Question 1

Tests of relationship and predication are divided into bivariate and multivariate tests. Bivariate correlations look at the relationship of two variables-one is a predictor and the other is the criterion. What does the correlation coefficient describe?

Accepted Answer

The correlation coefficient describes the relationship between the predictor variable and the criterion variable in terms of both strength and direction. Correlation cannot be used to determine causal relationships; it is used to determine relationships between variables when there is no other obvious explanation.

Question 2

One of the values used in bivariate correlation tests is the coefficient of determination. If there is a correlation coefficient of .50 between education and income, what is the coefficient of determination?

Accepted Answer

The coefficient of determination is the square of the correlation coefficient. Squares of decimals are always smaller than the original value; 5 squared is 25, and the decimal point remains in the same place. This number suggests that 25% (the equivalent percentage) of variability in income is explained by education.

Question 3

Test construction requires that tests have both validity and reliability. What is the Spearman-Brown prophecy formula used for in this context?

Accepted Answer

The Spearman-Brown prophecy formula is used to predict increase in reliability of a test if it were longer than it is. Specifically, the formula is used in the context of split-half reliability to account for the reduction in reliability resulting from this method of testing score consistency.

Question 4

Multiple regression tests are used when there are several predictors (or independent variables) and one or more criterion (dependent variables). What does multicollinearity refer to?

Accepted Answer

Multicollinearity is a problem that can occur in multiple regression tests when predictors are highly correlated and therefore redundant.

Question 5

Professor Plum just finished scoring the midterms of the 15 students in his statistics class. He put these results on the blackboard for the entire class to see:
100% 3 students
90% 4 students
80% 1 student
70% 7 students
Then he asked the one student who got 80% whether his score was the mean, median, mode, or some combination thereof. The bewildered student answered "I don't have a clue!" Can you pick the correct answer from the following four choices?

Accepted Answer

80 is the median, but not the mean or the mode. The median is the score that separates the top half of the group from the bottom half-7 students got a grade lower than 80 and 7 students got a grade higher than 80. The mean, or average score, equals the sum of all of the scores divided by the number of people, which in this case is 82. The mode is simply the score that occurred most frequently, which is 70.

Question 6

Which of the following statements about scientific research studies is the most accurate?

Accepted Answer

It is impossible to control for every variable, so if one looks hard enough, a flaw can be found in almost any study. Since a perfect study is more wishful thinking than reality, whether or not a hypothesis is confirmed could be a coincidence.

Question 7

Suppose you were given the results of a large sample of 11-year-old students, all of whom were given the 4th Edition of the Wechsler Intelligence Scale for Children (WISC-IV). The data were analyzed and, unsurprisingly, a statistically significant correlation was found between scores on the Vocabulary subtest of the WISC-IV and WISC Total IQ. A statistically significant correlation was also found between WISC-IV Vocabulary scores and total years of parents' schooling. What do these findings tell you?

Accepted Answer

Even if variable A has a significant correlation with both variable B and variable C, it does not necessarily follow that the same relationship exists between variable B and variable C. For example, there may be a very high correlation between being a girl and need for therapy for depression in adolescence and also a very high correlation between being a girl and having played with dolls as a youngster, but no significant correlation between playing or not playing with dolls and later developing or not developing depression. Also, significant correlations indicate the likelihood of two conditions coexisting; they do not indicate an absolute relationship where neither condition can exist without the other.

Question 8

What type of information would you not ordinarily find included in the manual accompanying a standardized test?

Accepted Answer

Test manuals do not ordinarily include guidelines to assist users in conducting their own validation study with the test. There are organizations with well-trained psychologists and statistics experts that do provide this type of service, but for a substantial extra fee that is not included in the price of the test.

Question 9

If the scores on a particular test fit the normal bell shaped curve distribution, what percentage of scores should fall between +2 and - 2 standard deviations from the mean?

Accepted Answer

On a normal bell shaped curve distribution, approximately 95% of the total number of scores falls within the boundaries of -2.0 and +2.0 standard deviations. The standard deviation serves as a yardstick indicating how far an individual's score on a particular area of functioning deviates from the mean, the average score.

Question 10

Psychologists often rely on rating scales for the assessment of children's behavior. One potential problem with rating scales is known as the "halo effect." Which of the following scenarios best describes ratings possibly contaminated by the halo effect?

Accepted Answer

A halo effect refers to the tendency, after assigning a high rating for an individual in one area, to assign high ratings for that person in every other area, even when the two areas of behavior are not related. Unless the rater is careful to evaluate each area of behavior on its own, the overall rating will be inflated and inaccurate.

Question 11

All of the students in Dr. Wright's psychology class had to conduct their own research study and prepare a report describing it. The instructor, after carefully reading all of the papers, gave one of the following students an A. The remaining three students all got lower grades. Based solely on the information given here, who was the happy student who got an A?

Accepted Answer

Since one of the students described got an A, by process of elimination, it had to be Tyrell. Danielle's study used too small of a sample. Janelle's study used a biased sample, and Lydell should have worked out his methodology more carefully before conducting the study instead of just acknowledging the deficiencies after the study. On the other hand, just because Tyrell's paper was shorter does not mean both his study and report were not high quality.

Question 12

A study showing evidence of substantial correlations between the scores of two alternate forms of the same test, with half of the subjects given one form first and half given the other form first, would be evidence of what?

Accepted Answer

A study showing evidence of substantial correlations between two different forms of a test provides evidence of the test's reliability-how consistent of a measure it is. Validity refers to the degree to which the test is actually measuring what it is supposed to be measuring. A test can still be a reliable measure of whatever it is intended to measure, but not a valid one. In regard to the above example, halo and practice effects are not a factor.

Question 13

The statistical technique known as item analysis provides information about what aspect of a test's utility?

Accepted Answer

The statistical technique known as item analysis provides information about a test's internal consistency. On a test where the level of internal consistency is high, those who get the highest total scores would be consistently getting certain items right that examinees with low overall test scores get wrong. In other words, the test is not loaded down with dead weight items that have little or no utility in differentiating the upper end of the test performance range from the lower end.

Question 14

If a study shows statistically significant differences between the MMPI-A scores of adolescents in a psychiatric hospital and those of adolescents who are not hospitalized, the findings would be evidence of what type of validity?

Accepted Answer

The findings would be evidence of concurrent validity since the comparison is between two sets of results occurring at the same time. If instead, the study were to look at just the hospitalized group and compare MMPI-A scores on admission to a variable to be measured later- like length of hospitalization-that would be a predictive validity study. Content and construct validity are ascertained by focusing on the test itself rather than how performance on the test relates to an outside criterion.

Quiz 8: Research Stats