t-test in Matlab - no data!

Hi,

Is there a way in Matlab to perform t-test (where we check for variance first and then perform the test and find the p-value) using means and std (no data)?

Thanks

Answer 1

[Aashray]

Hi Seldeeno,

Yes, you need to perform a "t-test" in MATLAB using summary statistics, using means, standard deviations, and sample sizes of both groups.

The following steps can be followed for performing the “t-test”:

• Define the means, standard deviations, and sample sizes for both groups.
• Conduct an "F-test" to determine whether the variances of the two groups are similar. This will help decide whether to use the standard "t-test" assuming equal variances or opt for Welch's t-test if the variances differ.
• Calculate the t-statistic and determine degrees of freedom.
• Compute the p-value to evaluate the statistical significance of the difference between the groups.

The following piece of code demonstrates a part of the problem:

F = (std1^2) / (std2^2); 
pValueF = 2 * min(fcdf(F, df1, df2), 1 - fcdf(F, df1, df2)); % df1 and df2 are degrees of freedom respectively 
equalVar = pValueF > alpha; % alpha is significance level 

% n1 and n2 are group sizes respectively 
if equalVar  
    sp = sqrt(((n1 - 1) * std1^2 + (n2 - 1) * std2^2) / (n1 + n2 - 2));  
    tStat = (mean1 - mean2) / (sp * sqrt(1/n1 + 1/n2));  
    df = n1 + n2 - 2;  
else  
    % Welch's t-test for unequal variances  
    tStat = (mean1 - mean2) / sqrt((std1^2/n1) + (std2^2/n2));  
    df = ((std1^2/n1 + std2^2/n2)^2) / (((std1^2/n1)^2 / (n1 - 1)) + ((std2^2/n2)^2 / (n2 - 1)));  
end

The following functions are helpful while performing the statistical calculations:

• “fcdf()”: Computes the cumulative distribution function for the F-distribution
• "sqrt()": Calculates the square root
• "tcdf()": Computes the cumulative distribution function for the t-distribution