Tests and Measurements and Statistics in Psychology

MODULE 1

Descriptive and Correlation Statistics

One of the most important tools used in modern psychology is statistics. Statistical concepts were first used by Galton in 1885, but the use of these techniques was not widespread until after 1920. Today, however, a knowledge of basic statistics is important to an understanding of even elementary psychology. This module introduces the use of statistics to summarize and analyze data. The concepts taught here will make it possible for you to understand how psychologists measure individual differences.

As you read the text, try to answer the following questions.

As in any empirical science, valid conclusions in psychology depend on the accumulation of amounts of data. For example, in the administration of a test, or in the performance of an experiment, many scores must be obtained and quantities measured. If one hundred subjects are tested, one hundred scores must be obtained and analyzed.

Figure 1. Raw data of test results
55576545454555466576 76767766666563767668 67657567647867557665 56565776678665765778 56555576957677666546

Score 3
4
5
6
7
8
9
Figure 2 Frequency distribution
Frequency
1
6
28
36
24
4
1

Data must be organized in a meaningful way before it can be analyzed

Figure 1 lists some test scores as an example of raw data with which a psychologist works. In this case, each score represents the number of questions answered correctly by a subject on a memory span test of nine items. Usually we find it convenient to summarize such data in either a table or a graph. Figure 2 shows the same raw scores summarized in a frequency distribution tab/e.

The frequency distribution reveals some things that are not readily apparent from the raw data. For example, there seems to be a "bunching" of subjects around the score of 6. This can be seen even more clearly if we convert our frequency distribution into a frequency polygon is in Figure 3. Figure 3 shows the same data as Figures 1 and 2, but here the scores are represented as points on a graph.

Figure 3 Frequency polygon

Measures of Central Tendency
Still another way to summarize data is to determine the central tendency of the test population. There are many ways to describe the central tendency but we will only deal with the three that are used most frequently.

Mean. The mean is the arithmetic average of the scores. This is the most familiar and frequently used measure of central tendency To obtain it, just add up all the scores and divide by the number of scores. In the example above there are 100 scores, the sum of which is 592. To obtain the mean, divide 592 by 100. In this example, the mean is 5.92.

Mode. The mode is defined as the score that occurs most frequently. Figure 3 shows that the score of 6 was made most frequently (36 times) so the mode of this distribution is 6.

Median. Median means middle. It is defined as a theoretical point above and below which exactly 50 percent of the cases lie. The exact method for calculating the median is too complicated to teach here but a simpler technique called the mid-point method can be used to obtain an approximate value for the median. First arrange the scores in order. Then count from one end until you reach the middle score. Since there are 100 scores in this problem, the middle score would be between the 50th and 51st score. If we count from either end we would find both of these scores to be 6. Therefore 6 is the approximate value of the median.

CORRELATION

Psychologists frequently wish to quantify the relationship, if any, between two sets of events or observations. One might wish to know if college grades can be used to predict financial success, or whether there is any relationship between any two sorts of behavior, such as frustration and aggression. Such questions as these can be investigated by correlational studies.

The typical graph used in a correlation study is a scatter plot. In this graph the range of one test (one type of observation) is marked along one axis and the range of another test "another observation) is marked along the other axis. The graph in Figure 4 correlates hypothetical scores on two final examinations given to the same 80 students who enrolled in both Introductory Psychology and Introductory Sociology.

Figure 4. A scatter tally represents a person located according to his score on a psychology test and his score on a sociology test.

Each tally represents one student's scores on the two tests. For example, the two tallies in the column to the far right indicate that only two students scored above 90% on the psychology test. One scored between 70% and 79% on the sociology test and the other scored between 60% and 69%

Correlation is not sufficient evidence of cause and effect

Let us consider the range of correlations that may be encountered and how they would appear in a scatter plot.

Figure 5. Scatter plot patterns

Perfect Positive
Suppose we were comparing people's height in inches with their height in centimeters. Both of these units measure the same thing (distance). Thus if we were to determine the height of 10 people using these measures we would find a perfect correlation (see Figure 5-A).

High Positive
If two different measures are closely related, they produce a high positive correlation. For example, intelligence test scores and reading skills often show a high positive correlation. Figure 5, Graph B, indicates such a correlation,

Low Positive Correlation
There are many measures which tend to vary together. e the relationship between height and weight. Generally, taller people weigh more, but there are exceptions; the correlation is generally positive, in that increases in one measure are associated with increases in the other, but the amounts of increases differ. The scatter plot of such a relationship would be indicative of a low positive correlation. Figure 5-C shows the distribution of heights and weights that may be found in a typical college classroom .

Zero Correlation
Most measures are totally unrelated. In a correlation of height and hair color, the points of the scatter plot would be random and no trend could be obtained in any direction. Figure D shows a zero correlation.

Negative Correlations

The negative correlations are interpreted in exactly the same way as the positive correlations but the relationships are in the reverse direction. That is, higher scores on one measure tend to be associated with low scores on a second measure. An example of a highly negative correlation may be the age of an automobile and its selling price. Figure F indicates that generally older cars (high age) cost less (low dollars).

Final Note
We must be careful to remember that correlation does not assume causation. For example, there is a low positive correlation between the rate at which ice cream melts on the sidewalks of New York and the number of deaths in Bombay, India. But one does not cause the other. Both are affected by a third factor, that is, the increase in temperature during summer months. Correlation may lead us to search for a common cause but it does not prove that a cause and effect relationship exists.

RELIABILITY AND VALIDITY

Much psychological information comes from the administration of tests. Two criteria that a test must meet are reliability and validity. These are both measured by correlations.

A test is reliable when repeated administrations to the same subjects have a high positive correlation. Reliability, then, is a measure of the degree to which repeated measurements of the same subject give the same reading. For example, if we gave students in a group an intelligence test on Monday and then re-tested them several weeks later, would their second scores be very similar to the first scores? If they are, then we could say that the test is reliable.

A test is valid when it can be proved that it does, in fact, measure what it claims to measure. Suppose that in a course in psychology the instructor gave a final examination on which all of the test questions required a knowledge of algebra. The students would probably protest that such a test was not a valid measure of their understanding of psychology. Unfortunately, the question of validity is not often that obvious. Many tests have "face" validity. That is, they look like they should measure what they claim but in fact they may not. For example, it may be that our psychology instructor, in response to student criticism, made up a new final but this time he used a number of very tricky questions which were really designed to test the students' intelligence rather than their knowledge of psychology. His students could claim that this final is just as invalid as the algebra test.

The best way to validate a test is to compare it with another measure (or criterion) of the same event. For example, if someone has devised an industrial aptitude test which supposedly predicts how well someone will do on a job, one could correlate these test scores with performance ratings they received from their supervisors. Both tests may be invalid, but each test deserves more confidence if a comparison of results shows a high positive correlation.


MODULE 1
PROGRESS CHECK 1

Now test yourself without looking back.

1. The most frequent score in a given distribution is called the_______________________________
2. The number that represents the sum of all scores divided by the number of scores is the___________________
3. The point in the distribution at which exactly 50% of the scores are higher and 50% are lower is called the________________________________

4. If a high score on one test can be used to predict a low score on another test, the correlation between the tests is_________________________________

5. Reliability is a measure of the degree to which_________________________

6. Define validity_______________________.

7. (graph here)

The graph above shows two (polygons).

8. Name the correlation indicated by each of the following:


a.__________________

b.__________________

c.__________________

d.__________________

ANSWER KEY


MODULE 1
EXERCISES

Table A
SCOREFREQUENCY
9
8
7
6
5
4
3
2
1
1
2
4
8
10
8
4
2
1

Table A

The value that occurs most frequently in a distribution is the mode.
The mode in the frequency distribution in Table A is _________. 2

The median is the midpoint in a distribution. The median in the distribution in Table A is_______________________________ 5

The mean is the arithmetic average of all the scores. Calculate below the mean for the distribution in Table A.
a. The sum of the scores is__________________

b. The number of scores is__________________

c. The sum divided by the number of scores is_________________________________4

The table below represents a__________________________. 3


SCORE FREQUENCY
13
12
11
10
8
7
6
5
4
3
2
1
2
3
9
15
38
45
32
20
11
6
2
1

Table B

ANSWERS

1) 8
2) 5
3) frequency distribution
4) a.200
b.40
c. 5
5) 5

In correlational studies researchers try to discover relationships between sets of data. In a graph used in a correlation study you might expect to find:

a. the subjects listed along one axis.
b. the scores achieved on one test along one axis and
the scores achieved on the other test along the other axis.

c. the total score from both tests along the bottom axis.

The following sets of scores were achieved by six students on two tests. Put dots in the scatter plot below for these results.

Subject Test 1 Test 2
A
B
C
D
E
F
3
4
3
2
2
2
4
5
5
4
2
3


_________________________________________________5

The graph you have completed is called a:
a. frequency polygon.
b. skewed curve.
c. scatter plot.
d. histogram.

________________________________1

Two sets of data are positively correlated when a high score in one set tends to be associated with a high score in the other and low scores in one set tend to be associated with a low score in the other. Which plot shows a positive correlation?

_______________________________________________2

A high positive correlation indicates that more accurate predictions can be made than in a low positive correlation. Write high positive or low positive beside each of the scatter plots below.

ANSWERS

1 C
2 B
3 B
4 a. low positive
b. high positive


.

If a scatter plot shows a high negative correlation, then which of the following would be true?
a. a high score on one test implies a high score on the other test.
b. a high score on one test implies a low score on the other test.
c. predictions can be more accurate than with a low negative correlation._________________________________3

A perfect positive correlation has a correlation coefficient of + 1.00. A perfect negative correlation has a correlation coefficient . Write the appropriate coefficient beside each of the following scatter plots.

.

All correlation coefficients fall and + 1.00. No correlation is indicated by a coefficient of 0. Which of the following coefficients would indicate a low negative correlation?

a. 0
b. +1.00
c. -.8l0
d. -.40

___________________________________________2

ANSWERS
1. a. +1.00
b. -1.00
2 d
3 b, c
4 a.perfect positive
b. perfect negative



Name the type of correlation indicated by each plot in the diagram above>\

a. ______________________________ .

b. ____________________________ .

c. _____________________________ .

_______________________________________7 .

From this list select the correlation coefficient that might refer to each plot in the diagram in the previous exercise: +3.00, +1.00, +.80, +.40, 0, - .40, - .80, - 1.00.

a.__________________________
b.______________________________
c.________________________________
________________________________________________4

Reliability is a measure of the degree to which repeated measurement of the same subject gives the same reading. If a test was reliable, the correlation between two measurements would be

a. high
b. low
c. zero

__________________________________1

Validity is determined by the degree to which a test measures what it is intended to measure. A test is valid when:
a. there is a high correlation between it and another measure known to be valid.
b. repeated testing of the same subjects gives the same score.
c. (both)

____________________________________________6

Write a definition of each of the following.
a. Reliability________________________________________________3

b. Validity___________________________________________________5

NOW TAKE PROGRESS CHECK 2

ANSWERS

1. a
2. correlation
3 a degree to which repeated measurement of the same subject gives the same results
4 a. +.80
b - .40
c.+1.00
5. degree to which a test measures what it is intended to measure
6 a
7 a high positive
b. low positive
c. perfect positive


MODULE 1
PROGRESS CHECK 2

1.
90, 73, 82, 86, 78, 91, 72, 84, 74, 90, 87

Using the scores listed above, find the following:

a. Mean__________

b. Mode__________

c. Median_________

2.
SCORE FREQUENCY
24
23
22
21
20
19
18
17
16
15
14
13
12
11
10
1
3
5
7
4
8
12
18
12
5
6
6
3
4
1

Draw a frequency polygon for the distribution above.

3. Write the degree of correlation indicated by a coefficient of:

a. + 1.00____________________________

b. 0____________________

c. - .40__________________

d. - .80______________________

4. What sort of graph is used for correlation studies?______________________

5. A check for the reliability of a test is accomplished by:
a. administering it twice to see if the scores are highly correlated.
b. using two different tests to if their scores are highly correlated.
c. using a different test for each group.
d. comparing test results with results of another test known to be valid.

6. If the correlation between two sets of scores is high positive, then a high score on one test can be used to predict_________________________________


7. Match.

1 ) Reliability __________

2) Validity________

3) Correlation coefficients__________

a. The degree to which a test measures what it is intended to measure
b. A measure of relationship between two tests or sets of observations

c. A measure of the degree to which repeated testing of the same subjects gives the same results

d. Used only in cases of significant differences



ANSWER KEY PAGE

Unit 7 Table of Contents

October 15, 2007