negative and tied

ranks for variable pair.

SPSS provides the Wilcoxon in the form of a z score to be

reported as:

Wilcoxon z = -5.18, p < .001, n = 41

Why 41 and not 48?

SPSS ignored tied ranks…

Page 11: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTS

Q3: Is the average level of risk recorded in the POST assessment different from the risk recorded in the PRE assessment?

RISKY SHIFT

Wilcoxon testRATINGS + WITHIN Ss + 2 GROUPS

Second, we need to see whether there is any difference between

POST and PRE, hence Wilcoxon test.

In SPSS, Wilcoxon test is hidden away underneath nonparametric

tests > 2 related samples.

Page 12: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTS

Q3: Is the average level of risk recorded in the POST assessment different from the risk recorded in the PRE assessment?

RISKY SHIFT

Wilcoxon testRATINGS + WITHIN Ss + 2 GROUPS

Again, SPSS would like to know what the test variables are, and in our case

we have a pair of variables.

Put premean and postmean into the test pair(s) box.

Under Test Type, make sure the Wilcoxon test is ticked because there

are other tests you can do.

Page 13: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTS

Q3: Is the average level of risk recorded in the POST assessment different from the risk recorded in the PRE assessment?

RISKY SHIFT

Wilcoxon testRATINGS + WITHIN Ss + 2 GROUPS

Ranks

36a 21.94 790.00

4b 7.50 30.00

8c

48

Negative Ranks

Positive Ranks

Ties

Total

post mean1_5- pre mean1_5

N Mean Rank Sum of Ranks

post mean1_5 < pre mean1_5a.

post mean1_5 > pre mean1_5b.

post mean1_5 = pre mean1_5c.

Test Statisticsb

-5.150a

.000

Z

Asymp. Sig. (2-tailed)

postmean1_5 -

pre mean1_5

Based on positive ranks.a.

Wilcoxon Signed Ranks Testb.

SPSS provides us with the positive, negative and tied

ranks for variable pair.

SPSS provides the Wilcoxon in the form of a z score to be

reported as:

Wilcoxon z = -5.15, p < .001, n = 40

Why 40 and not 48?

SPSS ignored tied ranks…

Page 14: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTS

Q3: Is the average level of risk recorded in the POST assessment different from the risk recorded in the PRE assessment?

RISKY SHIFT

Wilcoxon z = -5.15, p < .001, n = 40

Q1: Is the average level of risk different across the PRE, GROUP and POST assessments?

Friedman's test X2 (2) = 39.77, p < .001

Q2: Is the average level of risk recorded in the PRE assessment different from the risk recorded in the GROUP assessment?

Wilcoxon z = -5.18, p < .001, n = 41

YES.

YES.

YES.

…but what are the direction of these effects?

Page 15: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTSRISKY SHIFT

DOWNLOAD AND SAVE THE FILE ‘RISKYINEXCEL’

To calculate mean:

=AVERAGE(cellab:cellcd)

To calculate standard deviation:

=STDEV(cellab:cellcd)

To calculate standard error:

=(cellSTDEV)/sqrt (no. observations)

To start graphing:

Select Insert → Column →

Page 16: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTSRISKY SHIFT

SURPRISE!

(So much easier than Excel 2003)

Page 17: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTSRISKY SHIFT

MAKING ERROR BARS

-Select any part of the graph-Click on Layout (under Chart Tools on the toolbar)

-Then Error Bars-Then More Error Bars Options

Make sure you select all three standard errors.

Page 18: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTSRISKY SHIFT

TIDYING UP

Get rid of any unnecessary bits (e.g. background lines, legend)Change colours to greyscale

Change the y-axis to reflect full range of possible answersMake sure you have axis labels

Page 19: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

RESULTSRISKY SHIFT

LOW

HIGH

Q3: Is the average level of risk recorded in the POST

assessment different from the risk recorded in the PRE

assessment?

Q2: Is the average level of risk recorded in the PRE

assessment different from the risk recorded in the GROUP

assessment?

PRE > GROUP PRE > POST

Lower risk (higher score) in PRE Lower risk (higher score) in PRE

Page 20: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

DISCUSSION

GET TOGETHER IN GROUPS OF THREE OR FOUR AND REFLECT ON TODAY’S EXPERIENCE USING THE FOLLOWING QUESTIONS

Are there any problems with interpreting the

data?

What implications do the data have for the studies

outlined in the intro?

What do the data actually tell me with respect to my experimental hypotheses?

RISKY SHIFT

Page 21: RISKY SHIFT: DATA ANALYSIS Week 6 Practical. WEEK 6 PRACTICALRISKY SHIFT WEEK 1 WEEK 2 WEEK 3 WEEK 4 WEEK 5 WEEK 6 WEEK 7 WEEK 8 WEEK 9 WEEK 10 LECTURE

LEARNING OUTCOMES

BY THE END OF THE SESSION, YOU SHOULD BE ABLE TO:

Think about the meaning(s) of your results, how they relate to past research and how they could be flawed.

Make a graph to show the results with Excel 2007.

Use SPSS to test the first, second and third experimental hypotheses of Risky Shift experiment and produce related graphs.

Make a start on writing up your RESULTS and DISCUSSION sections for your lab report.

RISKY SHIFT