Introduction to Research Design and Statistics

Testing Means (t-distribution)

The t-Distribution

The t-distribution is used to test groups differences when samples are small or when the population variance is unknown. The t-distribution is a family of curves that are symmetric and appear similar to the normal distribution, in fact, for extremely large samples the t and the normal distributions are identical. This is why statisticians say that the t-distribution is asymptotically normal. The t-distribution was first described by a British statistician named W.S. Gossett who worked for the Guinness Brewery in Dublin, Ireland. He published his results under the nom de plume of Student.

Each member of the family of curves in the t-distribution is a function of a parameter called the degrees of freedom. The t-distribution could be tabled in the same way that the normal distribution is, except that it would require a separate table for each of the curves. The t-distribution is tabled with several different probability levels as columns and degrees of freedom as rows. As part of the computation of the t-tests formulas, you will be given formulas the degrees of freedom.

t-test for a Single Sample

Where is the standard error of the mean.

Sampling Distribution of a Mean

Hypotheses

2-tail - H₀: μ = value H₁: μ ≠ value
1-tail - H₀: μ <= value H₁: μ > value

1-tail - H₀: μ >= value H₁: μ < value

Example

A school district gives a state mandated standardized reading test. The mean of the test for all third graders is 65. A class of 20 third-graders has been taught reading for the past three years using the McNeil Method. The class mean is 70.5 with a variance of 121.

Is the class mean significantly different from the state mean?

First, compute the standard deviation from the variance:
s = SQRT(s²) = SQRT(121) = 11

Next, compute the standard error of the mean:
se = s/SQRT(n) = 11/SQRT(20) = 11/4.47 = 2.46

Compute the t-test:
t = (Xbar - μ)/se = (70.5 - 65)/2.46 = +2.24

The degrees of freedom, df = n -1 = 20 -1 = 19

The critical value for t at alpha = 0.05 is ±2.093

Thus, it is concluded that the class of third graders scored significantly higher than the state average.

Robustness

The ability of a statistical test to yield meaningful results when the data fail to meet underlying assumptions of the test.

N<15: Okay to use t-distribution if data are close to normal. If data are clearly nonnormal or if outliers are present, do not use t-distribution.
15<n<40: Okay to use t-distribution except in the presence of outliers or strong skewness.
N>40: t-distribution can be used even for clearly skewed distributions.

t-test for Independent Groups (Between-Subjects)

Sampling Distribution of the Difference between Means

Hypotheses

2-tail - H₀: μ₁ = μ₂ H₁: μ₁ ≠ μ₂

1-tail - H₀: μ₁ <= μ₂ H₁: μ₁ > μ₂

1-tail - H₀: μ₁ >= μ₂ H₁: μ₁ < μ₂

Assumptions

Independence

Observations are independent of each other.
Insured by SRS and random assignment of subjects to groups.

Normality

Observations are sampled from normal populations.
Observe histograms for each group.

Homogeneity of Variance

Samples come from populations with equal variances.
Inspect standard deviations to see that the ratio largest to smallest is no more than two.
If there is a problem with heterogeneity of variance, use the version of the t-test for unequal variances.

Formulas

Theses formulas are for the independent groups t-test with equal variances (pooled variances).

Pooled Variance

Standard Error of Differences between Means

t-test

Example

Consider the hypothetical posttest scores for first-graders in a teacher-expectancy study. Students were assigned at random to either the experimental or the control group. The students in the experimental group were described to teachers as being intellectually "gifted."

Control
Group Experimental
Group
102 107
99 125
90 111
121 117
114 122

Control Group	Experimental Group
102	107
99	125
90	111
121	117
114	122

Stata Examples

input group score x1
1 102  1
1  99  1
1  90  1
1 121  1
1 114  1
2 107 -1
2 125 -1
2 111 -1
2 117 -1
2 122 -1
end
 
ttest score, by(group)

Two-sample t test with equal variances

------------------------------------------------------------------------------
   Group |     Obs        Mean    Std. Err.   Std. Dev.   [95% Conf. Interval]
---------+--------------------------------------------------------------------
       1 |       5       105.2    5.508176    12.31666    89.90685    120.4931
       2 |       5       116.4    3.340659     7.46994    107.1248    125.6752
---------+--------------------------------------------------------------------
combined |      10       110.8    3.564641    11.27239    102.7362    118.8638
---------+--------------------------------------------------------------------
    diff |               -11.2    6.442049               -26.05539    3.655392
------------------------------------------------------------------------------
Degrees of freedom: 8

                      Ho: mean(1) - mean(2) = diff = 0

     Ha: diff < 0               Ha: diff ~= 0              Ha: diff > 0
       t =  -1.7386                t =  -1.7386              t =  -1.7386
   P < t =   0.0602          P > |t| =   0.1203          P > t =   0.9398
  
ttest score, by(group) unequal

Two-sample t test with unequal variances

------------------------------------------------------------------------------
   Group |     Obs        Mean    Std. Err.   Std. Dev.   [95% Conf. Interval]
---------+--------------------------------------------------------------------
       1 |       5       105.2    5.508176    12.31666    89.90685    120.4931
       2 |       5       116.4    3.340659     7.46994    107.1248    125.6752
---------+--------------------------------------------------------------------
combined |      10       110.8    3.564641    11.27239    102.7362    118.8638
---------+--------------------------------------------------------------------
    diff |               -11.2    6.442049               -26.62627    4.226274
------------------------------------------------------------------------------
Satterthwaite's degrees of freedom:  6.59196

                      Ho: mean(1) - mean(2) = diff = 0

     Ha: diff < 0               Ha: diff ~= 0              Ha: diff > 0
       t =  -1.7386                t =  -1.7386              t =  -1.7386
   P < t =   0.0642          P > |t| =   0.1283          P > t =   0.9358   
    
   
use http://www.gseis.ucla.edu/courses/data/hsb2
 

ttest write, by(female)

Two-sample t test with equal variances

------------------------------------------------------------------------------
   Group |     Obs        Mean    Std. Err.   Std. Dev.   [95% Conf. Interval]
---------+--------------------------------------------------------------------
    male |      91    50.12088    1.080274    10.30516    47.97473    52.26703
  female |     109    54.99083    .7790686    8.133715    53.44658    56.53507
---------+--------------------------------------------------------------------
combined |     200      52.775    .6702372    9.478586    51.45332    54.09668
---------+--------------------------------------------------------------------
    diff |           -4.869947    1.304191               -7.441835   -2.298059
------------------------------------------------------------------------------
Degrees of freedom: 198

                  Ho: mean(male) - mean(female) = diff = 0

     Ha: diff < 0               Ha: diff ~= 0              Ha: diff > 0
       t =  -3.7341                t =  -3.7341              t =  -3.7341
   P < t =   0.0001          P > |t| =   0.0002          P > t =   0.9999
 
ttest read, by(female)

Two-sample t test with equal variances

------------------------------------------------------------------------------
   Group |     Obs        Mean    Std. Err.   Std. Dev.   [95% Conf. Interval]
---------+--------------------------------------------------------------------
    male |      91    52.82418    1.101403    10.50671    50.63605     55.0123
  female |     109    51.73394    .9633659    10.05783    49.82439     53.6435
---------+--------------------------------------------------------------------
combined |     200       52.23    .7249921    10.25294    50.80035    53.65965
---------+--------------------------------------------------------------------
    diff |            1.090231    1.457507               -1.783997    3.964459
------------------------------------------------------------------------------
Degrees of freedom: 198

                  Ho: mean(male) - mean(female) = diff = 0

     Ha: diff < 0               Ha: diff ~= 0              Ha: diff > 0
       t =   0.7480                t =   0.7480              t =   0.7480
   P < t =   0.7723          P > |t| =   0.4553          P > t =   0.2277

Go to the next screen

UCLA Department of Education

Phil Ender, 14Nov00