SPSS Instructions for Introduction to Biostatistics

SPSS Instructions for Introduction to Biostatistics

Larry Winner

Department of Statistics

University of Florida

SPSS Windows• Data View

– Used to display data

– Columns represent variables

– Rows represent individual units or groups of units that share common values of variables

• Variable View– Used to display information on variables in dataset

– TYPE: Allows for various styles of displaying

– LABEL: Allows for longer description of variable name

– VALUES: Allows for longer description of variable levels

– MEASURE: Allows choice of measurement scale

• Output View– Displays Results of analyses/graphs

Data Entry Tips I

• For variables that are not identifiers (such as name, county, school, etc), use numeric values for levels and use the VALUES option in VARIABLE VIEW to give their levels. Some procedures require numeric labels for levels. SPSS will print the VALUES on output

• For large datasets, use a spreadsheet such as EXCEL which is more flexible for data entry, and import the file into SPSS

• Give descriptive LABEL to variable names in the VARIABLE VIEW

• Keep in mind that Columns are Variables, you don’t want multiple columns with the same variable

Data Entry/Analysis Tips II

• When re-analyzing previously published data, it is often possible to have only a few outcomes (especially with categorical data), with many individuals sharing the same outcomes (as in contingency tables)

• For ease of data entry:– Create one line for each combination of factor levels

– Create a new variable representing a COUNT of the number of individuals sharing this “outcome”

• When analyzing data Click on:– DATA WEIGHT CASES WEIGHT CASES BY

– Click on the variable representing COUNT

– All subsequent analyses treat that outcome as if it occurred COUNT times

Example 1.3 - Grapefruit Juice Study

crcl38667499806480

120

To import an EXCEL file, click on:

FILE OPEN DATA then change FILES OF TYPE to EXCEL (.xls)

To import a TEXT or DATA file, click on:

FILE OPEN DATA then change FILES OF TYPE to TEXT (.txt) or

DATA (.dat)

You will be prompted through a series of dialog boxes to import dataset

Descriptive Statistics-Numeric Data

• After Importing your dataset, and providing names to variables, click on:

• ANALYZE DESCRIPTIVE STATISTICS DESCRIPTIVES

• Choose any variables to be analyzed and place them in box on right

• Options include:

n

S

Sn

yyS

yn

yy

n

ii

n

ii

n

ii

:Mean S.E.

:Variance 1

:deviation Std.

:Sum :Mean

21

2

1

1

Example 1.3 - Grapefruit Juice Study

Descriptive Statistics

8 38 120 621 77.63 8.63 24.401 595.411

8

CRCL

Valid N (listwise)

Statistic Statistic Statistic Statistic Statistic Std. Error Statistic Statistic

N Minimum Maximum Sum Mean Std.Deviation

Variance

Descriptive Statistics-General Data


• ANALYZE DESCRIPTIVE STATISTICS FREQUENCIES

• Choose any variables to be analyzed and place them in box on right

• Options include (For Categorical Variables):– Frequency Tables

– Pie Charts, Bar Charts

• Options include (For Numeric Variables)– Frequency Tables (Useful for discrete data)

– Measures of Central Tendency, Dispersion, Percentiles

– Pie Charts, Histograms

Example 1.4 - Smoking Status

SMKSTTS

1990 37.9 37.9 37.9

1063 20.3 20.3 58.2

609 11.6 11.6 69.8

1332 25.4 25.4 95.2

253 4.8 4.8 100.0

5247 100.0 100.0

Never Smoked

Quit > 10 Years Ago

Quit < 10 Years Ago

Current Cigarette Smoker

Other Tobacco User

Total

ValidFrequency Percent Valid Percent

CumulativePercent

Vertical Bar Charts and Pie Charts


• GRAPHS BAR… SIMPLE (Summaries for Groups of Cases) DEFINE

• Bars Represent N of Cases (or % of Cases)

• Put the variable of interest as the CATEGORY AXIS

• GRAPHS PIE… (Summaries for Groups of Cases) DEFINE

• Slices Represent N of Cases (or % of Cases)

• Put the variable of interest as the DEFINE SLICES BY

Example 1.5 - Antibiotic Study

OUTCOME

54321

Co

un

t

80

60

40

20

0

5

4

3

2

1

Histograms


• GRAPHS HISTOGRAM

• Select Variable to be plotted

• Click on DISPLAY NORMAL CURVE if you want a normal curve superimposed (see Chapter 3).

Example 1.6 - Drug Approval Times

MONTHS

30

20

10

0

Std. Dev = 20.97

Mean = 32.1

N = 175.00

Side-by-Side Bar Charts


• GRAPHS BAR… Clustered (Summaries for Groups of Cases) DEFINE

• Bars Represent N of Cases (or % of Cases)

• CATEGORY AXIS: Variable that represents groups to be compared (independent variable)

• DEFINE CLUSTERS BY: Variable that represents outcomes of interest (dependent variable)

Example 1.7 - Streptomycin Study

TRT

21

Co

un

t30

20

10

0

OUTCOME

1

2

3

4

5

6

Scatterplots


• GRAPHS SCATTER SIMPLE DEFINE

• For Y-AXIS, choose the Dependent (Response) Variable

• For X-AXIS, choose the Independent (Explanatory) Variable

Example 1.8 - Theophylline Clearance

DRUG

3.53.02.52.01.51.0.5

TH

CL

RN

CE8

7

6

5

4

3

2

1

0

Scatterplots with 2 Independent Variables


• GRAPHS SCATTER SIMPLE DEFINE

• For Y-AXIS, choose the Dependent Variable

• For X-AXIS, choose the Independent Variable with the most levels

• For SET MARKERS BY, choose the Independent Variable with the fewest levels


SUBJECT

1614121086420

TH

CL

RN

CE

8

7

6

5

4

3

2

1

0

DRUG

Tagamet

Pepcid

Placebo

Contingency Tables for Conditional Probabilities


• ANALYZE DESCRIPTIVE STATISTICS CROSSTABS

• For ROWS, select the variable you are conditioning on (Independent Variable)

• For COLUMNS, select the variable you are finding the conditional probability of (Dependent Variable)

• Click on CELLS

• Click on ROW Percentages

Example 1.10 - Alcohol & Mortality

WINE * DEATH Crosstabulation

10535 2155 12690

83.0% 17.0% 100.0%

521 74 595

87.6% 12.4% 100.0%

11056 2229 13285

83.2% 16.8% 100.0%

Count

% within WINE

Count

% within WINE

Count

% within WINE

0

1

WINE

Total

0 1

DEATH

Total

Independent Sample t-Test


• ANALYZE COMPARE MEANS INDEPENDENT SAMPLES T-TEST

• For TEST VARIABLE, Select the dependent (response) variable(s)

• For GROUPING VARIABLE, Select the independent variable. Then define the names of the 2 levels to be compared (this can be used even when the full dataset has more than 2 levels for independent variable).

Example 3.5 - Levocabastine in Renal Patients

Group Statistics

6 563.83 172.032 70.232

6 499.67 131.409 53.647

GROUPNon-Dialysis

Hemodialysis

AUCN Mean Std. Deviation

Std. ErrorMean

Independent Samples Test

.204 .661 .726 10 .484 64.17 88.377 -132.750 261.083

.726 9.353 .486 64.17 88.377 -134.613 262.946

Equal variancesassumed

Equal variancesnot assumed

AUCF Sig.

Levene's Test forEquality of Variances

t df Sig. (2-tailed)Mean

DifferenceStd. ErrorDifference Lower Upper

95% ConfidenceInterval of the

Difference

t-test for Equality of Means

Wilcoxon Rank-Sum/Mann-Whitney Tests


• ANALYZE NONPARAMETRIC TESTS 2 INDEPENDENT SAMPLES

• For TEST VARIABLE, Select the dependent (response) variable(s)

• For GROUPING VARIABLE, Select the independent variable. Then define the names of the 2 levels to be compared (this can be used even when the full dataset has more than 2 levels for independent variable).

• Click on MANN-WHITNEY U

Example 3.6 - Levocabastine in Renal Patients

Ranks

6 7.50 45.00

6 5.50 33.00

12

GROUPNon-Dialysis

Hemodialysis

Total

AUCN Mean Rank Sum of Ranks

Test Statisticsb

12.000

33.000

-.962

.336

.394a

Mann-Whitney U

Wilcoxon W

Z

Asymp. Sig. (2-tailed)

Exact Sig. [2*(1-tailedSig.)]

AUC

Not corrected for ties.a.

Grouping Variable: GROUPb.

Paired t-test


• ANALYZE COMPARE MEANS PAIRED SAMPLES T-TEST

• For PAIRED VARIABLES, Select the two dependent (response) variables (the analysis will be based on first variable minus second variable)

Example 3.7 - Cmax in SRC&IRC Codeine

Paired Samples Statistics

217.838 13 79.7792 22.1268

138.815 13 59.3635 16.4645

SRC

IRC

Pair1

Mean N Std. DeviationStd. Error

Mean

Paired Samples Correlations

13 .746 .003SRC & IRCPair 1N Correlation Sig.

Paired Samples Test

79.023 53.0959 14.7262 46.938 111.109 5.366 12 .000SRC - IRCPair 1Mean Std. Deviation

Std. ErrorMean Lower Upper

95% ConfidenceInterval of the

Difference

Paired Differences

t df Sig. (2-tailed)

Wilcoxon Signed-Rank Test


• ANALYZE NONPARAMETRIC TESTS 2 RELATED SAMPLES

• For PAIRED VARIABLES, Select the two dependent (response) variables (be careful in determining which order the differences are being obtained, it will be clear on output)

• Click on WILCOXON Option

Example 3.8 - t1/2SS

in SRC&IRC Codeine

Ranks

9a 6.89 62.00

3b 5.33 16.00

1c

13

Negative Ranks

Positive Ranks

Ties

Total

IRC - SRCN Mean Rank Sum of Ranks

IRC < SRCa.

IRC > SRCb.

IRC = SRCc.

Test Statisticsb

-1.805a

.071

Z

Asymp. Sig. (2-tailed)

IRC - SRC

Based on positive ranks.a.

Wilcoxon Signed Ranks Testb.

Relative Risks and Odds Ratios



• For ROWS, Select the Independent Variable

• For COLUMNS, Select the Dependent Variable

• Under STATISTICS, Click on RISK

• Under CELLS, Click on OBSERVED and ROW PERCENTAGES

• NOTE: You will want to code the data so that the outcome present (Success) category has the lower value (e.g. 1) and the outcome absent (Failure) category has the higher value (e.g. 2). Similar for Exposure present category (e.g. 1) and exposure absent (e.g. 2). Use Value Labels to keep output straight.

Example 5.1 - Pamidronate Study

PAMIDREV * SKLEVREV Crosstabulation

47 149 196

24.0% 76.0% 100.0%

74 107 181

40.9% 59.1% 100.0%

121 256 377

32.1% 67.9% 100.0%

Count

% within PAMIDREV

Count

% within PAMIDREV

Count

% within PAMIDREV

Pamidronate

Placebo

PAMIDREV

Total

Yes No

SKLEVREV

Total

Risk Estimate

.456 .293 .710

.587 .432 .795

1.286 1.113 1.486

377

Odds Ratio for PAMIDREV(Pamidronate / Placebo)

For cohort SKLEVREV =Yes

For cohort SKLEVREV =No

N of Valid Cases

Value Lower Upper

95% ConfidenceInterval

Example 5.2 - Lip CancerPIPESREV * LIPCREV Crosstabulation

339 149 488

69.5% 30.5% 100.0%

198 351 549

36.1% 63.9% 100.0%

537 500 1037

51.8% 48.2% 100.0%

Count

% within PIPESREV

Count

% within PIPESREV

Count

% within PIPESREV

Yes

No

PIPESREV

Total

Yes No

LIPCREV

Total

Risk Estimate

4.033 3.111 5.229

1.926 1.698 2.185

.478 .412 .554

1037

Odds Ratio forPIPESREV (Yes / No)

For cohort LIPCREV =Yes

For cohort LIPCREV = No

N of Valid Cases

Value Lower Upper

95% ConfidenceInterval

Fisher’s Exact Test





• Under STATISTICS, Click on CHI-SQUARE

• Under CELLS, Click on OBSERVED and ROW PERCENTAGES


Example 5.5 - Antiseptic ExperimentTRTREV * DEATHREV Crosstabulation

6 34 40

15.0% 85.0% 100.0%

16 19 35

45.7% 54.3% 100.0%

22 53 75

29.3% 70.7% 100.0%

Count

% within TRTREV

Count

% within TRTREV

Count

% within TRTREV

Antiseptic

Control

TRTREV

Total

Death No Death

DEATHREV

Total

Chi-Square Tests

8.495b 1 .004

7.078 1 .008

8.687 1 .003

.005 .004

8.382 1 .004

75

Pearson Chi-Square

Continuity Correctiona

Likelihood Ratio

Fisher's Exact Test

Linear-by-LinearAssociation

N of Valid Cases

Value dfAsymp. Sig.

(2-sided)Exact Sig.(2-sided)

Exact Sig.(1-sided)

Computed only for a 2x2 tablea.

0 cells (.0%) have expected count less than 5. The minimum expected count is10.27.

b.

McNemar’s Test



• For ROWS, Select the outcome for condition/time 1

• For COLUMNS, Select the outcome for condition/time 2

• Under STATISTICS, Click on MCNEMAR

• Under CELLS, Click on OBSERVED and TOTAL PERCENTAGES


Example 5.6 - Report of Implant Leak

SELFREV * SURGREV Crosstabulation

69 28 97

41.8% 17.0% 58.8%

5 63 68

3.0% 38.2% 41.2%

74 91 165

44.8% 55.2% 100.0%

Count

% of Total

Count

% of Total

Count

% of Total

Present

Absent

SELFREV

Total

Present Absent

SURGREV

Total

Chi-Square Tests

.000a

165

McNemar Test

N of Valid Cases

ValueExact Sig.(2-sided)

Binomial distribution used.a.

P-value

Cochran Mantel-Haenszel Test





• For LAYERS, Select the Strata Variable

• Under STATISTICS, Click on COCHRAN’S AND MANTEL-HAENSZEL STATISTICS


Example 5.7 Smoking/Death by Age

SMOKEREV * DEATHREV * AGE Crosstabulation

Count

647 39990 40637

204 20132 20336

851 60122 60973

857 32894 33751

394 21671 22065

1251 54565 55816

855 20739 21594

488 19790 20278

1343 40529 41872

643 11197 11840

766 16499 17265

1409 27696 29105

Smoke

No Smoke

SMOKEREV

Total

Smoke

No Smoke

SMOKEREV

Total

Smoke

No Smoke

SMOKEREV

Total

Smoke

No Smoke

SMOKEREV

Total

AGE50-54

55-59

60-64

65-69

Death No Death

DEATHREV

Total

Mantel-Haenszel Common Odds Ratio Estimate

1.457

.377

.031

.000

1.372

1.548

.316

.437

Estimate

ln(Estimate)

Std. Error of ln(Estimate)

Asymp. Sig. (2-sided)

Lower Bound

Upper Bound

Common OddsRatio

Lower Bound

Upper Bound

ln(CommonOdds Ratio)

Asymp. 95% ConfidenceInterval

The Mantel-Haenszel common odds ratio estimate is asymptotically normallydistributed under the common odds ratio of 1.000 assumption. So is the natural log ofthe estimate.

Chi-Square Test





• Under STATISTICS, Click on CHI-SQUARE

• Under CELLS, Click on OBSERVED, EXPECTED, ROW PERCENTAGES, and ADJUSTED STANDARDIZED RESIDUALS

• NOTE: Large ADJUSTED STANDARDIZED RESIDUALS (in absolute value) show which cells are inconsistent with null hypothesis of independence. A common rule of thumb is seeing which if any cells have values >3 in absolute value

Example 5.8 - Marital Status & CancerMARITAL * CANCREV Crosstabulation

29 47 76

38.1 37.9 76.0

38.2% 61.8% 100.0%

-2.3 2.3

116 108 224

112.3 111.7 224.0

51.8% 48.2% 100.0%

.7 -.7

67 56 123

61.6 61.4 123.0

54.5% 45.5% 100.0%

1.1 -1.1

5 5 10

5.0 5.0 10.0

50.0% 50.0% 100.0%

.0 .0

217 216 433

217.0 216.0 433.0

50.1% 49.9% 100.0%

Count

Expected Count

% within MARITAL

Adjusted Residual

Count

Expected Count

% within MARITAL

Adjusted Residual

Count

Expected Count

% within MARITAL

Adjusted Residual

Count

Expected Count

% within MARITAL

Adjusted Residual

Count

Expected Count

% within MARITAL

Single

Married

Widowed

Div/Sep

MARITAL

Total

Cancer No Cancer

CANCREV

Total

Chi-Square Tests

5.530a 3 .137

5.572 3 .134

3.631 1 .057

433

Pearson Chi-Square

Likelihood Ratio

Linear-by-LinearAssociation

N of Valid Cases

Value dfAsymp. Sig.

(2-sided)

1 cells (12.5%) have expected count less than 5. Theminimum expected count is 4.99.

a.

Goodman & Kruskal’s / Kendall’s b





• Under STATISTICS, Click on GAMMA and KENDALL’S b

Examples 5.9,10 - Nicotine Patch/Exhaustion

DOSE * EXHSTN Crosstabulation

Count

16 2 18

16 2 18

13 4 17

14 4 18

59 12 71

1

2

3

4

DOSE

Total

1 2

EXHSTN

Total

Symmetric Measures

.124 .104 1.166 .243

.269 .220 1.166 .243

71

Kendall's tau-b

Gamma

Ordinal byOrdinal

N of Valid Cases

ValueAsymp.

Std. Errora

Approx. Tb

Approx. Sig.

Not assuming the null hypothesis.a.

Using the asymptotic standard error assuming the null hypothesis.b.

Kruskal-Wallis Test


• ANALYZE NONPARAMETRIC TESTS k INDEPENDENT SAMPLES

• For TEST VARIABLE, Select Dependent Variable

• For GROUPING VARIABLE, Select Independent Variable, then define range of levels of variable (Minimum and Maximum)

• Click on KRUSKAL-WALLIS H

Example 5.11 - Antibiotic Delivery

Ranks

181 285.65

181 261.15

179 266.14

541

DELIVERY1

2

3

Total

OUTCOMEN Mean Rank

Test Statisticsa,b

2.755

2

.252

Chi-Square

df

Asymp. Sig.

OUTCOME

Kruskal Wallis Testa.

Grouping Variable: DELIVERYb.

Note: This statistic makes the adjustment for ties. See Hollander and Wolfe (1973), p. 140.

Cohen’s



• For ROWS, Select Rater 1

• For COLUMNS, Select Rater 2

• Under STATISTICS, Click on KAPPA

• Under CELLS, Click on TOTAL Percentages to get the observed percentages in each cell (the first number under observed count in Table 5.17).

Example 5.12 - Siskel & EbertSISKEL * EBERT Crosstabulation

24 8 13 45

15.0% 5.0% 8.1% 28.1%

8 13 11 32

5.0% 8.1% 6.9% 20.0%

10 9 64 83

6.3% 5.6% 40.0% 51.9%

42 30 88 160

26.3% 18.8% 55.0% 100.0%

Count

% of Total

Count

% of Total

Count

% of Total

Count

% of Total

-1

0

1

SISKEL

Total

-1 0 1

EBERT

Total

Symmetric Measures

.389 .060 6.731 .000

160

KappaMeasure of Agreement

N of Valid Cases

ValueAsymp.

Std. Errora

Approx. Tb

Approx. Sig.

Not assuming the null hypothesis.a.

Using the asymptotic standard error assuming the null hypothesis.b.

1-Factor ANOVA - Independent Samples (Parallel Groups)


• ANALYZE COMPARE MEANS ONE-WAY ANOVA

• For DEPENDENT LIST, Click on the Dependent Variable

• For FACTOR, Click on the Independent Variable

• To obtain Pairwise Comparisons of Treatment Means:– Click on POST HOC

– Then TUKEY and BONFERRONI (among many other choices)

Examples 6.1,2 - HIV Clinical TrialANOVA

CD4

7074.600 2 3537.300 8.901 .000

106108.0 267 397.408

113182.6 269

Between Groups

Within Groups

Total

Sum ofSquares df Mean Square F Sig.

Multiple Comparisons

Dependent Variable: CD4

7.10000* 2.971749 .046 .09601 14.10399

12.50000* 2.971749 .000 5.49601 19.50399

-7.10000* 2.971749 .046 -14.10399 -.09601

5.40000 2.971749 .166 -1.60399 12.40399

-12.50000* 2.971749 .000 -19.50399 -5.49601

-5.40000 2.971749 .166 -12.40399 1.60399

7.10000 2.971749 .053 -.05942 14.25942

12.50000* 2.971749 .000 5.34058 19.65942

-7.10000 2.971749 .053 -14.25942 .05942

5.40000 2.971749 .211 -1.75942 12.55942

-12.50000* 2.971749 .000 -19.65942 -5.34058

-5.40000 2.971749 .211 -12.55942 1.75942

(J) TRTSZ

ZZ

SZZ

ZZ

SZZ

SZ

SZ

ZZ

SZZ

ZZ

SZZ

SZ

(I) TRTSZZ

SZ

ZZ

SZZ

SZ

ZZ

Tukey HSD

Bonferroni

MeanDifference

(I-J) Std. Error Sig. Lower Bound Upper Bound

95% Confidence Interval

The mean difference is significant at the .05 level.*.

CD4

90 -.30000

90 5.10000

90 12.20000

.166 1.000

TRTZZ

SZ

SZZ

Sig.

Tukey HSDaN 1 2

Subset for alpha = .05

Means for groups in homogeneous subsets are displayed.

Uses Harmonic Mean Sample Size = 90.000.a.

Kruskal-Wallis Test


• ANALYZE NONPARAMETRIC TESTS k INDEPENDENT SAMPLES

• For TEST VARIABLE, Select Dependent Variable

• For GROUPING VARIABLE, Select Independent Variable, then define range of levels of variable (Minimum and Maximum)

• Click on KRUSKAL-WALLIS H

Example 6.2(a) - Thalidomide and HIV-1

Ranks

8 24.44

8 21.63

8 6.56

8 13.38

32

TRT1

2

3

4

Total

WTGAINN Mean Rank

Test Statisticsa,b

18.070

3

.000

Chi-Square

df

Asymp. Sig.

WTGAIN

Kruskal Wallis Testa.

Grouping Variable: TRTb.

Randomized Block Design - F-test• After Importing your dataset, and providing names to

variables, click on:• ANALYZE GENERAL LINEAR MODEL UNIVARIATE

• Assign the DEPENDENT VARIABLE

• Assign the TREATMENT variable as a FIXED FACTOR

• Assign the BLOCK variable as a RANDOM FACTOR

• Click on MODEL, then CUSTOM, under BUILD TERMS choose MAIN EFFECTS, move both factors to MODEL list

• Click on POST HOC and select the TREATMENT factor for POST HOC TESTS and BONFERRONI and TUKEY (among many choices)

• For PLOTS, Select the BLOCK factor for HORIZONTAL AXIS and the TREATMENT factor for SEPARATE LINES, click ADD

Example 6.3 - Theophylline ClearanceTests of Between-Subjects Effects

Dependent Variable: THEOPHCL

336.713 1 336.713 60.955 .000

71.811 13 5.524a

7.005 2 3.503 10.591 .000

8.599 26 .331b

71.811 13 5.524 16.703 .000

8.599 26 .331b

SourceHypothesis

Error

Intercept

Hypothesis

Error

DRUG

Hypothesis

Error

SUBJECT

Type III Sumof Squares df Mean Square F Sig.

MS(SUBJECT)a.

MS(Error)b. Multiple Comparisons

Dependent Variable: THEOPHCL

-.0800 .21736 .928 -.6201 .4601

.8236* .21736 .002 .2835 1.3637

.0800 .21736 .928 -.4601 .6201

.9036* .21736 .001 .3635 1.4437

-.8236* .21736 .002 -1.3637 -.2835

-.9036* .21736 .001 -1.4437 -.3635

-.0800 .21736 1.000 -.6362 .4762

.8236* .21736 .002 .2674 1.3798

.0800 .21736 1.000 -.4762 .6362

.9036* .21736 .001 .3474 1.4598

-.8236* .21736 .002 -1.3798 -.2674

-.9036* .21736 .001 -1.4598 -.3474

(J) DRUGFamotidine

Placebo

Cimetidine

Placebo

Cimetidine

Famotidine

Famotidine

Placebo

Cimetidine

Placebo

Cimetidine

Famotidine

(I) DRUGCimetidine

Famotidine

Placebo

Cimetidine

Famotidine

Placebo

Tukey HSD

Bonferroni

MeanDifference

(I-J) Std. Error Sig. Lower Bound Upper Bound

95% Confidence Interval

Based on observed means.

The mean difference is significant at the .05 level.*.


THEOPHCL

14 2.2557

14 3.0793

14 3.1593

1.000 .928

DRUGPlacebo

Cimetidine

Famotidine

Sig.

Tukey HSDa,bN 1 2

Subset

Means for groups in homogeneous subsets are displayed.Based on Type III Sum of SquaresThe error term is Mean Square(Error) = .331.

Uses Harmonic Mean Sample Size = 14.000.a.

Alpha = .05.b.

Estimated Marginal Means of THEOPHCL

SUBJECT

1413121110987654321

Est

ima

ted

Ma

rgin

al M

ea

ns

7

6

5

4

3

2

1

0

DRUG

Cimetidine

Famotidine

Placebo

Randomized Block Design - Friedman’s test


• ANALYZE NONPARAMETRIC TESTS k RELATED SAMPLES

• For TEST VARIABLES, select the variables representing the treatments (each line is a subject/block)

• Click on FRIEDMAN

Example 6.4 - Absorption of Valproate Depakote

Ranks

1.73

2.95

1.32

CAPFAST

CAPNFAST

ECFAST

Mean Rank

Test Statisticsa

11

17.550

2

.000

N

Chi-Square

df

Asymp. Sig.

Friedman Testa.

Note: This makes an adjustment for ties, see Hollander and Wolfe (1973), p. 140.

2-Way ANOVA


• ANALYZE GENERAL LINEAR MODEL UNIVARIATE

• Assign the DEPENDENT VARIABLE

• Assign the FACTOR A variable as a FIXED FACTOR

• Assign the FACTOR B variable as a FIXED FACTOR

• Click on MODEL, then CUSTOM, select FULL FACTORIAL

• Click on POST HOC and select the both factors for POST HOC TESTS and BONFERRONI and TUKEY (among many choices)

• For PLOTS, Select FACTOR B for HORIZONTAL AXIS and FACTOR A for SEPARATE LINES, click ADD

Example 6.5 - Nortriptyline ClearanceTests of Between-Subjects Effects

Dependent Variable: CLRNCE

23.164a 3 7.721 1.444 .267

1131.008 1 1131.008 211.532 .000

.450 1 .450 .084 .775

20.402 1 20.402 3.816 .068

2.312 1 2.312 .432 .520

85.548 16 5.347

1239.720 20

108.712 19

SourceCorrected Model

Intercept

ETHNIC

GENDER

ETHNIC * GENDER

Error

Total

Corrected Total

Type III Sumof Squares df Mean Square F Sig.

R Squared = .213 (Adjusted R Squared = .066)a.

Estimated Marginal Means of CLRNCE

ETHNIC

21

Est

ima

ted

Ma

rgin

al M

ea

ns

9.5

9.0

8.5

8.0

7.5

7.0

6.5

6.0

GENDER

1

2

Linear Regression


• ANALYZE REGRESSION LINEAR

• Select the DEPENDENT VARIABLE

• Select the INDEPENDENT VARAIABLE(S)

• Click on STATISTICS, then ESTIMATES, CONFIDENCE INTERVALS, MODEL FIT

• For histogram of residuals, click on PLOTS, and HISTOGRAM under STANDARDIZED RESIDUAL PLOTS

Examples 7.1-7.6 - Gemfibrozil ClearanceCoefficientsa

460.828 54.338 8.481 .000 345.010 576.646

-3.215 1.181 -.575 -2.723 .016 -5.732 -.698

(Constant)

CLCR

Model1

B Std. Error

UnstandardizedCoefficients

Beta

StandardizedCoefficients

t Sig. Lower Bound Upper Bound

95% Confidence Interval for B

Dependent Variable: CLGMa.

Regression Standardized Residual

1.501.00.500.00-.50-1.00-1.50

Histogram

Dependent Variable: CLGM

Fre

qu

en

cy

6

5

4

3

2

1

0

Std. Dev = .97

Mean = 0.00

N = 17.00

Examples 7.1-7.6 - Gemfibrozil ClearanceANOVAb

107168.2 1 107168.158 7.413 .016a

216865.8 15 14457.723

324034.0 16

Regression

Residual

Total

Model1


Predictors: (Constant), CLCRa.

Dependent Variable: CLGMb.

Model Summaryb

.575a .331 .286 120.240Model1

R R SquareAdjustedR Square

Std. Error ofthe Estimate

Predictors: (Constant), CLCRa.

Dependent Variable: CLGMb.

Example 7.8 - TB/Thalidomide in HIV

Coefficientsa

2.662 .635 4.190 .000

.597 .116 .692 5.161 .000

-.330 .258 -.167 -1.279 .211

-.571 .262 -.289 -2.179 .038

(Constant)

LN_X

DRUG

TB

Model1

B Std. Error

UnstandardizedCoefficients

Beta

StandardizedCoefficients

t Sig.

Dependent Variable: LN_Ya.

ANOVAb

16.698 3 5.566 10.696 .000a

14.571 28 .520

31.270 31

Regression

Residual

Total

Model1


Predictors: (Constant), TB, DRUG, LN_Xa.

Dependent Variable: LN_Yb.

Useful Regression Plots

• Scatterplot with Fitted (Least Squares) Line– GRAPHS INTERACTIVE SCATTERPLOT

– Select DEPENDENT VARIABLE for UP/DOWN AXIS

– Select INDEPENDENT VARIABLE for RIGHT/LEFT AXIS

– Click on FIT Tab, then REGRESSION for METHOD

– NOTE: Be certain both variables are SCALE in VARIABLE VIEW under MEASURE

• Partial Regression Plots (Multiple Regression) to observe association of each Independent Variable with Y, controlling for all others– Fit REGRESSION model with all Independent Variables

– Click PLOTS, then PRODUCE ALL PARTIAL PLOTS

Example 7.1 - Gemfibrozil Scatterplot

Linear Regression

20 40 60

clcr

200

300

400

500

600

clg

m

clgm = 460.83 + -3.22 * clcrR-Square = 0.33

Logistic Regression


• ANALYZE REGRESSION BINARY LOGISTIC


• Select the INDEPENDENT VARAIABLE(S) as COVARIATES

• For a 95% CI for the odds ratio, click on OPTIONS, then CI for exp(B)

• Declare any CATEGORICAL COVARIATES (Independent variables whose levels are categorical, not numeric)

Example 8.1 - Navelbine Toxicity

Variables in the Equation

.488 .052 88.238 1 .000 1.628 1.471 1.803

-6.381 .690 85.498 1 .000 .002

DOSE

Constant

Step1

a

B S.E. Wald df Sig. Exp(B) Lower Upper

95.0% C.I.for EXP(B)

Variable(s) entered on step 1: DOSE.a.

Omnibus Tests of Model Coefficients

210.310 1 .000

210.310 1 .000

210.310 1 .000

Step

Block

Model

Step 1Chi-square df Sig.

Omnibus test for all regression coefficients (like F in linear regression)

Example 8.2 - CHD, BP, Cholesterol


6.394 1.475 18.792 1 .000 598.277 33.218 10775.391

3.454 .838 17.008 1 .000 31.631 6.126 163.319

-24.020 3.699 42.158 1 .000 .000

LOG10SC

LOG10BP

Constant

Step1

a

B S.E. Wald df Sig. Exp(B) Lower Upper

95.0% C.I.for EXP(B)

Variable(s) entered on step 1: LOG10SC, LOG10BP.a.

Omnibus Tests of Model Coefficients

42.566 2 .000

42.566 2 .000

42.566 2 .000

Step

Block

Model

Step 1Chi-square df Sig.

Nonlinear Regression


• ANALYZE REGRESSION NONLINEAR


• Define the MODEL EXPRESSION as a function of the INDEPENDENT VARIABLE(s) and unknown PARAMETERS

• Define the PARAMETERS and give them STARTING VALUES (this may take several attempts)

Example 8.3 - MK-639 in AIDS PatientsNonlinear Regression Summary Statistics Dependent Variable RNACHNG

Source DF Sum of Squares Mean Square

Regression 3 24.97099 8.32366 Residual 2 .02783 .01391 Uncorrected Total 5 24.99881

(Corrected Total) 4 10.83973

R squared = 1 - Residual SS / Corrected SS = .99743

Asymptotic 95 % Asymptotic Confidence Interval Parameter Estimate Std. Error Lower Upper

A 3.521788512 .121466117 2.999161991 4.044415032 B 35.598069675 7.532265897 3.189345253 68.006794097 C 18374.392967 82.899219276 18017.706415 18731.079519

Parameters,, :Model 60

CBAAUCxCx

Axy hBB

B

Survival Analysis -Kaplan-Meier Estimates and Log-Rank Test


• ANALYZE SURVIVAL KAPLAN-MEIER

• Select the variable representing the survival TIME of individual

• Select the variable representing the STATUS of individual (whether or not event has occured). NOTE: If the variable is an indicator that the observation was CENSORED, then a value of 0 for that variable will mean the event has occured.

• Select the variable representing the FACTOR containing the groups to be compared

• Click on COMPARE FACTOR, select LOG-RANK, and POOL ACROSS STRATA

Examples 9.1-2 - Navelbine and Taxol in MiceSurvival Analysis for TIME

Factor REGIMEN = 1

Time Status Cumulative Standard Cumulative Number Survival Error Events Remaining

6 0 .9796 .0202 1 48 8 0 .9592 .0283 2 47 22 0 .9388 .0342 3 46 32 0 4 45 32 0 .8980 .0432 5 44 35 0 .8776 .0468 6 43 41 0 .8571 .0500 7 42 46 0 .8367 .0528 8 41 54 0 .8163 .0553 9 40

Factor REGIMEN = 2

Time Status Cumulative Standard Cumulative Number Survival Error Events Remaining

8 0 .9333 .0644 1 14 10 0 .8667 .0878 2 13 27 0 .8000 .1033 3 12 31 0 .7333 .1142 4 11 34 0 .6667 .1217 5 10 35 0 .6000 .1265 6 9 39 0 .5333 .1288 7 8 47 0 .4667 .1288 8 7 57 0 .4000 .1265 9 6

Examples 9.1-2 - Navelbine and Taxol in MiceSurvival Functions

TIME

706050403020100

Cu

m S

urv

iva

l

1.1

1.0

.9

.8

.7

.6

.5

.4

.3

REGIMEN

2

2-censored

1

1-censored

Test Statistics for Equality of Survival Distributions for REGIMEN Statistic df Significance Log Rank 10.93 1 .0009

This is the square of the Z-statistic in text, and is a chi-square statistic

Relative Risk Regression (Cox Model)


• ANALYZE SURVIVAL COX REGRESSION

• Select the variable representing the survival TIME of individual

• Select the variable representing the STATUS of individual (whether or not event has occured). NOTE: If the variable is an indicator that the observation was CENSORED, then a value of 0 for that variable will mean the event has occured.

• Select the variable(s) representing the COVARIATES (Independent Variables in Model)

• Identify any CATEGORICAL COVARIATES including Dummy/Indicator variables

• K-M PLOTS can be obtained, with separate SURVIVAL curves by categories

Example 9.3 - 6MP vs Placebo


-1.509 .410 13.578 1 .000 .221 .099 .493TRTB SE Wald df Sig. Exp(B) Lower Upper

95.0% CI for Exp(B)

Survival Function for patterns 1 - 2

REMSTIME

3020100-10

Cu

m S

urv

iva

l

1.2

1.0

.8

.6

.4

.2

0.0

TRT

Placebo

6MP

SPSS Instructions for Introduction to Biostatistics

Technology