STATS THE PAIRED SAMPLES T-TEST

0.0(0)
studied byStudied by 0 people
learnLearn
examPractice Test
spaced repetitionSpaced Repetition
heart puzzleMatch
flashcardsFlashcards
Card Sorting

1/17

flashcard set

Earn XP

Description and Tags

aka "Related Samples" or "Correlated Groups" t test

Study Analytics
Name
Mastery
Learn
Test
Matching
Spaced

No study sessions yet.

18 Terms

1
New cards
<p>DV is… Fill out the table</p>

DV is… Fill out the table

Interval/ ratio

<p><strong>Interval/ ratio</strong></p>
2
New cards

When do we use a Paired Samples t-test?

DV is interval/ ratio and…

  • Within-subjects (concurrent-measures or repeated-measures designs)

  • Natural pairs (examples)— looking for DIFFERENCES, not correlations

    • Are older siblings more conscientious than younger ones (in same family)?

    • Does one person in a couple have better memory for relationship events than their partner does?

    • Do brothers & sisters in the same family differ in how permissive they think their parent was with them?

  • Experimental matching

    • Pairs are formed on a matching variable. Within each pair, one subject is randomly assigned to the experimental condition & the other is assigned to the control condition.

3
New cards

Experimental Matching: Example

Does a new way of teaching 3rd graders multiplication work better than the old way? Rank order the children on math achievement scores prior to the study:

<p>Does a new way of teaching 3rd graders multiplication work better than the old way? <strong>Rank order</strong> the children on math achievement <strong>scores</strong> <u>prior to the study</u>:</p>
4
New cards

WHY do we use a Paired Samples t-test?

  • Eliminates some of the “noise” due to disturbance variables

  • How does this affect power?

    • In a paired-samples t test, individual differences (a component of sampling error) are removed (by subtracting between pairs of scores)

    • This will reduce the size of the denominator. What remains for the eSEM is pure experimental error, which means that…

    • tobs will be larger!

  • Paired samples t test is generally more powerful than an independent samples t test

<ul><li><p><strong>Eliminates some of the “noise” due to disturbance variables</strong></p></li><li><p>How does this affect power?</p><ul><li><p>In a <strong>paired-samples t test, individual differences</strong> (a component of sampling error) are <strong>removed</strong> (by subtracting between pairs of scores)</p></li><li><p>This will reduce the size of the denominator. <strong>What remains for the eSEM is pure experimental error,</strong> which means that…</p></li><li><p>tobs will be larger!</p></li></ul></li><li><p>Paired samples t test is <strong>generally more powerful</strong> than an independent samples t test</p></li></ul><p></p>
5
New cards

Example of how a Paired Samples t-test Reduces Noise

  • Studies show that exercise increases blood flow to brain areas involved in memory.

  • Research question: Does aerobic exercise aid memory more than anaerobic exercise does?

  • Aerobic Condition:

    • Run in place for 10 min.

    • Take memory test

  • Anaerobic Condition:

    • Lift weights for 10 min.

    • Take memory test

  • If conducted as a between-groups study:

    • The two groups may differ with respect to individual differences that affect the DV:

      • Amount of sleep previous night

      • Ability to concentrate

      • Memory skill

  • If conducted as a within-groups study:

    • These individual differences are exactly the same in each condition (b/c the same people are in each condition).

    • Can remove this error (thus shrinking the denominator of the t test).

<ul><li><p>Studies show that exercise increases blood flow to brain areas involved in memory.</p></li><li><p>Research question: Does aerobic exercise aid memory more than anaerobic exercise does?</p></li><li><p>Aerobic Condition:</p><ul><li><p>Run in place for 10 min.</p></li><li><p>Take memory test</p></li></ul></li><li><p>Anaerobic Condition:</p><ul><li><p>Lift weights for 10 min.</p></li><li><p>Take memory test</p></li></ul></li><li><p>If conducted as a <strong>between-groups study:</strong></p><ul><li><p>The two <u>groups may differ with respect to individual differences</u> that <u>affect the DV</u>:</p><ul><li><p>Amount of sleep previous night</p></li><li><p>Ability to concentrate</p></li><li><p>Memory skill</p></li></ul></li></ul></li><li><p>If conducted as a <strong>within-groups study:</strong></p><ul><li><p>These <u>individual differences are exactly the same in each condition</u> <em>(b/c the same people are in each condition)</em>.</p></li><li><p>Can <strong>remove this error (thus shrinking the denominator</strong> of the t test).</p></li></ul></li></ul><p></p>
6
New cards

The statistical cost of the correlated samples t-test

Correlated samples (aka paired samples) t test is more powerful ONLY if we reduce enough error variance to offset the decrease in df (we usually do!)

<p>Correlated samples (aka paired samples) t test is <strong>more powerful ONLY if we reduce enough error variance to offset the decrease in df</strong> (we usually do!)</p>
7
New cards

HOW Do We Conduct a Paired Samples t test?

A paired samples t test is exactly like a one-sample t test, but using difference scores.

Population (bowl of difference scores), i.e. each score is a difference between a person’s score in Conditions A and B

<p>A paired samples t test is exactly like a one-sample t test, but <u>using difference scores</u>.</p><p>Population (bowl of difference scores), i.e. <u>each score is a difference between a person’s score in Conditions A and B</u></p>
8
New cards

Creating the Sampling Distribution

From population of difference scores:

Take all possible random samples (of size N) of difference scores, compute the mean of each sample, and plot them.

Distribution of means of difference scores for each sample

<p>From population of difference scores:</p><p><strong>Take all possible random samples (of size N) of difference scores</strong>, compute the <strong>mean</strong> of each sample, and <strong>plot them</strong>.</p><p>Distribution of means of difference scores for each sample</p>
9
New cards

Steps in Hypothesis Testing: Answering “Is there an effect?”

Establish H0 and H1 (If there’s no treatment effect, what should be the mean of the difference (D) scores in the population?)

  • H0: µD = 0

  • H1: µD ≠ 0

Collect data

Determine the relevant sampling distribution

  • Compute standard error of sampling distribution

  • Establish a (usually = .05)

  • Compute critical rejection points that define the “reject H0” and “fail to reject H0” regions

  • Compute test statistic (t in this case)

  • Draw a conclusion

10
New cards

Formula for t

knowt flashcard image
11
New cards

Compute the Paired Samples t test (DV = # correct on memory test); Data Collection

knowt flashcard image
12
New cards

Visualizing the Sampling Distribution

knowt flashcard image
13
New cards

Our 3 Qs:

Is there an effect?

  • Yes! Our t was statistically significant.

What is the nature of the effect?

  • Look at the means: Aerobic = 11.67, Anaerobic = 7.00

  • Memory was significantly better following aerobic than anaerobic activity. Include:

    • DV (memory)

    • Both levels of IV (aerobic, anaerobic)

    • Significance (write the word!)

    • Direction/Nature (better, less, more, slower)

      How strong is the effect? Cohen’s d and eta-squared

14
New cards

Measures of Effect Size

Cohen’s D

  • How many SDs the mean difference scores shifted above or below zero (zero = the difference specified in the null hypo)

Eta-squared

  • Proportion of variability in the DV associated with the IV after the effects of individual differences have been removed.

<p><strong>Cohen’s D</strong></p><ul><li><p><em><u>How many SDs the mean difference scores shifted above or below zero</u> </em>(zero = the difference specified in the null hypo)</p></li></ul><p><strong>Eta-squared</strong></p><ul><li><p><em>Proportion of variability in the DV associated with the IV <u>after the effects of individual differences have been removed.</u></em></p></li></ul><p></p>
15
New cards

Confidence Intervals

Where tcrit is found in the t table where a = .05 (for 95% CI)

  • We are 95% confident that the true mean difference in the population falls within this interval. The CI does not contain zero, consistent with our decision to reject the null hypothesis.

  • True mean difference: The difference in means that exists in the POPULATION, not the sample.

You are 95% confident that the sample difference you pulled comes from a population that has a mean difference somewhere between 3.09 and 6.25.

<p>Where tcrit is found in the t table where a = .05 (for 95% CI)</p><ul><li><p>We are <strong>95% confident that the true mean difference in the population falls within this interval</strong>. The CI does not contain zero, <u>consistent with our decision to reject the null</u> hypothesis.</p></li><li><p><em>True mean difference: The <u>difference in means</u> that exists in the <u>POPULATION</u>, not the sample.</em></p></li></ul><p><span style="color: #000000"><em>You are 95% confident</em> that the <em>sample difference you pulled comes from a population that has a mean difference somewhere between 3.09 and 6.25.</em></span></p><p></p>
16
New cards

Steps for conducting Paired Samples t-test

knowt flashcard image
17
New cards

Assumptions of the Paired Samples t test

  • Difference scores are independently and randomly sampled from the population.

  • Population of difference scores is normally distributed.

    • The t test is robust to violations of this assumption, especially if N > 40.

18
New cards

Sample APA-Style Results Section

A paired samples t test examining the difference in memory after aerobic exercise versus anaerobic exercise was statistically significant, t(5) = 7.59, p = .001, d = 3.10, eta-squared = .92. Participants recalled significantly more items following aerobic exercise (M = 11.67, SD = 1.63) than following the anaerobic exercise (M = 7.00, SD = 1.79); 95% CI [3.09, 6,25].