Qualitative Designs
Correlational Designs
Between Groups Designs
Within Group Designs
Bivariate Stats, Effect Size, and Power
100
What type of data is collected in qualitative research?
Non-numerical (ex. interviews, observations, focus groups)
100
What are the two types of variables needed to use Pearson's correlation (r)?
Two quantitative (numerical) variables
100
What is the purpose of a means plot, and can we determine whether the differences are statistically significant from the means plot alone?
A means plot allows researchers to visually see the average scores across different groups and to compare mean differences.
NO, we cannot determine statistical significance from the means plot alone (you would need to further conduct a BG ANOVA test).
100
True or false: The H₀ (Null hypothesis) of a Between Groups and Within Groups ANOVA is the same.
True! H₀ = There is no significant mean difference between participants across groups/conditions.
100
Which variables are qualitative, and which are quantitative?
1. Hair Color (1=Black, 2=Brown, 3=Blonde, 4=Other)
2. Attractiveness (On a scale of 1-10)
3. Relationship Status (0=Single, 1=Taken)
4. Happiness (On a scale of 1-10)
Hair Color and Relationship Status are Qualitative, Attractiveness and Happiness are Quantitative
200
A researcher wants to examine whether a person's employment status (employed or unemployed) is related to their mode of transportation (car, bus, or walking). Which statistical test should they use and why?
Pearson's chi-square (X2), because both variables are qualitative/categorical.
200
A researcher finds r = -.40 between hours spent gaming and exam scores. What is the direction and strength of this correlation and what would the graph look like?
The direction is negative and of moderate strength.
The graph would show a linear relationship with a downward trend from left to right, with points moderately spread out around the slope line.
200
What is one advantage and one disadvantage of using a between-groups (BG) design?
+ One advantage is that participants are only in one condition, so each score is independent, and there is no need to worry about scores affecting each other.
- Disadvantages of BG designs: Require more participants, individual differences, environmental variables, attrition (loss of subjects)
Most major issue: large differences between participants that are not caused by the independent variable (IV) can affect the results of the study (confounding variables)
200
When do we use a Paired Samples t-test?
To compare the means of two measurements taken from the same subjects. (e.g. before and after studies)
200
What is power set to in psychology?
Power is set to .80, the inverse of 𝛽 (a standard of .20)
300
What is the purpose of a contingency table in chi-square research?
To summarize data for two qualitative variables and determine their pattern of relationship.
300
What are three things a scatterplot tell us about the relationship between two variables?
1. Relationship between variables
2. Direction of correlation
3. Strength/size of correlation
300
A researcher examines whether three different music conditions (classical, pop, and silence) produce different mean memory test scores. Results show F(2, 57) = 4.85, p=.011.
State the research hypothesis (RH) and determine whether to reject or retain H0 and why?
RH: There is a significant mean difference in memory test scores based on music condition.
Reject H0 because p is less than .05 (p<.05). There is a statistically significant overall mean difference among the groups.
300
A researcher was interested in how happiness changes before, right after, and 1 week after going to the concert. They measured people's happiness levels before, right after, and 1 week after going to the concert.
1. What type of variables are these two?
2. What analysis should the researcher run? Why?
1. Qualitative (Time: Before, right after, 1 week after) and Quantitative (Happiness score)
2. Within Group/Repeated Measures ANOVA - There are 3 levels of our qualitative variable and measures the same participants across time.
300
A researcher predicts there is a positive correlation between hours of sleep and exam scores. She finds r(48) = -.35, p = .013. A literature review shows that over 300 prior studies found a positive linear relationship between sleep and exam scores.
Is this a Type III error or contrary results?
Type III error. Her result is statistically significant but in the opposite direction of her prediction. Because the literature agrees with her RQ/RH (positive), her RQ/RH was probably correct and her finding is the likely mistake.
400
A chi-square test finds X2(2, N = 80) = 8.25, p <.01. Should the researcher reject or retain H0 and why?
Reject H0 because p < .01. There is a statistically significant pattern of relationship between the two categorical variables.
400
A researcher finds r = 0.35, p = .43 between professor age and student evaluation scores. Should they reject or retain H0 and is this significant, why?
Retain (fail to reject) H0 because p is greater than .05 (p > .05). The negative correlation is NOT statistically significant.
400
A researcher compares mean anxiety scores among three independent treatment groups. A between-groups (BG) ANOVA finds F(2, 57) = 6.42, p=.003, η²=.18.
Write an APA-style interpretation of the results, including the significance and effect size.
There is a statistically significant mean difference in anxiety scores based on treatment group, F(2, 57) = 6.42, p=.003, η²=.18. The effect size is large because η²=.18 is greater than .14.
400
A researcher was interested in how student motivation differs between their 1st, 2nd, 3rd, and 4th years in college.
Based on the WG/RM ANOVA results:
1. What is the H0?
2. Do we reject/retain the H0? Why or why not?
H0 = There is no difference in motivation for students in their 1st, 2nd, 3rd, and 4th year in college.
We reject H0 because p < .001, which is less than α = .05. There is a significant difference in motivation across at least one pair of years.
400
A researcher examined whether students who studied using flashcards, practice tests, or rereading would differ in exam scores. She found that students who used practice tests scored highest (M = 84.2), followed by flashcards (M = 78.6) and rereading (M = 74.1), F(2, 57) = 6.50, p = .003.
Should we worry about the statistical power of this study? Why or why not?
No. Power is about Type II error, which is the chance of retaining H₀ when it should be rejected. Because p = .003 is less than .05, the researcher rejected H₀, so a Type II error isn’t possible. The concerns with a significant result are Type I and Type III errors, not power.
500
A researcher compares four treatment groups using chi-square. What is the difference between an omnibus test and pairwise comparisons, and how many comparisons are possible?
The omnibus test examines the overall pattern across groups. Pairwise tests compare specific pairs (whether each pair of IV levels/groups differs in response pattern).
Four treatment groups would have 6 possible comparisons: k(k-1)/2 = 4(3)/2 = 6, where K = # of IV levels or groups.
500
A researcher examines the relationship between hours of sleep and stress levels among 60 college students. Results show r(58) = -.52, p <.01.
What is the research hypothesis (RH) and an APA-style interpretation of the results?
RH: There is a significant negative linear relationship between hours of sleep and stress levels.
APA Write-Up: There is a statistically significant, strong negative linear relationship between hours of sleep and stress levels, r(58) = -.52, p <.01. As hours of sleep increase, stress levels tend to decrease.
500
Post hoc tests:
A researcher compares mean stress scores across three exercise groups. These are the Tukey's HSD results. Which groups have statistically significant mean differences and which do not?
Tukey’s HSD post hoc tests revealed that there IS a significant difference in stress scores between yoga vs. running tea (p=.032) and running vs. walking (p=.008) as both p-values are less than .05 (p<.05).
There is NOT a significant difference in stress scores between Yoga vs. walking (as p-value is greater than .05 (p<.05).
500
A researcher ran a Tukey's LSD Post Hoc to compare motivation in the same students across 1-4 years in college.
The following were their hypotheses:
H1: 1st Year and 2nd Year Motivation will be the same.
H2: 2nd Year and 3rd Year Motivation will be the same.
H3: 4th Year Motivation is greater than 1st Year Motivation.
Which hypotheses are supported, and which are not, and why?
H1: Not supported. 1st Year and 2nd Year motivation were significantly different from one another (p = .001)
H2: Supported. 2nd Year and 3rd Year motivation were not significantly different from one another (p = .471)
H3: Not supported. 1st Year and 4th Year motivation were not significantly different from one naother (p = .373).
500
You obtained r(31) = .30, p > .05
1. What is the sample size (S) for this post hoc power analysis?
2. What power level were you working at, and what does that tell you about your Type II error rate?
1. Sample size (S) = df + 2 (31+2) = 33
2. In the r = .30 column, S = 33 falls in the .40 power row. Power = .40, so the chance of a Type II error is 1 − .40 = .60.