1. Graphpad Prism 1
Introduction

2. Download & Install
You can obtain GraphPad/Prism by going to “Software Download Services” in MOLE.
Look for “Graphpad Prism for Windows or Mac”

3. Uniqueness of Prism
Prism caters for analysis and graphs for scientific publication, especially for laboratory and biomedical research.
Data are usually entered and manipulated using spreadsheet such as Microsoft Excel. Data needed for analysis are copied into specific tables within Prism.
Specific analysis requires specific tables. So you must know exactly what analysis is required.

4. The 6 tables within Prism.

5. XY Data Tables

6. Column tables

7. Grouped tables

8. Contingency tables

9. Survival tables

10. Parts of whole tables

11. Choosing the appropriate statistical tests
Use these tables to choose the appropriate statistical tests.

12. Parametric Statistical Tests
Independent
Dependent
Test
Qualitative
Dichotomus
Quantitative
Normally distributed data
Student's t Test
(more than 3 categories)
ANOVA
Repeated measurement of the
Paired t Test
same individual & item (e.g. Hb level before & after treatment). Normally distributed data
Quantitative -
continous
Pearson Correlation
& Linear Regresssion

13. Non-parametric Statistical Tests
Independent
Dependent
Test
Qualitative
Dichotomus
Quantitative
Data not normally distributed
Wilcoxon Rank Sum
Test or U Mann- Whitney Test
Kruskal-Wallis One
Polinomial
Way ANOVA Test
Repeated measurement of the
same individual & item
Wilcoxon Rank Sign
Test
Quantitative -
continous/ordina l
continous
Spearman/Kendall
Rank Correlation

14. Statistical Tests for Qualitative Data
Variable 1
Variable 2
Criteria
Type of Test
Qualitative
Sample size > 20 dan no
expected value < 5
Chi Square Test (X2)
Sample size > 30
Proportionate Test
Dichotomus
Sample size > 40 but with at
least one expected value < 5
X2 Test with Yates Correction
Sample size < 20 or (< 40 but
with at least one expected value < 5)
Fisher Test

15. Data – Factors Related to SGA

16. A study to identify factors that can cause small for gestational age (SGA) was conducted. Among the factors studied were the mothers’ body mass index (BMI). It is believed that mothers with lower BMI were of higher risk to get SGA babies.
4. Conduct the appropriate statistical test to test whether there is any association between BMI and OUTCOME.
5. Conduct the appropriate statistical test to find any association between OBESCLAS (Underweight/Normal/Overweight) and BIRTHWGT.
6. Assuming that both variables mBMI & BIRTHWGT are normally distributed, conduct an appropriate statistical test to prove the association between the two variables.
– Demonstrate the association using the appropriate chart. Determine the coefficient of determination.
7. Conduct Simple Linear Regression using BIRTHWGT as the dependent variable. Try to come out with a formula that will predict the baby’s birthweight based on the mother’s BMI.
1. Create a new variable mBMI (Mothers’ Body Mass Index) from the mothers’ HEIGHT (in metre) & WEIGHT (first trimester weight in kg). mBMI = weight in kg/(height in metre)2. Calculate the following for mBMI;
Mean
Standard deviation
2. Create a new variable OBESCLAS (Classification of Obesity) from mBMI. Use the following cutoff point;
<20 = Underweight
– 20 – = Normal
25 or larger = Overweight
Create a frequency table for OBESCLAS.
3. Conduct the appropriate statistical test to test whether there is any association between OBESCLAS (Underweight/ Normal/Overweight) and OUTCOME.
– y = a + bx

17. Exercise 1 & 2
1. Create a new variable mBMI (Mothers’ Body Mass Index) from the mothers’ HEIGHT (in metre) & WEIGHT (first trimester weight in kg). mBMI = weight in kg/(height in metre)2. Calculate the following for mBMI;
Mean
Standard deviation
2. Create a new variable OBESCLAS (Classification of Obesity) from mBMI. Use the following cutoff point;
<20 = Underweight
– 20 – = Normal
25 or larger = Overweight
Create a frequency table for OBESCLAS.

18. Compute BMI
Drag down to fill up the cells

19. Recode BMI into OBESCLAS
Type
=IF(F2<20,"Underweigh t",IF(F2>25,"Overweight ","Normal")) in cell G2 and press Enter.
Then drag down cell G2 until G101 to fill up the rest of the cells.
Try to “understand” the code below:
If (“condition 1 satisfied” , then put Value 1 , otherwise put Value2). (Comma is very important here!)
Instead of Value 2 you can introduce another IF! Can you see how it works?

20. Recode BMI into OBESCLAS

21. Recode BMI into 1,2 or 3
We should also recode BMI into numeric OBESCLAS2 for import into Prism. Prism doesn’t accept string data.
• =IF(F2<20,“1",IF(F2>25,“3
",“2")) in cell H2 and press Enter.
Then drag down cell H2 until H101 to fill up the rest of the cells.

22. Pivot Table: Frequency (Count) of data by category of BMI
Select column “OBESCLAS”
Click “Insert” -> “Pivot Table”
Click OK and it will open a new sheet in Excel.
Drag “OBESCLAS” in “Row” once.
Drag “OBESCLAS” in Sigma value.
You will see a table of count of data.
To switch to percentage, you can right click on one cell of the table and “show value as percentage Total”.
You can create a Pie chart by clicking on the table and going in Pie Chart.
You can also right click on one of the slices of the Pie Chart and click on “Add Data Labels”.
You can also modify the title of the graph by clicking on “Total” and changing the text.

24. Import Excel Data Into Prism
Select all the data from •
Excel. Copy.
Open Prism, select “Columns”, “Enter replicate values..” & click “Create”

25. Paste Into Prism Click the cell between “Group A” and row Y and paste.

26. Checking Normality Click on the “Analyze” button.
Select “Column Statistics”.
Select the variables with continuous data.
Then click “OK”.

27. Click on the following;
Test if the values from a Gaussian distribution.

28. Only Height is normally distributed
But for the purpose of today’s exercise, we are going to ASS-U-ME that all these continuous variables are normally distributed.

29. Question 1 – BMI Column Statistics also generates the Mean & S.D.;

30. Frequency Distribution
Go back to the data by clicking on the data table on left side of screen. Then click on the “Analyze” button again.
Select “Frequency Distribution”
Tick on OBESCLAS2. Then click on “OK”.

31. Frequency Distribution
Then click on OK again. You will get the following frequency distribution table.

32. Question 2 – Obese Classification
UW – 17%
N – 40%
OW – 43%

33. Exercise 3
3. Conduct the appropriate statistical test to test whether there is any association between OBESCLAS
SGA
Normal
TOTAL
UnderW
OverW
TOTAL 50 50
100
(Underweight/Normal/ Overweight) and OUTCOME.
Therefore most suitable analysis is Pearson Chi- square.
Variable 1
Variable 2
Criteria
Type of Test
Qualitative
Sample size > 20 dan no
expected value < 5
Chi Square Test (X2)
Dichotomus
Sample size > 30
Proportionate Test
Sample size > 40 but with at
least one expected value < 5
X2 Test with Yates Correction
Sample size < 20 or (< 40 but
with at least one expected value < 5)
Fisher Test

34. Pivot Table in Excel Click on “Insert”, “Pivot Table” in Excel.
Select all your earlier Excel data.

35. Pivot Table
On the right side of the screen, pull FREQ into values, OBESCLAS into row labels and OUTCOME into column labels.
Now select the created contingency table (excluding the “Grand Total”), and copy it using Ctrl-C.

36. Paste Pivot Table Into Prism
Click “New”, “New Data Table”.
Select “Contingency”, “Start with an empty table”.
Then paste the pivot table into Prism.

37. The Pasted Pivot Table

38. Chi-Square Analysis
Click on “Analyze”, “Contingency table analysis”, then “Chi- square”, then OK again twice.

39. Chi-Square Results from Prism
Contingency 60
SGA
Normal
Frequency 40 20
Normal Overweight Underweight
Mothers' Weight Classification
Prism only states that there is a significant association (p
< ) between mother’s weight classification and small for gestational age.
But it doesn’t show which group has the higher rate of SGA.

40. Combine Results From Excel & Prism
There is a significant difference (p<0.0001) of SGA rates between underweight, normal and overweight mothers.
Underweight mothers has a higher rate (94%) of SGA, compared to normal mothers (58%) and overweight mothers (26%).

41. Underweight vs Normal?
There is a significant difference (p<0.01) of SGA rates between underweight and normal mothers.
Underweight mothers has a significantly higher rate (94%) of SGA, compared to normal mothers (58%).

42. Question 3

43. Question 3

44. Exercise 4
4. Conduct the appropriate statistical test to test whether there is any association between BMI and OUTCOME.
Basically we are comparing the mean BMI of SGA mothers against BMI of Normal mothers.
Therefore the appropriate test is Student’s t-test.
Independent
Dependent
Test
Qualitative
Dichotomus
Quantitative
Data not normally distributed
Wilcoxon Rank Sum
Test or U Mann- Whitney Test
Kruskal-Wallis One
Polinomial
Way ANOVA Test
Repeated measurement of the
same individual & item
Wilcoxon Rank Sign
Test
Quantitative -
continous/ordina l
continous
Spearman/Kendall
Rank Correlation

45. Copy BMI Column Into Prism
Click “New”, “New Data Table”.
Select “Column”, “Enter replicate values into stacked columns”.
Then paste the BMI of SGA mothers into column A & BMI of Normal mothers into column B.

46. The Pasted BMI Data

47. Student’s T-Test
Click on “Analyze”, “Column analysis”, then “t-tests”, then OK again.
Tick “Unpaired”, “Yes, parametric”, then “equal SDs”, then OK again.

48. T-Test Results from Prism
Prism states that there is a significant mean difference of BMI (p < ) between SGA mother’s (22.52) and normal mothers (26.46). Therefore mean BMI of SGA mothers is significantly lower than the normal mothers.
And it also proves that there is equal variances of the two means.

49. BMI

50. Question 4

51. Question 4

52. Exercise 5
5. Conduct the appropriate statistical test to find any association between OBESCLAS (Underweight/Normal/Overweight) and BIRTHWGT.
Basically we are comparing the mean BIRTHWEIGHT of underweight mothers, normal weight mothers and overweight mothers.
Therefore the appropriate test is Analysis of Variance (ANOVA).

53. Sort Excel Data By BMI To Facilitate Copy & Paste

54. Copy Birth Weight Column Into Prism
Click “New”, “New Data Table”.
Select “Column”, “Enter replicate values into stacked columns”.
Then paste the babies’ birth weight of underweight mothers into column A, babies’ birth weight of normal weight mothers into column B & babies birth weight of overweight mothers in column C.

55. The Pasted Birth Weight Data

56. ANOVA
Click on “Analyze”, “Column analysis”, then “One-way ANOVA”, then OK again.
Tick “No matching”, “Yes, ANOVA”, then click “Multiple
Comparison” tab.
Click OK

57. ANOVA – post hoc
Click OK

58. ANOVA Results from Prism
Prism states that there is a significant mean difference of mean birth weight (p < ) between underweight mothers’ (2.187), normal mothers ‘(2.768) & overweight mothers’(3.245).
Unfortunately it also proves that there is unequal variances of the three means. So it fails the homogeneity of variances assumption.

59. ANOVA Results – post hoc
Post-hoc tests indicate there is significant difference of birth weight between ALL the three groups. Underweight mothers’ have the lowest mean birth weight of 2.187kg.

60. Compare Babies Birth Weight by Mother's Weight
ANOVA 5 4 3 2 1
Underweight Normal Overweight
Compare Babies Birth Weight by Mother's Weight
Birth weight

61. Question 5

62. Question 5

63. Exercise 6
6. Assuming that both variables mBMI & BIRTHWGT are normally distributed, conduct an appropriate statistical test to prove the association between the two variables.
– Demonstrate the association using the appropriate chart. Determine the coefficient of determination.

64. Pearson Correlation
Qualitative
Dichotomus
Quantitative
Normally distributed data
Student's t Test
Polinomial
ANOVA
Repeated measurement of the
same individual & item (e.g. Hb level before & after treatment). Normally distributed data
Paired t Test
Quantitative -
continous
Pearson Correlation
& Linear Regresssion
mBMI and birth weight are both normally distributed continuous data. Since the aim is to measure the strength and direction of the association between these two continuous variable, therefore Pearson Correlation is the most appropriate test.

65. Copy BMI & Birth Weight Into Prism
Click “New”, “New Data Table”.
Select “XY”, “Enter and plot a single Y value for each point”.
Then paste the BMI into column X & BIRTHWGT into column A.

66. The Pasted BMI & Birth weight Data
BMI is coded as X since it is the risk factor.
Birth weight is coded as Y since it is the outcome of interest.
Risk factor first, then Outcome.
X comes first before Y.
Capisce? (Understand?)

67. Pearson’s Correlation
Click on “Analyze”, “XY analysis”, then “Correlation”, then OK again.
Tick “Compute r between two selected data sets”, “Yes, Pearson correlation coefficients”, then “Two-tailed”, then OK again.

68. Correlation Results from Prism
Prism states that there is a significant, positive & fair (r=0.4812) correlation between mothers’ BMI and babies’ birth weight. Therefore as BMI increases, the birth weight also increases.
23.15% (r2=0.2315)
variability of the birth weight is determined by the variability of the mothers’ BMI.

69. Scatter Diagram - BMI vs Birth weight5 4 3 2 1
Birth weight 10 20 30 40 50
BMI

70. Question 6

71. Question 6

72. Exercise 7
7. Conduct Simple Linear Regression using BIRTHWGT as the dependent variable. Try to come out with a formula that will predict the baby’s birth weight based on the mother’s BMI.
– y = a + bx

73. Simple Linear Regression
Qualitative
Dichotomus
Quantitative
Normally distributed data
Student's t Test
Polinomial
ANOVA
Repeated measurement of the
same individual & item (e.g. Hb level before & after treatment). Normally distributed data
Paired t Test
Quantitative -
continous
Pearson Correlation
& Linear Regresssion
mBMI and birth weight are both normally distributed continuous data. Since the aim is to come out with a regression formula between these two continuous variable, therefore Simple Linear Regression is the most appropriate test.

74. Reuse BMI & Birth weight Data
BMI is coded as X since it is the risk factor.
Birth weight is coded as Y since it is the outcome of interest.
Since the SLR uses the same variables, we will reuse the XY table from Exercise 6.

75. Simple Linear Regression
Click on “SLR” icon, it is just above the “Analyze” icon.
Just change the range so that the line will start at the y axis (X=0).
We can set the line to end at the maximum value (it is X=41 in this exercise).
Click OK

76. SLR Results from Prism
Prism states that there is a significant regression coefficient (b= ).
The constant (a) is 1.081
23.15% (r2=0.2315)
variability of the birth weight is determined by the variability of the mothers’ BMI.
BW = BMI
For every increase of BMI of 1 unit, BW increases 0.07kg.

77. Scatter Diagram - BMI vs Birth weight5 4 3 2 1
Birth weight 10 20 30 40 50
BMI

78. Question 7

79. Question 7 Slight difference of the constant value. Prism calculated
1.081 instead of Maybe it was due to decimal difference of the BMI upon import.

80. Question 7 d
r t a m
i l @
g m a
i l
c o m .
Birth weight

81. The End