Statistics with Octa-

ve/Matlab
RMPE Handouts
Rakesh Sengupta

Rakesh Sengupta, Center for Creative Cognition, SRU

[email protected]
Contenido
4 Experimental designs
1.1 Key Components of Experimental Design . . . . . . . . . . . . . . . . . . . . . . . 4
1.2 Control Conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Counterbalancing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.4 Randomization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.5 Threshold Measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
1.6 ROC Curve . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

7 Statistics and Inference

2.1 Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Histogram and Density Plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.3 Normality Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.4 T-Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.5 Correlation and Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.6 One-Way ANOVA . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.7 Post-Hoc Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

2
This page is intentionally left blank.
Experimental designs
1.1 Key Components of Experimental Design
Hypothesis: A clear, testable statement predicting the outcome of the experiment.
Independent Variable (IV): The variable that is manipulated by the experimenter.
Dependent Variable (DV): The variable that is measured and is expected to change due to
manipulations of the IV.
Control Variables: Variables that are kept constant to prevent them from influencing the
DV.
Experimental Group: The group exposed to the treatment or condition.
Control Group: The group not exposed to the treatment, serving as a baseline for comparison.

1.2 Control Conditions

Definition: Control conditions are used to isolate the effect of the independent variable by
providing a baseline to compare against the experimental condition.
Purpose: - Ensure that observed effects are due to the IV and not extraneous factors. - Increase
the internal validity of the experiment.
Example: In a study measuring the effect of a new drug on memory, the control group would
receive a placebo.

1.3 Counterbalancing
Definition: A method used to control for order effects in repeated measures designs by varying
the order of conditions for different participants.
Types: - Complete Counterbalancing: All possible orders of conditions are used. - Partial
Counterbalancing: A subset of all possible orders is used.
Example: If testing the effect of three different types of noise on reading comprehension,
counterbalance the order in which participants experience each type of noise.

1.4 Randomization
Definition: The process of randomly assigning participants to different conditions or groups
to ensure each participant has an equal chance of being assigned to any group.
Purpose: - Minimize bias. - Control for confounding variables. - Ensure groups are comparable
at the start of the experiment.
Example: Randomly assigning participants to either a treatment or control group in a clinical
trial.

4
 Threshold Measurements • 5

1.5 Threshold Measurements

Definition: Methods used to determine the smallest detectable level of a stimulus.
Methods: - Method of Constant Stimuli: Presenting stimuli of varying intensities in ran-
dom order. - Method of Limits: Gradually increasing or decreasing the stimulus intensity
until it is detectable. - Method of Adjustment: Participants adjust the stimulus intensity to
the threshold level.
Example: Determining the minimum volume at which a tone can be heard.

6. Signal Detection Theory (SDT)

Definition: A framework for measuring the ability to differentiate between signal and noise.
Components: - Hit: Correctly identifying the presence of a signal. - Miss: Failing to identify
the presence of a signal. - False Alarm: Incorrectly identifying the presence of a signal when
there is none. - Correct Rejection: Correctly identifying the absence of a signal.
d’: A measure of sensitivity, representing the distance between the signal and noise distributions.

d ′ = Z(hit rate) − Z(false alarm rate)

1.6 ROC Curve

Definition: A graphical plot illustrating the performance of a binary classifier system as its
discrimination threshold is varied.
Axes: - X-axis: False Positive Rate (FPR) - Y-axis: True Positive Rate (TPR)
Purpose: - Visualize trade-offs between sensitivity and specificity. - Compare the performance
of different classifiers.
Example: Plotting the ROC curve to evaluate the accuracy of a medical diagnostic test.

Figura 1.1: Example of an ROC Curve

This page is intentionally left blank.
Statistics and Inference
2.1 Descriptive Statistics
Mean: The average value of a dataset.
∑
xi
Mean =
n
Median: The middle value in a dataset when it is ordered.
Mode: The most frequently occurring value in a dataset.
Variance: A measure of how much values in a dataset differ from the mean.
∑
(xi − Mean)2
Variance =
n−1
Standard Deviation: The square root of the variance, indicating how spread out the values
are. √
Standard Deviation = Variance

Listing 2.1: Descriptive Statistics

data = randn ( 1 0 0 , 1 ) ;
mean_value = mean( data ) ;
median_value = median ( data ) ;
mode_value = mode ( data ) ;
v a r i a n c e _ v a l u e = v a r ( data ) ;
std_dev_value = std ( data ) ;

2.2 Histogram and Density Plot

Histogram: A graphical representation of the distribution of a dataset using bars.
Density Plot: A smoothed version of the histogram that shows the probability density function
of the data.
Listing 2.2: Histogram and Density Plot
h i s t ( data ) ; % Histogram
[ f , x i ] = k s d e n s i t y ( data ) ; % D e n s i t y p l o t
plot ( x i , f ) ;

2.3 Normality Test

Shapiro-Wilk Test: A test to check if a dataset is normally distributed.
H0 : Data is normally distributed

7
8 • T-Test

H1 : Data is not normally distributed

p-value: If the p-value is less than a significance level (e.g., 0.05), reject H0 .
Listing 2.3: Normality Test
pkg load s t a t i s t i c s ; % Load t h e s t a t i s t i c s p a c k a g e
[ h , p_value ] = s w t e s t ( data ) ;

2.4 T-Test
One-Sample T-Test: Tests if the mean of a single sample is equal to a known value.

H0 : µ = µ0

Two-Sample T-Test: Tests if the means of two independent samples are equal.

H0 : µ1 = µ2

Listing 2.4: T-Test

d a t a 1 = randn ( 1 0 0 , 1 ) ;
d a t a 2 = randn ( 1 0 0 , 1 ) + 0 . 5 ;
[ h , p_value ] = t t e s t ( data1 , 0 ) ; % One−sample t − t e s t
[ h , p_value ] = t t e s t 2 ( data1 , da t a 2 ) ; % Two−sample t − t e s t

2.5 Correlation and Regression

Correlation Coefficient (r): Measures the strength and direction of a linear relationship
between two variables.
cov(x, y)
r=
σx σy
Linear Regression: Fits a line to the data that best predicts the dependent variable (y) from
the independent variable (x).
y = b0 + b1 x

Listing 2.5: Correlation and Regression

x = randn ( 1 0 0 , 1 ) ;
y = 2 ∗ x + randn ( 1 0 0 , 1 ) ;
correlation = corr (x , y ) ;
model = p o l y f i t ( x , y , 1 ) ;

2.6 One-Way ANOVA

One-Way ANOVA: Tests if there are any statistically significant differences between the
means of three or more independent groups.

H0 : µ1 = µ2 = µ3 = . . . = µk
 Post-Hoc Test • 9

H1 : At least one group mean is different

F-Statistic: Ratio of between-group variance to within-group variance.
Listing 2.6: One-Way ANOVA
group1 = randn ( 3 0 , 1 ) ;
group2 = randn ( 3 0 , 1 ) + 1 ;
group3 = randn ( 3 0 , 1 ) + 2 ;
data = [ group1 ; group2 ; group3 ] ;
g r o u p s = [ ones ( 3 0 , 1 ) ; 2∗ ones ( 3 0 , 1 ) ; 3∗ ones ( 3 0 , 1 ) ] ;
[ p , t a b l e , s t a t s ] = anova1 ( data , g r o u p s ) ;

2.7 Post-Hoc Test

Tukey’s HSD (Honestly Significant Difference): Used after ANOVA to find which specific
groups’ means are different. - Comparison of all pairs: Controls for Type I error across
multiple comparisons.
Listing 2.7: Post-Hoc Test
[ c , m, h , nms ] = multcompare ( s t a t s ) ;